* [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes
[not found] <5BFACB1451C1459BA8562A5561D0A4E6@stanford.edu>
@ 2010-09-02 0:08 ` David Cross
2010-09-02 1:37 ` Greg KH
` (3 more replies)
0 siblings, 4 replies; 12+ messages in thread
From: David Cross @ 2010-09-02 0:08 UTC (permalink / raw)
To: greg; +Cc: linux-kernel
> From: Greg KH [mailto:greg@kroah.com]
> Sent: Tuesday, August 31, 2010 9:32 AM
> To: David Cross
> Cc: gregkh@suse.de; linux-kernel@vger.kernel.org
> Subject: Re: [PATCH] West Bridge Astoria Driver 2.6.35
>
> On Tue, Aug 31, 2010 at 09:17:18AM -0700, David Cross wrote:
> >
> > > I get a build error when I try to build this driver:
> > >
> > drivers/staging/westbridge/astoria/block/../include/linux/westbridge/c>
> > yasmisc.h:521:2: error: expected declaration specifiers or '...' before >
> > 'cy_as_hal_device_tag'
> >
> > > It looks like the #include mess isn't working quite properly.
> >
> > > Also note that this driver is building on an x86-64 platform, which is
> > > something that you probably don't want :)
> >
> > > So, for now, I'll just mark the driver as CONFIG_BROKEN and can you send
> > > me some Kconfig patches against the next linux-next release which should
> > > have this driver in it, so that it will build properly?
> >
> > Sure, I can do that. I have since changed the Kconfig structure a bit. It
> > actually will build properly with the correct .config as it is though. I
> can
> > send you the .config if you would like.
>
> It's up to the Kconfig rules to ensure that there can never be a
> "non-correct" .config file, so that needs to be fixed so that the build
> will never be broken no matter what type of options are selected.
>
> Care to send a patch for that?
This patch actually contains quite a few fixes, but it addresses the
Kconfig issue as well. The linux-next tree does not seem to have a
config for the zoom2, and trying to build it for that board seems to
make the compilation break. As such, the only thing that I tested was
compilation using the two different HALs (one of which is added in this
patch). Please let me know if there are problems or questions with this.
Thanks,
David
Signed-off-by: David Cross <david.cross@cypress.com>
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/arch/arm/mach-omap2/gpmc.c linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c
--- linux-next-vanilla/arch/arm/mach-omap2/gpmc.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c 2010-09-01 16:10:21.000000000 -0700
@@ -133,6 +133,7 @@ void gpmc_cs_write_reg(int cs, int idx,
reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
__raw_writel(val, reg_addr);
}
+EXPORT_SYMBOL(gpmc_cs_write_reg);
u32 gpmc_cs_read_reg(int cs, int idx)
{
@@ -141,6 +142,7 @@ u32 gpmc_cs_read_reg(int cs, int idx)
reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
return __raw_readl(reg_addr);
}
+EXPORT_SYMBOL(gpmc_cs_read_reg);
/* TODO: Add support for gpmc_fck to clock framework and use it */
unsigned long gpmc_get_fclk_period(void)
@@ -294,6 +296,7 @@ int gpmc_cs_set_timings(int cs, const st
return 0;
}
+EXPORT_SYMBOL(gpmc_cs_set_timings);
static void gpmc_cs_enable_mem(int cs, u32 base, u32 size)
{
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 2010-08-27 17:52:14.000000000 -0700
@@ -0,0 +1,1851 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+ #define CYASSTORAGE_WRITE_EP_NUM (4)
+ #define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ cy_as_hal_print_message(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg with address");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ cy_as_hal_print_message(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+uint32_t cy_as_hal_gpmc_init(cy_as_omap_dev_kernel *dev_p)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ unsigned int cs_mem_base;
+ unsigned int cs_vma_base;
+
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ cy_as_hal_print_message(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, SZ_16K);
+ cy_as_hal_print_message(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&cs_mem_base) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ cy_as_hal_print_message(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ cs_mem_base, AST_GPMC_CS);
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(cs_mem_base, SZ_16K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ cs_mem_base);
+ goto out_free_cs;
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ cs_vma_base = (u32)ioremap_nocache(cs_mem_base, SZ_16K);
+ if (!cs_vma_base) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", cs_mem_base);
+
+ goto out_release_mem_region;
+ }
+ cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ cs_mem_base, cs_vma_base);
+
+ dev_p->m_phy_addr_base = (void *) cs_mem_base;
+ dev_p->m_vma_addr_base = (void *) cs_vma_base;
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_on;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_on;
+ timings.adv_rd_off = WB_GPMC_ADV_t_off;
+ timings.adv_wr_off = WB_GPMC_ADV_t_off;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_on;
+ timings.oe_off = WB_GPMC_OE_t_off;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_on;
+ timings.we_off = WB_GPMC_WE_t_off;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(0) |
+ GPMC_CONFIG1_DEVICESIZE_16
+ ));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(cs_mem_base, SZ_16K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ cy_as_hal_print_message(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ cy_as_hal_print_message(KERN_INFO"pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ cy_as_hal_print_message(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(cy_as_omap_dev_kernel *dev_p)
+{
+ int i, err;
+
+ cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init(dev_p);
+
+ cy_as_hal_config_c_s_mux();
+
+ return err;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (dev_p->m_vma_addr_base != 0)
+ iounmap((void *)dev_p->m_vma_addr_base);
+
+ if (dev_p->m_phy_addr_base != 0)
+ release_mem_region((void *)dev_p->m_phy_addr_base, SZ_16K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ uint16_t write_addr;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk("%s: bad TAG parameter passed\n", __func__) ;
+ return ;
+ }
+ #endif
+
+ write_addr = (void *)(dev_p->m_vma_addr_base + CYAS_DEV_CALC_ADDR (addr));
+ writew ((unsigned short)data, write_addr) ;
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+ uint32_t read_addr = 0;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk("%s: bad TAG parameter passed\n", __func__) ;
+ return dev_p->m_sig;
+ }
+ #endif
+
+ read_addr = (void *) (dev_p->m_vma_addr_base + CYAS_DEV_CALC_ADDR(addr));
+ data = (unsigned short) readw (read_addr) ;
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, i, size;
+ uint16_t *dptr;
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+ register void *read_addr ;
+ register uint16_t a,b,c,d,e,f,g,h ;
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = (uint16_t *) end_points[ep].data_p;
+
+ read_addr = dev_p->m_vma_addr_base + CYAS_DEV_CALC_EP_ADDR(ep) ;
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * Now, read the data from the device
+ */
+ for(i = size/16 ; i > 0 ; i--) {
+ a = (unsigned short) readw (read_addr) ;
+ b = (unsigned short) readw (read_addr) ;
+ c = (unsigned short) readw (read_addr) ;
+ d = (unsigned short) readw (read_addr) ;
+ e = (unsigned short) readw (read_addr) ;
+ f = (unsigned short) readw (read_addr) ;
+ g = (unsigned short) readw (read_addr) ;
+ h = (unsigned short) readw (read_addr) ;
+ *dptr++ = a ;
+ *dptr++ = b ;
+ *dptr++ = c ;
+ *dptr++ = d ;
+ *dptr++ = e ;
+ *dptr++ = f ;
+ *dptr++ = g ;
+ *dptr++ = h ;
+ }
+
+ switch ((size & 0xF)/2) {
+ case 7:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 6:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 5:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 4:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 3:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 2:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 1:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ /* Odd sized packet */
+ uint16_t d = (unsigned short) readw (read_addr) ;
+ *((uint8_t *)dptr) = (d & 0xff) ;
+ }
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0, i = 0;
+ uint32_t size;
+ uint16_t *dptr;
+ register void *write_addr ;
+ register uint16_t a,b,c,d ;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ write_addr = (void *) (dev_p->m_vma_addr_base + CYAS_DEV_CALC_EP_ADDR(ep)) ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size){
+ /*
+ * Now, write the data to the device
+ */
+ for(i = size/8 ; i > 0 ; i--) {
+ a = *dptr++ ;
+ b = *dptr++ ;
+ c = *dptr++ ;
+ d = *dptr++ ;
+ writew (a, write_addr) ;
+ writew (b, write_addr) ;
+ writew (c, write_addr) ;
+ writew (d, write_addr) ;
+ }
+
+ switch ((size & 7)/2) {
+ case 3:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ case 2:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ case 1:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ uint16_t v = *((uint8_t *)dptr) ;
+ writew (v, write_addr);
+ }
+ }
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if(state) {
+ __gpio_set_value(AST_WAKEUP, 1);
+ } else {
+ __gpio_set_value(AST_WAKEUP, 0);
+ }
+
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+ (void) is_standby_wakeup;
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ uint32_t err = 0;
+ /* No debug mode support through argument as of now */
+ (void)debug;
+
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ; /* No debug mode support through argument as of now */
+ (void)debug;
+
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /* allocate memory for OMAP HAL*/
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_CRAM_HAL_SIG;
+
+ /* initialize OMAP hardware and StartOMAPKernelall gpio pins */
+ err = cy_as_hal_processor_hw_init(dev_p);
+ if(err)
+ goto bus_acc_error;
+
+ {
+ uint16_t val1, val2, val3, val4, val5;
+ printk("%s: testing register reads\n", __func__);
+ val1 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0x80);
+ val2 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xC0);
+ val3 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xFF);
+ val4 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xA5);
+ val5 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0x5A);
+
+ printk("%s: testing register reads 0x%x 0x%x 0x%x 0x%x 0x%x\n",
+ __func__, val1, val2, val3, val4, val5);
+ }
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, CY_AS_MEM_CM_WB_CFG_ID);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, "
+ "CY_AS_MEM_CM_WB_CFG_ID %4.4x", data16[0]);
+ goto bus_acc_error;
+ }
+
+ cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 1969-12-31 16:00:00.000000000 -0800
@@ -1,2450 +0,0 @@
-/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
-
-#include <linux/fs.h>
-#include <linux/ioport.h>
-#include <linux/timer.h>
-#include <linux/gpio.h>
-#include <linux/interrupt.h>
-#include <linux/delay.h>
-#include <linux/scatterlist.h>
-#include <linux/mm.h>
-#include <linux/irq.h>
-#include <linux/slab.h>
-#include <linux/sched.h>
-/* include seems broken moving for patch submission
- * #include <mach/mux.h>
- * #include <mach/gpmc.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
- * #include <linux/westbridge/cyaserr.h>
- * #include <linux/westbridge/cyasregs.h>
- * #include <linux/westbridge/cyasdma.h>
- * #include <linux/westbridge/cyasintr.h>
- */
-#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
-#include "../../../include/linux/westbridge/cyaserr.h"
-#include "../../../include/linux/westbridge/cyasregs.h"
-#include "../../../include/linux/westbridge/cyasdma.h"
-#include "../../../include/linux/westbridge/cyasintr.h"
-
-#define HAL_REV "1.1.0"
-
-/*
- * uncomment to enable 16bit pnand interface
- */
-#define PNAND_16BIT_MODE
-
-/*
- * selects one of 3 versions of pnand_lbd_read()
- * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
- * reads through the gpmc CONTROLLER REGISTERS
- * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
- * same speed as the above
- * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
- */
-#define PNAND_LBD_READ_NO_PFE
-/* #define ENABLE_GPMC_PF_ENGINE */
-/* #define PFE_LBD_READ_V2 */
-
-/*
- * westbrige astoria ISR options to limit number of
- * back to back DMA transfers per ISR interrupt
- */
-#define MAX_DRQ_LOOPS_IN_ISR 4
-
-/*
- * debug prints enabling
- *#define DBGPRN_ENABLED
- *#define DBGPRN_DMA_SETUP_RD
- *#define DBGPRN_DMA_SETUP_WR
- */
-
-
-/*
- * For performance reasons, we handle storage endpoint transfers upto 4 KB
- * within the HAL itself.
- */
- #define CYASSTORAGE_WRITE_EP_NUM (4)
- #define CYASSTORAGE_READ_EP_NUM (8)
-
-/*
- * size of DMA packet HAL can accept from Storage API
- * HAL will fragment it into smaller chunks that the P port can accept
- */
-#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
-
-/*
- * P port MAX DMA packet size according to interface/ep configurartion
- */
-#define HAL_DMA_PKT_SZ 512
-
-#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
- ((ep) == 6) || ((ep) == 8))
-
-/*
- * persistant, stores current GPMC interface cfg mode
- */
-static uint8_t pnand_16bit;
-
-/*
- * keep processing new WB DRQ in ISR untill all handled (performance feature)
- */
-#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
-
-
-/*
- * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
- */
-#define CASDO 0x05
-#define CASDI 0x85
-#define RDPAGE_B1 0x00
-#define RDPAGE_B2 0x30
-#define PGMPAGE_B1 0x80
-#define PGMPAGE_B2 0x10
-
-/*
- * The type of DMA operation, per endpoint
- */
-typedef enum cy_as_hal_dma_type {
- cy_as_hal_read,
- cy_as_hal_write,
- cy_as_hal_none
-} cy_as_hal_dma_type ;
-
-
-/*
- * SG list halpers defined in scaterlist.h
-#define sg_is_chain(sg) ((sg)->page_link & 0x01)
-#define sg_is_last(sg) ((sg)->page_link & 0x02)
-#define sg_chain_ptr(sg) \
- ((struct scatterlist *) ((sg)->page_link & ~0x03))
-*/
-typedef struct cy_as_hal_endpoint_dma {
- cy_bool buffer_valid ;
- uint8_t *data_p ;
- uint32_t size ;
- /*
- * sg_list_enabled - if true use, r/w DMA transfers use sg list,
- * FALSE use pointer to a buffer
- * sg_p - pointer to the owner's sg list, of there is such
- * (like blockdriver)
- * dma_xfer_sz - size of the next dma xfer on P port
- * seg_xfer_cnt - counts xfered bytes for in current sg_list
- * memory segment
- * req_xfer_cnt - total number of bytes transfered so far in
- * current request
- * req_length - total request length
- */
- bool sg_list_enabled;
- struct scatterlist *sg_p ;
- uint16_t dma_xfer_sz;
- uint32_t seg_xfer_cnt;
- uint16_t req_xfer_cnt;
- uint16_t req_length;
- cy_as_hal_dma_type type ;
- cy_bool pending ;
-} cy_as_hal_endpoint_dma ;
-
-/*
- * The list of OMAP devices (should be one)
- */
-static cy_as_omap_dev_kernel *m_omap_list_p;
-
-/*
- * The callback to call after DMA operations are complete
- */
-static cy_as_hal_dma_complete_callback callback;
-
-/*
- * Pending data size for the endpoints
- */
-static cy_as_hal_endpoint_dma end_points[16] ;
-
-/*
- * Forward declaration
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
-
-static uint16_t intr_sequence_num;
-static uint8_t intr__enable;
-spinlock_t int_lock ;
-
-static u32 iomux_vma;
-static u32 csa_phy;
-
-/*
- * gpmc I/O registers VMA
- */
-static u32 gpmc_base ;
-
-/*
- * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
- */
-static u32 gpmc_data_vma;
-static u32 ndata_reg_vma;
-static u32 ncmd_reg_vma;
-static u32 naddr_reg_vma;
-
-/*
- * fwd declarations
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static inline u16 __attribute__((always_inline))
- ast_p_nand_casdo_read(u8 reg_addr8);
-static inline void __attribute__((always_inline))
- ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
-
-/*
- * prints given number of omap registers
- */
-static void cy_as_hal_print_omap_regs(char *name_prefix,
- u8 name_base, u32 virt_base, u16 count)
-{
- u32 reg_val, reg_addr;
- u16 i;
- cy_as_hal_print_message(KERN_INFO "\n");
- for (i = 0; i < count; i++) {
-
- reg_addr = virt_base + (i*4);
- /* use virtual addresses here*/
- reg_val = __raw_readl(reg_addr);
- cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
- name_prefix, name_base+i,
- reg_addr, reg_val);
- }
-}
-
-/*
- * setMUX function for a pad + additional pad flags
- */
-static u16 omap_cfg_reg_L(u32 pad_func_index)
-{
- static u8 sanity_check = 1;
-
- u32 reg_vma;
- u16 cur_val, wr_val, rdback_val;
-
- /*
- * do sanity check on the omap_mux_pin_cfg[] table
- */
- cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg ");
- if (sanity_check) {
- if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
-
- cy_as_hal_print_message(KERN_INFO
- "table is good.\n");
- } else {
- cy_as_hal_print_message(KERN_WARNING
- "table is bad, fix it");
- }
- /*
- * do it only once
- */
- sanity_check = 0;
- }
-
- /*
- * get virtual address to the PADCNF_REG
- */
- reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
-
- /*
- * add additional USER PU/PD/EN flags
- */
- wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
- cur_val = IORD16(reg_vma);
-
- /*
- * PADCFG regs 16 bit long, packed into 32 bit regs,
- * can also be accessed as u16
- */
- IOWR16(reg_vma, wr_val);
- rdback_val = IORD16(reg_vma);
-
- /*
- * in case if the caller wants to save the old value
- */
- return wr_val;
-}
-
-#define BLKSZ_4K 0x1000
-
-/*
- * switch GPMC DATA bus mode
- */
-void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
-{
- uint32_t tmp32;
-
- /*
- * disable gpmc CS4 operation 1st
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
- GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * GPMC NAND data bus can be 8 or 16 bit wide
- */
- if (dbus16_enabled) {
- DBGPRN("enabling 16 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2) |
- GPMC_CONFIG1_DEVICESIZE_16)
- );
- } else {
- DBGPRN(KERN_INFO "enabling 8 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2))
- );
- }
-
- /*
- * re-enable astoria CS operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- *remember the state
- */
- pnand_16bit = dbus16_enabled;
-}
-
-static int cy_as_hal_gpmc_init(void)
-{
- u32 tmp32;
- int err;
- struct gpmc_timings timings;
- /*
- * get GPMC i/o registers base(already been i/o mapped
- * in kernel, no need for separate i/o remap)
- */
- gpmc_base = phys_to_virt(OMAP34XX_GPMC_BASE);
- DBGPRN(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
- gpmc_base, __raw_readl(gpmc_base)
- );
-
- /*
- * these are globals are full VMAs of the gpmc_base above
- */
- ncmd_reg_vma = GPMC_VMA(GPMC_NAND_CMD);
- naddr_reg_vma = GPMC_VMA(GPMC_NAND_ADDR);
- ndata_reg_vma = GPMC_VMA(GPMC_NAND_DATA);
-
- /*
- * request GPMC CS for ASTORIA request
- */
- if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
- cy_as_hal_print_message(KERN_ERR "error failed to request"
- "ncs4 for ASTORIA\n");
- return -1;
- } else {
- DBGPRN(KERN_INFO "got phy_addr:%x for "
- "GPMC CS%d GPMC_CFGREG7[CS4]\n",
- csa_phy, AST_GPMC_CS);
- }
-
- /*
- * request VM region for 4K addr space for chip select 4 phy address
- * technically we don't need it for NAND devices, but do it anyway
- * so that data read/write bus cycle can be triggered by reading
- * or writing this mem region
- */
- if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
- err = -EBUSY;
- cy_as_hal_print_message(KERN_ERR "error MEM region "
- "request for phy_addr:%x failed\n",
- csa_phy);
- goto out_free_cs;
- }
-
- /*
- * REMAP mem region associated with our CS
- */
- gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
- if (!gpmc_data_vma) {
- err = -ENOMEM;
- cy_as_hal_print_message(KERN_ERR "error- ioremap()"
- "for phy_addr:%x failed", csa_phy);
-
- goto out_release_mem_region;
- }
- cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
- csa_phy, gpmc_data_vma);
-
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- memset(&timings, 0, sizeof(timings));
-
- /* cs timing */
- timings.cs_on = WB_GPMC_CS_t_o_n;
- timings.cs_wr_off = WB_GPMC_BUSCYC_t;
- timings.cs_rd_off = WB_GPMC_BUSCYC_t;
-
- /* adv timing */
- timings.adv_on = WB_GPMC_ADV_t_o_n;
- timings.adv_rd_off = WB_GPMC_BUSCYC_t;
- timings.adv_wr_off = WB_GPMC_BUSCYC_t;
-
- /* oe timing */
- timings.oe_on = WB_GPMC_OE_t_o_n;
- timings.oe_off = WB_GPMC_OE_t_o_f_f;
- timings.access = WB_GPMC_RD_t_a_c_c;
- timings.rd_cycle = WB_GPMC_BUSCYC_t;
-
- /* we timing */
- timings.we_on = WB_GPMC_WE_t_o_n;
- timings.we_off = WB_GPMC_WE_t_o_f_f;
- timings.wr_access = WB_GPMC_WR_t_a_c_c;
- timings.wr_cycle = WB_GPMC_BUSCYC_t;
-
- timings.page_burst_access = WB_GPMC_BUSCYC_t;
- timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
- gpmc_cs_set_timings(AST_GPMC_CS, &timings);
-
- cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
- GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
-
- /*
- * DISABLE cs4, NOTE GPMC REG7 is already configured
- * at this point by gpmc_cs_request
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
- ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (AS_CS_MASK | AS_CS_BADDR));
-
- /*
- * by default configure GPMC into 8 bit mode
- * (to match astoria default mode)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- /*
- * ENABLE astoria cs operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- * No method currently exists to write this register through GPMC APIs
- * need to change WAIT2 polarity
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
- IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
-
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
-
- return 0;
-
-out_release_mem_region:
- release_mem_region(csa_phy, BLKSZ_4K);
-
-out_free_cs:
- gpmc_cs_free(AST_GPMC_CS);
-
- return err;
-}
-
-/*
- * west bridge astoria ISR (Interrupt handler)
- */
-static irqreturn_t cy_astoria_int_handler(int irq,
- void *dev_id, struct pt_regs *regs)
-{
- cy_as_omap_dev_kernel *dev_p;
- uint16_t read_val = 0 ;
- uint16_t mask_val = 0 ;
-
- /*
- * debug stuff, counts number of loops per one intr trigger
- */
- uint16_t drq_loop_cnt = 0;
- uint8_t irq_pin;
- /*
- * flags to watch
- */
- const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
- CY_AS_MEM_P0_INTR_REG_MBINT |
- CY_AS_MEM_P0_INTR_REG_PMINT |
- CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
-
- /*
- * sample IRQ pin level (just for statistics)
- */
- irq_pin = __gpio_get_value(AST_INT);
-
- /*
- * this one just for debugging
- */
- intr_sequence_num++ ;
-
- /*
- * astoria device handle
- */
- dev_p = dev_id;
-
- /*
- * read Astoria intr register
- */
- read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) ;
-
- /*
- * save current mask value
- */
- mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG) ;
-
- DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
- intr_sequence_num, read_val);
-
- /*
- * Disable WB interrupt signal generation while we are in ISR
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
- /*
- * this is a DRQ Interrupt
- */
- if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
-
- do {
- /*
- * handle DRQ interrupt
- */
- drq_loop_cnt++;
-
- cy_handle_d_r_q_interrupt(dev_p) ;
-
- /*
- * spending to much time in ISR may impact
- * average system performance
- */
- if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
- break;
-
- /*
- * Keep processing if there is another DRQ int flag
- */
- } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) &
- CY_AS_MEM_P0_INTR_REG_DRQINT);
- }
-
- if (read_val & sentinel)
- cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
-
- DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
- "int_pin:%d DRQ_jobs:%d\n",
- intr_sequence_num,
- mask_val,
- irq_pin,
- drq_loop_cnt);
-
- /*
- * re-enable WB hw interrupts
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
-
- return IRQ_HANDLED ;
-}
-
-static int cy_as_hal_configure_interrupts(void *dev_p)
-{
- int result;
- int irq_pin = AST_INT;
-
- set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
-
- /*
- * for shared IRQS must provide non NULL device ptr
- * othervise the int won't register
- * */
- result = request_irq(OMAP_GPIO_IRQ(irq_pin),
- (irq_handler_t)cy_astoria_int_handler,
- IRQF_SHARED, "AST_INT#", dev_p);
-
- if (result == 0) {
- /*
- * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
- * assigned to this interrupt
- * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
- * of AST_INT GPIO IRQ_ENABLE FLAG
- */
- cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
- "%d assigned IRQ #%d IRQEN1=%d\n",
- irq_pin,
- OMAP_GPIO_IRQ(irq_pin),
- OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
- );
- } else {
- cy_as_hal_print_message("cyasomaphal: interrupt "
- "failed to register\n");
- gpio_free(irq_pin);
- cy_as_hal_print_message(KERN_WARNING
- "ASTORIA: can't get assigned IRQ"
- "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
- }
-
- return result;
-}
-
-/*
- * initialize OMAP pads/pins to user defined functions
- */
-static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- /*
- * browse through the table an dinitiaze the pins
- */
- u32 in_level = 0;
- u16 tmp16, mux_val;
-
- while (pad_cfg_tab->name != NULL) {
-
- if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
-
- pad_cfg_tab->valid = 1;
- mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
-
- /*
- * always set drv level before changing out direction
- */
- __gpio_set_value(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /*
- * "0" - OUT, "1", input omap_set_gpio_direction
- * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
- */
- if (pad_cfg_tab->dir)
- gpio_direction_input(pad_cfg_tab->pin_num);
- else
- gpio_direction_output(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /* sample the pin */
- in_level = __gpio_get_value(pad_cfg_tab->pin_num);
-
- cy_as_hal_print_message(KERN_INFO "configured %s to "
- "OMAP pad_%d, DIR=%d "
- "DOUT=%d, DIN=%d\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num,
- pad_cfg_tab->dir,
- pad_cfg_tab->drv,
- in_level
- );
- } else {
- /*
- * get the pad_mux value to check on the pin_function
- */
- cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
- "for signal %s, its already taken\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- }
-
- tmp16 = *(u16 *)PADCFG_VMA
- (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
-
- cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
- "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
- pad_cfg_tab->pin_num, tmp16,
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
- );
-
- /*
- * next pad_cfg deriptor
- */
- pad_cfg_tab++;
- }
-
- cy_as_hal_print_message(KERN_INFO"pads configured\n");
-}
-
-
-/*
- * release gpios taken by the module
- */
-static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- while (pad_cfg_tab->name != NULL) {
-
- if (pad_cfg_tab->valid) {
- gpio_free(pad_cfg_tab->pin_num);
- pad_cfg_tab->valid = 0;
- cy_as_hal_print_message(KERN_INFO "GPIO_%d "
- "released from %s\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- } else {
- cy_as_hal_print_message(KERN_INFO "no release "
- "for %s, GPIO_%d, wasn't acquired\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num);
- }
- pad_cfg_tab++;
- }
-}
-
-void cy_as_hal_config_c_s_mux(void)
-{
- /*
- * FORCE the GPMC CS4 pin (it is in use by the zoom system)
- */
- omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
-}
-EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
-
-/*
- * inits all omap h/w
- */
-uint32_t cy_as_hal_processor_hw_init(void)
-{
- int i, err;
-
- cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
-
- iomux_vma = (u32)ioremap_nocache(
- (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
- cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
- iomux_vma, IORD32(iomux_vma));
-
- /*
- * remap gpio banks
- */
- for (i = 0; i < 6; i++) {
- gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
- gpio_vma_tab[i].phy_addr,
- gpio_vma_tab[i].size);
-
- cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
- gpio_vma_tab[i].name,
- (u32)gpio_vma_tab[i].virt_addr);
- };
-
- /*
- * force OMAP_GPIO_126 to rleased state,
- * will be configured to drive reset
- */
- gpio_free(AST_RESET);
-
- /*
- *same thing with AStoria CS pin
- */
- gpio_free(AST_CS);
-
- /*
- * initialize all the OMAP pads connected to astoria
- */
- cy_as_hal_init_user_pads(user_pad_cfg);
-
- err = cy_as_hal_gpmc_init();
- if (err < 0)
- cy_as_hal_print_message(KERN_INFO"gpmc init failed:%d", err);
-
- cy_as_hal_config_c_s_mux();
-
- return gpmc_data_vma;
-}
-EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
-
-void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
-{
- /*
- * free omap hw resources
- */
- if (gpmc_data_vma != 0)
- iounmap((void *)gpmc_data_vma);
-
- if (csa_phy != 0)
- release_mem_region(csa_phy, BLKSZ_4K);
-
- gpmc_cs_free(AST_GPMC_CS);
-
- free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
-
- cy_as_hal_release_user_pads(user_pad_cfg);
-}
-
-/*
- * These are the functions that are not part of the
- * HAL layer, but are required to be called for this HAL
- */
-
-/*
- * Called On AstDevice LKM exit
- */
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
-
- /*
- * TODO: Need to disable WB interrupt handlere 1st
- */
- if (0 == dev_p)
- return 1 ;
-
- cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
- if (dev_p->m_sig != CY_AS_OMAP_KERNEL_HAL_SIG) {
- cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
- pgm, __func__) ;
- return 1 ;
- }
-
- /*
- * disable interrupt
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
-#if 0
- if (dev_p->thread_flag == 0) {
- dev_p->thread_flag = 1 ;
- wait_for_completion(&dev_p->thread_complete) ;
- cy_as_hal_print_message("cyasomaphal:"
- "done cleaning thread\n");
- cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
- }
-#endif
-
- cy_as_hal_omap_hardware_deinit(dev_p);
-
- /*
- * Rearrange the list
- */
- if (m_omap_list_p == dev_p)
- m_omap_list_p = dev_p->m_next_p ;
-
- cy_as_hal_free(dev_p) ;
-
- cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
- return 0;
-}
-
-int omap_start_intr(cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
- int ret = 0 ;
- const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
- CY_AS_MEM_P0_INTR_REG_MBINT ;
-
- /*
- * register for interrupts
- */
- ret = cy_as_hal_configure_interrupts(dev_p) ;
-
- /*
- * enable only MBox & DRQ interrupts for now
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask) ;
-
- return 1 ;
-}
-
-/*
- * Below are the functions that communicate with the WestBridge device.
- * These are system dependent and must be defined by the HAL layer
- * for a given system.
- */
-
-/*
- * GPMC NAND command+addr write phase
- */
-static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
-{
- /*
- * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
- */
- u32 tmpa32 = ((row_addr << 16) | col_addr);
- u8 RA2 = (u8)(row_addr >> 16);
-
- if (!pnand_16bit) {
- /*
- * GPMC PNAND 8bit BUS
- */
- /*
- * CMD1
- */
- IOWR8(ncmd_reg_vma, cmdb1);
-
- /*
- *pnand bus: <CA0,CA1,RA0,RA1>
- */
- IOWR32(naddr_reg_vma, tmpa32);
-
- /*
- * <RA2> , always zero
- */
- IOWR8(naddr_reg_vma, RA2);
-
- } else {
- /*
- * GPMC PNAND 16bit BUS , in 16 bit mode CMD
- * and ADDR sent on [d7..d0]
- */
- uint8_t CA0, CA1, RA0, RA1;
- CA0 = tmpa32 & 0x000000ff;
- CA1 = (tmpa32 >> 8) & 0x000000ff;
- RA0 = (tmpa32 >> 16) & 0x000000ff;
- RA1 = (tmpa32 >> 24) & 0x000000ff;
-
- /*
- * can't use 32 bit writes here omap will not serialize
- * them to lower half in16 bit mode
- */
-
- /*
- *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
- */
- IOWR8(ncmd_reg_vma, cmdb1);
- IOWR8(naddr_reg_vma, CA0);
- IOWR8(naddr_reg_vma, CA1);
- IOWR8(naddr_reg_vma, RA0);
- IOWR8(naddr_reg_vma, RA1);
- IOWR8(naddr_reg_vma, RA2);
- }
-}
-
-/*
- * spin until r/b goes high
- */
-inline int wait_rn_b_high(void)
-{
- u32 w_spins = 0;
-
- /*
- * TODO: note R/b may go low here, need to spin until high
- * while (omap_get_gpio_datain(AST_RnB) == 0) {
- * w_spins++;
- * }
- * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- *
- * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- * w_spins++;
- * }
- * printk("<1>RnB=0!:%d\n",w_spins);
- * }
- */
- return w_spins;
-}
-
-#ifdef ENABLE_GPMC_PF_ENGINE
-/* #define PFE_READ_DEBUG
- * PNAND block read with OMAP PFE enabled
- * status: Not tested, NW, broken , etc
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint8_t bytes_in_fifo;
-
- /* debug vars*/
-#ifdef PFE_READ_DEBUG
- uint32_t loop_limit;
- uint16_t bytes_read = 0;
-#endif
-
- /*
- * configure the prefetch engine
- */
- uint32_t tmp32;
- uint32_t pfe_status;
-
- /*
- * DISABLE GPMC CS4 operation 1st, this is
- * in case engine is be already disabled
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
-
-#ifdef PFE_READ_DEBUG
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
- if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
- printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
- tmp32, GPMC_PREFETCH_CONFIG1_VAL);
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
- }
-
- /*
- *sanity check 2
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
- if (tmp32 != (count))
- printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
- "!= VAL written:%d\n", tmp32, count);
-#endif
-
- /*
- * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
- * on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
-
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * start the prefetch engine
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- while (1) {
- /*
- * GPMC PFE service loop
- */
- do {
- /*
- * spin until PFE fetched some
- * PNAND bus words in the FIFO
- */
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- bytes_in_fifo = (pfe_status >> 24) & 0x7f;
- } while (bytes_in_fifo == 0);
-
- /* whole 32 bit words in fifo */
- w32cnt = bytes_in_fifo >> 2;
-
-#if 0
- /*
- *NOTE: FIFO_PTR indicates number of NAND bus words bytes
- * already received in the FIFO and available to be read
- * by DMA or MPU whether COUNTVAL indicates number of BUS
- * words yet to be read from PNAND bus words
- */
- printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
- pfe_status, bytes_in_fifo,
- (pfe_status & 0x3fff), w32cnt);
-#endif
-
- while (w32cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- if ((pfe_status & 0x3fff) == 0) {
- /*
- * PFE acc angine done, there still may be data leftover
- * in the FIFO re-read FIFO BYTE counter (check for
- * leftovers from 32 bit read accesses above)
- */
- bytes_in_fifo = (IORD32(
- GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- /*
- * NOTE we may still have one word left in the fifo
- * read it out
- */
- ptr8 = ptr32;
- switch (bytes_in_fifo) {
-
- case 0:
- /*
- * nothing to do we already read the
- * FIFO out with 32 bit accesses
- */
- break;
- case 1:
- /*
- * this only possible
- * for 8 bit pNAND only
- */
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 2:
- /*
- * this one can occur in either modes
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- break;
-
- case 3:
- /*
- * this only possible for 8 bit pNAND only
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- ptr8 += 2;
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 4:
- /*
- * shouldn't happen, but has been seen
- * in 8 bit mode
- */
- *ptr32 = IORD32(gpmc_data_vma);
- break;
-
- default:
- printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
- bytes_in_fifo);
- break;
- }
- /*
- * read is completed, get out of the while(1) loop
- */
- break;
- }
- }
-}
-#endif
-
-#ifdef PFE_LBD_READ_V2
-/*
- * PFE engine assisted reads with the 64 byte blocks
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint8_t rd_cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint16_t reminder;
- uint32_t pfe_status;
-
- /*
- * ISSUE PNAND CMD+ADDR
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * setup PFE block
- * count - OMAP number of bytes to access on pnand bus
- */
-
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- do {
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
- rd_cnt = pfe_status >> (24+2);
-
- while (rd_cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- } while (pfe_status & 0x3fff);
-
- /*
- * read out the leftover
- */
- ptr8 = ptr32;
- rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- while (rd_cnt--)
- *ptr8++ = IORD8(gpmc_data_vma);
-}
-#endif
-
-#ifdef PNAND_LBD_READ_NO_PFE
-/*
- * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
- * the original working code, works at max speed for 8 bit xfers
- * for 16 bit the bus diagram has gaps
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint16_t *ptr16;
- uint16_t remainder;
-
- DBGPRN("<1> %s(): NO_PFE\n", __func__);
-
- ptr32 = buff;
- /* number of whole 32 bit words in the transfer */
- w32cnt = count >> 2;
-
- /* remainder, in bytes(0..3) */
- remainder = count & 03;
-
- /*
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * read data by 32 bit chunks
- */
- while (w32cnt--)
- *ptr32++ = IORD32(ndata_reg_vma);
-
- /*
- * now do the remainder(it can be 0, 1, 2 or 3)
- * same code for both 8 & 16 bit bus
- * do 1 or 2 MORE words
- */
- ptr16 = (uint16_t *)ptr32;
-
- switch (remainder) {
- case 1:
- /* read one 16 bit word
- * IN 8 BIT WE NEED TO READ even number of bytes
- */
- case 2:
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- *ptr16++ = IORD16(ndata_reg_vma);
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- default:
- /*
- * remainder is 0
- */
- break;
- }
-}
-#endif
-
-/*
- * uses LBD mode to write N bytes into astoria
- * Status: Working, however there are 150ns idle
- * timeafter every 2 (16 bit or 4(8 bit) bus cycles
- */
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint16_t remainder;
- uint8_t *ptr8;
- uint16_t *ptr16;
- uint32_t *ptr32;
-
- remainder = count & 03;
- w32cnt = count >> 2;
- ptr32 = buff;
- ptr8 = buff;
-
- /*
- * send: CMDB1, CA0,CA1,RA0,RA1,RA2
- */
- nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
-
- /*
- * blast the data out in 32bit chunks
- */
- while (w32cnt--)
- IOWR32(ndata_reg_vma, *ptr32++);
-
- /*
- * do the reminder if there is one
- * same handling for both 8 & 16 bit pnand: mode
- */
- ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
-
- switch (remainder) {
- case 1:
- /*
- * read one 16 bit word
- */
- case 2:
- IOWR16(ndata_reg_vma, *ptr16);
- break;
-
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- IOWR16(ndata_reg_vma, *ptr16++);
- IOWR16(ndata_reg_vma, *ptr16);
- break;
- default:
- /*
- * reminder is 0
- */
- break;
- }
- /*
- * finally issue a PGM cmd
- */
- IOWR8(ncmd_reg_vma, PGMPAGE_B2);
-}
-
-/*
- * write Astoria register
- */
-static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
-{
- unsigned long flags;
- u16 addr16;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t rdreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&rdreg_usage_cnt) != 0) {
- cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
- "* cy_as_hal_write_register usage:%d\n",
- atomic_read(&rdreg_usage_cnt));
- }
-
- atomic_inc(&rdreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- */
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
-
- /*
- * this should be sent on the 16 bit bus
- */
- IOWR16(ndata_reg_vma, data);
- } else {
- /*
- * 8 bit nand mode GPMC will automatically
- * seriallize 16bit or 32 bit writes into
- * 8 bit onesto the lower 8 bit in LE order
- */
- addr16 = 0x0c00 | reg_addr8;
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR16(naddr_reg_vma, addr16);
- IOWR16(ndata_reg_vma, data);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&rdreg_usage_cnt);
- local_irq_restore(flags);
-}
-
-
-/*
- * read astoria register via pNAND interface
- */
-static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
-{
- u16 data;
- u16 addr16;
- unsigned long flags;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t wrreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&wrreg_usage_cnt) != 0) {
- /*
- * if it gets here ( from other threads), this function needs
- * need spin_lock_irq save() protection
- */
- cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
- "cy_as_hal_write_register usage:%d\n",
- atomic_read(&wrreg_usage_cnt));
- }
- atomic_inc(&wrreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- * CMD1, CA1, CA2,
- */
-
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
- IOWR8(ncmd_reg_vma, 0x00E0);
-
- udelay(1);
-
- /*
- * much faster through the gPMC Register space
- */
- data = IORD16(ndata_reg_vma);
- } else {
- /*
- * 8 BIT gpmc NAND mode
- * CMD1, CA1, CA2, CMD2
- */
- addr16 = 0x0c00 | reg_addr8;
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR16(naddr_reg_vma, addr16);
- IOWR8(ncmd_reg_vma, 0xE0);
- udelay(1);
- data = IORD16(ndata_reg_vma);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&wrreg_usage_cnt);
- local_irq_restore(flags);
-
- return data;
-}
-
-
-/*
- * This function must be defined to write a register within the WestBridge
- * device. The addr value is the address of the register to write with
- * respect to the base address of the WestBridge device.
- */
-void cy_as_hal_write_register(
- cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data)
-{
- ast_p_nand_casdi_write((u8)addr, data);
-}
-
-/*
- * This function must be defined to read a register from the WestBridge
- * device. The addr value is the address of the register to read with
- * respect to the base address of the WestBridge device.
- */
-uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
-{
- uint16_t data = 0 ;
-
- /*
- * READ ASTORIA REGISTER USING CASDO
- */
- data = ast_p_nand_casdo_read((u8)addr);
-
- return data ;
-}
-
-/*
- * preps Ep pointers & data counters for next packet
- * (fragment of the request) xfer returns true if
- * there is a next transfer, and false if all bytes in
- * current request have been xfered
- */
-static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
-{
-
- if (!end_points[ep].sg_list_enabled) {
- /*
- * no further transfers for non storage EPs
- * (like EP2 during firmware download, done
- * in 64 byte chunks)
- */
- if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
- DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
- __func__, end_points[ep].req_length, ep);
-
- /*
- * no more transfers, we are done with the request
- */
- return false;
- }
-
- /*
- * calculate size of the next DMA xfer, corner
- * case for non-storage EPs where transfer size
- * is not egual N * HAL_DMA_PKT_SZ xfers
- */
- if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
- >= HAL_DMA_PKT_SZ) {
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
- } else {
- /*
- * that would be the last chunk less
- * than P-port max size
- */
- end_points[ep].dma_xfer_sz = end_points[ep].req_length -
- end_points[ep].req_xfer_cnt;
- }
-
- return true;
- }
-
- /*
- * for SG_list assisted dma xfers
- * are we done with current SG ?
- */
- if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
- /*
- * was it the Last SG segment on the list ?
- */
- if (sg_is_last(end_points[ep].sg_p)) {
- DBGPRN("<1> %s: EP:%d completed,"
- "%d bytes xfered\n",
- __func__,
- ep,
- end_points[ep].req_xfer_cnt
- );
-
- return false;
- } else {
- /*
- * There are more SG segments in current
- * request's sg list setup new segment
- */
-
- end_points[ep].seg_xfer_cnt = 0;
- end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
- /* set data pointer for next DMA sg transfer*/
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- DBGPRN("<1> %s new SG:_va:%p\n\n",
- __func__, end_points[ep].data_p);
- }
-
- }
-
- /*
- * for sg list xfers it will always be 512 or 1024
- */
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
-
- /*
- * next transfer is required
- */
-
- return true;
-}
-
-/*
- * Astoria DMA read request, APP_CPU reads from WB ep buffer
- */
-static void cy_service_e_p_dma_read_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- uint16_t v, size;
- void *dptr;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
-
- /*
- * get the XFER size frtom WB eP DMA REGISTER
- */
- v = cy_as_hal_read_register(tag, ep_dma_reg);
-
- /*
- * amount of data in EP buff in bytes
- */
- size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
-
- /*
- * memory pointer for this DMA packet xfer (sub_segment)
- */
- dptr = end_points[ep].data_p;
-
- DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
- ep,
- size,
- intr_sequence_num,
- dptr
- );
-
- cy_as_hal_assert(size != 0);
-
- if (size) {
- /*
- * the actual WB-->OMAP memory "soft" DMA xfer
- */
- p_nand_lbd_read(col_addr, row_addr, size, dptr);
- }
-
- /*
- * clear DMAVALID bit indicating that the data has been read
- */
- cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
-
- /*
- * pre-advance data pointer (if it's outside sg
- * list it will be reset anyway
- */
- end_points[ep].data_p += size;
-
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * we have more data to read in this request,
- * setup next dma packet due tell WB how much
- * data we are going to xfer next
- */
- v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, ep_dma_reg, v);
- } else {
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
- end_points[ep].req_xfer_cnt);
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * omap_cpu needs to transfer data to ASTORIA EP buffer
- */
-static void cy_service_e_p_dma_write_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- uint16_t addr;
- uint16_t v = 0;
- uint32_t size;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- void *dptr;
-
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- /*
- * note: size here its the size of the dma transfer could be
- * anything > 0 && < P_PORT packet size
- */
- size = end_points[ep].dma_xfer_sz ;
- dptr = end_points[ep].data_p ;
-
- /*
- * perform the soft DMA transfer, soft in this case
- */
- if (size)
- p_nand_lbd_write(col_addr, row_addr, size, dptr);
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
- /*
- * pre-advance data pointer
- * (if it's outside sg list it will be reset anyway)
- */
- end_points[ep].data_p += size;
-
- /*
- * now clear DMAVAL bit to indicate we are done
- * transferring data and that the data can now be
- * sent via USB to the USB host, sent to storage,
- * or used internally.
- */
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
- cy_as_hal_write_register(tag, addr, size) ;
-
- /*
- * finally, tell the USB subsystem that the
- * data is gone and we can accept the
- * next request if one exists.
- */
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * There is more data to go. Re-init the WestBridge DMA side
- */
- v = end_points[ep].dma_xfer_sz |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v) ;
- } else {
-
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- /*
- * this callback will wake up the process that might be
- * sleeping on the EP which data is being transferred
- */
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * HANDLE DRQINT from Astoria (called in AS_Intr context
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
-{
- uint16_t v ;
- static uint8_t service_ep = 2 ;
-
- /*
- * We've got DRQ INT, read DRQ STATUS Register */
- v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_DRQ) ;
-
- if (v == 0) {
-#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
-#endif
- return;
- }
-
- /*
- * Now, pick a given DMA request to handle, for now, we just
- * go round robin. Each bit position in the service_mask
- * represents an endpoint from EP2 to EP15. We rotate through
- * each of the endpoints to find one that needs to be serviced.
- */
- while ((v & (1 << service_ep)) == 0) {
-
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
- if (end_points[service_ep].type == cy_as_hal_write) {
- /*
- * handle DMA WRITE REQUEST: app_cpu will
- * write data into astoria EP buffer
- */
- cy_service_e_p_dma_write_request(dev_p, service_ep) ;
- } else if (end_points[service_ep].type == cy_as_hal_read) {
- /*
- * handle DMA READ REQUEST: cpu will
- * read EP buffer from Astoria
- */
- cy_service_e_p_dma_read_request(dev_p, service_ep) ;
- }
-#ifndef WESTBRIDGE_NDEBUG
- else
- cy_as_hal_print_message("cyashalomap:interrupt,"
- " w/o pending DMA job,"
- "-check DRQ_MASK logic\n") ;
-#endif
-
- /*
- * Now bump the EP ahead, so other endpoints get
- * a shot before the one we just serviced
- */
- if (end_points[service_ep].type == cy_as_hal_none) {
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
-}
-
-void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
-{
- DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
- if (end_points[ep].pending)
- cy_as_hal_write_register(tag,
- CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
-
- end_points[ep].buffer_valid = cy_false ;
- end_points[ep].type = cy_as_hal_none;
-}
-
-/*
- * enables/disables SG list assisted DMA xfers for the given EP
- * sg_list assisted XFERS can use physical addresses of mem pages in case if the
- * xfer is performed by a h/w DMA controller rather then the CPU on P port
- */
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
-{
- end_points[ep].sg_list_enabled = sg_xfer_enabled;
- DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
- ep, end_points[ep].sg_list_enabled);
-}
-EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
-
-/*
- * This function must be defined to transfer a block of data to
- * the WestBridge device. This function can use the burst write
- * (DMA) capabilities of WestBridge to do this, or it can just copy
- * the data using writes.
- */
-void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr = 0 ;
- uint16_t v = 0;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem. Another DMA request
- * is coming down before the previous one has completed.
- */
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_write ;
- end_points[ep].pending = cy_true;
-
- /*
- * total length of the request
- */
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * smaller xfers for non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- /*
- * check the EP transfer mode uses sg_list rather then a memory buffer
- * block devices pass it to the HAL, so the hAL could get to the real
- * physical address for each segment and set up a DMA controller
- * hardware ( if there is one)
- */
- if (end_points[ep].sg_list_enabled) {
- /*
- * buf - pointer to the SG list
- * data_p - data pointer to the 1st DMA segment
- * seg_xfer_cnt - keeps track of N of bytes sent in current
- * sg_list segment
- * req_xfer_cnt - keeps track of the total N of bytes
- * transferred for the request
- */
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].req_xfer_cnt = 0;
-
-#ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
- "req_sz:%d, maxsz:%d\n",
- __func__,
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
-#endif
-
- } else {
- /*
- * setup XFER for non sg_list assisted EPs
- */
-
- #ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("<1>%s non storage or sz < 512:"
- "EP:%d, sz:%d\n", __func__, ep, size);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * will keep track No of bytes xferred for the request
- */
- end_points[ep].req_xfer_cnt = 0;
- }
-
- /*
- * Tell WB we are ready to send data on the given endpoint
- */
- v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
- | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- cy_as_hal_write_register(tag, addr, v) ;
-}
-
-/*
- * This function must be defined to transfer a block of data from
- * the WestBridge device. This function can use the burst read
- * (DMA) capabilities of WestBridge to do this, or it can just
- * copy the data using reads.
- */
-void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr ;
- uint16_t v ;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem.
- * Another DMA request is coming down before the
- * previous one has completed. we should not get
- * new requests if current is still in process
- */
-
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
-
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_read ;
- end_points[ep].pending = cy_true;
- end_points[ep].req_xfer_cnt = 0;
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * so that we could handle small xfers on in case
- * of non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- if (end_points[ep].sg_list_enabled) {
- /*
- * Handle sg-list assisted EPs
- * seg_xfer_cnt - keeps track of N of sent packets
- * buf - pointer to the SG list
- * data_p - data pointer for the 1st DMA segment
- */
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
-
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
- "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
- #endif
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- } else {
- /*
- * Non sg list EP passed void *buf rather then scatterlist *sg
- */
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("%s:non-sg_list EP:%d,"
- "RQ_sz:%d, maxsz:%d\n",
- __func__, ep, size, maxsize);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * Program the EP DMA register for Storage endpoints only.
- */
- if (is_storage_e_p(ep)) {
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- }
- }
-}
-
-/*
- * This function must be defined to allow the WB API to
- * register a callback function that is called when a
- * DMA transfer is complete.
- */
-void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb)
-{
- DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
- __func__, (uint32_t)cb);
- callback = cb ;
-}
-
-/*
- * This function must be defined to return the maximum size of
- * DMA request that can be handled on the given endpoint. The
- * return value should be the maximum size in bytes that the DMA
- * module can handle.
- */
-uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep)
-{
- /*
- * Storage reads and writes are always done in 512 byte blocks.
- * So, we do the count handling within the HAL, and save on
- * some of the data transfer delay.
- */
- if ((ep == CYASSTORAGE_READ_EP_NUM) ||
- (ep == CYASSTORAGE_WRITE_EP_NUM)) {
- /* max DMA request size HAL can handle by itself */
- return CYASSTORAGE_MAX_XFER_SIZE;
- } else {
- /*
- * For the USB - Processor endpoints, the maximum transfer
- * size depends on the speed of USB operation. So, we use
- * the following constant to indicate to the API that
- * splitting of the data into chunks less that or equal to
- * the max transfer size should be handled internally.
- */
-
- /* DEFINED AS 0xffffffff in cyasdma.h */
- return CY_AS_DMA_MAX_SIZE_HW_SIZE;
- }
-}
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the WestBridge device. Generally this is done via a GPIO of
- * some type.
- */
-cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
-{
- /*
- * Not supported as of now.
- */
- return cy_false ;
-}
-
-void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
-{
- cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
- cy_as_hal_print_message("please check the input voltage levels");
- cy_as_hal_print_message("and clock, and restart the system\n") ;
-}
-
-/*
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- */
-
-/*
- * This function is required by the API to allocate memory.
- * This function is expected to work exactly like malloc().
- */
-void *cy_as_hal_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required by the API to free memory allocated
- * with CyAsHalAlloc(). This function is'expected to work exacly
- * like free().
- */
-void cy_as_hal_free(void *mem_p)
-{
- kfree(mem_p) ;
-}
-
-/*
- * Allocator that can be used in interrupt context.
- * We have to ensure that the kmalloc call does not
- * sleep in this case.
- */
-void *cy_as_hal_c_b_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required to set a block of memory to a
- * specific value. This function is expected to work exactly
- * like memset()
- */
-void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
-{
- memset(ptr, value, cnt) ;
-}
-
-/*
- * This function is expected to create a sleep channel.
- * The data structure that represents the sleep channel object
- * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
- * passed as a pointer, and allpocated by the caller
- * (typically as a local var on the stack) "Create" word should read as
- * "SleepOn", this func doesn't actually create anything
- */
-cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- init_waitqueue_head(&channel->wq) ;
- return cy_true ;
-}
-
-/*
- * for this particular HAL it doesn't actually destroy anything
- * since no actual sleep object is created in CreateSleepChannel()
- * sleep channel is given by the pointer in the argument.
- */
-cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- return cy_true ;
-}
-
-/*
- * platform specific wakeable Sleep implementation
- */
-cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
-{
- wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
- return cy_true ;
-}
-
-/*
- * wakes up the process waiting on the CHANNEL
- */
-cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
-{
- wake_up_interruptible_all(&channel->wq);
- return cy_true ;
-}
-
-uint32_t cy_as_hal_disable_interrupts()
-{
- if (0 == intr__enable)
- ;
-
- intr__enable++ ;
- return 0 ;
-}
-
-void cy_as_hal_enable_interrupts(uint32_t val)
-{
- intr__enable-- ;
- if (0 == intr__enable)
- ;
-}
-
-/*
- * Sleep atleast 150ns, cpu dependent
- */
-void cy_as_hal_sleep150(void)
-{
- uint32_t i, j;
-
- j = 0;
- for (i = 0; i < 1000; i++)
- j += (~i);
-}
-
-void cy_as_hal_sleep(uint32_t ms)
-{
- cy_as_hal_sleep_channel channel;
-
- cy_as_hal_create_sleep_channel(&channel) ;
- cy_as_hal_sleep_on(&channel, ms) ;
- cy_as_hal_destroy_sleep_channel(&channel) ;
-}
-
-cy_bool cy_as_hal_is_polling()
-{
- return cy_false;
-}
-
-void cy_as_hal_c_b_free(void *ptr)
-{
- cy_as_hal_free(ptr);
-}
-
-/*
- * suppose to reinstate the astoria registers
- * that may be clobbered in sleep mode
- */
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
- (void) is_standby_wakeup;
-}
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
-}
-
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
-{
- /*
- * we are in asynchronous mode. so no need to handle this
- */
- return true;
-}
-
-/*
- * init OMAP h/w resources
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug)
-{
- cy_as_omap_dev_kernel *dev_p ;
- int i;
- u16 data16[4];
- u8 pncfg_reg;
-
- /*
- * No debug mode support through argument as of now
- */
- (void)debug;
-
- DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
-
- /*
- * Initialize the HAL level endpoint DMA data.
- */
- for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
- end_points[i].data_p = 0 ;
- end_points[i].pending = cy_false ;
- end_points[i].size = 0 ;
- end_points[i].type = cy_as_hal_none ;
- end_points[i].sg_list_enabled = cy_false;
-
- /*
- * by default the DMA transfers to/from the E_ps don't
- * use sg_list that implies that the upper devices like
- * blockdevice have to enable it for the E_ps in their
- * initialization code
- */
- }
-
- /*
- * allocate memory for OMAP HAL
- */
- dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
- sizeof(cy_as_omap_dev_kernel)) ;
- if (dev_p == 0) {
- cy_as_hal_print_message("out of memory allocating OMAP"
- "device structure\n") ;
- return 0 ;
- }
-
- dev_p->m_sig = CY_AS_OMAP_KERNEL_HAL_SIG;
-
- /*
- * initialize OMAP hardware and StartOMAPKernelall gpio pins
- */
- dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
-
- /*
- * Now perform a hard reset of the device to have
- * the new settings take effect
- */
- __gpio_set_value(AST_WAKEUP, 1);
-
- /*
- * do Astoria h/w reset
- */
- DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
-
- /*
- * NEGATIVE PULSE on RST pin
- */
- __gpio_set_value(AST_RESET, 0);
- mdelay(1);
- __gpio_set_value(AST_RESET, 1);
- mdelay(50);
-
- /*
- * note AFTER reset PNAND interface is 8 bit mode
- * so if gpmc Is configured in 8 bit mode upper half will be FF
- */
- pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
-#ifdef PNAND_16BIT_MODE
-
- /*
- * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
- */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
-
- /*
- * now in order to continue to talk to astoria
- * sw OMAP GPMC into 16 bit mode as well
- */
- cy_as_hal_gpmc_enable_16bit_bus(cy_true);
-#else
- /* Astoria and GPMC are already in 8 bit mode, jsut initialize PNAND_CFG */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
-#endif
-
- /*
- * NOTE: if you want to capture bus activity on the LA,
- * don't use printks in between the activities you want to capture.
- * prinks may take milliseconds, and the data of interest
- * will fall outside the LA capture window/buffer
- */
- data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
- data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
- if (data16[0] != 0xA200) {
- /*
- * astoria device is not found
- */
- printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
- printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
- data16[0], data16[0]);
- goto bus_acc_error;
- }
-
- cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
- "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
- "PNAND_CFG after RST:%4.4x\n "
- "CY_AS_MEM_PNAND_CFG"
- "after cfg_wr:%4.4x\n\n",
- data16[0], pncfg_reg, data16[1]);
-
- dev_p->thread_flag = 1 ;
- spin_lock_init(&int_lock) ;
- dev_p->m_next_p = m_omap_list_p ;
-
- m_omap_list_p = dev_p ;
- *tag = dev_p;
-
- cy_as_hal_configure_interrupts((void *)dev_p);
-
- cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
- ", kernel HZ:%d\n", dev_p, HZ);
-
- /*
- *make processor to storage endpoints SG assisted by default
- */
- cy_as_hal_set_ep_dma_mode(4, true);
- cy_as_hal_set_ep_dma_mode(8, true);
-
- return 1 ;
-
- /*
- * there's been a NAND bus access error or
- * astoria device is not connected
- */
-bus_acc_error:
- /*
- * at this point hal tag hasn't been set yet
- * so the device will not call omap_stop
- */
- cy_as_hal_omap_hardware_deinit(dev_p);
- cy_as_hal_free(dev_p) ;
- return 0;
-}
-
-#else
-/*
- * Some compilers do not like empty C files, so if the OMAP hal is not being
- * compiled, we compile this single function. We do this so that for a
- * given target HAL there are not multiple sources for the HAL functions.
- */
-void my_o_m_a_p_kernel_hal_dummy_function(void)
-{
-}
-
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 2010-09-01 16:01:14.000000000 -0700
@@ -0,0 +1,2462 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+/*
+ * uncomment to enable 16bit pnand interface
+ */
+#define PNAND_16BIT_MODE
+
+/*
+ * selects one of 3 versions of pnand_lbd_read()
+ * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
+ * reads through the gpmc CONTROLLER REGISTERS
+ * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
+ * same speed as the above
+ * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
+ */
+#define PNAND_LBD_READ_NO_PFE
+/* #define ENABLE_GPMC_PF_ENGINE */
+/* #define PFE_LBD_READ_V2 */
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+ #define CYASSTORAGE_WRITE_EP_NUM (4)
+ #define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+/*
+ * persistant, stores current GPMC interface cfg mode
+ */
+static uint8_t pnand_16bit;
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
+ */
+#define CASDO 0x05
+#define CASDI 0x85
+#define RDPAGE_B1 0x00
+#define RDPAGE_B2 0x30
+#define PGMPAGE_B1 0x80
+#define PGMPAGE_B2 0x10
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+static u32 csa_phy;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
+ */
+static u32 gpmc_data_vma;
+static u32 ndata_reg_vma;
+static u32 ncmd_reg_vma;
+static u32 naddr_reg_vma;
+
+/*
+ * fwd declarations
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static inline u16 __attribute__((always_inline))
+ ast_p_nand_casdo_read(u8 reg_addr8);
+static inline void __attribute__((always_inline))
+ ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ cy_as_hal_print_message(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg ");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ cy_as_hal_print_message(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+#define BLKSZ_4K 0x1000
+
+/*
+ * switch GPMC DATA bus mode
+ */
+void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
+{
+ uint32_t tmp32;
+
+ /*
+ * disable gpmc CS4 operation 1st
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
+ GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * GPMC NAND data bus can be 8 or 16 bit wide
+ */
+ if (dbus16_enabled) {
+ DBGPRN("enabling 16 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2) |
+ GPMC_CONFIG1_DEVICESIZE_16)
+ );
+ } else {
+ DBGPRN(KERN_INFO "enabling 8 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2))
+ );
+ }
+
+ /*
+ * re-enable astoria CS operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ *remember the state
+ */
+ pnand_16bit = dbus16_enabled;
+}
+
+static int cy_as_hal_gpmc_init(void)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ cy_as_hal_print_message(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ //gpmc_base = phys_to_virt(OMAP34XX_GPMC_BASE);
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, BLKSZ_4K);
+ cy_as_hal_print_message(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ /*
+ * these are globals are full VMAs of the gpmc_base above
+ */
+ ncmd_reg_vma = GPMC_VMA(AST_GPMC_NAND_CMD);
+ naddr_reg_vma = GPMC_VMA(AST_GPMC_NAND_ADDR);
+ ndata_reg_vma = GPMC_VMA(AST_GPMC_NAND_DATA);
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ cy_as_hal_print_message(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ csa_phy, AST_GPMC_CS);
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ csa_phy);
+ goto out_free_cs;
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
+ if (!gpmc_data_vma) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", csa_phy);
+
+ goto out_release_mem_region;
+ }
+ cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ csa_phy, gpmc_data_vma);
+
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_o_n;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_o_n;
+ timings.adv_rd_off = WB_GPMC_BUSCYC_t;
+ timings.adv_wr_off = WB_GPMC_BUSCYC_t;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_o_n;
+ timings.oe_off = WB_GPMC_OE_t_o_f_f;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_o_n;
+ timings.we_off = WB_GPMC_WE_t_o_f_f;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ /*
+ * DISABLE cs4, NOTE GPMC REG7 is already configured
+ * at this point by gpmc_cs_request
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
+ ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (AS_CS_MASK | AS_CS_BADDR));
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ /*
+ * ENABLE astoria cs operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ cy_as_hal_print_message(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ cy_as_hal_print_message(KERN_INFO"pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ cy_as_hal_print_message(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(void)
+{
+ int i, err;
+
+ cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init();
+ if (err < 0)
+ cy_as_hal_print_message(KERN_INFO"gpmc init failed:%d", err);
+
+ cy_as_hal_config_c_s_mux();
+
+ return gpmc_data_vma;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (gpmc_data_vma != 0)
+ iounmap((void *)gpmc_data_vma);
+
+ if (csa_phy != 0)
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_PNAND_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+/*
+ * GPMC NAND command+addr write phase
+ */
+static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
+{
+ /*
+ * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
+ */
+ u32 tmpa32 = ((row_addr << 16) | col_addr);
+ u8 RA2 = (u8)(row_addr >> 16);
+
+ if (!pnand_16bit) {
+ /*
+ * GPMC PNAND 8bit BUS
+ */
+ /*
+ * CMD1
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+
+ /*
+ *pnand bus: <CA0,CA1,RA0,RA1>
+ */
+ IOWR32(naddr_reg_vma, tmpa32);
+
+ /*
+ * <RA2> , always zero
+ */
+ IOWR8(naddr_reg_vma, RA2);
+
+ } else {
+ /*
+ * GPMC PNAND 16bit BUS , in 16 bit mode CMD
+ * and ADDR sent on [d7..d0]
+ */
+ uint8_t CA0, CA1, RA0, RA1;
+ CA0 = tmpa32 & 0x000000ff;
+ CA1 = (tmpa32 >> 8) & 0x000000ff;
+ RA0 = (tmpa32 >> 16) & 0x000000ff;
+ RA1 = (tmpa32 >> 24) & 0x000000ff;
+
+ /*
+ * can't use 32 bit writes here omap will not serialize
+ * them to lower half in16 bit mode
+ */
+
+ /*
+ *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+ IOWR8(naddr_reg_vma, CA0);
+ IOWR8(naddr_reg_vma, CA1);
+ IOWR8(naddr_reg_vma, RA0);
+ IOWR8(naddr_reg_vma, RA1);
+ IOWR8(naddr_reg_vma, RA2);
+ }
+}
+
+/*
+ * spin until r/b goes high
+ */
+inline int wait_rn_b_high(void)
+{
+ u32 w_spins = 0;
+
+ /*
+ * TODO: note R/b may go low here, need to spin until high
+ * while (omap_get_gpio_datain(AST_RnB) == 0) {
+ * w_spins++;
+ * }
+ * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ *
+ * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ * w_spins++;
+ * }
+ * printk("<1>RnB=0!:%d\n",w_spins);
+ * }
+ */
+ return w_spins;
+}
+
+#ifdef ENABLE_GPMC_PF_ENGINE
+/* #define PFE_READ_DEBUG
+ * PNAND block read with OMAP PFE enabled
+ * status: Not tested, NW, broken , etc
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint8_t bytes_in_fifo;
+
+ /* debug vars*/
+#ifdef PFE_READ_DEBUG
+ uint32_t loop_limit;
+ uint16_t bytes_read = 0;
+#endif
+
+ /*
+ * configure the prefetch engine
+ */
+ uint32_t tmp32;
+ uint32_t pfe_status;
+
+ /*
+ * DISABLE GPMC CS4 operation 1st, this is
+ * in case engine is be already disabled
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+
+#ifdef PFE_READ_DEBUG
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
+ if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
+ printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
+ tmp32, GPMC_PREFETCH_CONFIG1_VAL);
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
+ }
+
+ /*
+ *sanity check 2
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
+ if (tmp32 != (count))
+ printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
+ "!= VAL written:%d\n", tmp32, count);
+#endif
+
+ /*
+ * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
+ * on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * start the prefetch engine
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ while (1) {
+ /*
+ * GPMC PFE service loop
+ */
+ do {
+ /*
+ * spin until PFE fetched some
+ * PNAND bus words in the FIFO
+ */
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ bytes_in_fifo = (pfe_status >> 24) & 0x7f;
+ } while (bytes_in_fifo == 0);
+
+ /* whole 32 bit words in fifo */
+ w32cnt = bytes_in_fifo >> 2;
+
+#if 0
+ /*
+ *NOTE: FIFO_PTR indicates number of NAND bus words bytes
+ * already received in the FIFO and available to be read
+ * by DMA or MPU whether COUNTVAL indicates number of BUS
+ * words yet to be read from PNAND bus words
+ */
+ printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
+ pfe_status, bytes_in_fifo,
+ (pfe_status & 0x3fff), w32cnt);
+#endif
+
+ while (w32cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ if ((pfe_status & 0x3fff) == 0) {
+ /*
+ * PFE acc angine done, there still may be data leftover
+ * in the FIFO re-read FIFO BYTE counter (check for
+ * leftovers from 32 bit read accesses above)
+ */
+ bytes_in_fifo = (IORD32(
+ GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ /*
+ * NOTE we may still have one word left in the fifo
+ * read it out
+ */
+ ptr8 = ptr32;
+ switch (bytes_in_fifo) {
+
+ case 0:
+ /*
+ * nothing to do we already read the
+ * FIFO out with 32 bit accesses
+ */
+ break;
+ case 1:
+ /*
+ * this only possible
+ * for 8 bit pNAND only
+ */
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 2:
+ /*
+ * this one can occur in either modes
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ break;
+
+ case 3:
+ /*
+ * this only possible for 8 bit pNAND only
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ ptr8 += 2;
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 4:
+ /*
+ * shouldn't happen, but has been seen
+ * in 8 bit mode
+ */
+ *ptr32 = IORD32(gpmc_data_vma);
+ break;
+
+ default:
+ printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
+ bytes_in_fifo);
+ break;
+ }
+ /*
+ * read is completed, get out of the while(1) loop
+ */
+ break;
+ }
+ }
+}
+#endif
+
+#ifdef PFE_LBD_READ_V2
+/*
+ * PFE engine assisted reads with the 64 byte blocks
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint8_t rd_cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint16_t reminder;
+ uint32_t pfe_status;
+
+ /*
+ * ISSUE PNAND CMD+ADDR
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * setup PFE block
+ * count - OMAP number of bytes to access on pnand bus
+ */
+
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ do {
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
+ rd_cnt = pfe_status >> (24+2);
+
+ while (rd_cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ } while (pfe_status & 0x3fff);
+
+ /*
+ * read out the leftover
+ */
+ ptr8 = ptr32;
+ rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ while (rd_cnt--)
+ *ptr8++ = IORD8(gpmc_data_vma);
+}
+#endif
+
+#ifdef PNAND_LBD_READ_NO_PFE
+/*
+ * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
+ * the original working code, works at max speed for 8 bit xfers
+ * for 16 bit the bus diagram has gaps
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint16_t *ptr16;
+ uint16_t remainder;
+
+ DBGPRN("<1> %s(): NO_PFE\n", __func__);
+
+ ptr32 = buff;
+ /* number of whole 32 bit words in the transfer */
+ w32cnt = count >> 2;
+
+ /* remainder, in bytes(0..3) */
+ remainder = count & 03;
+
+ /*
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * read data by 32 bit chunks
+ */
+ while (w32cnt--)
+ *ptr32++ = IORD32(ndata_reg_vma);
+
+ /*
+ * now do the remainder(it can be 0, 1, 2 or 3)
+ * same code for both 8 & 16 bit bus
+ * do 1 or 2 MORE words
+ */
+ ptr16 = (uint16_t *)ptr32;
+
+ switch (remainder) {
+ case 1:
+ /* read one 16 bit word
+ * IN 8 BIT WE NEED TO READ even number of bytes
+ */
+ case 2:
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ *ptr16++ = IORD16(ndata_reg_vma);
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ default:
+ /*
+ * remainder is 0
+ */
+ break;
+ }
+}
+#endif
+
+/*
+ * uses LBD mode to write N bytes into astoria
+ * Status: Working, however there are 150ns idle
+ * timeafter every 2 (16 bit or 4(8 bit) bus cycles
+ */
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint16_t remainder;
+ uint8_t *ptr8;
+ uint16_t *ptr16;
+ uint32_t *ptr32;
+
+ remainder = count & 03;
+ w32cnt = count >> 2;
+ ptr32 = buff;
+ ptr8 = buff;
+
+ /*
+ * send: CMDB1, CA0,CA1,RA0,RA1,RA2
+ */
+ nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
+
+ /*
+ * blast the data out in 32bit chunks
+ */
+ while (w32cnt--)
+ IOWR32(ndata_reg_vma, *ptr32++);
+
+ /*
+ * do the reminder if there is one
+ * same handling for both 8 & 16 bit pnand: mode
+ */
+ ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
+
+ switch (remainder) {
+ case 1:
+ /*
+ * read one 16 bit word
+ */
+ case 2:
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ IOWR16(ndata_reg_vma, *ptr16++);
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+ default:
+ /*
+ * reminder is 0
+ */
+ break;
+ }
+ /*
+ * finally issue a PGM cmd
+ */
+ IOWR8(ncmd_reg_vma, PGMPAGE_B2);
+}
+
+/*
+ * write Astoria register
+ */
+static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
+{
+ unsigned long flags;
+ u16 addr16;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t rdreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&rdreg_usage_cnt) != 0) {
+ cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
+ "* cy_as_hal_write_register usage:%d\n",
+ atomic_read(&rdreg_usage_cnt));
+ }
+
+ atomic_inc(&rdreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ */
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+
+ /*
+ * this should be sent on the 16 bit bus
+ */
+ IOWR16(ndata_reg_vma, data);
+ } else {
+ /*
+ * 8 bit nand mode GPMC will automatically
+ * seriallize 16bit or 32 bit writes into
+ * 8 bit onesto the lower 8 bit in LE order
+ */
+ addr16 = 0x0c00 | reg_addr8;
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR16(ndata_reg_vma, data);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&rdreg_usage_cnt);
+ local_irq_restore(flags);
+}
+
+
+/*
+ * read astoria register via pNAND interface
+ */
+static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
+{
+ u16 data;
+ u16 addr16;
+ unsigned long flags;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t wrreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&wrreg_usage_cnt) != 0) {
+ /*
+ * if it gets here ( from other threads), this function needs
+ * need spin_lock_irq save() protection
+ */
+ cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
+ "cy_as_hal_write_register usage:%d\n",
+ atomic_read(&wrreg_usage_cnt));
+ }
+ atomic_inc(&wrreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ * CMD1, CA1, CA2,
+ */
+
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+ IOWR8(ncmd_reg_vma, 0x00E0);
+
+ udelay(1);
+
+ /*
+ * much faster through the gPMC Register space
+ */
+ data = IORD16(ndata_reg_vma);
+ } else {
+ /*
+ * 8 BIT gpmc NAND mode
+ * CMD1, CA1, CA2, CMD2
+ */
+ addr16 = 0x0c00 | reg_addr8;
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR8(ncmd_reg_vma, 0xE0);
+ udelay(1);
+ data = IORD16(ndata_reg_vma);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&wrreg_usage_cnt);
+ local_irq_restore(flags);
+
+ return data;
+}
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ ast_p_nand_casdi_write((u8)addr, data);
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+
+ /*
+ * READ ASTORIA REGISTER USING CASDO
+ */
+ data = ast_p_nand_casdo_read((u8)addr);
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+ /* this is not defined currently*/
+ #ifdef WESTBRIDGE_PNAND_HAL_PREP_CHANGE
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+ #endif
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, size;
+ void *dptr;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = end_points[ep].data_p;
+
+ DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
+ ep,
+ size,
+ intr_sequence_num,
+ dptr
+ );
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * the actual WB-->OMAP memory "soft" DMA xfer
+ */
+ p_nand_lbd_read(col_addr, row_addr, size, dptr);
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0;
+ uint32_t size;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ void *dptr;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size)
+ p_nand_lbd_write(col_addr, row_addr, size, dptr);
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if(state) {
+ cy_as_hal_gpmc_enable_16bit_bus(cy_false);
+ __gpio_set_value(AST_WAKEUP, 1);
+ } else {
+ __gpio_set_value(AST_WAKEUP, 0);
+ }
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ (void) tag;
+ (void) is_standby_wakeup;
+
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ u8 pncfg_reg;
+
+ /*
+ * No debug mode support through argument as of now
+ */
+ (void)debug;
+
+ DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ;
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /*
+ * allocate memory for OMAP HAL
+ */
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_PNAND_HAL_SIG;
+
+ /*
+ * initialize OMAP hardware and StartOMAPKernelall gpio pins
+ */
+ dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
+
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+ /*
+ * note AFTER reset PNAND interface is 8 bit mode
+ * so if gpmc Is configured in 8 bit mode upper half will be FF
+ */
+ pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+#ifdef PNAND_16BIT_MODE
+
+ /*
+ * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
+ */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+
+ /*
+ * now in order to continue to talk to astoria
+ * sw OMAP GPMC into 16 bit mode as well
+ */
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+#else
+ /* Astoria and GPMC are already in 8 bit mode, jsut initialize PNAND_CFG */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
+#endif
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
+ data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
+ printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
+ data16[0], data16[0]);
+ goto bus_acc_error;
+ }
+
+ cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "PNAND_CFG after RST:%4.4x\n "
+ "CY_AS_MEM_PNAND_CFG"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], pncfg_reg, data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 2010-08-27 17:50:08.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_cram.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+ /* Debug printing enabled */
+
+ #define DBGPRN(...) printk(__VA_ARGS__)
+ #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+ /** NO DEBUG PRINTING **/
+ #define DBGPRN(...)
+ #define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 2010-09-01 16:29:59.000000000 -0700
@@ -0,0 +1,565 @@
+/*
+ OMAP3430 ZOOM MDK astoria interface defs(cyasmemmap.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+#include <mach/mux.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+
+#ifndef _INCLUDED_CYASMEMMAP_H_
+#define _INCLUDED_CYASMEMMAP_H_
+
+/* defines copied from OMAP kernel branch */
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
+#define OMAP34XX_MUX_MODE0 0
+#define OMAP34XX_MUX_MODE4 4
+#define OMAP3_INPUT_EN (1 << 8)
+#define OMAP34XX_PIN_INPUT_PULLUP (OMAP2_PULL_ENA | OMAP3_INPUT_EN \
+ | OMAP2_PULL_UP)
+
+/*
+ * for OMAP3430 <-> astoria : ADmux mode, 8 bit data path
+ * WB Signal- OMAP3430 signal COMMENTS
+ * --------------------------- --------------------
+ * CS_L -GPMC_nCS4_GPIO_53 ZOOM I SOM board
+ * signal: up_nCS_A_EXT
+ * AD[7:0]-upD[7:0] buffered on the
+ * transposer board
+ * GPMC_ADDR
+ * [A8:A1]->upD[7:0]
+ * INT# -GPMC_nWP_GPIO_62
+ * DACK -N/C not conected
+ * WAKEUP-GPIO_167
+ * RESET-GPIO_126
+ * R/B -GPMC_WAIT2_GPIO_64
+ * -------------------------------------------
+ * The address range for nCS1B is 0x06000000 - 0x07FF FFFF.
+*/
+
+/*
+ *OMAP_ZOOM LEDS
+ */
+#define LED_0 156
+#define LED_1 128
+#define LED_2 64
+#define LED_3 60
+
+#define HIGH 1
+#define LOW 1
+
+/*
+ *omap GPIO number
+ */
+#define AST_WAKEUP 167
+#define AST_RESET 126
+#define AST__rn_b 64
+
+/*
+ * NOTE THIS PIN IS USED AS WP for OMAP NAND
+ */
+#define AST_INT 62
+
+/* N4_OMAP3430_GPMC_A1
+ M4_OMAP3430_GPMC_A2
+ L4_OMAP3430_GPMC_A3
+ K4_OMAP3430_GPMC_A4
+ T3_OMAP3430_GPMC_A5
+ R3_OMAP3430_GPMC_A6
+ N3_OMAP3430_GPMC_A7
+ M3_OMAP3430_GPMC_A8
+ * as an I/O, it is actually controlled by GPMC
+ */
+#define AST_CS 55
+
+
+/*
+ *GPMC prefetch engine
+ */
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG1 0x01E0
+
+ /*32 bytes for 16 bit pnand mode*/
+ #define PFE_THRESHOLD 31
+
+ /*
+ * bit fields
+ * PF_ACCESSMODE - 0 - read mode, 1 - write mode
+ * PF_DMAMODE - 0 - default only intr line signal will be generated
+ * PF_SYNCHROMODE - default 0 - engin will start access as soon as
+ * ctrl re STARTENGINE is set
+ * PF_WAITPINSEL - FOR synchro mode selects WAIT pin whch edge
+ * will be monitored
+ * PF_EN_ENGINE - 1- ENABLES ENGINE, but it needs to be started after
+ * that C ctrl reg bit 0
+ * PF_FIFO_THRESHOLD - FIFO threshhold in number of BUS(8 or 16) words
+ * PF_WEIGHTED_PRIO - NUM of cycles granted to PFE if RND_ROBIN
+ * prioritization is enabled
+ * PF_ROUND_ROBIN - if enabled, gives priority to other CS, but
+ * reserves NUM of cycles for PFE's turn
+ * PF_ENGIN_CS_SEL - GPMC CS assotiated with PFE function
+ */
+ #define PF_ACCESSMODE (0 << 0)
+ #define PF_DMAMODE (0 << 2)
+ #define PF_SYNCHROMODE (0 << 3)
+ #define PF_WAITPINSEL (0x0 << 4)
+ #define PF_EN_ENGINE (1 << 7)
+ #define PF_FIFO_THRESHOLD (PFE_THRESHOLD << 8)
+ #define PF_WEIGHTED_PRIO (0x0 << 16)
+ #define PF_ROUND_ROBIN (0 << 23)
+ #define PF_ENGIN_CS_SEL (AST_GPMC_CS << 24)
+ #define PF_EN_OPTIM_ACC (0 << 27)
+ #define PF_CYCLEOPTIM (0x0 << 28)
+
+#define GPMC_PREFETCH_CONFIG1_VAL (PF_ACCESSMODE | \
+ PF_DMAMODE | PF_SYNCHROMODE | \
+ PF_WAITPINSEL | PF_EN_ENGINE | \
+ PF_FIFO_THRESHOLD | PF_FIFO_THRESHOLD | \
+ PF_WEIGHTED_PRIO | PF_ROUND_ROBIN | \
+ PF_ENGIN_CS_SEL | PF_EN_OPTIM_ACC | \
+ PF_CYCLEOPTIM)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG2 0x01E4
+ /*
+ * bit fields
+ * 14 bit field NOTE this counts is also
+ * is in number of BUS(8 or 16) words
+ */
+ #define PF_TRANSFERCOUNT (0x000)
+
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONTROL 0x01EC
+ /*
+ * bit fields , ONLY BIT 0 is implemented
+ * PFWE engin must be programmed with this bit = 0
+ */
+ #define PFPW_STARTENGINE (1 << 0)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_STATUS 0x01F0
+
+ /* */
+ #define PFE_FIFO_THRESHOLD (1 << 16)
+
+/*
+ * GPMC posted write/prefetch engine end
+ */
+
+
+/*
+ * chip select number on GPMC ( 0..7 )
+ */
+#define AST_GPMC_CS 4
+
+/*
+ * not connected
+ */
+#define AST_DACK 00
+
+
+/*
+ * Physical address above the NAND flash
+ * we use CS For mapping in OMAP3430 RAM space use 0x0600 0000
+ */
+#define CYAS_DEV_BASE_ADDR (0x20000000)
+#define CYAS_DEV_MAX_ADDR (0xFF)
+#define CYAS_DEV_ADDR_RANGE (CYAS_DEV_MAX_ADDR << 1)
+
+ /* in CRAM or PSRAM mode OMAP A1..An wires-> Astoria, there is no A0 line */
+ #define CYAS_DEV_CALC_ADDR(cyas_addr) (cyas_addr << 1)
+ #define CYAS_DEV_CALC_EP_ADDR(ep) (ep << 1)
+
+/*
+ *OMAP3430 i/o access macros
+ */
+#define IORD32(addr) (*(volatile u32 *)(addr))
+#define IOWR32(addr, val) (*(volatile u32 *)(addr) = val)
+
+#define IORD16(addr) (*(volatile u16 *)(addr))
+#define IOWR16(addr, val) (*(volatile u16 *)(addr) = val)
+
+#define IORD8(addr) (*(volatile u8 *)(addr))
+#define IOWR8(addr, val) (*(volatile u8 *)(addr) = val)
+
+/*
+ * local defines for accessing to OMAP GPIO ***
+ */
+#define CTLPADCONF_BASE_ADDR 0x48002000
+#define CTLPADCONF_SIZE 0x1000
+
+#define GPIO1_BASE_ADDR 0x48310000
+#define GPIO2_BASE_ADDR 0x49050000
+#define GPIO3_BASE_ADDR 0x49052000
+#define GPIO4_BASE_ADDR 0x49054000
+#define GPIO5_BASE_ADDR 0x49056000
+#define GPIO6_BASE_ADDR 0x49058000
+#define GPIO_SPACE_SIZE 0x1000
+
+
+/*
+ * OMAP3430 GPMC timing for pNAND interface
+ */
+#define GPMC_BASE 0x6E000000
+#define GPMC_REGION_SIZE 0x1000
+#define GPMC_CONFIG_REG (0x50)
+
+/*
+ * bit 0 in the GPMC_CONFIG_REG
+ */
+#define NAND_FORCE_POSTED_WRITE_B 1
+
+/*
+ * WAIT2STATUS, must be (1 << 10)
+ */
+#define AS_WAIT_PIN_MASK (1 << 10)
+
+/*
+ * GPMC_CONFIG(reg number [1..7] [for chip sel CS[0..7])
+ */
+#define GPMC_CFG_REG(N, CS) ((0x60 + (4*(N-1))) + (0x30*CS))
+
+/*
+ * we get "gpmc_base" from kernel
+ */
+#define GPMC_VMA(offset) (gpmc_base + offset)
+
+/*
+ * GPMC CS space VMA start address
+ */
+#define GPMC_CS_VMA(offset) (gpmc_data_vma + offset)
+
+/*
+ * PAD_CFG mux space VMA
+ */
+#define PADCFG_VMA(offset) (iomux_vma + offset)
+
+/*
+ * CONFIG1: by default, sngle access, async r/w RD_MULTIPLE[30]
+ * WR_MULTIPLE[28]; GPMC_FCL_DIV[1:0]
+ */
+#define GPMC_FCLK_DIV ((0) << 0)
+
+/*
+ * ADDITIONAL DIVIDER FOR ALL TIMING PARAMS
+ */
+#define TIME_GRAN_SCALE ((0) << 4)
+
+/*
+ * for use by gpmc_set_timings api, measured in ns, not clocks
+ */
+#define WB_GPMC_BUSCYC_t (7 * 6)
+#define WB_GPMC_CS_t_on (0)
+#define WB_GPMC_ADV_t_on (1 * 6)
+#define WB_GPMC_ADV_t_off (3 * 6)
+#define WB_GPMC_OE_t_on (2 * 6)
+#define WB_GPMC_OE_t_off (6 * 6)
+#define WB_GPMC_WE_t_on (2 * 6)
+#define WB_GPMC_WE_t_off (6 * 6)
+#define WB_GPMC_RDS_ADJ (1 * 6)
+#define WB_GPMC_RD_t_a_c_c (WB_GPMC_OE_t_off + WB_GPMC_RDS_ADJ)
+#define WB_GPMC_WR_t_a_c_c (WB_GPMC_BUSCYC_t)
+
+#define DIR_OUT 0
+#define DIR_INP 1
+#define DRV_HI 1
+#define DRV_LO 0
+
+/*
+ * GPMC_CONFIG7[cs] register bit fields3
+ * AS_CS_MASK - 3 bit mask for A26,A25,A24,
+ * AS_CS_BADDR - 6 BIT VALUE A29 ...A24
+ * CSVALID_B - CSVALID bit on GPMC_CONFIG7[cs] register
+ */
+#define AS_CS_MASK (0X7 << 8)
+#define AS_CS_BADDR 0x02
+#define CSVALID_B (1 << 6)
+
+/*
+ * DEFINE OMAP34XX GPIO OFFSETS (should have been defined in kernel /arch
+ * these are offsets from the BASE_ADDRESS of the GPIO BLOCK
+ */
+#define GPIO_REVISION 0x000
+#define GPIO_SYSCONFIG 0x010
+#define GPIO_SYSSTATUS1 0x014
+#define GPIO_IRQSTATUS1 0x0183
+#define GPIO_IRQENABLE1 0x01C
+#define GPIO_IRQSTATUS2 0x028
+#define GPIO_CTRL 0x030
+#define GPIO_OE 0x034
+#define GPIO_DATA_IN 0x038
+#define GPIO_DATA_OUT 0x03C
+#define GPIO_LEVELDETECT0 0x040
+#define GPIO_LEVELDETECT1 0x044
+#define GPIO_RISINGDETECT 0x048
+#define GPIO_FALLINGDETECT 0x04c
+#define GPIO_CLEAR_DATAOUT 0x090
+#define GPIO_SET_DATAOUT 0x094
+
+typedef struct {
+ char *name;
+ u32 phy_addr;
+ u32 virt_addr;
+ u32 size;
+} io2vma_tab_t;
+
+/*
+ * GPIO phy to translation VMA table
+ */
+static io2vma_tab_t gpio_vma_tab[6] = {
+ {"GPIO1_BASE_ADDR", GPIO1_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO2_BASE_ADDR", GPIO2_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO3_BASE_ADDR", GPIO3_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO4_BASE_ADDR", GPIO4_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO5_BASE_ADDR", GPIO5_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO6_BASE_ADDR", GPIO6_BASE_ADDR , 0 , GPIO_SPACE_SIZE}
+};
+/*
+ * name - USER signal name assigned to the pin ( for printks)
+ * mux_func - enum index NAME for the pad_cfg function
+ * pin_num - pin_number if mux_func is GPIO, if not a GPIO it is -1
+ * mux_ptr - pointer to the corresponding pad_cfg_reg
+ * (used for pad release )
+ * mux_save - preserve here original PAD_CNF value for this
+ * pin (used for pad release)
+ * dir - if GPIO: 0 - OUT , 1 - IN
+ * dir_save - save original pin direction
+ * drv - initial drive level "0" or "1"
+ * drv_save - save original pin drive level
+ * valid - 1 if successfuly configured
+*/
+typedef struct {
+ char *name;
+ u32 mux_func;
+ int pin_num;
+ u16 *mux_ptr;
+ u16 mux_save;
+ u8 dir;
+ u8 dir_save;
+ u8 drv;
+ u8 drv_save;
+ u8 valid;
+} user_pad_cfg_t;
+
+/*
+ * need to ensure that enums are in sync with the
+ * omap_mux_pin_cfg table, these enums designate
+ * functions that OMAP pads can be configured to
+ */
+enum {
+ B23_OMAP3430_GPIO_167,
+ D23_OMAP3430_GPIO_126,
+ H1_OMAP3430_GPIO_62,
+ H1_OMAP3430_GPMC_n_w_p,
+ T8_OMAP3430_GPMC_n_c_s4,
+ T8_OMAP3430_GPIO_55,
+ R25_OMAP3430_GPIO_156,
+ R27_OMAP3430_GPIO_128,
+ K8_OMAP3430_GPIO_64,
+ K8_GPMC_WAIT2,
+ G3_OMAP3430_GPIO_60,
+ G3_OMAP3430_n_b_e0_CLE,
+ C6_GPMC_WAIT3,
+ J1_OMAP3430_GPIO_61,
+ C6_OMAP3430_GPIO_65,
+ N4_OMAP3430_GPMC_A1,
+ M4_OMAP3430_GPMC_A2,
+ L4_OMAP3430_GPMC_A3,
+ K4_OMAP3430_GPMC_A4,
+ T3_OMAP3430_GPMC_A5,
+ R3_OMAP3430_GPMC_A6,
+ N3_OMAP3430_GPMC_A7,
+ M3_OMAP3430_GPMC_A8,
+ END_OF_TABLE
+};
+
+/*
+ * number of GPIOS we plan to grab
+ */
+#define GPIO_SLOTS 8
+
+/*
+ * user_pads_init() reads(and saves) from/to this table
+ * used in conjunction with omap_3430_mux_t table in .h file
+ * because the way it's done in the kernel code
+ * TODO: implement restore of the the original cfg and i/o regs
+ */
+
+static user_pad_cfg_t user_pad_cfg[] = {
+ /*
+ * name,pad_func,pin_num, mux_ptr, mux_sav, dir,
+ * dir_sav, drv, drv_save, valid
+ */
+ {"AST_WAKEUP", B23_OMAP3430_GPIO_167, 167, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST_RESET", D23_OMAP3430_GPIO_126, 126, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST__rn_b", K8_GPMC_WAIT2, 64, NULL, 0,
+ DIR_INP, 0, 0, 0, 0},
+ {"AST_INTR", H1_OMAP3430_GPIO_62, 62, NULL, 0,
+ DIR_INP, 0, DRV_HI, 0, 0},
+ {"AST_CS", T8_OMAP3430_GPMC_n_c_s4, 55, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"LED_0", R25_OMAP3430_GPIO_156, 156, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"LED_1", R27_OMAP3430_GPIO_128, 128, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"AST_CLE", G3_OMAP3430_n_b_e0_CLE , 60, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ /*
+ * Z terminator, must always be present
+ * for sanity check, don't remove
+ */
+ {NULL}
+};
+
+#define GPIO_BANK(pin) (pin >> 5)
+#define REG_WIDTH 32
+#define GPIO_REG_VMA(pin_num, offset) \
+ (gpio_vma_tab[GPIO_BANK(pin_num)].virt_addr + offset)
+
+/*
+ * OMAP GPIO_REG 32 BIT MASK for a bit or
+ * flag in gpio_No[0..191] apply it to a 32 b3it
+ * location to set clear or check on a corresponding
+ * gpio bit or flag
+ */
+#define GPIO_REG_MASK(pin_num) (1 << \
+ (pin_num - (GPIO_BANK(pin_num) * REG_WIDTH)))
+
+/*
+ * OMAP GPIO registers bitwise access macros
+ */
+
+#define OMAP_GPIO_BIT(pin_num, reg) \
+ ((*((u32 *)GPIO_REG_VMA(pin_num, reg)) \
+ & GPIO_REG_MASK(pin_num)) ? 1 : 0)
+
+#define RD_OMAP_GPIO_BIT(pin_num, v) OMAP_GPIO_BIT(pin_num, reg)
+
+/*
+ *these are superfast set/clr bitbang macro, 48ns cyc tyme
+ */
+#define OMAP_SET_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_SET_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+#define OMAP_CLR_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_CLEAR_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+
+#define WR_OMAP_GPIO_BIT(pin_num, v) \
+ (v ? (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_SET_DATAOUT) = GPIO_REG_MASK(pin_num)) \
+ : (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_CLEAR_DATAOUT) = GPIO_REG_MASK(pin_num)))
+
+/*
+ * Note this pin cfg mimicks similar implementation
+ * in linux kernel, which unfortunately doesn't allow
+ * us to dynamically insert new custom GPIO mux
+ * configurations all REG definitions used in this
+ * applications. to add a new pad_cfg function, insert
+ * a new ENUM and new pin_cfg entry in omap_mux_pin_cfg[]
+ * table below3
+ *
+ * offset - note this is a word offset since the
+ * SCM regs are 16 bit packed in one 32 bit word
+ * mux_val - just enough to describe pins used
+ */
+typedef struct {
+ char *name;
+ u16 offset;
+ u16 mux_val;
+} omap_3430_mux_t;
+
+/*
+ * "OUTIN" is configuration when DATA reg drives the
+ * pin but the level at the pin can be sensed
+ */
+#define PAD_AS_OUTIN (OMAP34XX_MUX_MODE4 | \
+ OMAP34XX_PIN_OUTPUT | OMAP34XX_PIN_INPUT)
+
+omap_3430_mux_t omap_mux_pin_cfg[] = {
+ /*
+ * B23_OMAP3430_GPIO_167 - GPIO func to PA3D 167 WB wakeup
+ * D23_OMAP3430_GPIO_126 - drive GPIO_126 ( AST RESET)
+ * H1_OMAP3430_GPIO_62 - need a pullup on this pin
+ * H1_OMAP3430_GPMC_n_w_p - GPMC NAND CTRL n_w_p out
+ * T8_OMAP3430_GPMC_n_c_s4" - T8 is controlled b_y GPMC NAND ctrl
+ * R25_OMAP3430_GPIO_156 - OMAPZOOM drive LED_0
+ * R27_OMAP3430_GPIO_128 - OMAPZOOM drive LED_1
+ * K8_OMAP3430_GPIO_64 - OMAPZOOM drive LED_2
+ * K8_GPMC_WAIT2 - GPMC WAIT2 function on PAD K8
+ * G3_OMAP3430_GPIO_60 - OMAPZOOM drive LED_3
+ * G3_OMAP3430_n_b_e0_CLE -GPMC NAND ctrl CLE signal
+ */
+
+ {"B23_OMAP3430_GPIO_167", 0x0130, (OMAP34XX_MUX_MODE4)},
+ {"D23_OMAP3430_GPIO_126", 0x0132, (OMAP34XX_MUX_MODE4)},
+ {"H1_OMAP3430_GPIO_62", 0x00CA, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP) },
+ {"H1_OMAP3430_GPMC_n_w_p", 0x00CA, (OMAP34XX_MUX_MODE0)},
+ {"T8_OMAP3430_GPMC_n_c_s4", 0x00B6, (OMAP34XX_MUX_MODE0) },
+ {"T8_OMAP3430_GPIO_55", 0x00B6, (OMAP34XX_MUX_MODE4) },
+ {"R25_OMAP3430_GPIO_156", 0x018C, (OMAP34XX_MUX_MODE4) },
+ {"R27_OMAP3430_GPIO_128", 0x0154, (OMAP34XX_MUX_MODE4) },
+ {"K8_OMAP3430_GPIO_64", 0x00d0, (OMAP34XX_MUX_MODE4) },
+ {"K8_GPMC_WAIT2", 0x00d0, (OMAP34XX_MUX_MODE0) },
+ {"G3_OMAP3430_GPIO_60", 0x00C6, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"G3_OMAP3430_n_b_e0_CLE", 0x00C6, (OMAP34XX_MUX_MODE0)},
+ {"C6_GPMC_WAIT3", 0x00d2, (OMAP34XX_MUX_MODE0)},
+ {"C6_OMAP3430_GPIO_65", 0x00d2, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"J1_OMAP3430_GPIO_61", 0x00C8, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP)},
+
+ {"N4_OMAP3430_GPMC_A1", 0x007A, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M4_OMAP3430_GPMC_A2", 0x007C, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"L4_OMAP3430_GPMC_A3", 0x007E, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"K4_OMAP3430_GPMC_A4", 0x0080, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"T3_OMAP3430_GPMC_A5", 0x0082, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"R3_OMAP3430_GPMC_A6", 0x0084, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"N3_OMAP3430_GPMC_A7", 0x0086, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M3_OMAP3430_GPMC_A8", 0x0088, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ /*
+ * don't remove, used for sanity check.
+ */
+ {"END_OF_TABLE"}
+};
+
+
+#endif /* _INCLUDED_CYASMEMMAP_H_ */
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 2010-08-26 10:57:38.000000000 -0700
@@ -0,0 +1,73 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_CRAM_HAL_SIG (0x2332)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_phy_addr_base;
+
+ void *m_vma_addr_base;
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,319 +0,0 @@
-/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-/*
- * This file contains the defintion of the hardware abstraction
- * layer on OMAP3430 talking to the West Bridge Astoria device
- */
-
-
-#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
-#define _INCLUDED_CYASHALOMAP_KERNEL_H_
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/wait.h>
-#include <linux/string.h>
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-typedef struct cy_as_hal_sleep_channel_t {
- wait_queue_head_t wq ;
-} cy_as_hal_sleep_channel ;
-
-/* moved to staging location, eventual location
- * considered is here
-#include <mach/westbridge/cyashaldef.h>
-#include <linux/westbridge/cyastypes.h>
-#include <linux/westbridge/cyas_cplus_start.h>
-*/
-#include "../cyashaldef.h"
-#include "../../../../../../../include/linux/westbridge/cyastypes.h"
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#include "cyasomapdev_kernel.h"
-
-/*
- * Below are the data structures that must be defined by the HAL layer
- */
-
-/*
- * The HAL layer must define a TAG for identifying a specific Astoria
- * device in the system. In this case the tag is a void * which is
- * really an OMAP device pointer
- */
-typedef void *cy_as_hal_device_tag ;
-
-
-/* This must be included after the CyAsHalDeviceTag type is defined */
-
-/* moved to staging location, eventual location
- * considered is here
- * #include <linux/westbridge/cyashalcb.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
-/*
- * Below are the functions that communicate with the West Bridge
- * device. These are system dependent and must be defined by
- * the HAL layer for a given system.
- */
-
-/*
- * This function must be defined to write a register within the Antioch
- * device. The addr value is the address of the register to write with
- * respect to the base address of the Antioch device.
- */
-void
-cy_as_hal_write_register(cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data) ;
-
-/*
- * This function must be defined to read a register from
- * the west bridge device. The addr value is the address of
- * the register to read with respect to the base address
- * of the west bridge device.
- */
-uint16_t
-cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
-
-/*
- * This function must be defined to transfer a block of data
- * to the west bridge device. This function can use the burst write
- * (DMA) capabilities of Antioch to do this, or it can just copy
- * the data using writes.
- */
-void
-cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to transfer a block of data
- * from the Antioch device. This function can use the burst
- * read (DMA) capabilities of Antioch to do this, or it can
- * just copy the data using reads.
- */
-void
-cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
- void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to cancel any pending DMA request.
- */
-void
-cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
-
-/*
- * This function must be defined to allow the Antioch API to
- * register a callback function that is called when a DMA transfer
- * is complete.
- */
-void
-cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb) ;
-
-/*
- * This function must be defined to return the maximum size of DMA
- * request that can be handled on the given endpoint. The return
- * value should be the maximum size in bytes that the DMA module can
- * handle.
- */
-uint32_t
-cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep) ;
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the Antioch device. Generally this is done via a GPIO of some
- * type.
- */
-cy_bool
-cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
-
-/*
- * This function is called when the Antioch PLL loses lock, because
- * of a problem in the supply voltage or the input clock.
- */
-void
-cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
-
-
-/**********************************************************************
- *
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- *
-***********************************************************************/
-
-/*
- * This function is required by the API to allocate memory. This function
- * is expected to work exactly like malloc().
- */
-void *
-cy_as_hal_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsHalAlloc(). This function is expected to work exacly like free().
- */
-void
-cy_as_hal_free(void *mem_p) ;
-
-/*
- * This function is required by the API to allocate memory during a
- * callback. This function must be able to provide storage at inturupt
- * time.
- */
-void *
-cy_as_hal_c_b_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsCBHalAlloc().
- */
-void
-cy_as_hal_c_b_free(void *ptr) ;
-
-/*
- * This function is required to set a block of memory to a specific
- * value. This function is expected to work exactly like memset()
- */
-void
-cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
-
-/*
- * This function is expected to create a sleep channel. The data
- * structure that represents the sleep channel is given by the
- * pointer in the argument.
- */
-cy_bool
-cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-/*
- * This function is expected to destroy a sleep channel. The data
- * structure that represents the sleep channel is given by
- * the pointer in the argument.
- */
-
-
-cy_bool
-cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-cy_bool
-cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
-
-cy_bool
-cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
-
-uint32_t
-cy_as_hal_disable_interrupts(void) ;
-
-void
-cy_as_hal_enable_interrupts(uint32_t);
-
-void
-cy_as_hal_sleep150(void);
-
-void
-cy_as_hal_sleep(uint32_t ms);
-
-cy_bool
-cy_as_hal_is_polling(void);
-
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup);
-
-/*
- * required only in spi mode
- */
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
-
-
-#ifndef NDEBUG
-#define cy_as_hal_assert(cond) if (!(cond))\
- printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
-#else
-#define cy_as_hal_assert(cond)
-#endif
-
-#define cy_as_hal_print_message printk
-
-/* removable debug printks */
-#ifndef WESTBRIDGE_NDEBUG
-#define DBG_PRINT_ENABLED
-#endif
-
-/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
-
-
-#ifdef DBG_PRINT_ENABLED
- /* Debug printing enabled */
-
- #define DBGPRN(...) printk(__VA_ARGS__)
- #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
- current->pid, __func__)
-
-#else
- /** NO DEBUG PRINTING **/
- #define DBGPRN(...)
- #define DBGPRN_FUNC_NAME
-
-#endif
-
-/*
-CyAsMiscSetLogLevel(uint8_t level)
-{
- debug_level = level ;
-}
-
-#ifdef CY_AS_LOG_SUPPORT
-
-void
-cy_as_log_debug_message(int level, const char *str)
-{
- if (level <= debug_level)
- cy_as_hal_print_message("log %d: %s\n", level, str) ;
-}
-*/
-
-
-/*
- * print buffer helper
- */
-void cyashal_prn_buf(void *buf, uint16_t offset, int len);
-
-/*
- * These are the functions that are not part of the HAL layer,
- * but are required to be called for this HAL.
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug) ;
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag) ;
-int omap_start_intr(cy_as_hal_device_tag tag) ;
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
-
-/* moved to staging location
-#include <linux/westbridge/cyas_cplus_end.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 2010-08-27 17:49:56.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_pnand.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+ /* Debug printing enabled */
+
+ #define DBGPRN(...) printk(__VA_ARGS__)
+ #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+ /** NO DEBUG PRINTING **/
+ #define DBGPRN(...)
+ #define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-09-01 16:04:25.000000000 -0700
@@ -31,6 +31,9 @@
#define _INCLUDED_CYASMEMMAP_H_
/* defines copied from OMAP kernel branch */
+
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
#define OMAP34XX_MUX_MODE0 0
#define OMAP34XX_MUX_MODE4 4
#define OMAP3_INPUT_EN (1 << 8)
@@ -250,9 +253,9 @@
/*
*gpmc nand registers for CS4
*/
-#define GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
#define GPMC_STAT_REG (0x54)
#define GPMC_ERR_TYPE (0x48)
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,72 +0,0 @@
-/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
-#define __CY_AS_OMAP_DEV_KERNEL_H__
-
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/completion.h>
-
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-
-/*
- * Constants
- */
-#define CY_AS_OMAP_KERNEL_HAL_SIG (0x1441)
-
-
-/*
- * Data structures
- */
-typedef struct cy_as_omap_dev_kernel {
- /* This is the signature for this data structure */
- unsigned int m_sig ;
-
- /* Address base of Antioch Device */
- void *m_addr_base;
-
- /* This is a pointer to the next Antioch device in the system */
- struct cy_as_omap_dev_kernel *m_next_p;
-
- /* This is for thread sync */
- struct completion thread_complete;
-
- /* This is for thread to wait for interrupts */
- cy_as_hal_sleep_channel thread_sc;
-
- /* This is for thread to exit upon StopOmapKernel */
- int thread_flag; /* set 1 to exit */
-
- int dma_ch;
-
- /* This is for dma sync */
- struct completion dma_complete;
-} cy_as_omap_dev_kernel;
-
-#endif
-
-/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 2010-08-20 14:36:23.000000000 -0700
@@ -0,0 +1,72 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_PNAND_HAL_SIG (0x1441)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_addr_base;
+
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-25 18:29:44.000000000 -0700
@@ -389,7 +389,7 @@ int cyasblkdev_media_changed(struct gend
struct cyasblkdev_blk_data *bd;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("cyasblkdev_media_changed() is called\n");
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev_media_changed() is called\n");
#endif
if (gd)
@@ -466,7 +466,7 @@ static int cyasblkdev_blk_prep_rq(
/* Check for excessive requests.*/
if (blk_rq_pos(req) + blk_rq_sectors(req) > get_capacity(req->rq_disk)) {
- cy_as_hal_print_message("cyasblkdev: bad request address\n");
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev: bad request address\n");
stat = BLKPREP_KILL;
}
@@ -493,7 +493,7 @@ static void cyasblkdev_issuecallback(
{
int retry_cnt = 0;
DBGPRN_FUNC_NAME;
-
+
if (status != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
@@ -509,7 +509,10 @@ static void cyasblkdev_issuecallback(
__func__, (unsigned int) gl_bd->queue.req, status,
(unsigned int) blk_rq_sectors(gl_bd->queue.req)) ;
#endif
-
+
+ if(rq_data_dir(gl_bd->queue.req) != READ) {
+ cy_as_release_common_lock();
+ }
/* note: blk_end_request w/o __ prefix should
* not require spinlocks on the queue*/
while (blk_end_request(gl_bd->queue.req,
@@ -533,7 +536,7 @@ static void cyasblkdev_issuecallback(
/* queue is not plugged */
gl_bd->queue.req = blk_fetch_request(gl_bd->queue.queue);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s blkdev_callback: "
+ cy_as_hal_print_message(KERN_INFO"%s blkdev_callback: "
"blk_fetch_request():%p\n",
__func__, gl_bd->queue.req);
#endif
@@ -543,7 +546,7 @@ static void cyasblkdev_issuecallback(
spin_unlock_irq(&gl_bd->lock);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s blkdev_callback: about to "
+ cy_as_hal_print_message(KERN_INFO"%s blkdev_callback: about to "
"call issue_fn:%p\n", __func__, gl_bd->queue.req);
#endif
@@ -585,7 +588,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->user_disk_0_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to disk 0 "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to disk 0 "
"for sector=%d, num_sectors=%d, unit_no=%d\n",
__func__, req_sector, (int) blk_rq_sectors(req),
lcl_unit_no);
@@ -598,7 +601,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->user_disk_1_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to disk 1 for "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to disk 1 for "
"sector=%d, num_sectors=%d, unit_no=%d\n", __func__,
req_sector, (int) blk_rq_sectors(req), lcl_unit_no);
#endif
@@ -609,7 +612,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->system_disk_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to system disk "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to system disk "
"for sector=%d, num_sectors=%d, unit_no=%d\n", __func__,
req_sector, (int) blk_rq_sectors(req), lcl_unit_no);
#endif
@@ -625,7 +628,7 @@ static int cyasblkdev_blk_issue_rq(
if (rq_data_dir(req) == READ) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: calling readasync() "
+ cy_as_hal_print_message(KERN_INFO"%s: calling readasync() "
"req_sector=0x%x, req_nr_sectors=0x%x, bd->sg:%x\n\n",
__func__, req_sector, req_nr_sectors, (uint32_t)bd->sg);
#endif
@@ -636,10 +639,10 @@ static int cyasblkdev_blk_issue_rq(
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:readasync() error %d at "
+ cy_as_hal_print_message(KERN_INFO"%s:readasync() error %d at "
"address %ld, unit no %d\n", __func__, ret,
blk_rq_pos(req), lcl_unit_no);
- cy_as_hal_print_message("%s:ending i/o request "
+ cy_as_hal_print_message(KERN_INFO"%s:ending i/o request "
"on reg:%x\n", __func__, (uint32_t)req);
#endif
@@ -651,13 +654,14 @@ static int cyasblkdev_blk_issue_rq(
bq->req = NULL ;
}
} else {
+ cy_as_acquire_common_lock();
ret = cy_as_storage_write_async(bd->dev_handle, bus_num, 0,
lcl_unit_no, req_sector, bd->sg, req_nr_sectors,
(cy_as_storage_callback)cyasblkdev_issuecallback);
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: write failed with "
+ cy_as_hal_print_message(KERN_INFO"%s: write failed with "
"error %d at address %ld, unit no %d\n",
__func__, ret, blk_rq_pos(req), lcl_unit_no);
#endif
@@ -690,7 +694,7 @@ static void cyasblkdev_storage_callback(
)
{
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: bus:%d, device:%d, evtype:%d, "
+ cy_as_hal_print_message(KERN_INFO"%s: bus:%d, device:%d, evtype:%d, "
"evdata:%p\n ", __func__, bus, device, evtype, evdata);
#endif
@@ -746,7 +750,7 @@ uint32_t cyasblkdev_get_vfat_offset(int
for (sect_no = 0; sect_no < SECTORS_TO_SCAN; sect_no++) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s before cyasstorageread "
+ cy_as_hal_print_message(KERN_INFO"%s before cyasstorageread "
"gl_bd->sg addr=0x%x\n", __func__,
(unsigned int) gl_bd->sg);
#endif
@@ -787,8 +791,8 @@ uint32_t cyasblkdev_get_vfat_offset(int
if (stat != 0) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s sector scan error\n",
- __func__);
+ cy_as_hal_print_message(KERN_INFO"%s sector scan error: %d\n",
+ __func__, stat);
#endif
break;
}
@@ -821,7 +825,7 @@ static int cyasblkdev_add_disks(int bus_
cy_as_storage_query_unit_data unit_data = {0} ;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:query device: "
+ cy_as_hal_print_message(KERN_INFO"%s:query device: "
"type:%d, removable:%d, writable:%d, "
"blksize %d, units:%d, locked:%d, "
"erase_sz:%d\n",
@@ -846,7 +850,7 @@ static int cyasblkdev_add_disks(int bus_
bd->dev_handle, bus_num, 0, 0, 0) ;
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s cannot release"
+ cy_as_hal_print_message(KERN_INFO"%s cannot release"
" storage\n", __func__) ;
#endif
goto out;
@@ -861,7 +865,7 @@ static int cyasblkdev_add_disks(int bus_
&unit_data, 0, 0) ;
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot query "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot query "
"%d device unit - reason code %d\n",
__func__, bus_num, ret) ;
#endif
@@ -876,7 +880,7 @@ static int cyasblkdev_add_disks(int bus_
if ((ret != CY_AS_ERROR_SUCCESS) &&
(ret != CY_AS_ERROR_ALREADY_PARTITIONED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cy_as_storage_"
+ cy_as_hal_print_message(KERN_INFO"%s: cy_as_storage_"
"create_p_partition after size > 0 check "
"failed with error code %d\n",
__func__, ret);
@@ -976,7 +980,7 @@ static int cyasblkdev_add_disks(int bus_
}
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: partition "
+ cy_as_hal_print_message(KERN_INFO"%s: partition "
"exists and sizes equal\n",
__func__);
#endif
@@ -1054,13 +1058,13 @@ static int cyasblkdev_add_disks(int bus_
#endif
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: setting gendisk disk "
+ cy_as_hal_print_message(KERN_INFO"%s: setting gendisk disk "
"capacity to %d\n", __func__, (int) disk_cap);
#endif
/* initializing bd->queue */
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: init bd->queue\n",
+ cy_as_hal_print_message(KERN_INFO"%s: init bd->queue\n",
__func__);
#endif
@@ -1111,7 +1115,8 @@ static int cyasblkdev_add_disks(int bus_
* public partition beginning */
if (vfat_search) {
bd->user_disk_0_first_sector =
- cyasblkdev_get_vfat_offset(0,
+ cyasblkdev_get_vfat_offset(
+ bd->user_disk_0_bus_num,
bd->user_disk_0_unit_no);
} else {
bd->user_disk_0_first_sector = 0;
@@ -1244,7 +1249,8 @@ static int cyasblkdev_add_disks(int bus_
if (vfat_search) {
bd->user_disk_1_first_sector =
cyasblkdev_get_vfat_offset(
- 1, bd->user_disk_1_unit_no);
+ bd->user_disk_1_bus_num,
+ bd->user_disk_1_unit_no);
} else {
bd->user_disk_1_first_sector
= 0;
@@ -1342,7 +1348,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s west bridge device handle:%x\n",
+ cy_as_hal_print_message(KERN_INFO"%s west bridge device handle:%x\n",
__func__, (uint32_t)bd->dev_handle);
#endif
@@ -1356,7 +1362,7 @@ static struct cyasblkdev_blk_data *cyasb
if ((stat != CY_AS_ERROR_SUCCESS) &&
(stat != CY_AS_ERROR_RESOURCE_NOT_OWNED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot release "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot release "
"resource bus 0 - reason code %d\n",
__func__, stat) ;
#endif
@@ -1366,7 +1372,7 @@ static struct cyasblkdev_blk_data *cyasb
if ((stat != CY_AS_ERROR_SUCCESS) &&
(stat != CY_AS_ERROR_RESOURCE_NOT_OWNED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot release "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot release "
"resource bus 0 - reason code %d\n",
__func__, stat) ;
#endif
@@ -1376,14 +1382,14 @@ static struct cyasblkdev_blk_data *cyasb
stat = cy_as_storage_start(bd->dev_handle, 0, 0x101) ;
if (stat != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot start storage "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot start storage "
"stack - reason code %d\n", __func__, stat) ;
#endif
goto out;
}
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: storage started:%d ok\n",
+ cy_as_hal_print_message(KERN_INFO"%s: storage started:%d ok\n",
__func__, stat);
#endif
@@ -1391,7 +1397,7 @@ static struct cyasblkdev_blk_data *cyasb
cyasblkdev_storage_callback);
if (stat != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot register callback "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot register callback "
"- reason code %d\n", __func__, stat) ;
#endif
goto out;
@@ -1405,7 +1411,7 @@ static struct cyasblkdev_blk_data *cyasb
bd->media_count[bus_num];
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot query %d, "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot query %d, "
"reason code: %d\n",
__func__, bus_num, stat) ;
#endif
@@ -1433,7 +1439,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else {
- cy_as_hal_print_message("%s: no available "
+ cy_as_hal_print_message(KERN_INFO"%s: no available "
"gen_disk for disk 0, "
"physically inconsistent\n", __func__);
}
@@ -1453,7 +1459,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else {
- cy_as_hal_print_message("%s: no available "
+ cy_as_hal_print_message(KERN_INFO"%s: no available "
"gen_disk for media, "
"physically inconsistent\n", __func__);
}
@@ -1461,13 +1467,13 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else if (total_media_count > 2) {
- cy_as_hal_print_message("%s: count corrupted = 0x%d\n",
+ cy_as_hal_print_message(KERN_INFO"%s: count corrupted = 0x%d\n",
__func__, total_media_count);
}
#endif
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: %d device(s) found\n",
+ cy_as_hal_print_message(KERN_INFO"%s: %d device(s) found\n",
__func__, total_media_count) ;
#endif
@@ -1476,7 +1482,7 @@ static struct cyasblkdev_blk_data *cyasb
stat = cy_as_storage_claim(bd->dev_handle,
bus_num, 0, 0, 0) ;
if (stat != CY_AS_ERROR_SUCCESS) {
- cy_as_hal_print_message("%s: cannot claim "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot claim "
"%d bus - reason code %d\n",
__func__, bus_num, stat) ;
goto out;
@@ -1570,7 +1576,7 @@ static int __init cyasblkdev_blk_init(vo
return 0;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("cyasblkdev init error:%d\n", res);
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev init error:%d\n", res);
#endif
return res;
}
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c 2010-09-01 15:56:46.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c 2010-09-01 15:57:26.000000000 -0700
@@ -95,9 +95,9 @@ static int cyasblkdev_prep_request(
DBGPRN_FUNC_NAME;
/* we only like normal block requests.*/
- if (!blk_fs_request(req)) {
+ if (req->cmd_type != REQ_TYPE_FS && !(req->cmd_flags & REQ_DISCARD)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:%x bad request received\n",
+ cy_as_hal_print_message(KERN_INFO"%s:%x bad request received\n",
__func__, current->pid) ;
#endif
@@ -192,7 +192,7 @@ static int cyasblkdev_queue_thread(void
"thread_sem->count=%d\n",
__func__, bq->thread_sem.count);
if (spin_is_locked(q->queue_lock)) {
- cy_as_hal_print_message("%s: queue_lock "
+ cy_as_hal_print_message(KERN_INFO"%s: queue_lock "
"is locked, need to release\n", __func__);
spin_unlock(q->queue_lock);
@@ -249,7 +249,7 @@ static int cyasblkdev_queue_thread(void
complete_and_exit(&bq->thread_complete, 0);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: is finished\n", __func__) ;
+ cy_as_hal_print_message(KERN_INFO"%s: is finished\n", __func__) ;
#endif
return 0;
@@ -274,7 +274,7 @@ static void cyasblkdev_request(struct re
if (!bq->req) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s wake_up(&bq->thread_wq)\n",
+ cy_as_hal_print_message(KERN_INFO"%s wake_up(&bq->thread_wq)\n",
__func__);
#endif
@@ -282,7 +282,7 @@ static void cyasblkdev_request(struct re
wake_up(&bq->thread_wq);
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: don't wake Q_thr, bq->req:%x\n",
+ cy_as_hal_print_message(KERN_INFO"%s: don't wake Q_thr, bq->req:%x\n",
__func__, (uint32_t)bq->req);
#endif
}
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h 2010-08-23 16:45:15.000000000 -0700
@@ -54,6 +54,8 @@ extern int cyasblkdev_init_queue(struct
extern void cyasblkdev_cleanup_queue(struct cyasblkdev_queue *);
extern void cyasblkdev_queue_suspend(struct cyasblkdev_queue *);
extern void cyasblkdev_queue_resume(struct cyasblkdev_queue *);
+extern void cy_as_acquire_common_lock(void);
+extern void cy_as_release_common_lock(void);
extern cy_as_device_handle cyasdevice_getdevhandle(void) ;
#define MOD_LOGS 1
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/device/cyasdevice.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/cyasdevice.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/device/cyasdevice.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/cyasdevice.c 2010-08-27 17:49:42.000000000 -0700
@@ -48,6 +48,8 @@ typedef struct cyasdevice {
cy_as_device_handle dev_handle;
/* Handle to the HAL */
cy_as_hal_device_tag hal_tag;
+ spinlock_t common_lock;
+ unsigned long flags;
} cyasdevice ;
/* global ptr to astoria device */
@@ -82,12 +84,19 @@ static void cyasdevice_deinit(cyasdevice
if (cy_as_dev->hal_tag) {
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
- if (stop_o_m_a_p_kernel(dev_handle_name,
+ if (cy_as_hal_omap_pnand_stop(dev_handle_name,
cy_as_dev->hal_tag) != 0)
cy_as_hal_print_message("<1>_cy_as_device: stopping "
- "OMAP kernel HAL failed\n");
+ "OMAP PNAND HAL failed\n");
#endif
+
+ #ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ if (cy_as_hal_omap_cram_stop(dev_handle_name,
+ cy_as_dev->hal_tag) != 0)
+ cy_as_hal_print_message("<1>_cy_as_device: stopping "
+ "OMAP CRAM HAL failed\n");
+ #endif
}
cy_as_hal_print_message("<1>_cy_as_device:HAL layer stopped\n") ;
@@ -126,6 +135,28 @@ static void cy_misc_callback(cy_as_devic
}
}
+void cy_as_acquire_common_lock()
+{
+ /*printk("%s: lock address is 0x%x, is_locked=0x%x\n", __func__,
+ &cy_as_device_controller->common_lock,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+ spin_lock_irqsave(&cy_as_device_controller->common_lock,
+ cy_as_device_controller->flags);/*
+ printk("%s: -- is_locked=0x%x\n", __func__,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+}
+EXPORT_SYMBOL(cy_as_acquire_common_lock);
+
+void cy_as_release_common_lock()
+{
+ /*printk("%s: lock address is 0x%x, is_locked=0x%x\n", __func__,
+ &cy_as_device_controller->common_lock,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+ spin_unlock_irqrestore(&cy_as_device_controller->common_lock,
+ cy_as_device_controller->flags);
+}
+EXPORT_SYMBOL(cy_as_release_common_lock);
+
/* reset astoria and reinit all regs */
#define PNAND_REG_CFG_INIT_VAL 0x0000
void hal_reset(cy_as_hal_device_tag tag)
@@ -188,7 +219,7 @@ static int cyasdevice_initialize(void)
cy_as_device_init_done = 0;
cy_as_misc_set_log_level(8);
-
+
cy_as_hal_print_message("<1>_cy_as_device initialize called\n") ;
if (cy_as_device_controller != 0) {
@@ -212,7 +243,7 @@ static int cyasdevice_initialize(void)
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
/* start OMAP HAL init instsnce */
- if (!start_o_m_a_p_kernel(dev_handle_name,
+ if (!cy_as_hal_omap_pnand_start(dev_handle_name,
&(cy_as_dev->hal_tag), cy_false)) {
cy_as_hal_print_message(
@@ -221,6 +252,17 @@ static int cyasdevice_initialize(void)
}
#endif
+ #ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ /* start OMAP HAL init instsnce */
+
+ if (!cy_as_hal_omap_cram_start(dev_handle_name,
+ &(cy_as_dev->hal_tag), cy_false)) {
+
+ cy_as_hal_print_message(
+ "<1>_cy_as_device: start OMAP34xx HAL failed\n") ;
+ goto done;
+ }
+ #endif
/* Now create the device */
if (cy_as_misc_create_device(&(cy_as_dev->dev_handle),
cy_as_dev->hal_tag) != CY_AS_ERROR_SUCCESS) {
@@ -331,9 +373,15 @@ static int cyasdevice_initialize(void)
((ver_data.is_debug_mode) ? "debug" : "release"),
ver_data.major, ver_data.minor, ver_data.build, str) ;
+ printk("%s: lock address is 0x%x\n", __func__, &cy_as_dev->common_lock);
+ spin_lock_init(&cy_as_dev->common_lock);
+ printk("%s: lock address is 0x%x\n", __func__, &cy_as_dev->common_lock);
+
/* done now */
cy_as_device_controller = cy_as_dev ;
-
+
+
+
return 0 ;
done:
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile
--- linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile 2010-08-19 14:30:50.000000000 -0700
@@ -12,12 +12,20 @@ obj-$(CONFIG_WESTBRIDGE_DEVICE_DRIVER)
ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL),y)
#moved for staging compatbility
#cyasdev-y := ../../../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o
-cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o \
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_pnand.o cyasdevice.o \
../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
../api/src/cyasstorage.o ../api/src/cyasusb.o
else
+
+ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL),y)
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_cram.o cyasdevice.o \
+ ../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
+ ../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
+ ../api/src/cyasstorage.o ../api/src/cyasusb.o
+else
# should not get here, need to be built with some hal
cyasdev-y := cyasdevice.o
endif
+endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.c 2010-08-26 11:18:18.000000000 -0700
@@ -224,6 +224,8 @@ static void cy_as_gadget_mtp_event_callb
cy_as_mtp_send_object_complete_data *send_obj_data =
(cy_as_mtp_send_object_complete_data *) evdata ;
+ cy_as_release_common_lock();
+
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
"<6>MTP EVENT: send_object_complete\n");
@@ -246,6 +248,8 @@ static void cy_as_gadget_mtp_event_callb
{
cy_as_mtp_get_object_complete_data *get_obj_data =
(cy_as_mtp_get_object_complete_data *) evdata ;
+;
+ cy_as_release_common_lock();
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
@@ -752,6 +756,9 @@ static int cyasgadget_queue(
_req->status = -EINPROGRESS;
_req->actual = 0;
+ if (as_req)
+ list_add_tail(&as_req->queue, &as_ep->queue);
+
spin_unlock_irqrestore(&cy_as_dev->lock, flags);
/* Call Async functions */
@@ -772,16 +779,6 @@ static int cyasgadget_queue(
else
_req->status = -EALREADY ;
} else if (as_ep->num == 0) {
- /*
- ret = cy_as_usb_write_data_async(cy_as_dev->dev_handle,
- as_ep->num, _req->length, _req->buf, cy_false,
- cyasgadget_setupwritecallback) ;
-
- if (ret != CY_AS_ERROR_SUCCESS)
- cy_as_hal_print_message("<1>_cy_as_gadget: "
- "cy_as_usb_write_data_async failed with error "
- "code %d\n", ret) ;
- */
if ((cy_as_dev->outsetupreq) && (_req->length)) {
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("<1>_cy_as_gadget: "
@@ -817,8 +814,7 @@ static int cyasgadget_queue(
"cy_as_usb_write_data_async failed with "
"error code %d\n", ret) ;
}
-
- } else if (list_empty(&as_ep->queue)) {
+ } else { /* read function not to EP 0*/
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("<1>_cy_as_gadget: "
"cy_as_usb_read_data_async being called since "
@@ -836,13 +832,6 @@ static int cyasgadget_queue(
_req->status = -EALREADY ;
}
- spin_lock_irqsave(&cy_as_dev->lock, flags);
-
- if (as_req)
- list_add_tail(&as_req->queue, &as_ep->queue);
-
- spin_unlock_irqrestore(&cy_as_dev->lock, flags);
-
return 0;
}
@@ -1146,11 +1135,9 @@ static int cyasgadget_ioctl(
struct inode *inode = mapping->host;
struct inode *alloc_inode =
file_to_allocate->f_path.dentry->d_inode;
- int cluster = 0;
uint32_t num_clusters = 0;
struct buffer_head bh;
struct kstat stat;
- struct iattr alloc_iattr;
int nr_pages = 0;
int ret_stat = 0;
@@ -1158,10 +1145,10 @@ static int cyasgadget_ioctl(
cy_as_hal_print_message("%s: fhandle is OK, "
"calling vfs_getattr\n", __func__);
#endif
-
+
ret_stat = vfs_getattr(file_to_allocate->f_path.mnt,
file_to_allocate->f_path.dentry, &stat);
-
+
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("%s: returned from "
"vfs_getattr() stat->blksize=0x%lx\n",
@@ -1171,15 +1158,11 @@ static int cyasgadget_ioctl(
/* TODO: get this from disk properties
* (from blockdevice)*/
#define SECTOR_SIZE 512
- if (stat.blksize != 0) {
num_clusters = (k_d.num_bytes) / SECTOR_SIZE;
- if (((k_d.num_bytes) % SECTOR_SIZE) != 0)
+ if (((k_d.num_bytes) % SECTOR_SIZE) != 0)
num_clusters++;
- } else {
- goto initsoj_safe_exit;
- }
-
+
bh.b_state = 0;
bh.b_blocknr = 0;
/* block size is arbitrary , we'll use sector size*/
@@ -1190,42 +1173,6 @@ static int cyasgadget_ioctl(
"size of %d\n", __func__,
num_clusters, bh.b_size);
#endif
- for (cluster = 0; cluster < num_clusters; cluster++) {
- ret_stat = fat_get_block(inode,
- cluster, &bh, 1);
- if (ret_stat) {
- cy_as_hal_print_message(
- "%s: unable to get fat block, "
- "ret_stat=0x%d\n",
- __func__, ret_stat);
- goto initsoj_safe_exit;
- }
- }
-
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: allocated clusters "
- "successfully (fat_get_block), check bmap..."
- "\n", __func__);
- #endif
-
- alloc_iattr.ia_valid = ATTR_SIZE;
- alloc_iattr.ia_size = k_d.num_bytes;
-
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: calling fat_notify_change "
- "(ia_valid:%d, ia_size:%d)\n", __func__,
- alloc_iattr.ia_valid,
- (int)alloc_iattr.ia_size);
- #endif
-
- /* adjust the filesize */
- ret_stat = alloc_inode->i_op->setattr(
- file_to_allocate->f_path.dentry, &alloc_iattr);
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: fat_setattr() "
- "returned 0x%x\n",
- __func__, ret_stat);
- #endif
/* clear dirty pages in page cache
* (if were any allocated) */
@@ -1246,7 +1193,7 @@ static int cyasgadget_ioctl(
"mpage_cleardirty() "
"for %d pages\n", __func__, nr_pages);
#endif
-
+
ret_stat = mpage_cleardirty(mapping, nr_pages);
/*fill up the the block table from the addr mapping */
@@ -1363,6 +1310,9 @@ static int cyasgadget_ioctl(
__func__);
#endif
sg_init_one(&sg, &blk_table, sizeof(blk_table));
+
+ cy_as_acquire_common_lock();
+
ret_stat = cy_as_mtp_init_send_object(dev->dev_handle,
(cy_as_mtp_block_table *)&sg,
k_d.num_bytes, 0, 0);
@@ -1379,7 +1329,6 @@ static int cyasgadget_ioctl(
__func__, alloc_filename);
} /* end if (file_to_allocate)*/
#endif
-initsoj_safe_exit:
ret_stat = 0;
retval = __put_user(ret_stat,
(uint32_t __user *)(&(usr_d->ret_val)));
@@ -1541,6 +1490,8 @@ initsoj_safe_exit:
k_d.num_bytes);
#endif
+ cy_as_acquire_common_lock();
+
ret_stat = cy_as_mtp_init_get_object(
dev->dev_handle,
(cy_as_mtp_block_table *)&sg,
@@ -2151,7 +2102,21 @@ static int cyasgadget_initialize(void)
/*strcpy(cy_as_dev->gadget.dev.bus_id, "cyasgadget");*/
cy_as_dev->gadget.dev.release = cyas_gadget_release;
cy_as_dev->gadget.name = cy_as_driver_name;
-
+
+ /* this "gadget" abstracts/virtualizes the controller */
+ dev_set_name(&cy_as_dev->gadget.dev, "cyas" );
+ cy_as_dev->gadget.dev.parent = cy_as_dev->cy_controller;
+
+ retval = device_register(&cy_as_dev->gadget.dev);
+ if (retval != 0) {
+ cy_as_hal_print_message( "%s: can not register device\n", __func__) ;
+ return retval;
+ }
+ #ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message( "%s: successfully registered device\n", __func__) ;
+ #endif
+
/* Get the device handle */
cy_as_dev->dev_handle = cyasdevice_getdevhandle() ;
if (0 == cy_as_dev->dev_handle) {
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.h 2010-08-25 18:28:27.000000000 -0700
@@ -83,8 +83,9 @@
#include <linux/uaccess.h> /* copy_*_user */
extern int mpage_cleardirty(struct address_space *mapping, int num_pages);
-extern int fat_get_block(struct inode *, sector_t , struct buffer_head *, int);
extern cy_as_device_handle *cyasdevice_getdevhandle(void) ;
+extern void cy_as_acquire_common_lock(void);
+extern void cy_as_release_common_lock(void);
/* Driver data structures and utilities */
typedef struct cyasgadget_ep {
@@ -119,6 +120,7 @@ typedef struct cyasgadget {
struct usb_gadget gadget;
spinlock_t lock;
struct cyasgadget_ep an_gadget_ep[16];
+ struct device *cy_controller;
struct usb_gadget_driver *driver;
/* Handle to the West Bridge device */
cy_as_device_handle dev_handle;
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-08-20 14:35:14.000000000 -0700
@@ -74,7 +74,7 @@
-/***** OMAP3430 Kernel HAL ****/
+/***** OMAP3430 PNAND HAL ****/
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
#ifdef CY_HAL_DEFINED
@@ -86,7 +86,24 @@
* considered is here
* #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
*/
- #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+
+#endif
+/*****************************/
+
+/***** OMAP3430 CRAM HAL ****/
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+ #ifdef CY_HAL_DEFINED
+ #error only one HAL layer can be defined
+ #endif
+
+ #define CY_HAL_DEFINED
+/* moved to staging location, eventual implementation
+ * considered is here
+ * #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+*/
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
#endif
/*****************************/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/Kconfig linux-next-incl-sdk/drivers/staging/westbridge/Kconfig
--- linux-next-vanilla/drivers/staging/westbridge/Kconfig 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/Kconfig 2010-09-01 15:45:46.000000000 -0700
@@ -4,7 +4,7 @@
menuconfig WESTBRIDGE
tristate "West Bridge support"
- depends on HAS_IOMEM && BROKEN
+ depends on WESTBRIDGE_HAL_SELECTED
help
This selects West Bridge Peripheral controller support.
@@ -12,7 +12,7 @@ menuconfig WESTBRIDGE
menuconfig WESTBRIDGE_ASTORIA
bool "West Bridge Astoria support"
- depends on WESTBRIDGE != n
+ depends on WESTBRIDGE != n && WESTBRIDGE_HAL_SELECTED
help
This option enables support for West Bridge Astoria
@@ -20,11 +20,43 @@ if WESTBRIDGE_ASTORIA
source "drivers/staging/westbridge/astoria/Kconfig"
endif #WESTBRIDGE_ASTORIA
-menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+#menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+# bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
+# depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+# help
+# Include the OMAP3430 HAL for PNAND interface
+
+config WESTBRIDGE_HAL_SELECTED
+ boolean
+
+choice
+ prompt "West Bridge HAL"
+ help
+ West Bridge HAL/processor interface to be used
+
+#
+# HAL Layers
+#
+
+config MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
- depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ bool "WESTBRIDGE OMAP3430 Astoria CRAM HAL"
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_NO_WESTBRIDGE
+ bool "no West Bridge HAL selected"
help
- Include the OMAP3430 Linux Based HAL
+ Do not include any HAL layer(de-activates West Bridge option)
+endchoice
config WESTBRIDGE_DEBUG
bool "West Bridge debugging"
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/fs/mpage.c linux-next-incl-sdk/fs/mpage.c
--- linux-next-vanilla/fs/mpage.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/fs/mpage.c 2010-09-01 16:12:40.000000000 -0700
@@ -716,3 +716,49 @@ int mpage_writepage(struct page *page, g
return ret;
}
EXPORT_SYMBOL(mpage_writepage);
+
+int mpage_cleardirty(struct address_space *mapping, int num_pages)
+{
+ int ret = 0;
+ int nr_pages;
+ struct pagevec pvec;
+ pgoff_t index = 0;
+ pgoff_t end;
+
+ pagevec_init(&pvec, 0);
+ index = mapping->writeback_index;
+ end = index + num_pages;
+
+ while ((index <= end) &&
+ (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
+ PAGECACHE_TAG_DIRTY,
+ min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+ unsigned i;
+
+ for (i = 0; i < nr_pages; i++) {
+ struct page *page = pvec.pages[i];
+
+ lock_page(page);
+ ret = clear_page_dirty_for_io(page);
+ if (page_has_private(page))
+ do_invalidatepage(page, 0);
+
+ cancel_dirty_page(page, PAGE_CACHE_SIZE);
+
+ remove_from_page_cache(page);
+ ClearPageMappedToDisk(page);
+ page_cache_release(page); /* pagecache ref */
+ unlock_page(page);
+
+ if (!ret) {
+ printk(KERN_INFO "mpage_cleardirty: clear_page_dirty_for_io returned %d\n", ret);
+ return ret;
+ }
+ }
+ pagevec_release(&pvec);
+ cond_resched();
+ }
+
+ return ret;
+}
+EXPORT_SYMBOL(mpage_cleardirty);
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes
2010-09-02 0:08 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes David Cross
@ 2010-09-02 1:37 ` Greg KH
2010-09-02 22:43 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes David Cross
` (2 subsequent siblings)
3 siblings, 0 replies; 12+ messages in thread
From: Greg KH @ 2010-09-02 1:37 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Wed, Sep 01, 2010 at 05:08:21PM -0700, David Cross wrote:
>
> > From: Greg KH [mailto:greg@kroah.com]
> > Sent: Tuesday, August 31, 2010 9:32 AM
> > To: David Cross
> > Cc: gregkh@suse.de; linux-kernel@vger.kernel.org
> > Subject: Re: [PATCH] West Bridge Astoria Driver 2.6.35
> >
> > On Tue, Aug 31, 2010 at 09:17:18AM -0700, David Cross wrote:
> > >
> > > > I get a build error when I try to build this driver:
> > > >
> > > drivers/staging/westbridge/astoria/block/../include/linux/westbridge/c>
> > > yasmisc.h:521:2: error: expected declaration specifiers or '...' before >
> > > 'cy_as_hal_device_tag'
> > >
> > > > It looks like the #include mess isn't working quite properly.
> > >
> > > > Also note that this driver is building on an x86-64 platform, which is
> > > > something that you probably don't want :)
> > >
> > > > So, for now, I'll just mark the driver as CONFIG_BROKEN and can you send
> > > > me some Kconfig patches against the next linux-next release which should
> > > > have this driver in it, so that it will build properly?
> > >
> > > Sure, I can do that. I have since changed the Kconfig structure a bit. It
> > > actually will build properly with the correct .config as it is though. I
> > can
> > > send you the .config if you would like.
> >
> > It's up to the Kconfig rules to ensure that there can never be a
> > "non-correct" .config file, so that needs to be fixed so that the build
> > will never be broken no matter what type of options are selected.
> >
> > Care to send a patch for that?
>
> This patch actually contains quite a few fixes, but it addresses the
> Kconfig issue as well. The linux-next tree does not seem to have a
> config for the zoom2, and trying to build it for that board seems to
> make the compilation break. As such, the only thing that I tested was
> compilation using the two different HALs (one of which is added in this
> patch). Please let me know if there are problems or questions with this.
> Thanks,
> David
Hm, I need a single-patch-per-change, as described in the
Documentation/SubmittingPatches file. So, could you break this up into
the logical changes, so that I can apply them all? That makes it easier
for everyone involved (reviewing, bisecting, etc.) and it's what is
needed for the rest of the kernel as well.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes
2010-09-02 0:08 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes David Cross
2010-09-02 1:37 ` Greg KH
@ 2010-09-02 22:43 ` David Cross
2010-09-05 5:17 ` Greg KH
2010-09-07 19:22 ` [PATCH] west bridge, kconfig and hal fixes David Cross
2010-09-02 22:47 ` [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty David Cross
2010-09-02 22:49 ` [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates David Cross
3 siblings, 2 replies; 12+ messages in thread
From: David Cross @ 2010-09-02 22:43 UTC (permalink / raw)
To: greg; +Cc: linux-kernel
On Wed, 2010-09-01 at 17:08 -0700, David Cross wrote:
From: Greg KH [mailto:greg@kroah.com]
> Sent: Tuesday, August 31, 2010 9:32 AM
> To: David Cross
> Cc: gregkh@suse.de; linux-kernel@vger.kernel.org
> Subject: Re: [PATCH] West Bridge Astoria Driver 2.6.35
>
> On Tue, Aug 31, 2010 at 09:17:18AM -0700, David Cross wrote:
> >
> > > I get a build error when I try to build this driver:
> > >
> > drivers/staging/westbridge/astoria/block/../include/linux/westbridge/c>
> > yasmisc.h:521:2: error: expected declaration specifiers or '...' before >
> > 'cy_as_hal_device_tag'
> >
> > > It looks like the #include mess isn't working quite properly.
> >
> > > Also note that this driver is building on an x86-64 platform, which is
> > > something that you probably don't want :)
> >
> > > So, for now, I'll just mark the driver as CONFIG_BROKEN and can you send
> > > me some Kconfig patches against the next linux-next release which should
> > > have this driver in it, so that it will build properly?
> >
> > Sure, I can do that. I have since changed the Kconfig structure a bit. It
> > actually will build properly with the correct .config as it is though. I
> can
> > send you the .config if you would like.
>
> It's up to the Kconfig rules to ensure that there can never be a
> "non-correct" .config file, so that needs to be fixed so that the build
> will never be broken no matter what type of options are selected.
>
> Care to send a patch for that?
This patch actually contains the KConfig changes necessary an additional
HAL layer. The Kconfig changes are pretty closely related to this HAL
layer, as such this patch is difficult to logically separate. The HAL
layers also rely on the export of some gpmc functions, which are added
to this patch. Although I don't own this driver, it is difficult for me
to understand why there would be issues with exporting these symbols.
The linux-next tree does not seem to have a config for the zoom2, and
trying to build it for that board seems to make the compilation break.
As such, the only thing that I tested was compilation using the two
different HALs (one of which is added in this patch). Please let me know
if there are problems or questions with this.
Thanks,
David
Signed-off-by: David Cross <david.cross@cypress.com>
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/arch/arm/mach-omap2/gpmc.c linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c
--- linux-next-vanilla/arch/arm/mach-omap2/gpmc.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c 2010-09-01 16:10:21.000000000 -0700
@@ -133,6 +133,7 @@ void gpmc_cs_write_reg(int cs, int idx,
reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
__raw_writel(val, reg_addr);
}
+EXPORT_SYMBOL(gpmc_cs_write_reg);
u32 gpmc_cs_read_reg(int cs, int idx)
{
@@ -141,6 +142,7 @@ u32 gpmc_cs_read_reg(int cs, int idx)
reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
return __raw_readl(reg_addr);
}
+EXPORT_SYMBOL(gpmc_cs_read_reg);
/* TODO: Add support for gpmc_fck to clock framework and use it */
unsigned long gpmc_get_fclk_period(void)
@@ -294,6 +296,7 @@ int gpmc_cs_set_timings(int cs, const st
return 0;
}
+EXPORT_SYMBOL(gpmc_cs_set_timings);
static void gpmc_cs_enable_mem(int cs, u32 base, u32 size)
{
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 2010-08-27 17:52:14.000000000 -0700
@@ -0,0 +1,1851 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+ #define CYASSTORAGE_WRITE_EP_NUM (4)
+ #define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ cy_as_hal_print_message(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg with address");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ cy_as_hal_print_message(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+uint32_t cy_as_hal_gpmc_init(cy_as_omap_dev_kernel *dev_p)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ unsigned int cs_mem_base;
+ unsigned int cs_vma_base;
+
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ cy_as_hal_print_message(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, SZ_16K);
+ cy_as_hal_print_message(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&cs_mem_base) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ cy_as_hal_print_message(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ cs_mem_base, AST_GPMC_CS);
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(cs_mem_base, SZ_16K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ cs_mem_base);
+ goto out_free_cs;
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ cs_vma_base = (u32)ioremap_nocache(cs_mem_base, SZ_16K);
+ if (!cs_vma_base) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", cs_mem_base);
+
+ goto out_release_mem_region;
+ }
+ cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ cs_mem_base, cs_vma_base);
+
+ dev_p->m_phy_addr_base = (void *) cs_mem_base;
+ dev_p->m_vma_addr_base = (void *) cs_vma_base;
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_on;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_on;
+ timings.adv_rd_off = WB_GPMC_ADV_t_off;
+ timings.adv_wr_off = WB_GPMC_ADV_t_off;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_on;
+ timings.oe_off = WB_GPMC_OE_t_off;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_on;
+ timings.we_off = WB_GPMC_WE_t_off;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(0) |
+ GPMC_CONFIG1_DEVICESIZE_16
+ ));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(cs_mem_base, SZ_16K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ cy_as_hal_print_message(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ cy_as_hal_print_message(KERN_INFO"pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ cy_as_hal_print_message(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(cy_as_omap_dev_kernel *dev_p)
+{
+ int i, err;
+
+ cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init(dev_p);
+
+ cy_as_hal_config_c_s_mux();
+
+ return err;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (dev_p->m_vma_addr_base != 0)
+ iounmap((void *)dev_p->m_vma_addr_base);
+
+ if (dev_p->m_phy_addr_base != 0)
+ release_mem_region((void *)dev_p->m_phy_addr_base, SZ_16K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ uint16_t write_addr;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk("%s: bad TAG parameter passed\n", __func__) ;
+ return ;
+ }
+ #endif
+
+ write_addr = (void *)(dev_p->m_vma_addr_base + CYAS_DEV_CALC_ADDR (addr));
+ writew ((unsigned short)data, write_addr) ;
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+ uint32_t read_addr = 0;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk("%s: bad TAG parameter passed\n", __func__) ;
+ return dev_p->m_sig;
+ }
+ #endif
+
+ read_addr = (void *) (dev_p->m_vma_addr_base + CYAS_DEV_CALC_ADDR(addr));
+ data = (unsigned short) readw (read_addr) ;
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, i, size;
+ uint16_t *dptr;
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+ register void *read_addr ;
+ register uint16_t a,b,c,d,e,f,g,h ;
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = (uint16_t *) end_points[ep].data_p;
+
+ read_addr = dev_p->m_vma_addr_base + CYAS_DEV_CALC_EP_ADDR(ep) ;
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * Now, read the data from the device
+ */
+ for(i = size/16 ; i > 0 ; i--) {
+ a = (unsigned short) readw (read_addr) ;
+ b = (unsigned short) readw (read_addr) ;
+ c = (unsigned short) readw (read_addr) ;
+ d = (unsigned short) readw (read_addr) ;
+ e = (unsigned short) readw (read_addr) ;
+ f = (unsigned short) readw (read_addr) ;
+ g = (unsigned short) readw (read_addr) ;
+ h = (unsigned short) readw (read_addr) ;
+ *dptr++ = a ;
+ *dptr++ = b ;
+ *dptr++ = c ;
+ *dptr++ = d ;
+ *dptr++ = e ;
+ *dptr++ = f ;
+ *dptr++ = g ;
+ *dptr++ = h ;
+ }
+
+ switch ((size & 0xF)/2) {
+ case 7:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 6:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 5:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 4:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 3:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 2:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 1:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ /* Odd sized packet */
+ uint16_t d = (unsigned short) readw (read_addr) ;
+ *((uint8_t *)dptr) = (d & 0xff) ;
+ }
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0, i = 0;
+ uint32_t size;
+ uint16_t *dptr;
+ register void *write_addr ;
+ register uint16_t a,b,c,d ;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ write_addr = (void *) (dev_p->m_vma_addr_base + CYAS_DEV_CALC_EP_ADDR(ep)) ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size){
+ /*
+ * Now, write the data to the device
+ */
+ for(i = size/8 ; i > 0 ; i--) {
+ a = *dptr++ ;
+ b = *dptr++ ;
+ c = *dptr++ ;
+ d = *dptr++ ;
+ writew (a, write_addr) ;
+ writew (b, write_addr) ;
+ writew (c, write_addr) ;
+ writew (d, write_addr) ;
+ }
+
+ switch ((size & 7)/2) {
+ case 3:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ case 2:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ case 1:
+ writew (*dptr, write_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ uint16_t v = *((uint8_t *)dptr) ;
+ writew (v, write_addr);
+ }
+ }
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if(state) {
+ __gpio_set_value(AST_WAKEUP, 1);
+ } else {
+ __gpio_set_value(AST_WAKEUP, 0);
+ }
+
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+ (void) is_standby_wakeup;
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ uint32_t err = 0;
+ /* No debug mode support through argument as of now */
+ (void)debug;
+
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ; /* No debug mode support through argument as of now */
+ (void)debug;
+
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /* allocate memory for OMAP HAL*/
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_CRAM_HAL_SIG;
+
+ /* initialize OMAP hardware and StartOMAPKernelall gpio pins */
+ err = cy_as_hal_processor_hw_init(dev_p);
+ if(err)
+ goto bus_acc_error;
+
+ {
+ uint16_t val1, val2, val3, val4, val5;
+ printk("%s: testing register reads\n", __func__);
+ val1 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0x80);
+ val2 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xC0);
+ val3 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xFF);
+ val4 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0xA5);
+ val5 = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, 0x5A);
+
+ printk("%s: testing register reads 0x%x 0x%x 0x%x 0x%x 0x%x\n",
+ __func__, val1, val2, val3, val4, val5);
+ }
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p, CY_AS_MEM_CM_WB_CFG_ID);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, "
+ "CY_AS_MEM_CM_WB_CFG_ID %4.4x", data16[0]);
+ goto bus_acc_error;
+ }
+
+ cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 1969-12-31 16:00:00.000000000 -0800
@@ -1,2450 +0,0 @@
-/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
-
-#include <linux/fs.h>
-#include <linux/ioport.h>
-#include <linux/timer.h>
-#include <linux/gpio.h>
-#include <linux/interrupt.h>
-#include <linux/delay.h>
-#include <linux/scatterlist.h>
-#include <linux/mm.h>
-#include <linux/irq.h>
-#include <linux/slab.h>
-#include <linux/sched.h>
-/* include seems broken moving for patch submission
- * #include <mach/mux.h>
- * #include <mach/gpmc.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
- * #include <linux/westbridge/cyaserr.h>
- * #include <linux/westbridge/cyasregs.h>
- * #include <linux/westbridge/cyasdma.h>
- * #include <linux/westbridge/cyasintr.h>
- */
-#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
-#include "../../../include/linux/westbridge/cyaserr.h"
-#include "../../../include/linux/westbridge/cyasregs.h"
-#include "../../../include/linux/westbridge/cyasdma.h"
-#include "../../../include/linux/westbridge/cyasintr.h"
-
-#define HAL_REV "1.1.0"
-
-/*
- * uncomment to enable 16bit pnand interface
- */
-#define PNAND_16BIT_MODE
-
-/*
- * selects one of 3 versions of pnand_lbd_read()
- * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
- * reads through the gpmc CONTROLLER REGISTERS
- * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
- * same speed as the above
- * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
- */
-#define PNAND_LBD_READ_NO_PFE
-/* #define ENABLE_GPMC_PF_ENGINE */
-/* #define PFE_LBD_READ_V2 */
-
-/*
- * westbrige astoria ISR options to limit number of
- * back to back DMA transfers per ISR interrupt
- */
-#define MAX_DRQ_LOOPS_IN_ISR 4
-
-/*
- * debug prints enabling
- *#define DBGPRN_ENABLED
- *#define DBGPRN_DMA_SETUP_RD
- *#define DBGPRN_DMA_SETUP_WR
- */
-
-
-/*
- * For performance reasons, we handle storage endpoint transfers upto 4 KB
- * within the HAL itself.
- */
- #define CYASSTORAGE_WRITE_EP_NUM (4)
- #define CYASSTORAGE_READ_EP_NUM (8)
-
-/*
- * size of DMA packet HAL can accept from Storage API
- * HAL will fragment it into smaller chunks that the P port can accept
- */
-#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
-
-/*
- * P port MAX DMA packet size according to interface/ep configurartion
- */
-#define HAL_DMA_PKT_SZ 512
-
-#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
- ((ep) == 6) || ((ep) == 8))
-
-/*
- * persistant, stores current GPMC interface cfg mode
- */
-static uint8_t pnand_16bit;
-
-/*
- * keep processing new WB DRQ in ISR untill all handled (performance feature)
- */
-#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
-
-
-/*
- * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
- */
-#define CASDO 0x05
-#define CASDI 0x85
-#define RDPAGE_B1 0x00
-#define RDPAGE_B2 0x30
-#define PGMPAGE_B1 0x80
-#define PGMPAGE_B2 0x10
-
-/*
- * The type of DMA operation, per endpoint
- */
-typedef enum cy_as_hal_dma_type {
- cy_as_hal_read,
- cy_as_hal_write,
- cy_as_hal_none
-} cy_as_hal_dma_type ;
-
-
-/*
- * SG list halpers defined in scaterlist.h
-#define sg_is_chain(sg) ((sg)->page_link & 0x01)
-#define sg_is_last(sg) ((sg)->page_link & 0x02)
-#define sg_chain_ptr(sg) \
- ((struct scatterlist *) ((sg)->page_link & ~0x03))
-*/
-typedef struct cy_as_hal_endpoint_dma {
- cy_bool buffer_valid ;
- uint8_t *data_p ;
- uint32_t size ;
- /*
- * sg_list_enabled - if true use, r/w DMA transfers use sg list,
- * FALSE use pointer to a buffer
- * sg_p - pointer to the owner's sg list, of there is such
- * (like blockdriver)
- * dma_xfer_sz - size of the next dma xfer on P port
- * seg_xfer_cnt - counts xfered bytes for in current sg_list
- * memory segment
- * req_xfer_cnt - total number of bytes transfered so far in
- * current request
- * req_length - total request length
- */
- bool sg_list_enabled;
- struct scatterlist *sg_p ;
- uint16_t dma_xfer_sz;
- uint32_t seg_xfer_cnt;
- uint16_t req_xfer_cnt;
- uint16_t req_length;
- cy_as_hal_dma_type type ;
- cy_bool pending ;
-} cy_as_hal_endpoint_dma ;
-
-/*
- * The list of OMAP devices (should be one)
- */
-static cy_as_omap_dev_kernel *m_omap_list_p;
-
-/*
- * The callback to call after DMA operations are complete
- */
-static cy_as_hal_dma_complete_callback callback;
-
-/*
- * Pending data size for the endpoints
- */
-static cy_as_hal_endpoint_dma end_points[16] ;
-
-/*
- * Forward declaration
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
-
-static uint16_t intr_sequence_num;
-static uint8_t intr__enable;
-spinlock_t int_lock ;
-
-static u32 iomux_vma;
-static u32 csa_phy;
-
-/*
- * gpmc I/O registers VMA
- */
-static u32 gpmc_base ;
-
-/*
- * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
- */
-static u32 gpmc_data_vma;
-static u32 ndata_reg_vma;
-static u32 ncmd_reg_vma;
-static u32 naddr_reg_vma;
-
-/*
- * fwd declarations
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static inline u16 __attribute__((always_inline))
- ast_p_nand_casdo_read(u8 reg_addr8);
-static inline void __attribute__((always_inline))
- ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
-
-/*
- * prints given number of omap registers
- */
-static void cy_as_hal_print_omap_regs(char *name_prefix,
- u8 name_base, u32 virt_base, u16 count)
-{
- u32 reg_val, reg_addr;
- u16 i;
- cy_as_hal_print_message(KERN_INFO "\n");
- for (i = 0; i < count; i++) {
-
- reg_addr = virt_base + (i*4);
- /* use virtual addresses here*/
- reg_val = __raw_readl(reg_addr);
- cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
- name_prefix, name_base+i,
- reg_addr, reg_val);
- }
-}
-
-/*
- * setMUX function for a pad + additional pad flags
- */
-static u16 omap_cfg_reg_L(u32 pad_func_index)
-{
- static u8 sanity_check = 1;
-
- u32 reg_vma;
- u16 cur_val, wr_val, rdback_val;
-
- /*
- * do sanity check on the omap_mux_pin_cfg[] table
- */
- cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg ");
- if (sanity_check) {
- if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
-
- cy_as_hal_print_message(KERN_INFO
- "table is good.\n");
- } else {
- cy_as_hal_print_message(KERN_WARNING
- "table is bad, fix it");
- }
- /*
- * do it only once
- */
- sanity_check = 0;
- }
-
- /*
- * get virtual address to the PADCNF_REG
- */
- reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
-
- /*
- * add additional USER PU/PD/EN flags
- */
- wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
- cur_val = IORD16(reg_vma);
-
- /*
- * PADCFG regs 16 bit long, packed into 32 bit regs,
- * can also be accessed as u16
- */
- IOWR16(reg_vma, wr_val);
- rdback_val = IORD16(reg_vma);
-
- /*
- * in case if the caller wants to save the old value
- */
- return wr_val;
-}
-
-#define BLKSZ_4K 0x1000
-
-/*
- * switch GPMC DATA bus mode
- */
-void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
-{
- uint32_t tmp32;
-
- /*
- * disable gpmc CS4 operation 1st
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
- GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * GPMC NAND data bus can be 8 or 16 bit wide
- */
- if (dbus16_enabled) {
- DBGPRN("enabling 16 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2) |
- GPMC_CONFIG1_DEVICESIZE_16)
- );
- } else {
- DBGPRN(KERN_INFO "enabling 8 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2))
- );
- }
-
- /*
- * re-enable astoria CS operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- *remember the state
- */
- pnand_16bit = dbus16_enabled;
-}
-
-static int cy_as_hal_gpmc_init(void)
-{
- u32 tmp32;
- int err;
- struct gpmc_timings timings;
- /*
- * get GPMC i/o registers base(already been i/o mapped
- * in kernel, no need for separate i/o remap)
- */
- gpmc_base = phys_to_virt(OMAP34XX_GPMC_BASE);
- DBGPRN(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
- gpmc_base, __raw_readl(gpmc_base)
- );
-
- /*
- * these are globals are full VMAs of the gpmc_base above
- */
- ncmd_reg_vma = GPMC_VMA(GPMC_NAND_CMD);
- naddr_reg_vma = GPMC_VMA(GPMC_NAND_ADDR);
- ndata_reg_vma = GPMC_VMA(GPMC_NAND_DATA);
-
- /*
- * request GPMC CS for ASTORIA request
- */
- if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
- cy_as_hal_print_message(KERN_ERR "error failed to request"
- "ncs4 for ASTORIA\n");
- return -1;
- } else {
- DBGPRN(KERN_INFO "got phy_addr:%x for "
- "GPMC CS%d GPMC_CFGREG7[CS4]\n",
- csa_phy, AST_GPMC_CS);
- }
-
- /*
- * request VM region for 4K addr space for chip select 4 phy address
- * technically we don't need it for NAND devices, but do it anyway
- * so that data read/write bus cycle can be triggered by reading
- * or writing this mem region
- */
- if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
- err = -EBUSY;
- cy_as_hal_print_message(KERN_ERR "error MEM region "
- "request for phy_addr:%x failed\n",
- csa_phy);
- goto out_free_cs;
- }
-
- /*
- * REMAP mem region associated with our CS
- */
- gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
- if (!gpmc_data_vma) {
- err = -ENOMEM;
- cy_as_hal_print_message(KERN_ERR "error- ioremap()"
- "for phy_addr:%x failed", csa_phy);
-
- goto out_release_mem_region;
- }
- cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
- csa_phy, gpmc_data_vma);
-
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- memset(&timings, 0, sizeof(timings));
-
- /* cs timing */
- timings.cs_on = WB_GPMC_CS_t_o_n;
- timings.cs_wr_off = WB_GPMC_BUSCYC_t;
- timings.cs_rd_off = WB_GPMC_BUSCYC_t;
-
- /* adv timing */
- timings.adv_on = WB_GPMC_ADV_t_o_n;
- timings.adv_rd_off = WB_GPMC_BUSCYC_t;
- timings.adv_wr_off = WB_GPMC_BUSCYC_t;
-
- /* oe timing */
- timings.oe_on = WB_GPMC_OE_t_o_n;
- timings.oe_off = WB_GPMC_OE_t_o_f_f;
- timings.access = WB_GPMC_RD_t_a_c_c;
- timings.rd_cycle = WB_GPMC_BUSCYC_t;
-
- /* we timing */
- timings.we_on = WB_GPMC_WE_t_o_n;
- timings.we_off = WB_GPMC_WE_t_o_f_f;
- timings.wr_access = WB_GPMC_WR_t_a_c_c;
- timings.wr_cycle = WB_GPMC_BUSCYC_t;
-
- timings.page_burst_access = WB_GPMC_BUSCYC_t;
- timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
- gpmc_cs_set_timings(AST_GPMC_CS, &timings);
-
- cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
- GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
-
- /*
- * DISABLE cs4, NOTE GPMC REG7 is already configured
- * at this point by gpmc_cs_request
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
- ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (AS_CS_MASK | AS_CS_BADDR));
-
- /*
- * by default configure GPMC into 8 bit mode
- * (to match astoria default mode)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- /*
- * ENABLE astoria cs operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- * No method currently exists to write this register through GPMC APIs
- * need to change WAIT2 polarity
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
- IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
-
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
-
- return 0;
-
-out_release_mem_region:
- release_mem_region(csa_phy, BLKSZ_4K);
-
-out_free_cs:
- gpmc_cs_free(AST_GPMC_CS);
-
- return err;
-}
-
-/*
- * west bridge astoria ISR (Interrupt handler)
- */
-static irqreturn_t cy_astoria_int_handler(int irq,
- void *dev_id, struct pt_regs *regs)
-{
- cy_as_omap_dev_kernel *dev_p;
- uint16_t read_val = 0 ;
- uint16_t mask_val = 0 ;
-
- /*
- * debug stuff, counts number of loops per one intr trigger
- */
- uint16_t drq_loop_cnt = 0;
- uint8_t irq_pin;
- /*
- * flags to watch
- */
- const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
- CY_AS_MEM_P0_INTR_REG_MBINT |
- CY_AS_MEM_P0_INTR_REG_PMINT |
- CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
-
- /*
- * sample IRQ pin level (just for statistics)
- */
- irq_pin = __gpio_get_value(AST_INT);
-
- /*
- * this one just for debugging
- */
- intr_sequence_num++ ;
-
- /*
- * astoria device handle
- */
- dev_p = dev_id;
-
- /*
- * read Astoria intr register
- */
- read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) ;
-
- /*
- * save current mask value
- */
- mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG) ;
-
- DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
- intr_sequence_num, read_val);
-
- /*
- * Disable WB interrupt signal generation while we are in ISR
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
- /*
- * this is a DRQ Interrupt
- */
- if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
-
- do {
- /*
- * handle DRQ interrupt
- */
- drq_loop_cnt++;
-
- cy_handle_d_r_q_interrupt(dev_p) ;
-
- /*
- * spending to much time in ISR may impact
- * average system performance
- */
- if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
- break;
-
- /*
- * Keep processing if there is another DRQ int flag
- */
- } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) &
- CY_AS_MEM_P0_INTR_REG_DRQINT);
- }
-
- if (read_val & sentinel)
- cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
-
- DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
- "int_pin:%d DRQ_jobs:%d\n",
- intr_sequence_num,
- mask_val,
- irq_pin,
- drq_loop_cnt);
-
- /*
- * re-enable WB hw interrupts
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
-
- return IRQ_HANDLED ;
-}
-
-static int cy_as_hal_configure_interrupts(void *dev_p)
-{
- int result;
- int irq_pin = AST_INT;
-
- set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
-
- /*
- * for shared IRQS must provide non NULL device ptr
- * othervise the int won't register
- * */
- result = request_irq(OMAP_GPIO_IRQ(irq_pin),
- (irq_handler_t)cy_astoria_int_handler,
- IRQF_SHARED, "AST_INT#", dev_p);
-
- if (result == 0) {
- /*
- * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
- * assigned to this interrupt
- * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
- * of AST_INT GPIO IRQ_ENABLE FLAG
- */
- cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
- "%d assigned IRQ #%d IRQEN1=%d\n",
- irq_pin,
- OMAP_GPIO_IRQ(irq_pin),
- OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
- );
- } else {
- cy_as_hal_print_message("cyasomaphal: interrupt "
- "failed to register\n");
- gpio_free(irq_pin);
- cy_as_hal_print_message(KERN_WARNING
- "ASTORIA: can't get assigned IRQ"
- "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
- }
-
- return result;
-}
-
-/*
- * initialize OMAP pads/pins to user defined functions
- */
-static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- /*
- * browse through the table an dinitiaze the pins
- */
- u32 in_level = 0;
- u16 tmp16, mux_val;
-
- while (pad_cfg_tab->name != NULL) {
-
- if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
-
- pad_cfg_tab->valid = 1;
- mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
-
- /*
- * always set drv level before changing out direction
- */
- __gpio_set_value(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /*
- * "0" - OUT, "1", input omap_set_gpio_direction
- * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
- */
- if (pad_cfg_tab->dir)
- gpio_direction_input(pad_cfg_tab->pin_num);
- else
- gpio_direction_output(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /* sample the pin */
- in_level = __gpio_get_value(pad_cfg_tab->pin_num);
-
- cy_as_hal_print_message(KERN_INFO "configured %s to "
- "OMAP pad_%d, DIR=%d "
- "DOUT=%d, DIN=%d\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num,
- pad_cfg_tab->dir,
- pad_cfg_tab->drv,
- in_level
- );
- } else {
- /*
- * get the pad_mux value to check on the pin_function
- */
- cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
- "for signal %s, its already taken\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- }
-
- tmp16 = *(u16 *)PADCFG_VMA
- (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
-
- cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
- "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
- pad_cfg_tab->pin_num, tmp16,
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
- );
-
- /*
- * next pad_cfg deriptor
- */
- pad_cfg_tab++;
- }
-
- cy_as_hal_print_message(KERN_INFO"pads configured\n");
-}
-
-
-/*
- * release gpios taken by the module
- */
-static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- while (pad_cfg_tab->name != NULL) {
-
- if (pad_cfg_tab->valid) {
- gpio_free(pad_cfg_tab->pin_num);
- pad_cfg_tab->valid = 0;
- cy_as_hal_print_message(KERN_INFO "GPIO_%d "
- "released from %s\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- } else {
- cy_as_hal_print_message(KERN_INFO "no release "
- "for %s, GPIO_%d, wasn't acquired\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num);
- }
- pad_cfg_tab++;
- }
-}
-
-void cy_as_hal_config_c_s_mux(void)
-{
- /*
- * FORCE the GPMC CS4 pin (it is in use by the zoom system)
- */
- omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
-}
-EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
-
-/*
- * inits all omap h/w
- */
-uint32_t cy_as_hal_processor_hw_init(void)
-{
- int i, err;
-
- cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
-
- iomux_vma = (u32)ioremap_nocache(
- (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
- cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
- iomux_vma, IORD32(iomux_vma));
-
- /*
- * remap gpio banks
- */
- for (i = 0; i < 6; i++) {
- gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
- gpio_vma_tab[i].phy_addr,
- gpio_vma_tab[i].size);
-
- cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
- gpio_vma_tab[i].name,
- (u32)gpio_vma_tab[i].virt_addr);
- };
-
- /*
- * force OMAP_GPIO_126 to rleased state,
- * will be configured to drive reset
- */
- gpio_free(AST_RESET);
-
- /*
- *same thing with AStoria CS pin
- */
- gpio_free(AST_CS);
-
- /*
- * initialize all the OMAP pads connected to astoria
- */
- cy_as_hal_init_user_pads(user_pad_cfg);
-
- err = cy_as_hal_gpmc_init();
- if (err < 0)
- cy_as_hal_print_message(KERN_INFO"gpmc init failed:%d", err);
-
- cy_as_hal_config_c_s_mux();
-
- return gpmc_data_vma;
-}
-EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
-
-void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
-{
- /*
- * free omap hw resources
- */
- if (gpmc_data_vma != 0)
- iounmap((void *)gpmc_data_vma);
-
- if (csa_phy != 0)
- release_mem_region(csa_phy, BLKSZ_4K);
-
- gpmc_cs_free(AST_GPMC_CS);
-
- free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
-
- cy_as_hal_release_user_pads(user_pad_cfg);
-}
-
-/*
- * These are the functions that are not part of the
- * HAL layer, but are required to be called for this HAL
- */
-
-/*
- * Called On AstDevice LKM exit
- */
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
-
- /*
- * TODO: Need to disable WB interrupt handlere 1st
- */
- if (0 == dev_p)
- return 1 ;
-
- cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
- if (dev_p->m_sig != CY_AS_OMAP_KERNEL_HAL_SIG) {
- cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
- pgm, __func__) ;
- return 1 ;
- }
-
- /*
- * disable interrupt
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
-#if 0
- if (dev_p->thread_flag == 0) {
- dev_p->thread_flag = 1 ;
- wait_for_completion(&dev_p->thread_complete) ;
- cy_as_hal_print_message("cyasomaphal:"
- "done cleaning thread\n");
- cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
- }
-#endif
-
- cy_as_hal_omap_hardware_deinit(dev_p);
-
- /*
- * Rearrange the list
- */
- if (m_omap_list_p == dev_p)
- m_omap_list_p = dev_p->m_next_p ;
-
- cy_as_hal_free(dev_p) ;
-
- cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
- return 0;
-}
-
-int omap_start_intr(cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
- int ret = 0 ;
- const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
- CY_AS_MEM_P0_INTR_REG_MBINT ;
-
- /*
- * register for interrupts
- */
- ret = cy_as_hal_configure_interrupts(dev_p) ;
-
- /*
- * enable only MBox & DRQ interrupts for now
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask) ;
-
- return 1 ;
-}
-
-/*
- * Below are the functions that communicate with the WestBridge device.
- * These are system dependent and must be defined by the HAL layer
- * for a given system.
- */
-
-/*
- * GPMC NAND command+addr write phase
- */
-static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
-{
- /*
- * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
- */
- u32 tmpa32 = ((row_addr << 16) | col_addr);
- u8 RA2 = (u8)(row_addr >> 16);
-
- if (!pnand_16bit) {
- /*
- * GPMC PNAND 8bit BUS
- */
- /*
- * CMD1
- */
- IOWR8(ncmd_reg_vma, cmdb1);
-
- /*
- *pnand bus: <CA0,CA1,RA0,RA1>
- */
- IOWR32(naddr_reg_vma, tmpa32);
-
- /*
- * <RA2> , always zero
- */
- IOWR8(naddr_reg_vma, RA2);
-
- } else {
- /*
- * GPMC PNAND 16bit BUS , in 16 bit mode CMD
- * and ADDR sent on [d7..d0]
- */
- uint8_t CA0, CA1, RA0, RA1;
- CA0 = tmpa32 & 0x000000ff;
- CA1 = (tmpa32 >> 8) & 0x000000ff;
- RA0 = (tmpa32 >> 16) & 0x000000ff;
- RA1 = (tmpa32 >> 24) & 0x000000ff;
-
- /*
- * can't use 32 bit writes here omap will not serialize
- * them to lower half in16 bit mode
- */
-
- /*
- *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
- */
- IOWR8(ncmd_reg_vma, cmdb1);
- IOWR8(naddr_reg_vma, CA0);
- IOWR8(naddr_reg_vma, CA1);
- IOWR8(naddr_reg_vma, RA0);
- IOWR8(naddr_reg_vma, RA1);
- IOWR8(naddr_reg_vma, RA2);
- }
-}
-
-/*
- * spin until r/b goes high
- */
-inline int wait_rn_b_high(void)
-{
- u32 w_spins = 0;
-
- /*
- * TODO: note R/b may go low here, need to spin until high
- * while (omap_get_gpio_datain(AST_RnB) == 0) {
- * w_spins++;
- * }
- * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- *
- * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- * w_spins++;
- * }
- * printk("<1>RnB=0!:%d\n",w_spins);
- * }
- */
- return w_spins;
-}
-
-#ifdef ENABLE_GPMC_PF_ENGINE
-/* #define PFE_READ_DEBUG
- * PNAND block read with OMAP PFE enabled
- * status: Not tested, NW, broken , etc
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint8_t bytes_in_fifo;
-
- /* debug vars*/
-#ifdef PFE_READ_DEBUG
- uint32_t loop_limit;
- uint16_t bytes_read = 0;
-#endif
-
- /*
- * configure the prefetch engine
- */
- uint32_t tmp32;
- uint32_t pfe_status;
-
- /*
- * DISABLE GPMC CS4 operation 1st, this is
- * in case engine is be already disabled
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
-
-#ifdef PFE_READ_DEBUG
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
- if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
- printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
- tmp32, GPMC_PREFETCH_CONFIG1_VAL);
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
- }
-
- /*
- *sanity check 2
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
- if (tmp32 != (count))
- printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
- "!= VAL written:%d\n", tmp32, count);
-#endif
-
- /*
- * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
- * on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
-
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * start the prefetch engine
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- while (1) {
- /*
- * GPMC PFE service loop
- */
- do {
- /*
- * spin until PFE fetched some
- * PNAND bus words in the FIFO
- */
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- bytes_in_fifo = (pfe_status >> 24) & 0x7f;
- } while (bytes_in_fifo == 0);
-
- /* whole 32 bit words in fifo */
- w32cnt = bytes_in_fifo >> 2;
-
-#if 0
- /*
- *NOTE: FIFO_PTR indicates number of NAND bus words bytes
- * already received in the FIFO and available to be read
- * by DMA or MPU whether COUNTVAL indicates number of BUS
- * words yet to be read from PNAND bus words
- */
- printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
- pfe_status, bytes_in_fifo,
- (pfe_status & 0x3fff), w32cnt);
-#endif
-
- while (w32cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- if ((pfe_status & 0x3fff) == 0) {
- /*
- * PFE acc angine done, there still may be data leftover
- * in the FIFO re-read FIFO BYTE counter (check for
- * leftovers from 32 bit read accesses above)
- */
- bytes_in_fifo = (IORD32(
- GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- /*
- * NOTE we may still have one word left in the fifo
- * read it out
- */
- ptr8 = ptr32;
- switch (bytes_in_fifo) {
-
- case 0:
- /*
- * nothing to do we already read the
- * FIFO out with 32 bit accesses
- */
- break;
- case 1:
- /*
- * this only possible
- * for 8 bit pNAND only
- */
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 2:
- /*
- * this one can occur in either modes
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- break;
-
- case 3:
- /*
- * this only possible for 8 bit pNAND only
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- ptr8 += 2;
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 4:
- /*
- * shouldn't happen, but has been seen
- * in 8 bit mode
- */
- *ptr32 = IORD32(gpmc_data_vma);
- break;
-
- default:
- printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
- bytes_in_fifo);
- break;
- }
- /*
- * read is completed, get out of the while(1) loop
- */
- break;
- }
- }
-}
-#endif
-
-#ifdef PFE_LBD_READ_V2
-/*
- * PFE engine assisted reads with the 64 byte blocks
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint8_t rd_cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint16_t reminder;
- uint32_t pfe_status;
-
- /*
- * ISSUE PNAND CMD+ADDR
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * setup PFE block
- * count - OMAP number of bytes to access on pnand bus
- */
-
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- do {
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
- rd_cnt = pfe_status >> (24+2);
-
- while (rd_cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- } while (pfe_status & 0x3fff);
-
- /*
- * read out the leftover
- */
- ptr8 = ptr32;
- rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- while (rd_cnt--)
- *ptr8++ = IORD8(gpmc_data_vma);
-}
-#endif
-
-#ifdef PNAND_LBD_READ_NO_PFE
-/*
- * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
- * the original working code, works at max speed for 8 bit xfers
- * for 16 bit the bus diagram has gaps
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint16_t *ptr16;
- uint16_t remainder;
-
- DBGPRN("<1> %s(): NO_PFE\n", __func__);
-
- ptr32 = buff;
- /* number of whole 32 bit words in the transfer */
- w32cnt = count >> 2;
-
- /* remainder, in bytes(0..3) */
- remainder = count & 03;
-
- /*
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * read data by 32 bit chunks
- */
- while (w32cnt--)
- *ptr32++ = IORD32(ndata_reg_vma);
-
- /*
- * now do the remainder(it can be 0, 1, 2 or 3)
- * same code for both 8 & 16 bit bus
- * do 1 or 2 MORE words
- */
- ptr16 = (uint16_t *)ptr32;
-
- switch (remainder) {
- case 1:
- /* read one 16 bit word
- * IN 8 BIT WE NEED TO READ even number of bytes
- */
- case 2:
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- *ptr16++ = IORD16(ndata_reg_vma);
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- default:
- /*
- * remainder is 0
- */
- break;
- }
-}
-#endif
-
-/*
- * uses LBD mode to write N bytes into astoria
- * Status: Working, however there are 150ns idle
- * timeafter every 2 (16 bit or 4(8 bit) bus cycles
- */
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint16_t remainder;
- uint8_t *ptr8;
- uint16_t *ptr16;
- uint32_t *ptr32;
-
- remainder = count & 03;
- w32cnt = count >> 2;
- ptr32 = buff;
- ptr8 = buff;
-
- /*
- * send: CMDB1, CA0,CA1,RA0,RA1,RA2
- */
- nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
-
- /*
- * blast the data out in 32bit chunks
- */
- while (w32cnt--)
- IOWR32(ndata_reg_vma, *ptr32++);
-
- /*
- * do the reminder if there is one
- * same handling for both 8 & 16 bit pnand: mode
- */
- ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
-
- switch (remainder) {
- case 1:
- /*
- * read one 16 bit word
- */
- case 2:
- IOWR16(ndata_reg_vma, *ptr16);
- break;
-
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- IOWR16(ndata_reg_vma, *ptr16++);
- IOWR16(ndata_reg_vma, *ptr16);
- break;
- default:
- /*
- * reminder is 0
- */
- break;
- }
- /*
- * finally issue a PGM cmd
- */
- IOWR8(ncmd_reg_vma, PGMPAGE_B2);
-}
-
-/*
- * write Astoria register
- */
-static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
-{
- unsigned long flags;
- u16 addr16;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t rdreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&rdreg_usage_cnt) != 0) {
- cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
- "* cy_as_hal_write_register usage:%d\n",
- atomic_read(&rdreg_usage_cnt));
- }
-
- atomic_inc(&rdreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- */
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
-
- /*
- * this should be sent on the 16 bit bus
- */
- IOWR16(ndata_reg_vma, data);
- } else {
- /*
- * 8 bit nand mode GPMC will automatically
- * seriallize 16bit or 32 bit writes into
- * 8 bit onesto the lower 8 bit in LE order
- */
- addr16 = 0x0c00 | reg_addr8;
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR16(naddr_reg_vma, addr16);
- IOWR16(ndata_reg_vma, data);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&rdreg_usage_cnt);
- local_irq_restore(flags);
-}
-
-
-/*
- * read astoria register via pNAND interface
- */
-static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
-{
- u16 data;
- u16 addr16;
- unsigned long flags;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t wrreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&wrreg_usage_cnt) != 0) {
- /*
- * if it gets here ( from other threads), this function needs
- * need spin_lock_irq save() protection
- */
- cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
- "cy_as_hal_write_register usage:%d\n",
- atomic_read(&wrreg_usage_cnt));
- }
- atomic_inc(&wrreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- * CMD1, CA1, CA2,
- */
-
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
- IOWR8(ncmd_reg_vma, 0x00E0);
-
- udelay(1);
-
- /*
- * much faster through the gPMC Register space
- */
- data = IORD16(ndata_reg_vma);
- } else {
- /*
- * 8 BIT gpmc NAND mode
- * CMD1, CA1, CA2, CMD2
- */
- addr16 = 0x0c00 | reg_addr8;
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR16(naddr_reg_vma, addr16);
- IOWR8(ncmd_reg_vma, 0xE0);
- udelay(1);
- data = IORD16(ndata_reg_vma);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&wrreg_usage_cnt);
- local_irq_restore(flags);
-
- return data;
-}
-
-
-/*
- * This function must be defined to write a register within the WestBridge
- * device. The addr value is the address of the register to write with
- * respect to the base address of the WestBridge device.
- */
-void cy_as_hal_write_register(
- cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data)
-{
- ast_p_nand_casdi_write((u8)addr, data);
-}
-
-/*
- * This function must be defined to read a register from the WestBridge
- * device. The addr value is the address of the register to read with
- * respect to the base address of the WestBridge device.
- */
-uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
-{
- uint16_t data = 0 ;
-
- /*
- * READ ASTORIA REGISTER USING CASDO
- */
- data = ast_p_nand_casdo_read((u8)addr);
-
- return data ;
-}
-
-/*
- * preps Ep pointers & data counters for next packet
- * (fragment of the request) xfer returns true if
- * there is a next transfer, and false if all bytes in
- * current request have been xfered
- */
-static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
-{
-
- if (!end_points[ep].sg_list_enabled) {
- /*
- * no further transfers for non storage EPs
- * (like EP2 during firmware download, done
- * in 64 byte chunks)
- */
- if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
- DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
- __func__, end_points[ep].req_length, ep);
-
- /*
- * no more transfers, we are done with the request
- */
- return false;
- }
-
- /*
- * calculate size of the next DMA xfer, corner
- * case for non-storage EPs where transfer size
- * is not egual N * HAL_DMA_PKT_SZ xfers
- */
- if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
- >= HAL_DMA_PKT_SZ) {
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
- } else {
- /*
- * that would be the last chunk less
- * than P-port max size
- */
- end_points[ep].dma_xfer_sz = end_points[ep].req_length -
- end_points[ep].req_xfer_cnt;
- }
-
- return true;
- }
-
- /*
- * for SG_list assisted dma xfers
- * are we done with current SG ?
- */
- if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
- /*
- * was it the Last SG segment on the list ?
- */
- if (sg_is_last(end_points[ep].sg_p)) {
- DBGPRN("<1> %s: EP:%d completed,"
- "%d bytes xfered\n",
- __func__,
- ep,
- end_points[ep].req_xfer_cnt
- );
-
- return false;
- } else {
- /*
- * There are more SG segments in current
- * request's sg list setup new segment
- */
-
- end_points[ep].seg_xfer_cnt = 0;
- end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
- /* set data pointer for next DMA sg transfer*/
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- DBGPRN("<1> %s new SG:_va:%p\n\n",
- __func__, end_points[ep].data_p);
- }
-
- }
-
- /*
- * for sg list xfers it will always be 512 or 1024
- */
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
-
- /*
- * next transfer is required
- */
-
- return true;
-}
-
-/*
- * Astoria DMA read request, APP_CPU reads from WB ep buffer
- */
-static void cy_service_e_p_dma_read_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- uint16_t v, size;
- void *dptr;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
-
- /*
- * get the XFER size frtom WB eP DMA REGISTER
- */
- v = cy_as_hal_read_register(tag, ep_dma_reg);
-
- /*
- * amount of data in EP buff in bytes
- */
- size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
-
- /*
- * memory pointer for this DMA packet xfer (sub_segment)
- */
- dptr = end_points[ep].data_p;
-
- DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
- ep,
- size,
- intr_sequence_num,
- dptr
- );
-
- cy_as_hal_assert(size != 0);
-
- if (size) {
- /*
- * the actual WB-->OMAP memory "soft" DMA xfer
- */
- p_nand_lbd_read(col_addr, row_addr, size, dptr);
- }
-
- /*
- * clear DMAVALID bit indicating that the data has been read
- */
- cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
-
- /*
- * pre-advance data pointer (if it's outside sg
- * list it will be reset anyway
- */
- end_points[ep].data_p += size;
-
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * we have more data to read in this request,
- * setup next dma packet due tell WB how much
- * data we are going to xfer next
- */
- v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, ep_dma_reg, v);
- } else {
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
- end_points[ep].req_xfer_cnt);
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * omap_cpu needs to transfer data to ASTORIA EP buffer
- */
-static void cy_service_e_p_dma_write_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- uint16_t addr;
- uint16_t v = 0;
- uint32_t size;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- void *dptr;
-
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- /*
- * note: size here its the size of the dma transfer could be
- * anything > 0 && < P_PORT packet size
- */
- size = end_points[ep].dma_xfer_sz ;
- dptr = end_points[ep].data_p ;
-
- /*
- * perform the soft DMA transfer, soft in this case
- */
- if (size)
- p_nand_lbd_write(col_addr, row_addr, size, dptr);
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
- /*
- * pre-advance data pointer
- * (if it's outside sg list it will be reset anyway)
- */
- end_points[ep].data_p += size;
-
- /*
- * now clear DMAVAL bit to indicate we are done
- * transferring data and that the data can now be
- * sent via USB to the USB host, sent to storage,
- * or used internally.
- */
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
- cy_as_hal_write_register(tag, addr, size) ;
-
- /*
- * finally, tell the USB subsystem that the
- * data is gone and we can accept the
- * next request if one exists.
- */
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * There is more data to go. Re-init the WestBridge DMA side
- */
- v = end_points[ep].dma_xfer_sz |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v) ;
- } else {
-
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- /*
- * this callback will wake up the process that might be
- * sleeping on the EP which data is being transferred
- */
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * HANDLE DRQINT from Astoria (called in AS_Intr context
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
-{
- uint16_t v ;
- static uint8_t service_ep = 2 ;
-
- /*
- * We've got DRQ INT, read DRQ STATUS Register */
- v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_DRQ) ;
-
- if (v == 0) {
-#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
-#endif
- return;
- }
-
- /*
- * Now, pick a given DMA request to handle, for now, we just
- * go round robin. Each bit position in the service_mask
- * represents an endpoint from EP2 to EP15. We rotate through
- * each of the endpoints to find one that needs to be serviced.
- */
- while ((v & (1 << service_ep)) == 0) {
-
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
- if (end_points[service_ep].type == cy_as_hal_write) {
- /*
- * handle DMA WRITE REQUEST: app_cpu will
- * write data into astoria EP buffer
- */
- cy_service_e_p_dma_write_request(dev_p, service_ep) ;
- } else if (end_points[service_ep].type == cy_as_hal_read) {
- /*
- * handle DMA READ REQUEST: cpu will
- * read EP buffer from Astoria
- */
- cy_service_e_p_dma_read_request(dev_p, service_ep) ;
- }
-#ifndef WESTBRIDGE_NDEBUG
- else
- cy_as_hal_print_message("cyashalomap:interrupt,"
- " w/o pending DMA job,"
- "-check DRQ_MASK logic\n") ;
-#endif
-
- /*
- * Now bump the EP ahead, so other endpoints get
- * a shot before the one we just serviced
- */
- if (end_points[service_ep].type == cy_as_hal_none) {
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
-}
-
-void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
-{
- DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
- if (end_points[ep].pending)
- cy_as_hal_write_register(tag,
- CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
-
- end_points[ep].buffer_valid = cy_false ;
- end_points[ep].type = cy_as_hal_none;
-}
-
-/*
- * enables/disables SG list assisted DMA xfers for the given EP
- * sg_list assisted XFERS can use physical addresses of mem pages in case if the
- * xfer is performed by a h/w DMA controller rather then the CPU on P port
- */
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
-{
- end_points[ep].sg_list_enabled = sg_xfer_enabled;
- DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
- ep, end_points[ep].sg_list_enabled);
-}
-EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
-
-/*
- * This function must be defined to transfer a block of data to
- * the WestBridge device. This function can use the burst write
- * (DMA) capabilities of WestBridge to do this, or it can just copy
- * the data using writes.
- */
-void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr = 0 ;
- uint16_t v = 0;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem. Another DMA request
- * is coming down before the previous one has completed.
- */
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_write ;
- end_points[ep].pending = cy_true;
-
- /*
- * total length of the request
- */
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * smaller xfers for non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- /*
- * check the EP transfer mode uses sg_list rather then a memory buffer
- * block devices pass it to the HAL, so the hAL could get to the real
- * physical address for each segment and set up a DMA controller
- * hardware ( if there is one)
- */
- if (end_points[ep].sg_list_enabled) {
- /*
- * buf - pointer to the SG list
- * data_p - data pointer to the 1st DMA segment
- * seg_xfer_cnt - keeps track of N of bytes sent in current
- * sg_list segment
- * req_xfer_cnt - keeps track of the total N of bytes
- * transferred for the request
- */
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].req_xfer_cnt = 0;
-
-#ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
- "req_sz:%d, maxsz:%d\n",
- __func__,
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
-#endif
-
- } else {
- /*
- * setup XFER for non sg_list assisted EPs
- */
-
- #ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("<1>%s non storage or sz < 512:"
- "EP:%d, sz:%d\n", __func__, ep, size);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * will keep track No of bytes xferred for the request
- */
- end_points[ep].req_xfer_cnt = 0;
- }
-
- /*
- * Tell WB we are ready to send data on the given endpoint
- */
- v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
- | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- cy_as_hal_write_register(tag, addr, v) ;
-}
-
-/*
- * This function must be defined to transfer a block of data from
- * the WestBridge device. This function can use the burst read
- * (DMA) capabilities of WestBridge to do this, or it can just
- * copy the data using reads.
- */
-void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr ;
- uint16_t v ;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem.
- * Another DMA request is coming down before the
- * previous one has completed. we should not get
- * new requests if current is still in process
- */
-
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
-
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_read ;
- end_points[ep].pending = cy_true;
- end_points[ep].req_xfer_cnt = 0;
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * so that we could handle small xfers on in case
- * of non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- if (end_points[ep].sg_list_enabled) {
- /*
- * Handle sg-list assisted EPs
- * seg_xfer_cnt - keeps track of N of sent packets
- * buf - pointer to the SG list
- * data_p - data pointer for the 1st DMA segment
- */
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
-
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
- "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
- #endif
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- } else {
- /*
- * Non sg list EP passed void *buf rather then scatterlist *sg
- */
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("%s:non-sg_list EP:%d,"
- "RQ_sz:%d, maxsz:%d\n",
- __func__, ep, size, maxsize);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * Program the EP DMA register for Storage endpoints only.
- */
- if (is_storage_e_p(ep)) {
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- }
- }
-}
-
-/*
- * This function must be defined to allow the WB API to
- * register a callback function that is called when a
- * DMA transfer is complete.
- */
-void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb)
-{
- DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
- __func__, (uint32_t)cb);
- callback = cb ;
-}
-
-/*
- * This function must be defined to return the maximum size of
- * DMA request that can be handled on the given endpoint. The
- * return value should be the maximum size in bytes that the DMA
- * module can handle.
- */
-uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep)
-{
- /*
- * Storage reads and writes are always done in 512 byte blocks.
- * So, we do the count handling within the HAL, and save on
- * some of the data transfer delay.
- */
- if ((ep == CYASSTORAGE_READ_EP_NUM) ||
- (ep == CYASSTORAGE_WRITE_EP_NUM)) {
- /* max DMA request size HAL can handle by itself */
- return CYASSTORAGE_MAX_XFER_SIZE;
- } else {
- /*
- * For the USB - Processor endpoints, the maximum transfer
- * size depends on the speed of USB operation. So, we use
- * the following constant to indicate to the API that
- * splitting of the data into chunks less that or equal to
- * the max transfer size should be handled internally.
- */
-
- /* DEFINED AS 0xffffffff in cyasdma.h */
- return CY_AS_DMA_MAX_SIZE_HW_SIZE;
- }
-}
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the WestBridge device. Generally this is done via a GPIO of
- * some type.
- */
-cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
-{
- /*
- * Not supported as of now.
- */
- return cy_false ;
-}
-
-void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
-{
- cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
- cy_as_hal_print_message("please check the input voltage levels");
- cy_as_hal_print_message("and clock, and restart the system\n") ;
-}
-
-/*
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- */
-
-/*
- * This function is required by the API to allocate memory.
- * This function is expected to work exactly like malloc().
- */
-void *cy_as_hal_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required by the API to free memory allocated
- * with CyAsHalAlloc(). This function is'expected to work exacly
- * like free().
- */
-void cy_as_hal_free(void *mem_p)
-{
- kfree(mem_p) ;
-}
-
-/*
- * Allocator that can be used in interrupt context.
- * We have to ensure that the kmalloc call does not
- * sleep in this case.
- */
-void *cy_as_hal_c_b_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required to set a block of memory to a
- * specific value. This function is expected to work exactly
- * like memset()
- */
-void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
-{
- memset(ptr, value, cnt) ;
-}
-
-/*
- * This function is expected to create a sleep channel.
- * The data structure that represents the sleep channel object
- * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
- * passed as a pointer, and allpocated by the caller
- * (typically as a local var on the stack) "Create" word should read as
- * "SleepOn", this func doesn't actually create anything
- */
-cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- init_waitqueue_head(&channel->wq) ;
- return cy_true ;
-}
-
-/*
- * for this particular HAL it doesn't actually destroy anything
- * since no actual sleep object is created in CreateSleepChannel()
- * sleep channel is given by the pointer in the argument.
- */
-cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- return cy_true ;
-}
-
-/*
- * platform specific wakeable Sleep implementation
- */
-cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
-{
- wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
- return cy_true ;
-}
-
-/*
- * wakes up the process waiting on the CHANNEL
- */
-cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
-{
- wake_up_interruptible_all(&channel->wq);
- return cy_true ;
-}
-
-uint32_t cy_as_hal_disable_interrupts()
-{
- if (0 == intr__enable)
- ;
-
- intr__enable++ ;
- return 0 ;
-}
-
-void cy_as_hal_enable_interrupts(uint32_t val)
-{
- intr__enable-- ;
- if (0 == intr__enable)
- ;
-}
-
-/*
- * Sleep atleast 150ns, cpu dependent
- */
-void cy_as_hal_sleep150(void)
-{
- uint32_t i, j;
-
- j = 0;
- for (i = 0; i < 1000; i++)
- j += (~i);
-}
-
-void cy_as_hal_sleep(uint32_t ms)
-{
- cy_as_hal_sleep_channel channel;
-
- cy_as_hal_create_sleep_channel(&channel) ;
- cy_as_hal_sleep_on(&channel, ms) ;
- cy_as_hal_destroy_sleep_channel(&channel) ;
-}
-
-cy_bool cy_as_hal_is_polling()
-{
- return cy_false;
-}
-
-void cy_as_hal_c_b_free(void *ptr)
-{
- cy_as_hal_free(ptr);
-}
-
-/*
- * suppose to reinstate the astoria registers
- * that may be clobbered in sleep mode
- */
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
- (void) is_standby_wakeup;
-}
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
-}
-
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
-{
- /*
- * we are in asynchronous mode. so no need to handle this
- */
- return true;
-}
-
-/*
- * init OMAP h/w resources
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug)
-{
- cy_as_omap_dev_kernel *dev_p ;
- int i;
- u16 data16[4];
- u8 pncfg_reg;
-
- /*
- * No debug mode support through argument as of now
- */
- (void)debug;
-
- DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
-
- /*
- * Initialize the HAL level endpoint DMA data.
- */
- for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
- end_points[i].data_p = 0 ;
- end_points[i].pending = cy_false ;
- end_points[i].size = 0 ;
- end_points[i].type = cy_as_hal_none ;
- end_points[i].sg_list_enabled = cy_false;
-
- /*
- * by default the DMA transfers to/from the E_ps don't
- * use sg_list that implies that the upper devices like
- * blockdevice have to enable it for the E_ps in their
- * initialization code
- */
- }
-
- /*
- * allocate memory for OMAP HAL
- */
- dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
- sizeof(cy_as_omap_dev_kernel)) ;
- if (dev_p == 0) {
- cy_as_hal_print_message("out of memory allocating OMAP"
- "device structure\n") ;
- return 0 ;
- }
-
- dev_p->m_sig = CY_AS_OMAP_KERNEL_HAL_SIG;
-
- /*
- * initialize OMAP hardware and StartOMAPKernelall gpio pins
- */
- dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
-
- /*
- * Now perform a hard reset of the device to have
- * the new settings take effect
- */
- __gpio_set_value(AST_WAKEUP, 1);
-
- /*
- * do Astoria h/w reset
- */
- DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
-
- /*
- * NEGATIVE PULSE on RST pin
- */
- __gpio_set_value(AST_RESET, 0);
- mdelay(1);
- __gpio_set_value(AST_RESET, 1);
- mdelay(50);
-
- /*
- * note AFTER reset PNAND interface is 8 bit mode
- * so if gpmc Is configured in 8 bit mode upper half will be FF
- */
- pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
-#ifdef PNAND_16BIT_MODE
-
- /*
- * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
- */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
-
- /*
- * now in order to continue to talk to astoria
- * sw OMAP GPMC into 16 bit mode as well
- */
- cy_as_hal_gpmc_enable_16bit_bus(cy_true);
-#else
- /* Astoria and GPMC are already in 8 bit mode, jsut initialize PNAND_CFG */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
-#endif
-
- /*
- * NOTE: if you want to capture bus activity on the LA,
- * don't use printks in between the activities you want to capture.
- * prinks may take milliseconds, and the data of interest
- * will fall outside the LA capture window/buffer
- */
- data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
- data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
- if (data16[0] != 0xA200) {
- /*
- * astoria device is not found
- */
- printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
- printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
- data16[0], data16[0]);
- goto bus_acc_error;
- }
-
- cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
- "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
- "PNAND_CFG after RST:%4.4x\n "
- "CY_AS_MEM_PNAND_CFG"
- "after cfg_wr:%4.4x\n\n",
- data16[0], pncfg_reg, data16[1]);
-
- dev_p->thread_flag = 1 ;
- spin_lock_init(&int_lock) ;
- dev_p->m_next_p = m_omap_list_p ;
-
- m_omap_list_p = dev_p ;
- *tag = dev_p;
-
- cy_as_hal_configure_interrupts((void *)dev_p);
-
- cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
- ", kernel HZ:%d\n", dev_p, HZ);
-
- /*
- *make processor to storage endpoints SG assisted by default
- */
- cy_as_hal_set_ep_dma_mode(4, true);
- cy_as_hal_set_ep_dma_mode(8, true);
-
- return 1 ;
-
- /*
- * there's been a NAND bus access error or
- * astoria device is not connected
- */
-bus_acc_error:
- /*
- * at this point hal tag hasn't been set yet
- * so the device will not call omap_stop
- */
- cy_as_hal_omap_hardware_deinit(dev_p);
- cy_as_hal_free(dev_p) ;
- return 0;
-}
-
-#else
-/*
- * Some compilers do not like empty C files, so if the OMAP hal is not being
- * compiled, we compile this single function. We do this so that for a
- * given target HAL there are not multiple sources for the HAL functions.
- */
-void my_o_m_a_p_kernel_hal_dummy_function(void)
-{
-}
-
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 2010-09-01 16:01:14.000000000 -0700
@@ -0,0 +1,2462 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+/*
+ * uncomment to enable 16bit pnand interface
+ */
+#define PNAND_16BIT_MODE
+
+/*
+ * selects one of 3 versions of pnand_lbd_read()
+ * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
+ * reads through the gpmc CONTROLLER REGISTERS
+ * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
+ * same speed as the above
+ * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
+ */
+#define PNAND_LBD_READ_NO_PFE
+/* #define ENABLE_GPMC_PF_ENGINE */
+/* #define PFE_LBD_READ_V2 */
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+ #define CYASSTORAGE_WRITE_EP_NUM (4)
+ #define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+/*
+ * persistant, stores current GPMC interface cfg mode
+ */
+static uint8_t pnand_16bit;
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
+ */
+#define CASDO 0x05
+#define CASDI 0x85
+#define RDPAGE_B1 0x00
+#define RDPAGE_B2 0x30
+#define PGMPAGE_B1 0x80
+#define PGMPAGE_B2 0x10
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+static u32 csa_phy;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
+ */
+static u32 gpmc_data_vma;
+static u32 ndata_reg_vma;
+static u32 ncmd_reg_vma;
+static u32 naddr_reg_vma;
+
+/*
+ * fwd declarations
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static inline u16 __attribute__((always_inline))
+ ast_p_nand_casdo_read(u8 reg_addr8);
+static inline void __attribute__((always_inline))
+ ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ cy_as_hal_print_message(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg ");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ cy_as_hal_print_message(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+#define BLKSZ_4K 0x1000
+
+/*
+ * switch GPMC DATA bus mode
+ */
+void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
+{
+ uint32_t tmp32;
+
+ /*
+ * disable gpmc CS4 operation 1st
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
+ GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * GPMC NAND data bus can be 8 or 16 bit wide
+ */
+ if (dbus16_enabled) {
+ DBGPRN("enabling 16 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2) |
+ GPMC_CONFIG1_DEVICESIZE_16)
+ );
+ } else {
+ DBGPRN(KERN_INFO "enabling 8 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2))
+ );
+ }
+
+ /*
+ * re-enable astoria CS operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ *remember the state
+ */
+ pnand_16bit = dbus16_enabled;
+}
+
+static int cy_as_hal_gpmc_init(void)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ cy_as_hal_print_message(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ //gpmc_base = phys_to_virt(OMAP34XX_GPMC_BASE);
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, BLKSZ_4K);
+ cy_as_hal_print_message(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ /*
+ * these are globals are full VMAs of the gpmc_base above
+ */
+ ncmd_reg_vma = GPMC_VMA(AST_GPMC_NAND_CMD);
+ naddr_reg_vma = GPMC_VMA(AST_GPMC_NAND_ADDR);
+ ndata_reg_vma = GPMC_VMA(AST_GPMC_NAND_DATA);
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ cy_as_hal_print_message(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ csa_phy, AST_GPMC_CS);
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ csa_phy);
+ goto out_free_cs;
+ }
+
+ cy_as_hal_print_message(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
+ if (!gpmc_data_vma) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", csa_phy);
+
+ goto out_release_mem_region;
+ }
+ cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ csa_phy, gpmc_data_vma);
+
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_o_n;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_o_n;
+ timings.adv_rd_off = WB_GPMC_BUSCYC_t;
+ timings.adv_wr_off = WB_GPMC_BUSCYC_t;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_o_n;
+ timings.oe_off = WB_GPMC_OE_t_o_f_f;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_o_n;
+ timings.we_off = WB_GPMC_WE_t_o_f_f;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ /*
+ * DISABLE cs4, NOTE GPMC REG7 is already configured
+ * at this point by gpmc_cs_request
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
+ ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (AS_CS_MASK | AS_CS_BADDR));
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ /*
+ * ENABLE astoria cs operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ cy_as_hal_print_message(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ cy_as_hal_print_message(KERN_INFO"pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ cy_as_hal_print_message(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ cy_as_hal_print_message(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(void)
+{
+ int i, err;
+
+ cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init();
+ if (err < 0)
+ cy_as_hal_print_message(KERN_INFO"gpmc init failed:%d", err);
+
+ cy_as_hal_config_c_s_mux();
+
+ return gpmc_data_vma;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (gpmc_data_vma != 0)
+ iounmap((void *)gpmc_data_vma);
+
+ if (csa_phy != 0)
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_PNAND_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+/*
+ * GPMC NAND command+addr write phase
+ */
+static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
+{
+ /*
+ * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
+ */
+ u32 tmpa32 = ((row_addr << 16) | col_addr);
+ u8 RA2 = (u8)(row_addr >> 16);
+
+ if (!pnand_16bit) {
+ /*
+ * GPMC PNAND 8bit BUS
+ */
+ /*
+ * CMD1
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+
+ /*
+ *pnand bus: <CA0,CA1,RA0,RA1>
+ */
+ IOWR32(naddr_reg_vma, tmpa32);
+
+ /*
+ * <RA2> , always zero
+ */
+ IOWR8(naddr_reg_vma, RA2);
+
+ } else {
+ /*
+ * GPMC PNAND 16bit BUS , in 16 bit mode CMD
+ * and ADDR sent on [d7..d0]
+ */
+ uint8_t CA0, CA1, RA0, RA1;
+ CA0 = tmpa32 & 0x000000ff;
+ CA1 = (tmpa32 >> 8) & 0x000000ff;
+ RA0 = (tmpa32 >> 16) & 0x000000ff;
+ RA1 = (tmpa32 >> 24) & 0x000000ff;
+
+ /*
+ * can't use 32 bit writes here omap will not serialize
+ * them to lower half in16 bit mode
+ */
+
+ /*
+ *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+ IOWR8(naddr_reg_vma, CA0);
+ IOWR8(naddr_reg_vma, CA1);
+ IOWR8(naddr_reg_vma, RA0);
+ IOWR8(naddr_reg_vma, RA1);
+ IOWR8(naddr_reg_vma, RA2);
+ }
+}
+
+/*
+ * spin until r/b goes high
+ */
+inline int wait_rn_b_high(void)
+{
+ u32 w_spins = 0;
+
+ /*
+ * TODO: note R/b may go low here, need to spin until high
+ * while (omap_get_gpio_datain(AST_RnB) == 0) {
+ * w_spins++;
+ * }
+ * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ *
+ * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ * w_spins++;
+ * }
+ * printk("<1>RnB=0!:%d\n",w_spins);
+ * }
+ */
+ return w_spins;
+}
+
+#ifdef ENABLE_GPMC_PF_ENGINE
+/* #define PFE_READ_DEBUG
+ * PNAND block read with OMAP PFE enabled
+ * status: Not tested, NW, broken , etc
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint8_t bytes_in_fifo;
+
+ /* debug vars*/
+#ifdef PFE_READ_DEBUG
+ uint32_t loop_limit;
+ uint16_t bytes_read = 0;
+#endif
+
+ /*
+ * configure the prefetch engine
+ */
+ uint32_t tmp32;
+ uint32_t pfe_status;
+
+ /*
+ * DISABLE GPMC CS4 operation 1st, this is
+ * in case engine is be already disabled
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+
+#ifdef PFE_READ_DEBUG
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
+ if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
+ printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
+ tmp32, GPMC_PREFETCH_CONFIG1_VAL);
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
+ }
+
+ /*
+ *sanity check 2
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
+ if (tmp32 != (count))
+ printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
+ "!= VAL written:%d\n", tmp32, count);
+#endif
+
+ /*
+ * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
+ * on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * start the prefetch engine
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ while (1) {
+ /*
+ * GPMC PFE service loop
+ */
+ do {
+ /*
+ * spin until PFE fetched some
+ * PNAND bus words in the FIFO
+ */
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ bytes_in_fifo = (pfe_status >> 24) & 0x7f;
+ } while (bytes_in_fifo == 0);
+
+ /* whole 32 bit words in fifo */
+ w32cnt = bytes_in_fifo >> 2;
+
+#if 0
+ /*
+ *NOTE: FIFO_PTR indicates number of NAND bus words bytes
+ * already received in the FIFO and available to be read
+ * by DMA or MPU whether COUNTVAL indicates number of BUS
+ * words yet to be read from PNAND bus words
+ */
+ printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
+ pfe_status, bytes_in_fifo,
+ (pfe_status & 0x3fff), w32cnt);
+#endif
+
+ while (w32cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ if ((pfe_status & 0x3fff) == 0) {
+ /*
+ * PFE acc angine done, there still may be data leftover
+ * in the FIFO re-read FIFO BYTE counter (check for
+ * leftovers from 32 bit read accesses above)
+ */
+ bytes_in_fifo = (IORD32(
+ GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ /*
+ * NOTE we may still have one word left in the fifo
+ * read it out
+ */
+ ptr8 = ptr32;
+ switch (bytes_in_fifo) {
+
+ case 0:
+ /*
+ * nothing to do we already read the
+ * FIFO out with 32 bit accesses
+ */
+ break;
+ case 1:
+ /*
+ * this only possible
+ * for 8 bit pNAND only
+ */
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 2:
+ /*
+ * this one can occur in either modes
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ break;
+
+ case 3:
+ /*
+ * this only possible for 8 bit pNAND only
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ ptr8 += 2;
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 4:
+ /*
+ * shouldn't happen, but has been seen
+ * in 8 bit mode
+ */
+ *ptr32 = IORD32(gpmc_data_vma);
+ break;
+
+ default:
+ printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
+ bytes_in_fifo);
+ break;
+ }
+ /*
+ * read is completed, get out of the while(1) loop
+ */
+ break;
+ }
+ }
+}
+#endif
+
+#ifdef PFE_LBD_READ_V2
+/*
+ * PFE engine assisted reads with the 64 byte blocks
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint8_t rd_cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint16_t reminder;
+ uint32_t pfe_status;
+
+ /*
+ * ISSUE PNAND CMD+ADDR
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * setup PFE block
+ * count - OMAP number of bytes to access on pnand bus
+ */
+
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ do {
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
+ rd_cnt = pfe_status >> (24+2);
+
+ while (rd_cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ } while (pfe_status & 0x3fff);
+
+ /*
+ * read out the leftover
+ */
+ ptr8 = ptr32;
+ rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ while (rd_cnt--)
+ *ptr8++ = IORD8(gpmc_data_vma);
+}
+#endif
+
+#ifdef PNAND_LBD_READ_NO_PFE
+/*
+ * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
+ * the original working code, works at max speed for 8 bit xfers
+ * for 16 bit the bus diagram has gaps
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint16_t *ptr16;
+ uint16_t remainder;
+
+ DBGPRN("<1> %s(): NO_PFE\n", __func__);
+
+ ptr32 = buff;
+ /* number of whole 32 bit words in the transfer */
+ w32cnt = count >> 2;
+
+ /* remainder, in bytes(0..3) */
+ remainder = count & 03;
+
+ /*
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * read data by 32 bit chunks
+ */
+ while (w32cnt--)
+ *ptr32++ = IORD32(ndata_reg_vma);
+
+ /*
+ * now do the remainder(it can be 0, 1, 2 or 3)
+ * same code for both 8 & 16 bit bus
+ * do 1 or 2 MORE words
+ */
+ ptr16 = (uint16_t *)ptr32;
+
+ switch (remainder) {
+ case 1:
+ /* read one 16 bit word
+ * IN 8 BIT WE NEED TO READ even number of bytes
+ */
+ case 2:
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ *ptr16++ = IORD16(ndata_reg_vma);
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ default:
+ /*
+ * remainder is 0
+ */
+ break;
+ }
+}
+#endif
+
+/*
+ * uses LBD mode to write N bytes into astoria
+ * Status: Working, however there are 150ns idle
+ * timeafter every 2 (16 bit or 4(8 bit) bus cycles
+ */
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint16_t remainder;
+ uint8_t *ptr8;
+ uint16_t *ptr16;
+ uint32_t *ptr32;
+
+ remainder = count & 03;
+ w32cnt = count >> 2;
+ ptr32 = buff;
+ ptr8 = buff;
+
+ /*
+ * send: CMDB1, CA0,CA1,RA0,RA1,RA2
+ */
+ nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
+
+ /*
+ * blast the data out in 32bit chunks
+ */
+ while (w32cnt--)
+ IOWR32(ndata_reg_vma, *ptr32++);
+
+ /*
+ * do the reminder if there is one
+ * same handling for both 8 & 16 bit pnand: mode
+ */
+ ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
+
+ switch (remainder) {
+ case 1:
+ /*
+ * read one 16 bit word
+ */
+ case 2:
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ IOWR16(ndata_reg_vma, *ptr16++);
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+ default:
+ /*
+ * reminder is 0
+ */
+ break;
+ }
+ /*
+ * finally issue a PGM cmd
+ */
+ IOWR8(ncmd_reg_vma, PGMPAGE_B2);
+}
+
+/*
+ * write Astoria register
+ */
+static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
+{
+ unsigned long flags;
+ u16 addr16;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t rdreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&rdreg_usage_cnt) != 0) {
+ cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
+ "* cy_as_hal_write_register usage:%d\n",
+ atomic_read(&rdreg_usage_cnt));
+ }
+
+ atomic_inc(&rdreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ */
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+
+ /*
+ * this should be sent on the 16 bit bus
+ */
+ IOWR16(ndata_reg_vma, data);
+ } else {
+ /*
+ * 8 bit nand mode GPMC will automatically
+ * seriallize 16bit or 32 bit writes into
+ * 8 bit onesto the lower 8 bit in LE order
+ */
+ addr16 = 0x0c00 | reg_addr8;
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR16(ndata_reg_vma, data);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&rdreg_usage_cnt);
+ local_irq_restore(flags);
+}
+
+
+/*
+ * read astoria register via pNAND interface
+ */
+static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
+{
+ u16 data;
+ u16 addr16;
+ unsigned long flags;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t wrreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&wrreg_usage_cnt) != 0) {
+ /*
+ * if it gets here ( from other threads), this function needs
+ * need spin_lock_irq save() protection
+ */
+ cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
+ "cy_as_hal_write_register usage:%d\n",
+ atomic_read(&wrreg_usage_cnt));
+ }
+ atomic_inc(&wrreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ * CMD1, CA1, CA2,
+ */
+
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+ IOWR8(ncmd_reg_vma, 0x00E0);
+
+ udelay(1);
+
+ /*
+ * much faster through the gPMC Register space
+ */
+ data = IORD16(ndata_reg_vma);
+ } else {
+ /*
+ * 8 BIT gpmc NAND mode
+ * CMD1, CA1, CA2, CMD2
+ */
+ addr16 = 0x0c00 | reg_addr8;
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR8(ncmd_reg_vma, 0xE0);
+ udelay(1);
+ data = IORD16(ndata_reg_vma);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&wrreg_usage_cnt);
+ local_irq_restore(flags);
+
+ return data;
+}
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ ast_p_nand_casdi_write((u8)addr, data);
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+
+ /*
+ * READ ASTORIA REGISTER USING CASDO
+ */
+ data = ast_p_nand_casdo_read((u8)addr);
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+ /* this is not defined currently*/
+ #ifdef WESTBRIDGE_PNAND_HAL_PREP_CHANGE
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+ #endif
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, size;
+ void *dptr;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = end_points[ep].data_p;
+
+ DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
+ ep,
+ size,
+ intr_sequence_num,
+ dptr
+ );
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * the actual WB-->OMAP memory "soft" DMA xfer
+ */
+ p_nand_lbd_read(col_addr, row_addr, size, dptr);
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0;
+ uint32_t size;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ void *dptr;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size)
+ p_nand_lbd_write(col_addr, row_addr, size, dptr);
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if(state) {
+ cy_as_hal_gpmc_enable_16bit_bus(cy_false);
+ __gpio_set_value(AST_WAKEUP, 1);
+ } else {
+ __gpio_set_value(AST_WAKEUP, 0);
+ }
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ (void) tag;
+ (void) is_standby_wakeup;
+
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ u8 pncfg_reg;
+
+ /*
+ * No debug mode support through argument as of now
+ */
+ (void)debug;
+
+ DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ;
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /*
+ * allocate memory for OMAP HAL
+ */
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_PNAND_HAL_SIG;
+
+ /*
+ * initialize OMAP hardware and StartOMAPKernelall gpio pins
+ */
+ dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
+
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+ /*
+ * note AFTER reset PNAND interface is 8 bit mode
+ * so if gpmc Is configured in 8 bit mode upper half will be FF
+ */
+ pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+#ifdef PNAND_16BIT_MODE
+
+ /*
+ * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
+ */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+
+ /*
+ * now in order to continue to talk to astoria
+ * sw OMAP GPMC into 16 bit mode as well
+ */
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+#else
+ /* Astoria and GPMC are already in 8 bit mode, jsut initialize PNAND_CFG */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
+#endif
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
+ data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
+ printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
+ data16[0], data16[0]);
+ goto bus_acc_error;
+ }
+
+ cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "PNAND_CFG after RST:%4.4x\n "
+ "CY_AS_MEM_PNAND_CFG"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], pncfg_reg, data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 2010-08-27 17:50:08.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_cram.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+ /* Debug printing enabled */
+
+ #define DBGPRN(...) printk(__VA_ARGS__)
+ #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+ /** NO DEBUG PRINTING **/
+ #define DBGPRN(...)
+ #define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 2010-09-01 16:29:59.000000000 -0700
@@ -0,0 +1,565 @@
+/*
+ OMAP3430 ZOOM MDK astoria interface defs(cyasmemmap.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+#include <mach/mux.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+
+#ifndef _INCLUDED_CYASMEMMAP_H_
+#define _INCLUDED_CYASMEMMAP_H_
+
+/* defines copied from OMAP kernel branch */
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
+#define OMAP34XX_MUX_MODE0 0
+#define OMAP34XX_MUX_MODE4 4
+#define OMAP3_INPUT_EN (1 << 8)
+#define OMAP34XX_PIN_INPUT_PULLUP (OMAP2_PULL_ENA | OMAP3_INPUT_EN \
+ | OMAP2_PULL_UP)
+
+/*
+ * for OMAP3430 <-> astoria : ADmux mode, 8 bit data path
+ * WB Signal- OMAP3430 signal COMMENTS
+ * --------------------------- --------------------
+ * CS_L -GPMC_nCS4_GPIO_53 ZOOM I SOM board
+ * signal: up_nCS_A_EXT
+ * AD[7:0]-upD[7:0] buffered on the
+ * transposer board
+ * GPMC_ADDR
+ * [A8:A1]->upD[7:0]
+ * INT# -GPMC_nWP_GPIO_62
+ * DACK -N/C not conected
+ * WAKEUP-GPIO_167
+ * RESET-GPIO_126
+ * R/B -GPMC_WAIT2_GPIO_64
+ * -------------------------------------------
+ * The address range for nCS1B is 0x06000000 - 0x07FF FFFF.
+*/
+
+/*
+ *OMAP_ZOOM LEDS
+ */
+#define LED_0 156
+#define LED_1 128
+#define LED_2 64
+#define LED_3 60
+
+#define HIGH 1
+#define LOW 1
+
+/*
+ *omap GPIO number
+ */
+#define AST_WAKEUP 167
+#define AST_RESET 126
+#define AST__rn_b 64
+
+/*
+ * NOTE THIS PIN IS USED AS WP for OMAP NAND
+ */
+#define AST_INT 62
+
+/* N4_OMAP3430_GPMC_A1
+ M4_OMAP3430_GPMC_A2
+ L4_OMAP3430_GPMC_A3
+ K4_OMAP3430_GPMC_A4
+ T3_OMAP3430_GPMC_A5
+ R3_OMAP3430_GPMC_A6
+ N3_OMAP3430_GPMC_A7
+ M3_OMAP3430_GPMC_A8
+ * as an I/O, it is actually controlled by GPMC
+ */
+#define AST_CS 55
+
+
+/*
+ *GPMC prefetch engine
+ */
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG1 0x01E0
+
+ /*32 bytes for 16 bit pnand mode*/
+ #define PFE_THRESHOLD 31
+
+ /*
+ * bit fields
+ * PF_ACCESSMODE - 0 - read mode, 1 - write mode
+ * PF_DMAMODE - 0 - default only intr line signal will be generated
+ * PF_SYNCHROMODE - default 0 - engin will start access as soon as
+ * ctrl re STARTENGINE is set
+ * PF_WAITPINSEL - FOR synchro mode selects WAIT pin whch edge
+ * will be monitored
+ * PF_EN_ENGINE - 1- ENABLES ENGINE, but it needs to be started after
+ * that C ctrl reg bit 0
+ * PF_FIFO_THRESHOLD - FIFO threshhold in number of BUS(8 or 16) words
+ * PF_WEIGHTED_PRIO - NUM of cycles granted to PFE if RND_ROBIN
+ * prioritization is enabled
+ * PF_ROUND_ROBIN - if enabled, gives priority to other CS, but
+ * reserves NUM of cycles for PFE's turn
+ * PF_ENGIN_CS_SEL - GPMC CS assotiated with PFE function
+ */
+ #define PF_ACCESSMODE (0 << 0)
+ #define PF_DMAMODE (0 << 2)
+ #define PF_SYNCHROMODE (0 << 3)
+ #define PF_WAITPINSEL (0x0 << 4)
+ #define PF_EN_ENGINE (1 << 7)
+ #define PF_FIFO_THRESHOLD (PFE_THRESHOLD << 8)
+ #define PF_WEIGHTED_PRIO (0x0 << 16)
+ #define PF_ROUND_ROBIN (0 << 23)
+ #define PF_ENGIN_CS_SEL (AST_GPMC_CS << 24)
+ #define PF_EN_OPTIM_ACC (0 << 27)
+ #define PF_CYCLEOPTIM (0x0 << 28)
+
+#define GPMC_PREFETCH_CONFIG1_VAL (PF_ACCESSMODE | \
+ PF_DMAMODE | PF_SYNCHROMODE | \
+ PF_WAITPINSEL | PF_EN_ENGINE | \
+ PF_FIFO_THRESHOLD | PF_FIFO_THRESHOLD | \
+ PF_WEIGHTED_PRIO | PF_ROUND_ROBIN | \
+ PF_ENGIN_CS_SEL | PF_EN_OPTIM_ACC | \
+ PF_CYCLEOPTIM)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG2 0x01E4
+ /*
+ * bit fields
+ * 14 bit field NOTE this counts is also
+ * is in number of BUS(8 or 16) words
+ */
+ #define PF_TRANSFERCOUNT (0x000)
+
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONTROL 0x01EC
+ /*
+ * bit fields , ONLY BIT 0 is implemented
+ * PFWE engin must be programmed with this bit = 0
+ */
+ #define PFPW_STARTENGINE (1 << 0)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_STATUS 0x01F0
+
+ /* */
+ #define PFE_FIFO_THRESHOLD (1 << 16)
+
+/*
+ * GPMC posted write/prefetch engine end
+ */
+
+
+/*
+ * chip select number on GPMC ( 0..7 )
+ */
+#define AST_GPMC_CS 4
+
+/*
+ * not connected
+ */
+#define AST_DACK 00
+
+
+/*
+ * Physical address above the NAND flash
+ * we use CS For mapping in OMAP3430 RAM space use 0x0600 0000
+ */
+#define CYAS_DEV_BASE_ADDR (0x20000000)
+#define CYAS_DEV_MAX_ADDR (0xFF)
+#define CYAS_DEV_ADDR_RANGE (CYAS_DEV_MAX_ADDR << 1)
+
+ /* in CRAM or PSRAM mode OMAP A1..An wires-> Astoria, there is no A0 line */
+ #define CYAS_DEV_CALC_ADDR(cyas_addr) (cyas_addr << 1)
+ #define CYAS_DEV_CALC_EP_ADDR(ep) (ep << 1)
+
+/*
+ *OMAP3430 i/o access macros
+ */
+#define IORD32(addr) (*(volatile u32 *)(addr))
+#define IOWR32(addr, val) (*(volatile u32 *)(addr) = val)
+
+#define IORD16(addr) (*(volatile u16 *)(addr))
+#define IOWR16(addr, val) (*(volatile u16 *)(addr) = val)
+
+#define IORD8(addr) (*(volatile u8 *)(addr))
+#define IOWR8(addr, val) (*(volatile u8 *)(addr) = val)
+
+/*
+ * local defines for accessing to OMAP GPIO ***
+ */
+#define CTLPADCONF_BASE_ADDR 0x48002000
+#define CTLPADCONF_SIZE 0x1000
+
+#define GPIO1_BASE_ADDR 0x48310000
+#define GPIO2_BASE_ADDR 0x49050000
+#define GPIO3_BASE_ADDR 0x49052000
+#define GPIO4_BASE_ADDR 0x49054000
+#define GPIO5_BASE_ADDR 0x49056000
+#define GPIO6_BASE_ADDR 0x49058000
+#define GPIO_SPACE_SIZE 0x1000
+
+
+/*
+ * OMAP3430 GPMC timing for pNAND interface
+ */
+#define GPMC_BASE 0x6E000000
+#define GPMC_REGION_SIZE 0x1000
+#define GPMC_CONFIG_REG (0x50)
+
+/*
+ * bit 0 in the GPMC_CONFIG_REG
+ */
+#define NAND_FORCE_POSTED_WRITE_B 1
+
+/*
+ * WAIT2STATUS, must be (1 << 10)
+ */
+#define AS_WAIT_PIN_MASK (1 << 10)
+
+/*
+ * GPMC_CONFIG(reg number [1..7] [for chip sel CS[0..7])
+ */
+#define GPMC_CFG_REG(N, CS) ((0x60 + (4*(N-1))) + (0x30*CS))
+
+/*
+ * we get "gpmc_base" from kernel
+ */
+#define GPMC_VMA(offset) (gpmc_base + offset)
+
+/*
+ * GPMC CS space VMA start address
+ */
+#define GPMC_CS_VMA(offset) (gpmc_data_vma + offset)
+
+/*
+ * PAD_CFG mux space VMA
+ */
+#define PADCFG_VMA(offset) (iomux_vma + offset)
+
+/*
+ * CONFIG1: by default, sngle access, async r/w RD_MULTIPLE[30]
+ * WR_MULTIPLE[28]; GPMC_FCL_DIV[1:0]
+ */
+#define GPMC_FCLK_DIV ((0) << 0)
+
+/*
+ * ADDITIONAL DIVIDER FOR ALL TIMING PARAMS
+ */
+#define TIME_GRAN_SCALE ((0) << 4)
+
+/*
+ * for use by gpmc_set_timings api, measured in ns, not clocks
+ */
+#define WB_GPMC_BUSCYC_t (7 * 6)
+#define WB_GPMC_CS_t_on (0)
+#define WB_GPMC_ADV_t_on (1 * 6)
+#define WB_GPMC_ADV_t_off (3 * 6)
+#define WB_GPMC_OE_t_on (2 * 6)
+#define WB_GPMC_OE_t_off (6 * 6)
+#define WB_GPMC_WE_t_on (2 * 6)
+#define WB_GPMC_WE_t_off (6 * 6)
+#define WB_GPMC_RDS_ADJ (1 * 6)
+#define WB_GPMC_RD_t_a_c_c (WB_GPMC_OE_t_off + WB_GPMC_RDS_ADJ)
+#define WB_GPMC_WR_t_a_c_c (WB_GPMC_BUSCYC_t)
+
+#define DIR_OUT 0
+#define DIR_INP 1
+#define DRV_HI 1
+#define DRV_LO 0
+
+/*
+ * GPMC_CONFIG7[cs] register bit fields3
+ * AS_CS_MASK - 3 bit mask for A26,A25,A24,
+ * AS_CS_BADDR - 6 BIT VALUE A29 ...A24
+ * CSVALID_B - CSVALID bit on GPMC_CONFIG7[cs] register
+ */
+#define AS_CS_MASK (0X7 << 8)
+#define AS_CS_BADDR 0x02
+#define CSVALID_B (1 << 6)
+
+/*
+ * DEFINE OMAP34XX GPIO OFFSETS (should have been defined in kernel /arch
+ * these are offsets from the BASE_ADDRESS of the GPIO BLOCK
+ */
+#define GPIO_REVISION 0x000
+#define GPIO_SYSCONFIG 0x010
+#define GPIO_SYSSTATUS1 0x014
+#define GPIO_IRQSTATUS1 0x0183
+#define GPIO_IRQENABLE1 0x01C
+#define GPIO_IRQSTATUS2 0x028
+#define GPIO_CTRL 0x030
+#define GPIO_OE 0x034
+#define GPIO_DATA_IN 0x038
+#define GPIO_DATA_OUT 0x03C
+#define GPIO_LEVELDETECT0 0x040
+#define GPIO_LEVELDETECT1 0x044
+#define GPIO_RISINGDETECT 0x048
+#define GPIO_FALLINGDETECT 0x04c
+#define GPIO_CLEAR_DATAOUT 0x090
+#define GPIO_SET_DATAOUT 0x094
+
+typedef struct {
+ char *name;
+ u32 phy_addr;
+ u32 virt_addr;
+ u32 size;
+} io2vma_tab_t;
+
+/*
+ * GPIO phy to translation VMA table
+ */
+static io2vma_tab_t gpio_vma_tab[6] = {
+ {"GPIO1_BASE_ADDR", GPIO1_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO2_BASE_ADDR", GPIO2_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO3_BASE_ADDR", GPIO3_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO4_BASE_ADDR", GPIO4_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO5_BASE_ADDR", GPIO5_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO6_BASE_ADDR", GPIO6_BASE_ADDR , 0 , GPIO_SPACE_SIZE}
+};
+/*
+ * name - USER signal name assigned to the pin ( for printks)
+ * mux_func - enum index NAME for the pad_cfg function
+ * pin_num - pin_number if mux_func is GPIO, if not a GPIO it is -1
+ * mux_ptr - pointer to the corresponding pad_cfg_reg
+ * (used for pad release )
+ * mux_save - preserve here original PAD_CNF value for this
+ * pin (used for pad release)
+ * dir - if GPIO: 0 - OUT , 1 - IN
+ * dir_save - save original pin direction
+ * drv - initial drive level "0" or "1"
+ * drv_save - save original pin drive level
+ * valid - 1 if successfuly configured
+*/
+typedef struct {
+ char *name;
+ u32 mux_func;
+ int pin_num;
+ u16 *mux_ptr;
+ u16 mux_save;
+ u8 dir;
+ u8 dir_save;
+ u8 drv;
+ u8 drv_save;
+ u8 valid;
+} user_pad_cfg_t;
+
+/*
+ * need to ensure that enums are in sync with the
+ * omap_mux_pin_cfg table, these enums designate
+ * functions that OMAP pads can be configured to
+ */
+enum {
+ B23_OMAP3430_GPIO_167,
+ D23_OMAP3430_GPIO_126,
+ H1_OMAP3430_GPIO_62,
+ H1_OMAP3430_GPMC_n_w_p,
+ T8_OMAP3430_GPMC_n_c_s4,
+ T8_OMAP3430_GPIO_55,
+ R25_OMAP3430_GPIO_156,
+ R27_OMAP3430_GPIO_128,
+ K8_OMAP3430_GPIO_64,
+ K8_GPMC_WAIT2,
+ G3_OMAP3430_GPIO_60,
+ G3_OMAP3430_n_b_e0_CLE,
+ C6_GPMC_WAIT3,
+ J1_OMAP3430_GPIO_61,
+ C6_OMAP3430_GPIO_65,
+ N4_OMAP3430_GPMC_A1,
+ M4_OMAP3430_GPMC_A2,
+ L4_OMAP3430_GPMC_A3,
+ K4_OMAP3430_GPMC_A4,
+ T3_OMAP3430_GPMC_A5,
+ R3_OMAP3430_GPMC_A6,
+ N3_OMAP3430_GPMC_A7,
+ M3_OMAP3430_GPMC_A8,
+ END_OF_TABLE
+};
+
+/*
+ * number of GPIOS we plan to grab
+ */
+#define GPIO_SLOTS 8
+
+/*
+ * user_pads_init() reads(and saves) from/to this table
+ * used in conjunction with omap_3430_mux_t table in .h file
+ * because the way it's done in the kernel code
+ * TODO: implement restore of the the original cfg and i/o regs
+ */
+
+static user_pad_cfg_t user_pad_cfg[] = {
+ /*
+ * name,pad_func,pin_num, mux_ptr, mux_sav, dir,
+ * dir_sav, drv, drv_save, valid
+ */
+ {"AST_WAKEUP", B23_OMAP3430_GPIO_167, 167, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST_RESET", D23_OMAP3430_GPIO_126, 126, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST__rn_b", K8_GPMC_WAIT2, 64, NULL, 0,
+ DIR_INP, 0, 0, 0, 0},
+ {"AST_INTR", H1_OMAP3430_GPIO_62, 62, NULL, 0,
+ DIR_INP, 0, DRV_HI, 0, 0},
+ {"AST_CS", T8_OMAP3430_GPMC_n_c_s4, 55, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"LED_0", R25_OMAP3430_GPIO_156, 156, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"LED_1", R27_OMAP3430_GPIO_128, 128, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"AST_CLE", G3_OMAP3430_n_b_e0_CLE , 60, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ /*
+ * Z terminator, must always be present
+ * for sanity check, don't remove
+ */
+ {NULL}
+};
+
+#define GPIO_BANK(pin) (pin >> 5)
+#define REG_WIDTH 32
+#define GPIO_REG_VMA(pin_num, offset) \
+ (gpio_vma_tab[GPIO_BANK(pin_num)].virt_addr + offset)
+
+/*
+ * OMAP GPIO_REG 32 BIT MASK for a bit or
+ * flag in gpio_No[0..191] apply it to a 32 b3it
+ * location to set clear or check on a corresponding
+ * gpio bit or flag
+ */
+#define GPIO_REG_MASK(pin_num) (1 << \
+ (pin_num - (GPIO_BANK(pin_num) * REG_WIDTH)))
+
+/*
+ * OMAP GPIO registers bitwise access macros
+ */
+
+#define OMAP_GPIO_BIT(pin_num, reg) \
+ ((*((u32 *)GPIO_REG_VMA(pin_num, reg)) \
+ & GPIO_REG_MASK(pin_num)) ? 1 : 0)
+
+#define RD_OMAP_GPIO_BIT(pin_num, v) OMAP_GPIO_BIT(pin_num, reg)
+
+/*
+ *these are superfast set/clr bitbang macro, 48ns cyc tyme
+ */
+#define OMAP_SET_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_SET_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+#define OMAP_CLR_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_CLEAR_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+
+#define WR_OMAP_GPIO_BIT(pin_num, v) \
+ (v ? (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_SET_DATAOUT) = GPIO_REG_MASK(pin_num)) \
+ : (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_CLEAR_DATAOUT) = GPIO_REG_MASK(pin_num)))
+
+/*
+ * Note this pin cfg mimicks similar implementation
+ * in linux kernel, which unfortunately doesn't allow
+ * us to dynamically insert new custom GPIO mux
+ * configurations all REG definitions used in this
+ * applications. to add a new pad_cfg function, insert
+ * a new ENUM and new pin_cfg entry in omap_mux_pin_cfg[]
+ * table below3
+ *
+ * offset - note this is a word offset since the
+ * SCM regs are 16 bit packed in one 32 bit word
+ * mux_val - just enough to describe pins used
+ */
+typedef struct {
+ char *name;
+ u16 offset;
+ u16 mux_val;
+} omap_3430_mux_t;
+
+/*
+ * "OUTIN" is configuration when DATA reg drives the
+ * pin but the level at the pin can be sensed
+ */
+#define PAD_AS_OUTIN (OMAP34XX_MUX_MODE4 | \
+ OMAP34XX_PIN_OUTPUT | OMAP34XX_PIN_INPUT)
+
+omap_3430_mux_t omap_mux_pin_cfg[] = {
+ /*
+ * B23_OMAP3430_GPIO_167 - GPIO func to PA3D 167 WB wakeup
+ * D23_OMAP3430_GPIO_126 - drive GPIO_126 ( AST RESET)
+ * H1_OMAP3430_GPIO_62 - need a pullup on this pin
+ * H1_OMAP3430_GPMC_n_w_p - GPMC NAND CTRL n_w_p out
+ * T8_OMAP3430_GPMC_n_c_s4" - T8 is controlled b_y GPMC NAND ctrl
+ * R25_OMAP3430_GPIO_156 - OMAPZOOM drive LED_0
+ * R27_OMAP3430_GPIO_128 - OMAPZOOM drive LED_1
+ * K8_OMAP3430_GPIO_64 - OMAPZOOM drive LED_2
+ * K8_GPMC_WAIT2 - GPMC WAIT2 function on PAD K8
+ * G3_OMAP3430_GPIO_60 - OMAPZOOM drive LED_3
+ * G3_OMAP3430_n_b_e0_CLE -GPMC NAND ctrl CLE signal
+ */
+
+ {"B23_OMAP3430_GPIO_167", 0x0130, (OMAP34XX_MUX_MODE4)},
+ {"D23_OMAP3430_GPIO_126", 0x0132, (OMAP34XX_MUX_MODE4)},
+ {"H1_OMAP3430_GPIO_62", 0x00CA, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP) },
+ {"H1_OMAP3430_GPMC_n_w_p", 0x00CA, (OMAP34XX_MUX_MODE0)},
+ {"T8_OMAP3430_GPMC_n_c_s4", 0x00B6, (OMAP34XX_MUX_MODE0) },
+ {"T8_OMAP3430_GPIO_55", 0x00B6, (OMAP34XX_MUX_MODE4) },
+ {"R25_OMAP3430_GPIO_156", 0x018C, (OMAP34XX_MUX_MODE4) },
+ {"R27_OMAP3430_GPIO_128", 0x0154, (OMAP34XX_MUX_MODE4) },
+ {"K8_OMAP3430_GPIO_64", 0x00d0, (OMAP34XX_MUX_MODE4) },
+ {"K8_GPMC_WAIT2", 0x00d0, (OMAP34XX_MUX_MODE0) },
+ {"G3_OMAP3430_GPIO_60", 0x00C6, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"G3_OMAP3430_n_b_e0_CLE", 0x00C6, (OMAP34XX_MUX_MODE0)},
+ {"C6_GPMC_WAIT3", 0x00d2, (OMAP34XX_MUX_MODE0)},
+ {"C6_OMAP3430_GPIO_65", 0x00d2, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"J1_OMAP3430_GPIO_61", 0x00C8, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP)},
+
+ {"N4_OMAP3430_GPMC_A1", 0x007A, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M4_OMAP3430_GPMC_A2", 0x007C, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"L4_OMAP3430_GPMC_A3", 0x007E, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"K4_OMAP3430_GPMC_A4", 0x0080, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"T3_OMAP3430_GPMC_A5", 0x0082, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"R3_OMAP3430_GPMC_A6", 0x0084, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"N3_OMAP3430_GPMC_A7", 0x0086, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M3_OMAP3430_GPMC_A8", 0x0088, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ /*
+ * don't remove, used for sanity check.
+ */
+ {"END_OF_TABLE"}
+};
+
+
+#endif /* _INCLUDED_CYASMEMMAP_H_ */
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 2010-08-26 10:57:38.000000000 -0700
@@ -0,0 +1,73 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_CRAM_HAL_SIG (0x2332)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_phy_addr_base;
+
+ void *m_vma_addr_base;
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,319 +0,0 @@
-/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-/*
- * This file contains the defintion of the hardware abstraction
- * layer on OMAP3430 talking to the West Bridge Astoria device
- */
-
-
-#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
-#define _INCLUDED_CYASHALOMAP_KERNEL_H_
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/wait.h>
-#include <linux/string.h>
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-typedef struct cy_as_hal_sleep_channel_t {
- wait_queue_head_t wq ;
-} cy_as_hal_sleep_channel ;
-
-/* moved to staging location, eventual location
- * considered is here
-#include <mach/westbridge/cyashaldef.h>
-#include <linux/westbridge/cyastypes.h>
-#include <linux/westbridge/cyas_cplus_start.h>
-*/
-#include "../cyashaldef.h"
-#include "../../../../../../../include/linux/westbridge/cyastypes.h"
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#include "cyasomapdev_kernel.h"
-
-/*
- * Below are the data structures that must be defined by the HAL layer
- */
-
-/*
- * The HAL layer must define a TAG for identifying a specific Astoria
- * device in the system. In this case the tag is a void * which is
- * really an OMAP device pointer
- */
-typedef void *cy_as_hal_device_tag ;
-
-
-/* This must be included after the CyAsHalDeviceTag type is defined */
-
-/* moved to staging location, eventual location
- * considered is here
- * #include <linux/westbridge/cyashalcb.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
-/*
- * Below are the functions that communicate with the West Bridge
- * device. These are system dependent and must be defined by
- * the HAL layer for a given system.
- */
-
-/*
- * This function must be defined to write a register within the Antioch
- * device. The addr value is the address of the register to write with
- * respect to the base address of the Antioch device.
- */
-void
-cy_as_hal_write_register(cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data) ;
-
-/*
- * This function must be defined to read a register from
- * the west bridge device. The addr value is the address of
- * the register to read with respect to the base address
- * of the west bridge device.
- */
-uint16_t
-cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
-
-/*
- * This function must be defined to transfer a block of data
- * to the west bridge device. This function can use the burst write
- * (DMA) capabilities of Antioch to do this, or it can just copy
- * the data using writes.
- */
-void
-cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to transfer a block of data
- * from the Antioch device. This function can use the burst
- * read (DMA) capabilities of Antioch to do this, or it can
- * just copy the data using reads.
- */
-void
-cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
- void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to cancel any pending DMA request.
- */
-void
-cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
-
-/*
- * This function must be defined to allow the Antioch API to
- * register a callback function that is called when a DMA transfer
- * is complete.
- */
-void
-cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb) ;
-
-/*
- * This function must be defined to return the maximum size of DMA
- * request that can be handled on the given endpoint. The return
- * value should be the maximum size in bytes that the DMA module can
- * handle.
- */
-uint32_t
-cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep) ;
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the Antioch device. Generally this is done via a GPIO of some
- * type.
- */
-cy_bool
-cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
-
-/*
- * This function is called when the Antioch PLL loses lock, because
- * of a problem in the supply voltage or the input clock.
- */
-void
-cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
-
-
-/**********************************************************************
- *
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- *
-***********************************************************************/
-
-/*
- * This function is required by the API to allocate memory. This function
- * is expected to work exactly like malloc().
- */
-void *
-cy_as_hal_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsHalAlloc(). This function is expected to work exacly like free().
- */
-void
-cy_as_hal_free(void *mem_p) ;
-
-/*
- * This function is required by the API to allocate memory during a
- * callback. This function must be able to provide storage at inturupt
- * time.
- */
-void *
-cy_as_hal_c_b_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsCBHalAlloc().
- */
-void
-cy_as_hal_c_b_free(void *ptr) ;
-
-/*
- * This function is required to set a block of memory to a specific
- * value. This function is expected to work exactly like memset()
- */
-void
-cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
-
-/*
- * This function is expected to create a sleep channel. The data
- * structure that represents the sleep channel is given by the
- * pointer in the argument.
- */
-cy_bool
-cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-/*
- * This function is expected to destroy a sleep channel. The data
- * structure that represents the sleep channel is given by
- * the pointer in the argument.
- */
-
-
-cy_bool
-cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-cy_bool
-cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
-
-cy_bool
-cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
-
-uint32_t
-cy_as_hal_disable_interrupts(void) ;
-
-void
-cy_as_hal_enable_interrupts(uint32_t);
-
-void
-cy_as_hal_sleep150(void);
-
-void
-cy_as_hal_sleep(uint32_t ms);
-
-cy_bool
-cy_as_hal_is_polling(void);
-
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup);
-
-/*
- * required only in spi mode
- */
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
-
-
-#ifndef NDEBUG
-#define cy_as_hal_assert(cond) if (!(cond))\
- printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
-#else
-#define cy_as_hal_assert(cond)
-#endif
-
-#define cy_as_hal_print_message printk
-
-/* removable debug printks */
-#ifndef WESTBRIDGE_NDEBUG
-#define DBG_PRINT_ENABLED
-#endif
-
-/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
-
-
-#ifdef DBG_PRINT_ENABLED
- /* Debug printing enabled */
-
- #define DBGPRN(...) printk(__VA_ARGS__)
- #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
- current->pid, __func__)
-
-#else
- /** NO DEBUG PRINTING **/
- #define DBGPRN(...)
- #define DBGPRN_FUNC_NAME
-
-#endif
-
-/*
-CyAsMiscSetLogLevel(uint8_t level)
-{
- debug_level = level ;
-}
-
-#ifdef CY_AS_LOG_SUPPORT
-
-void
-cy_as_log_debug_message(int level, const char *str)
-{
- if (level <= debug_level)
- cy_as_hal_print_message("log %d: %s\n", level, str) ;
-}
-*/
-
-
-/*
- * print buffer helper
- */
-void cyashal_prn_buf(void *buf, uint16_t offset, int len);
-
-/*
- * These are the functions that are not part of the HAL layer,
- * but are required to be called for this HAL.
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug) ;
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag) ;
-int omap_start_intr(cy_as_hal_device_tag tag) ;
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
-
-/* moved to staging location
-#include <linux/westbridge/cyas_cplus_end.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 2010-08-27 17:49:56.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_pnand.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+ /* Debug printing enabled */
+
+ #define DBGPRN(...) printk(__VA_ARGS__)
+ #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+ /** NO DEBUG PRINTING **/
+ #define DBGPRN(...)
+ #define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-09-01 16:04:25.000000000 -0700
@@ -31,6 +31,9 @@
#define _INCLUDED_CYASMEMMAP_H_
/* defines copied from OMAP kernel branch */
+
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
#define OMAP34XX_MUX_MODE0 0
#define OMAP34XX_MUX_MODE4 4
#define OMAP3_INPUT_EN (1 << 8)
@@ -250,9 +253,9 @@
/*
*gpmc nand registers for CS4
*/
-#define GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
#define GPMC_STAT_REG (0x54)
#define GPMC_ERR_TYPE (0x48)
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,72 +0,0 @@
-/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
-#define __CY_AS_OMAP_DEV_KERNEL_H__
-
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/completion.h>
-
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-
-/*
- * Constants
- */
-#define CY_AS_OMAP_KERNEL_HAL_SIG (0x1441)
-
-
-/*
- * Data structures
- */
-typedef struct cy_as_omap_dev_kernel {
- /* This is the signature for this data structure */
- unsigned int m_sig ;
-
- /* Address base of Antioch Device */
- void *m_addr_base;
-
- /* This is a pointer to the next Antioch device in the system */
- struct cy_as_omap_dev_kernel *m_next_p;
-
- /* This is for thread sync */
- struct completion thread_complete;
-
- /* This is for thread to wait for interrupts */
- cy_as_hal_sleep_channel thread_sc;
-
- /* This is for thread to exit upon StopOmapKernel */
- int thread_flag; /* set 1 to exit */
-
- int dma_ch;
-
- /* This is for dma sync */
- struct completion dma_complete;
-} cy_as_omap_dev_kernel;
-
-#endif
-
-/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 2010-08-20 14:36:23.000000000 -0700
@@ -0,0 +1,72 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_PNAND_HAL_SIG (0x1441)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_addr_base;
+
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile
--- linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile 2010-08-19 14:30:50.000000000 -0700
@@ -12,12 +12,20 @@ obj-$(CONFIG_WESTBRIDGE_DEVICE_DRIVER)
ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL),y)
#moved for staging compatbility
#cyasdev-y := ../../../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o
-cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o \
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_pnand.o cyasdevice.o \
../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
../api/src/cyasstorage.o ../api/src/cyasusb.o
else
+
+ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL),y)
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_cram.o cyasdevice.o \
+ ../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
+ ../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
+ ../api/src/cyasstorage.o ../api/src/cyasusb.o
+else
# should not get here, need to be built with some hal
cyasdev-y := cyasdevice.o
endif
+endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-08-20 14:35:14.000000000 -0700
@@ -74,7 +74,7 @@
-/***** OMAP3430 Kernel HAL ****/
+/***** OMAP3430 PNAND HAL ****/
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
#ifdef CY_HAL_DEFINED
@@ -86,7 +86,24 @@
* considered is here
* #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
*/
- #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+
+#endif
+/*****************************/
+
+/***** OMAP3430 CRAM HAL ****/
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+ #ifdef CY_HAL_DEFINED
+ #error only one HAL layer can be defined
+ #endif
+
+ #define CY_HAL_DEFINED
+/* moved to staging location, eventual implementation
+ * considered is here
+ * #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+*/
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
#endif
/*****************************/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/Kconfig linux-next-incl-sdk/drivers/staging/westbridge/Kconfig
--- linux-next-vanilla/drivers/staging/westbridge/Kconfig 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/Kconfig 2010-09-01 15:45:46.000000000 -0700
@@ -4,7 +4,7 @@
menuconfig WESTBRIDGE
tristate "West Bridge support"
- depends on HAS_IOMEM && BROKEN
+ depends on WESTBRIDGE_HAL_SELECTED
help
This selects West Bridge Peripheral controller support.
@@ -12,7 +12,7 @@ menuconfig WESTBRIDGE
menuconfig WESTBRIDGE_ASTORIA
bool "West Bridge Astoria support"
- depends on WESTBRIDGE != n
+ depends on WESTBRIDGE != n && WESTBRIDGE_HAL_SELECTED
help
This option enables support for West Bridge Astoria
@@ -20,11 +20,43 @@ if WESTBRIDGE_ASTORIA
source "drivers/staging/westbridge/astoria/Kconfig"
endif #WESTBRIDGE_ASTORIA
-menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+#menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+# bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
+# depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+# help
+# Include the OMAP3430 HAL for PNAND interface
+
+config WESTBRIDGE_HAL_SELECTED
+ boolean
+
+choice
+ prompt "West Bridge HAL"
+ help
+ West Bridge HAL/processor interface to be used
+
+#
+# HAL Layers
+#
+
+config MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
- depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ bool "WESTBRIDGE OMAP3430 Astoria CRAM HAL"
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_NO_WESTBRIDGE
+ bool "no West Bridge HAL selected"
help
- Include the OMAP3430 Linux Based HAL
+ Do not include any HAL layer(de-activates West Bridge option)
+endchoice
config WESTBRIDGE_DEBUG
bool "West Bridge debugging"
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty
2010-09-02 0:08 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes David Cross
2010-09-02 1:37 ` Greg KH
2010-09-02 22:43 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes David Cross
@ 2010-09-02 22:47 ` David Cross
2010-09-05 5:20 ` Greg KH
2010-09-02 22:49 ` [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates David Cross
3 siblings, 1 reply; 12+ messages in thread
From: David Cross @ 2010-09-02 22:47 UTC (permalink / raw)
To: greg; +Cc: linux-kernel
This patch contains update to the gadget driver, some of which are based on feedback from the Linux community concerning the usage of fat_get_block.
It also contains the addition of mpage_clear_dirty, a function used to clear dirty pages from the page cache. These two patches are also interdependent
and difficult to separate. As such, they are included as one patch.
Thanks,
David
Signed-off-by: David Cross <david.cross@cypress.com>
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.c 2010-08-26 11:18:18.000000000 -0700
@@ -224,6 +224,8 @@ static void cy_as_gadget_mtp_event_callb
cy_as_mtp_send_object_complete_data *send_obj_data =
(cy_as_mtp_send_object_complete_data *) evdata ;
+ cy_as_release_common_lock();
+
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
"<6>MTP EVENT: send_object_complete\n");
@@ -246,6 +248,8 @@ static void cy_as_gadget_mtp_event_callb
{
cy_as_mtp_get_object_complete_data *get_obj_data =
(cy_as_mtp_get_object_complete_data *) evdata ;
+;
+ cy_as_release_common_lock();
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
@@ -752,6 +756,9 @@ static int cyasgadget_queue(
_req->status = -EINPROGRESS;
_req->actual = 0;
+ if (as_req)
+ list_add_tail(&as_req->queue, &as_ep->queue);
+
spin_unlock_irqrestore(&cy_as_dev->lock, flags);
/* Call Async functions */
@@ -772,16 +779,6 @@ static int cyasgadget_queue(
else
_req->status = -EALREADY ;
} else if (as_ep->num == 0) {
- /*
- ret = cy_as_usb_write_data_async(cy_as_dev->dev_handle,
- as_ep->num, _req->length, _req->buf, cy_false,
- cyasgadget_setupwritecallback) ;
-
- if (ret != CY_AS_ERROR_SUCCESS)
- cy_as_hal_print_message("<1>_cy_as_gadget: "
- "cy_as_usb_write_data_async failed with error "
- "code %d\n", ret) ;
- */
if ((cy_as_dev->outsetupreq) && (_req->length)) {
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("<1>_cy_as_gadget: "
@@ -817,8 +814,7 @@ static int cyasgadget_queue(
"cy_as_usb_write_data_async failed with "
"error code %d\n", ret) ;
}
-
- } else if (list_empty(&as_ep->queue)) {
+ } else { /* read function not to EP 0*/
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("<1>_cy_as_gadget: "
"cy_as_usb_read_data_async being called since "
@@ -836,13 +832,6 @@ static int cyasgadget_queue(
_req->status = -EALREADY ;
}
- spin_lock_irqsave(&cy_as_dev->lock, flags);
-
- if (as_req)
- list_add_tail(&as_req->queue, &as_ep->queue);
-
- spin_unlock_irqrestore(&cy_as_dev->lock, flags);
-
return 0;
}
@@ -1146,11 +1135,9 @@ static int cyasgadget_ioctl(
struct inode *inode = mapping->host;
struct inode *alloc_inode =
file_to_allocate->f_path.dentry->d_inode;
- int cluster = 0;
uint32_t num_clusters = 0;
struct buffer_head bh;
struct kstat stat;
- struct iattr alloc_iattr;
int nr_pages = 0;
int ret_stat = 0;
@@ -1158,10 +1145,10 @@ static int cyasgadget_ioctl(
cy_as_hal_print_message("%s: fhandle is OK, "
"calling vfs_getattr\n", __func__);
#endif
-
+
ret_stat = vfs_getattr(file_to_allocate->f_path.mnt,
file_to_allocate->f_path.dentry, &stat);
-
+
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message("%s: returned from "
"vfs_getattr() stat->blksize=0x%lx\n",
@@ -1171,15 +1158,11 @@ static int cyasgadget_ioctl(
/* TODO: get this from disk properties
* (from blockdevice)*/
#define SECTOR_SIZE 512
- if (stat.blksize != 0) {
num_clusters = (k_d.num_bytes) / SECTOR_SIZE;
- if (((k_d.num_bytes) % SECTOR_SIZE) != 0)
+ if (((k_d.num_bytes) % SECTOR_SIZE) != 0)
num_clusters++;
- } else {
- goto initsoj_safe_exit;
- }
-
+
bh.b_state = 0;
bh.b_blocknr = 0;
/* block size is arbitrary , we'll use sector size*/
@@ -1190,42 +1173,6 @@ static int cyasgadget_ioctl(
"size of %d\n", __func__,
num_clusters, bh.b_size);
#endif
- for (cluster = 0; cluster < num_clusters; cluster++) {
- ret_stat = fat_get_block(inode,
- cluster, &bh, 1);
- if (ret_stat) {
- cy_as_hal_print_message(
- "%s: unable to get fat block, "
- "ret_stat=0x%d\n",
- __func__, ret_stat);
- goto initsoj_safe_exit;
- }
- }
-
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: allocated clusters "
- "successfully (fat_get_block), check bmap..."
- "\n", __func__);
- #endif
-
- alloc_iattr.ia_valid = ATTR_SIZE;
- alloc_iattr.ia_size = k_d.num_bytes;
-
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: calling fat_notify_change "
- "(ia_valid:%d, ia_size:%d)\n", __func__,
- alloc_iattr.ia_valid,
- (int)alloc_iattr.ia_size);
- #endif
-
- /* adjust the filesize */
- ret_stat = alloc_inode->i_op->setattr(
- file_to_allocate->f_path.dentry, &alloc_iattr);
- #ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: fat_setattr() "
- "returned 0x%x\n",
- __func__, ret_stat);
- #endif
/* clear dirty pages in page cache
* (if were any allocated) */
@@ -1246,7 +1193,7 @@ static int cyasgadget_ioctl(
"mpage_cleardirty() "
"for %d pages\n", __func__, nr_pages);
#endif
-
+
ret_stat = mpage_cleardirty(mapping, nr_pages);
/*fill up the the block table from the addr mapping */
@@ -1363,6 +1310,9 @@ static int cyasgadget_ioctl(
__func__);
#endif
sg_init_one(&sg, &blk_table, sizeof(blk_table));
+
+ cy_as_acquire_common_lock();
+
ret_stat = cy_as_mtp_init_send_object(dev->dev_handle,
(cy_as_mtp_block_table *)&sg,
k_d.num_bytes, 0, 0);
@@ -1379,7 +1329,6 @@ static int cyasgadget_ioctl(
__func__, alloc_filename);
} /* end if (file_to_allocate)*/
#endif
-initsoj_safe_exit:
ret_stat = 0;
retval = __put_user(ret_stat,
(uint32_t __user *)(&(usr_d->ret_val)));
@@ -1541,6 +1490,8 @@ initsoj_safe_exit:
k_d.num_bytes);
#endif
+ cy_as_acquire_common_lock();
+
ret_stat = cy_as_mtp_init_get_object(
dev->dev_handle,
(cy_as_mtp_block_table *)&sg,
@@ -2151,7 +2102,21 @@ static int cyasgadget_initialize(void)
/*strcpy(cy_as_dev->gadget.dev.bus_id, "cyasgadget");*/
cy_as_dev->gadget.dev.release = cyas_gadget_release;
cy_as_dev->gadget.name = cy_as_driver_name;
-
+
+ /* this "gadget" abstracts/virtualizes the controller */
+ dev_set_name(&cy_as_dev->gadget.dev, "cyas" );
+ cy_as_dev->gadget.dev.parent = cy_as_dev->cy_controller;
+
+ retval = device_register(&cy_as_dev->gadget.dev);
+ if (retval != 0) {
+ cy_as_hal_print_message( "%s: can not register device\n", __func__) ;
+ return retval;
+ }
+ #ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message( "%s: successfully registered device\n", __func__) ;
+ #endif
+
/* Get the device handle */
cy_as_dev->dev_handle = cyasdevice_getdevhandle() ;
if (0 == cy_as_dev->dev_handle) {
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/gadget/cyasgadget.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/gadget/cyasgadget.h 2010-08-25 18:28:27.000000000 -0700
@@ -83,8 +83,9 @@
#include <linux/uaccess.h> /* copy_*_user */
extern int mpage_cleardirty(struct address_space *mapping, int num_pages);
-extern int fat_get_block(struct inode *, sector_t , struct buffer_head *, int);
extern cy_as_device_handle *cyasdevice_getdevhandle(void) ;
+extern void cy_as_acquire_common_lock(void);
+extern void cy_as_release_common_lock(void);
/* Driver data structures and utilities */
typedef struct cyasgadget_ep {
@@ -119,6 +120,7 @@ typedef struct cyasgadget {
struct usb_gadget gadget;
spinlock_t lock;
struct cyasgadget_ep an_gadget_ep[16];
+ struct device *cy_controller;
struct usb_gadget_driver *driver;
/* Handle to the West Bridge device */
cy_as_device_handle dev_handle;
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/fs/mpage.c linux-next-incl-sdk/fs/mpage.c
--- linux-next-vanilla/fs/mpage.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/fs/mpage.c 2010-09-01 16:12:40.000000000 -0700
@@ -716,3 +716,49 @@ int mpage_writepage(struct page *page, g
return ret;
}
EXPORT_SYMBOL(mpage_writepage);
+
+int mpage_cleardirty(struct address_space *mapping, int num_pages)
+{
+ int ret = 0;
+ int nr_pages;
+ struct pagevec pvec;
+ pgoff_t index = 0;
+ pgoff_t end;
+
+ pagevec_init(&pvec, 0);
+ index = mapping->writeback_index;
+ end = index + num_pages;
+
+ while ((index <= end) &&
+ (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
+ PAGECACHE_TAG_DIRTY,
+ min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+ unsigned i;
+
+ for (i = 0; i < nr_pages; i++) {
+ struct page *page = pvec.pages[i];
+
+ lock_page(page);
+ ret = clear_page_dirty_for_io(page);
+ if (page_has_private(page))
+ do_invalidatepage(page, 0);
+
+ cancel_dirty_page(page, PAGE_CACHE_SIZE);
+
+ remove_from_page_cache(page);
+ ClearPageMappedToDisk(page);
+ page_cache_release(page); /* pagecache ref */
+ unlock_page(page);
+
+ if (!ret) {
+ printk(KERN_INFO "mpage_cleardirty: clear_page_dirty_for_io returned %d\n", ret);
+ return ret;
+ }
+ }
+ pagevec_release(&pvec);
+ cond_resched();
+ }
+
+ return ret;
+}
+EXPORT_SYMBOL(mpage_cleardirty);
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates
2010-09-02 0:08 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes David Cross
` (2 preceding siblings ...)
2010-09-02 22:47 ` [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty David Cross
@ 2010-09-02 22:49 ` David Cross
2010-09-05 5:24 ` Greg KH
3 siblings, 1 reply; 12+ messages in thread
From: David Cross @ 2010-09-02 22:49 UTC (permalink / raw)
To: greg; +Cc: linux-kernel
This patch contains minor updates to the west bridge block and device
drivers, including the addition of a common lock and a minor update to
remove the blk_fs_request call based on a kernel update. These changes
could possibly be separated, but are very minor in scope and as such
have been submitted together.
Thanks,
David
Signed-off-by: David Cross <david.cross@cypress.com>
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-25 18:29:44.000000000 -0700
@@ -389,7 +389,7 @@ int cyasblkdev_media_changed(struct gend
struct cyasblkdev_blk_data *bd;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("cyasblkdev_media_changed() is called\n");
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev_media_changed() is called\n");
#endif
if (gd)
@@ -466,7 +466,7 @@ static int cyasblkdev_blk_prep_rq(
/* Check for excessive requests.*/
if (blk_rq_pos(req) + blk_rq_sectors(req) > get_capacity(req->rq_disk)) {
- cy_as_hal_print_message("cyasblkdev: bad request address\n");
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev: bad request address\n");
stat = BLKPREP_KILL;
}
@@ -493,7 +493,7 @@ static void cyasblkdev_issuecallback(
{
int retry_cnt = 0;
DBGPRN_FUNC_NAME;
-
+
if (status != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
cy_as_hal_print_message(
@@ -509,7 +509,10 @@ static void cyasblkdev_issuecallback(
__func__, (unsigned int) gl_bd->queue.req, status,
(unsigned int) blk_rq_sectors(gl_bd->queue.req)) ;
#endif
-
+
+ if(rq_data_dir(gl_bd->queue.req) != READ) {
+ cy_as_release_common_lock();
+ }
/* note: blk_end_request w/o __ prefix should
* not require spinlocks on the queue*/
while (blk_end_request(gl_bd->queue.req,
@@ -533,7 +536,7 @@ static void cyasblkdev_issuecallback(
/* queue is not plugged */
gl_bd->queue.req = blk_fetch_request(gl_bd->queue.queue);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s blkdev_callback: "
+ cy_as_hal_print_message(KERN_INFO"%s blkdev_callback: "
"blk_fetch_request():%p\n",
__func__, gl_bd->queue.req);
#endif
@@ -543,7 +546,7 @@ static void cyasblkdev_issuecallback(
spin_unlock_irq(&gl_bd->lock);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s blkdev_callback: about to "
+ cy_as_hal_print_message(KERN_INFO"%s blkdev_callback: about to "
"call issue_fn:%p\n", __func__, gl_bd->queue.req);
#endif
@@ -585,7 +588,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->user_disk_0_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to disk 0 "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to disk 0 "
"for sector=%d, num_sectors=%d, unit_no=%d\n",
__func__, req_sector, (int) blk_rq_sectors(req),
lcl_unit_no);
@@ -598,7 +601,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->user_disk_1_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to disk 1 for "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to disk 1 for "
"sector=%d, num_sectors=%d, unit_no=%d\n", __func__,
req_sector, (int) blk_rq_sectors(req), lcl_unit_no);
#endif
@@ -609,7 +612,7 @@ static int cyasblkdev_blk_issue_rq(
lcl_unit_no = gl_bd->system_disk_unit_no;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: request made to system disk "
+ cy_as_hal_print_message(KERN_INFO"%s: request made to system disk "
"for sector=%d, num_sectors=%d, unit_no=%d\n", __func__,
req_sector, (int) blk_rq_sectors(req), lcl_unit_no);
#endif
@@ -625,7 +628,7 @@ static int cyasblkdev_blk_issue_rq(
if (rq_data_dir(req) == READ) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: calling readasync() "
+ cy_as_hal_print_message(KERN_INFO"%s: calling readasync() "
"req_sector=0x%x, req_nr_sectors=0x%x, bd->sg:%x\n\n",
__func__, req_sector, req_nr_sectors, (uint32_t)bd->sg);
#endif
@@ -636,10 +639,10 @@ static int cyasblkdev_blk_issue_rq(
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:readasync() error %d at "
+ cy_as_hal_print_message(KERN_INFO"%s:readasync() error %d at "
"address %ld, unit no %d\n", __func__, ret,
blk_rq_pos(req), lcl_unit_no);
- cy_as_hal_print_message("%s:ending i/o request "
+ cy_as_hal_print_message(KERN_INFO"%s:ending i/o request "
"on reg:%x\n", __func__, (uint32_t)req);
#endif
@@ -651,13 +654,14 @@ static int cyasblkdev_blk_issue_rq(
bq->req = NULL ;
}
} else {
+ cy_as_acquire_common_lock();
ret = cy_as_storage_write_async(bd->dev_handle, bus_num, 0,
lcl_unit_no, req_sector, bd->sg, req_nr_sectors,
(cy_as_storage_callback)cyasblkdev_issuecallback);
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: write failed with "
+ cy_as_hal_print_message(KERN_INFO"%s: write failed with "
"error %d at address %ld, unit no %d\n",
__func__, ret, blk_rq_pos(req), lcl_unit_no);
#endif
@@ -690,7 +694,7 @@ static void cyasblkdev_storage_callback(
)
{
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: bus:%d, device:%d, evtype:%d, "
+ cy_as_hal_print_message(KERN_INFO"%s: bus:%d, device:%d, evtype:%d, "
"evdata:%p\n ", __func__, bus, device, evtype, evdata);
#endif
@@ -746,7 +750,7 @@ uint32_t cyasblkdev_get_vfat_offset(int
for (sect_no = 0; sect_no < SECTORS_TO_SCAN; sect_no++) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s before cyasstorageread "
+ cy_as_hal_print_message(KERN_INFO"%s before cyasstorageread "
"gl_bd->sg addr=0x%x\n", __func__,
(unsigned int) gl_bd->sg);
#endif
@@ -787,8 +791,8 @@ uint32_t cyasblkdev_get_vfat_offset(int
if (stat != 0) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s sector scan error\n",
- __func__);
+ cy_as_hal_print_message(KERN_INFO"%s sector scan error: %d\n",
+ __func__, stat);
#endif
break;
}
@@ -821,7 +825,7 @@ static int cyasblkdev_add_disks(int bus_
cy_as_storage_query_unit_data unit_data = {0} ;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:query device: "
+ cy_as_hal_print_message(KERN_INFO"%s:query device: "
"type:%d, removable:%d, writable:%d, "
"blksize %d, units:%d, locked:%d, "
"erase_sz:%d\n",
@@ -846,7 +850,7 @@ static int cyasblkdev_add_disks(int bus_
bd->dev_handle, bus_num, 0, 0, 0) ;
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s cannot release"
+ cy_as_hal_print_message(KERN_INFO"%s cannot release"
" storage\n", __func__) ;
#endif
goto out;
@@ -861,7 +865,7 @@ static int cyasblkdev_add_disks(int bus_
&unit_data, 0, 0) ;
if (ret != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot query "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot query "
"%d device unit - reason code %d\n",
__func__, bus_num, ret) ;
#endif
@@ -876,7 +880,7 @@ static int cyasblkdev_add_disks(int bus_
if ((ret != CY_AS_ERROR_SUCCESS) &&
(ret != CY_AS_ERROR_ALREADY_PARTITIONED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cy_as_storage_"
+ cy_as_hal_print_message(KERN_INFO"%s: cy_as_storage_"
"create_p_partition after size > 0 check "
"failed with error code %d\n",
__func__, ret);
@@ -976,7 +980,7 @@ static int cyasblkdev_add_disks(int bus_
}
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: partition "
+ cy_as_hal_print_message(KERN_INFO"%s: partition "
"exists and sizes equal\n",
__func__);
#endif
@@ -1054,13 +1058,13 @@ static int cyasblkdev_add_disks(int bus_
#endif
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: setting gendisk disk "
+ cy_as_hal_print_message(KERN_INFO"%s: setting gendisk disk "
"capacity to %d\n", __func__, (int) disk_cap);
#endif
/* initializing bd->queue */
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: init bd->queue\n",
+ cy_as_hal_print_message(KERN_INFO"%s: init bd->queue\n",
__func__);
#endif
@@ -1111,7 +1115,8 @@ static int cyasblkdev_add_disks(int bus_
* public partition beginning */
if (vfat_search) {
bd->user_disk_0_first_sector =
- cyasblkdev_get_vfat_offset(0,
+ cyasblkdev_get_vfat_offset(
+ bd->user_disk_0_bus_num,
bd->user_disk_0_unit_no);
} else {
bd->user_disk_0_first_sector = 0;
@@ -1244,7 +1249,8 @@ static int cyasblkdev_add_disks(int bus_
if (vfat_search) {
bd->user_disk_1_first_sector =
cyasblkdev_get_vfat_offset(
- 1, bd->user_disk_1_unit_no);
+ bd->user_disk_1_bus_num,
+ bd->user_disk_1_unit_no);
} else {
bd->user_disk_1_first_sector
= 0;
@@ -1342,7 +1348,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s west bridge device handle:%x\n",
+ cy_as_hal_print_message(KERN_INFO"%s west bridge device handle:%x\n",
__func__, (uint32_t)bd->dev_handle);
#endif
@@ -1356,7 +1362,7 @@ static struct cyasblkdev_blk_data *cyasb
if ((stat != CY_AS_ERROR_SUCCESS) &&
(stat != CY_AS_ERROR_RESOURCE_NOT_OWNED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot release "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot release "
"resource bus 0 - reason code %d\n",
__func__, stat) ;
#endif
@@ -1366,7 +1372,7 @@ static struct cyasblkdev_blk_data *cyasb
if ((stat != CY_AS_ERROR_SUCCESS) &&
(stat != CY_AS_ERROR_RESOURCE_NOT_OWNED)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot release "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot release "
"resource bus 0 - reason code %d\n",
__func__, stat) ;
#endif
@@ -1376,14 +1382,14 @@ static struct cyasblkdev_blk_data *cyasb
stat = cy_as_storage_start(bd->dev_handle, 0, 0x101) ;
if (stat != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot start storage "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot start storage "
"stack - reason code %d\n", __func__, stat) ;
#endif
goto out;
}
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: storage started:%d ok\n",
+ cy_as_hal_print_message(KERN_INFO"%s: storage started:%d ok\n",
__func__, stat);
#endif
@@ -1391,7 +1397,7 @@ static struct cyasblkdev_blk_data *cyasb
cyasblkdev_storage_callback);
if (stat != CY_AS_ERROR_SUCCESS) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot register callback "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot register callback "
"- reason code %d\n", __func__, stat) ;
#endif
goto out;
@@ -1405,7 +1411,7 @@ static struct cyasblkdev_blk_data *cyasb
bd->media_count[bus_num];
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: cannot query %d, "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot query %d, "
"reason code: %d\n",
__func__, bus_num, stat) ;
#endif
@@ -1433,7 +1439,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else {
- cy_as_hal_print_message("%s: no available "
+ cy_as_hal_print_message(KERN_INFO"%s: no available "
"gen_disk for disk 0, "
"physically inconsistent\n", __func__);
}
@@ -1453,7 +1459,7 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else {
- cy_as_hal_print_message("%s: no available "
+ cy_as_hal_print_message(KERN_INFO"%s: no available "
"gen_disk for media, "
"physically inconsistent\n", __func__);
}
@@ -1461,13 +1467,13 @@ static struct cyasblkdev_blk_data *cyasb
}
#ifndef WESTBRIDGE_NDEBUG
else if (total_media_count > 2) {
- cy_as_hal_print_message("%s: count corrupted = 0x%d\n",
+ cy_as_hal_print_message(KERN_INFO"%s: count corrupted = 0x%d\n",
__func__, total_media_count);
}
#endif
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: %d device(s) found\n",
+ cy_as_hal_print_message(KERN_INFO"%s: %d device(s) found\n",
__func__, total_media_count) ;
#endif
@@ -1476,7 +1482,7 @@ static struct cyasblkdev_blk_data *cyasb
stat = cy_as_storage_claim(bd->dev_handle,
bus_num, 0, 0, 0) ;
if (stat != CY_AS_ERROR_SUCCESS) {
- cy_as_hal_print_message("%s: cannot claim "
+ cy_as_hal_print_message(KERN_INFO"%s: cannot claim "
"%d bus - reason code %d\n",
__func__, bus_num, stat) ;
goto out;
@@ -1570,7 +1576,7 @@ static int __init cyasblkdev_blk_init(vo
return 0;
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("cyasblkdev init error:%d\n", res);
+ cy_as_hal_print_message(KERN_INFO"cyasblkdev init error:%d\n", res);
#endif
return res;
}
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c 2010-09-01 15:56:46.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.c 2010-09-01 15:57:26.000000000 -0700
@@ -95,9 +95,9 @@ static int cyasblkdev_prep_request(
DBGPRN_FUNC_NAME;
/* we only like normal block requests.*/
- if (!blk_fs_request(req)) {
+ if (req->cmd_type != REQ_TYPE_FS && !(req->cmd_flags & REQ_DISCARD)) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s:%x bad request received\n",
+ cy_as_hal_print_message(KERN_INFO"%s:%x bad request received\n",
__func__, current->pid) ;
#endif
@@ -192,7 +192,7 @@ static int cyasblkdev_queue_thread(void
"thread_sem->count=%d\n",
__func__, bq->thread_sem.count);
if (spin_is_locked(q->queue_lock)) {
- cy_as_hal_print_message("%s: queue_lock "
+ cy_as_hal_print_message(KERN_INFO"%s: queue_lock "
"is locked, need to release\n", __func__);
spin_unlock(q->queue_lock);
@@ -249,7 +249,7 @@ static int cyasblkdev_queue_thread(void
complete_and_exit(&bq->thread_complete, 0);
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: is finished\n", __func__) ;
+ cy_as_hal_print_message(KERN_INFO"%s: is finished\n", __func__) ;
#endif
return 0;
@@ -274,7 +274,7 @@ static void cyasblkdev_request(struct re
if (!bq->req) {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s wake_up(&bq->thread_wq)\n",
+ cy_as_hal_print_message(KERN_INFO"%s wake_up(&bq->thread_wq)\n",
__func__);
#endif
@@ -282,7 +282,7 @@ static void cyasblkdev_request(struct re
wake_up(&bq->thread_wq);
} else {
#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("%s: don't wake Q_thr, bq->req:%x\n",
+ cy_as_hal_print_message(KERN_INFO"%s: don't wake Q_thr, bq->req:%x\n",
__func__, (uint32_t)bq->req);
#endif
}
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_queue.h 2010-08-23 16:45:15.000000000 -0700
@@ -54,6 +54,8 @@ extern int cyasblkdev_init_queue(struct
extern void cyasblkdev_cleanup_queue(struct cyasblkdev_queue *);
extern void cyasblkdev_queue_suspend(struct cyasblkdev_queue *);
extern void cyasblkdev_queue_resume(struct cyasblkdev_queue *);
+extern void cy_as_acquire_common_lock(void);
+extern void cy_as_release_common_lock(void);
extern cy_as_device_handle cyasdevice_getdevhandle(void) ;
#define MOD_LOGS 1
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/device/cyasdevice.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/cyasdevice.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/device/cyasdevice.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/cyasdevice.c 2010-08-27 17:49:42.000000000 -0700
@@ -48,6 +48,8 @@ typedef struct cyasdevice {
cy_as_device_handle dev_handle;
/* Handle to the HAL */
cy_as_hal_device_tag hal_tag;
+ spinlock_t common_lock;
+ unsigned long flags;
} cyasdevice ;
/* global ptr to astoria device */
@@ -82,12 +84,19 @@ static void cyasdevice_deinit(cyasdevice
if (cy_as_dev->hal_tag) {
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
- if (stop_o_m_a_p_kernel(dev_handle_name,
+ if (cy_as_hal_omap_pnand_stop(dev_handle_name,
cy_as_dev->hal_tag) != 0)
cy_as_hal_print_message("<1>_cy_as_device: stopping "
- "OMAP kernel HAL failed\n");
+ "OMAP PNAND HAL failed\n");
#endif
+
+ #ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ if (cy_as_hal_omap_cram_stop(dev_handle_name,
+ cy_as_dev->hal_tag) != 0)
+ cy_as_hal_print_message("<1>_cy_as_device: stopping "
+ "OMAP CRAM HAL failed\n");
+ #endif
}
cy_as_hal_print_message("<1>_cy_as_device:HAL layer stopped\n") ;
@@ -126,6 +135,28 @@ static void cy_misc_callback(cy_as_devic
}
}
+void cy_as_acquire_common_lock()
+{
+ /*printk("%s: lock address is 0x%x, is_locked=0x%x\n", __func__,
+ &cy_as_device_controller->common_lock,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+ spin_lock_irqsave(&cy_as_device_controller->common_lock,
+ cy_as_device_controller->flags);/*
+ printk("%s: -- is_locked=0x%x\n", __func__,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+}
+EXPORT_SYMBOL(cy_as_acquire_common_lock);
+
+void cy_as_release_common_lock()
+{
+ /*printk("%s: lock address is 0x%x, is_locked=0x%x\n", __func__,
+ &cy_as_device_controller->common_lock,
+ spin_is_locked(&cy_as_device_controller->common_lock));*/
+ spin_unlock_irqrestore(&cy_as_device_controller->common_lock,
+ cy_as_device_controller->flags);
+}
+EXPORT_SYMBOL(cy_as_release_common_lock);
+
/* reset astoria and reinit all regs */
#define PNAND_REG_CFG_INIT_VAL 0x0000
void hal_reset(cy_as_hal_device_tag tag)
@@ -188,7 +219,7 @@ static int cyasdevice_initialize(void)
cy_as_device_init_done = 0;
cy_as_misc_set_log_level(8);
-
+
cy_as_hal_print_message("<1>_cy_as_device initialize called\n") ;
if (cy_as_device_controller != 0) {
@@ -212,7 +243,7 @@ static int cyasdevice_initialize(void)
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
/* start OMAP HAL init instsnce */
- if (!start_o_m_a_p_kernel(dev_handle_name,
+ if (!cy_as_hal_omap_pnand_start(dev_handle_name,
&(cy_as_dev->hal_tag), cy_false)) {
cy_as_hal_print_message(
@@ -221,6 +252,17 @@ static int cyasdevice_initialize(void)
}
#endif
+ #ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ /* start OMAP HAL init instsnce */
+
+ if (!cy_as_hal_omap_cram_start(dev_handle_name,
+ &(cy_as_dev->hal_tag), cy_false)) {
+
+ cy_as_hal_print_message(
+ "<1>_cy_as_device: start OMAP34xx HAL failed\n") ;
+ goto done;
+ }
+ #endif
/* Now create the device */
if (cy_as_misc_create_device(&(cy_as_dev->dev_handle),
cy_as_dev->hal_tag) != CY_AS_ERROR_SUCCESS) {
@@ -331,9 +373,15 @@ static int cyasdevice_initialize(void)
((ver_data.is_debug_mode) ? "debug" : "release"),
ver_data.major, ver_data.minor, ver_data.build, str) ;
+ printk("%s: lock address is 0x%x\n", __func__, &cy_as_dev->common_lock);
+ spin_lock_init(&cy_as_dev->common_lock);
+ printk("%s: lock address is 0x%x\n", __func__, &cy_as_dev->common_lock);
+
/* done now */
cy_as_device_controller = cy_as_dev ;
-
+
+
+
return 0 ;
done:
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes
2010-09-02 22:43 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes David Cross
@ 2010-09-05 5:17 ` Greg KH
2010-09-07 19:22 ` [PATCH] west bridge, kconfig and hal fixes David Cross
1 sibling, 0 replies; 12+ messages in thread
From: Greg KH @ 2010-09-05 5:17 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Thu, Sep 02, 2010 at 03:43:54PM -0700, David Cross wrote:
> On Wed, 2010-09-01 at 17:08 -0700, David Cross wrote:
> From: Greg KH [mailto:greg@kroah.com]
> > Sent: Tuesday, August 31, 2010 9:32 AM
> > To: David Cross
> > Cc: gregkh@suse.de; linux-kernel@vger.kernel.org
> > Subject: Re: [PATCH] West Bridge Astoria Driver 2.6.35
> >
> > On Tue, Aug 31, 2010 at 09:17:18AM -0700, David Cross wrote:
> > >
> > > > I get a build error when I try to build this driver:
> > > >
> > > drivers/staging/westbridge/astoria/block/../include/linux/westbridge/c>
> > > yasmisc.h:521:2: error: expected declaration specifiers or '...' before >
> > > 'cy_as_hal_device_tag'
> > >
> > > > It looks like the #include mess isn't working quite properly.
> > >
> > > > Also note that this driver is building on an x86-64 platform, which is
> > > > something that you probably don't want :)
> > >
> > > > So, for now, I'll just mark the driver as CONFIG_BROKEN and can you send
> > > > me some Kconfig patches against the next linux-next release which should
> > > > have this driver in it, so that it will build properly?
> > >
> > > Sure, I can do that. I have since changed the Kconfig structure a bit. It
> > > actually will build properly with the correct .config as it is though. I
> > can
> > > send you the .config if you would like.
> >
> > It's up to the Kconfig rules to ensure that there can never be a
> > "non-correct" .config file, so that needs to be fixed so that the build
> > will never be broken no matter what type of options are selected.
> >
> > Care to send a patch for that?
>
> This patch actually contains the KConfig changes necessary an additional
> HAL layer.
Note, when sending a patch, don't include all of the stuff above, that
makes no sense.
Also, your Subject: should be a bit cleaner. For example, I would edit
this subject from:
Subject: Re: [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes
to
Subject: [PATCH] Staging: westbridge: Kconfig and HAL fixes
The kernel version doesn't matter as I can't patch an old kernel, right?
:)
> The Kconfig changes are pretty closely related to this HAL
> layer, as such this patch is difficult to logically separate. The HAL
> layers also rely on the export of some gpmc functions, which are added
> to this patch. Although I don't own this driver, it is difficult for me
> to understand why there would be issues with exporting these symbols.
> The linux-next tree does not seem to have a config for the zoom2, and
> trying to build it for that board seems to make the compilation break.
> As such, the only thing that I tested was compilation using the two
> different HALs (one of which is added in this patch). Please let me know
> if there are problems or questions with this.
> Thanks,
> David
>
> Signed-off-by: David Cross <david.cross@cypress.com>
You really need all of this at once to fix the Kconfig issues? It's a
huge patch, what happened that made it needed?
> diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/arch/arm/mach-omap2/gpmc.c linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c
> --- linux-next-vanilla/arch/arm/mach-omap2/gpmc.c 2010-08-31 19:32:51.000000000 -0700
> +++ linux-next-incl-sdk/arch/arm/mach-omap2/gpmc.c 2010-09-01 16:10:21.000000000 -0700
> @@ -133,6 +133,7 @@ void gpmc_cs_write_reg(int cs, int idx,
> reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
> __raw_writel(val, reg_addr);
> }
> +EXPORT_SYMBOL(gpmc_cs_write_reg);
>
> u32 gpmc_cs_read_reg(int cs, int idx)
> {
> @@ -141,6 +142,7 @@ u32 gpmc_cs_read_reg(int cs, int idx)
> reg_addr = gpmc_base + GPMC_CS0_OFFSET + (cs * GPMC_CS_SIZE) + idx;
> return __raw_readl(reg_addr);
> }
> +EXPORT_SYMBOL(gpmc_cs_read_reg);
>
> /* TODO: Add support for gpmc_fck to clock framework and use it */
> unsigned long gpmc_get_fclk_period(void)
> @@ -294,6 +296,7 @@ int gpmc_cs_set_timings(int cs, const st
>
> return 0;
> }
> +EXPORT_SYMBOL(gpmc_cs_set_timings);
>
> static void gpmc_cs_enable_mem(int cs, u32 base, u32 size)
> {
Sorry, I can't change things outside of the drivers/staging/ directory
unless you get the maintainer/owner of this file to agree to export
these symbols.
Can you do that as a separate patch and get their ack? Then I can apply
it.
Also, how about "EXPORT_SYMBOL_GPL()" instead?
So, care to redo this patch?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty
2010-09-02 22:47 ` [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty David Cross
@ 2010-09-05 5:20 ` Greg KH
0 siblings, 0 replies; 12+ messages in thread
From: Greg KH @ 2010-09-05 5:20 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Thu, Sep 02, 2010 at 03:47:04PM -0700, David Cross wrote:
> This patch contains update to the gadget driver, some of which are based on feedback from the Linux community concerning the usage of fat_get_block.
> It also contains the addition of mpage_clear_dirty, a function used to clear dirty pages from the page cache. These two patches are also interdependent
> and difficult to separate. As such, they are included as one patch.
Please wrap your lines at 72 columns.
And is this really hard to separate? I can't take a patch that touches
anything outside of drivers/staging/ at the moment.
How about a patch here to fix up the locking issues, and the rest as a
separate patch?
Oh, you also added lines with trailing spaces and that's it, like the
following:
> @@ -1158,10 +1145,10 @@ static int cyasgadget_ioctl(
> cy_as_hal_print_message("%s: fhandle is OK, "
> "calling vfs_getattr\n", __func__);
> #endif
> -
> +
> ret_stat = vfs_getattr(file_to_allocate->f_path.mnt,
> file_to_allocate->f_path.dentry, &stat);
> -
> +
> #ifndef WESTBRIDGE_NDEBUG
> cy_as_hal_print_message("%s: returned from "
> "vfs_getattr() stat->blksize=0x%lx\n",
Why? I think you might want to make sure your editor doesn't do this in
the future.
Also, if you ran your patch through the scripts/checkpatch.pl script, it
would have caught this error.
Care to redo it?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates
2010-09-02 22:49 ` [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates David Cross
@ 2010-09-05 5:24 ` Greg KH
0 siblings, 0 replies; 12+ messages in thread
From: Greg KH @ 2010-09-05 5:24 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Thu, Sep 02, 2010 at 03:49:54PM -0700, David Cross wrote:
> This patch contains minor updates to the west bridge block and device
> drivers, including the addition of a common lock and a minor update to
> remove the blk_fs_request call based on a kernel update. These changes
> could possibly be separated, but are very minor in scope and as such
> have been submitted together.
Again, one patch per logical change please. That's the way patches for
the kernel work. So can you please separate these and resend them?
> Signed-off-by: David Cross <david.cross@cypress.com>
>
> diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c
> --- linux-next-vanilla/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-31 19:32:51.000000000 -0700
> +++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/block/cyasblkdev_block.c 2010-08-25 18:29:44.000000000 -0700
> @@ -389,7 +389,7 @@ int cyasblkdev_media_changed(struct gend
> struct cyasblkdev_blk_data *bd;
>
> #ifndef WESTBRIDGE_NDEBUG
> - cy_as_hal_print_message("cyasblkdev_media_changed() is called\n");
> + cy_as_hal_print_message(KERN_INFO"cyasblkdev_media_changed() is called\n");
Ick. How about a ' ' after KERN_INFO? And do you really always want to
do this? You can enable tracing to get this type of info.
> #endif
The whole NDEBUG stuff should be removed from the .c files and put in
the .h file if it's really needed.
> @@ -493,7 +493,7 @@ static void cyasblkdev_issuecallback(
> {
> int retry_cnt = 0;
> DBGPRN_FUNC_NAME;
> -
> +
> if (status != CY_AS_ERROR_SUCCESS) {
> #ifndef WESTBRIDGE_NDEBUG
> cy_as_hal_print_message(
You added a line of trailing spaces for no good reason :(
> @@ -509,7 +509,10 @@ static void cyasblkdev_issuecallback(
> __func__, (unsigned int) gl_bd->queue.req, status,
> (unsigned int) blk_rq_sectors(gl_bd->queue.req)) ;
> #endif
> -
> +
> + if(rq_data_dir(gl_bd->queue.req) != READ) {
> + cy_as_release_common_lock();
> + }
No need for the extra {}, please take a look at
Documentation/CodingStyle for how to do things for the kernel. Also a
space is needed after the 'if'.
Please run your patches through the scripts/checkpatch.pl tool before
sending them to me. Yes, there's lots of checkpatch issues in the
driver already, but don't add new ones now please.
Care to redo this?
thanks,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] west bridge, kconfig and hal fixes
2010-09-02 22:43 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes David Cross
2010-09-05 5:17 ` Greg KH
@ 2010-09-07 19:22 ` David Cross
2010-09-07 23:57 ` Greg KH
1 sibling, 1 reply; 12+ messages in thread
From: David Cross @ 2010-09-07 19:22 UTC (permalink / raw)
To: greg; +Cc: linux-kernel
This patch contains the kconfig changes necessary to fix build errors
that could come up in the linux-next version. It also includes an
additional HAL layer for the west bridge CRAM interface. The inclusion
of this interface support did require the reorganization of some of the
existing code, which is part of the reason for the size of the patch.
Moving files and directories makes this patch seem larger than it really
is. The Kconfig changes are closely related to the inclusion of the CRAM
HAL layer, and as such this patch is difficult to logically separate.
The linux-next tree does not seem to have a config for the zoom2, and
trying to build it for that board seems to make the compilation break.
As such, the only thing that I tested was compilation using the two
different HALs (one of which is added in this patch). Please let me know
if there are problems or questions with this.
Thanks,
David
Signed-off-by: David Cross <david.cross@cypress.com>
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_cram.c 2010-09-07 11:57:13.000000000 -0700
@@ -0,0 +1,1845 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_cram.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+#define CYASSTORAGE_WRITE_EP_NUM (4)
+#define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ printk(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ printk(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ printk(KERN_INFO " OMAP pins user_pad cfg with address");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ printk(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+uint32_t cy_as_hal_gpmc_init(cy_as_omap_dev_kernel *dev_p)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ unsigned int cs_mem_base;
+ unsigned int cs_vma_base;
+
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ printk(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, SZ_16K);
+ printk(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ printk(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&cs_mem_base) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ printk(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ cs_mem_base, AST_GPMC_CS);
+ }
+
+ printk(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(cs_mem_base, SZ_16K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ cs_mem_base);
+ goto out_free_cs;
+ }
+
+ printk(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ cs_vma_base = (u32)ioremap_nocache(cs_mem_base, SZ_16K);
+ if (!cs_vma_base) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", cs_mem_base);
+
+ goto out_release_mem_region;
+ }
+ printk(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ cs_mem_base, cs_vma_base);
+
+ dev_p->m_phy_addr_base = (void *) cs_mem_base;
+ dev_p->m_vma_addr_base = (void *) cs_vma_base;
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_on;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_on;
+ timings.adv_rd_off = WB_GPMC_ADV_t_off;
+ timings.adv_wr_off = WB_GPMC_ADV_t_off;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_on;
+ timings.oe_off = WB_GPMC_OE_t_off;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_on;
+ timings.we_off = WB_GPMC_WE_t_off;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(0) |
+ GPMC_CONFIG1_DEVICESIZE_16
+ ));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(cs_mem_base, SZ_16K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ printk(KERN_INFO "AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ printk(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ printk(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ printk(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ printk(KERN_INFO "pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ printk(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ printk(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(cy_as_omap_dev_kernel *dev_p)
+{
+ int i, err;
+
+ printk(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ printk(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ printk(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init(dev_p);
+
+ cy_as_hal_config_c_s_mux();
+
+ return err;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (dev_p->m_vma_addr_base != 0)
+ iounmap((void *)dev_p->m_vma_addr_base);
+
+ if (dev_p->m_phy_addr_base != 0)
+ release_mem_region((void *)dev_p->m_phy_addr_base, SZ_16K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ printk(KERN_INFO "OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ uint16_t write_addr;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk(KERN_INFO "%s: bad TAG parameter passed\n",
+ __func__) ;
+ return ;
+ }
+ #endif
+
+ write_addr = (void *)(dev_p->m_vma_addr_base +
+ CYAS_DEV_CALC_ADDR(addr));
+ writew((unsigned short)data, write_addr) ;
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+ uint32_t read_addr = 0;
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ #ifndef WESTBRIDGE_NDEBUG
+ if (dev_p->m_sig != CY_AS_OMAP_CRAM_HAL_SIG) {
+ printk(KERN_INFO "%s: bad TAG parameter passed\n",
+ __func__) ;
+ return dev_p->m_sig;
+ }
+ #endif
+
+ read_addr = (void *) (dev_p->m_vma_addr_base +
+ CYAS_DEV_CALC_ADDR(addr));
+ data = (unsigned short) readw(read_addr) ;
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, i, size;
+ uint16_t *dptr;
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+ register void *read_addr ;
+ register uint16_t a, b, c, d, e, f, g, h ;
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = (uint16_t *) end_points[ep].data_p;
+
+ read_addr = dev_p->m_vma_addr_base + CYAS_DEV_CALC_EP_ADDR(ep) ;
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * Now, read the data from the device
+ */
+ for (i = size/16 ; i > 0 ; i--) {
+ a = (unsigned short) readw(read_addr) ;
+ b = (unsigned short) readw(read_addr) ;
+ c = (unsigned short) readw(read_addr) ;
+ d = (unsigned short) readw(read_addr) ;
+ e = (unsigned short) readw(read_addr) ;
+ f = (unsigned short) readw(read_addr) ;
+ g = (unsigned short) readw(read_addr) ;
+ h = (unsigned short) readw(read_addr) ;
+ *dptr++ = a ;
+ *dptr++ = b ;
+ *dptr++ = c ;
+ *dptr++ = d ;
+ *dptr++ = e ;
+ *dptr++ = f ;
+ *dptr++ = g ;
+ *dptr++ = h ;
+ }
+
+ switch ((size & 0xF)/2) {
+ case 7:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 6:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 5:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 4:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 3:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 2:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ case 1:
+ *dptr = (unsigned short) readw(read_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ /* Odd sized packet */
+ uint16_t d = (unsigned short) readw(read_addr) ;
+ *((uint8_t *)dptr) = (d & 0xff) ;
+ }
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0, i = 0;
+ uint32_t size;
+ uint16_t *dptr;
+ register void *write_addr ;
+ register uint16_t a, b, c, d ;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ write_addr = (void *) (dev_p->m_vma_addr_base +
+ CYAS_DEV_CALC_EP_ADDR(ep)) ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size) {
+ /*
+ * Now, write the data to the device
+ */
+ for (i = size/8 ; i > 0 ; i--) {
+ a = *dptr++ ;
+ b = *dptr++ ;
+ c = *dptr++ ;
+ d = *dptr++ ;
+ writew(a, write_addr) ;
+ writew(b, write_addr) ;
+ writew(c, write_addr) ;
+ writew(d, write_addr) ;
+ }
+
+ switch ((size & 7)/2) {
+ case 3:
+ writew(*dptr, write_addr) ;
+ dptr++ ;
+ case 2:
+ writew(*dptr, write_addr) ;
+ dptr++ ;
+ case 1:
+ writew(*dptr, write_addr) ;
+ dptr++ ;
+ break ;
+ }
+
+ if (size & 1) {
+ uint16_t v = *((uint8_t *)dptr) ;
+ writew(v, write_addr);
+ }
+ }
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if (state)
+ __gpio_set_value(AST_WAKEUP, 1);
+ else
+ __gpio_set_value(AST_WAKEUP, 0);
+
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+ (void) is_standby_wakeup;
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ uint32_t err = 0;
+ /* No debug mode support through argument as of now */
+ (void)debug;
+
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ;
+ /* No debug mode support through argument as of now */
+ (void)debug;
+
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /* allocate memory for OMAP HAL*/
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_CRAM_HAL_SIG;
+
+ /* initialize OMAP hardware and StartOMAPKernelall gpio pins */
+ err = cy_as_hal_processor_hw_init(dev_p);
+ if (err)
+ goto bus_acc_error;
+
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = cy_as_hal_read_register(
+ (cy_as_hal_device_tag)dev_p, CY_AS_MEM_CM_WB_CFG_ID);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, "
+ "CY_AS_MEM_CM_WB_CFG_ID %4.4x", data16[0]);
+ goto bus_acc_error;
+ }
+
+ printk(KERN_INFO " register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ printk(KERN_INFO "OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_kernel.c 1969-12-31 16:00:00.000000000 -0800
@@ -1,2450 +0,0 @@
-/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
-
-#include <linux/fs.h>
-#include <linux/ioport.h>
-#include <linux/timer.h>
-#include <linux/gpio.h>
-#include <linux/interrupt.h>
-#include <linux/delay.h>
-#include <linux/scatterlist.h>
-#include <linux/mm.h>
-#include <linux/irq.h>
-#include <linux/slab.h>
-#include <linux/sched.h>
-/* include seems broken moving for patch submission
- * #include <mach/mux.h>
- * #include <mach/gpmc.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h>
- * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
- * #include <linux/westbridge/cyaserr.h>
- * #include <linux/westbridge/cyasregs.h>
- * #include <linux/westbridge/cyasdma.h>
- * #include <linux/westbridge/cyasintr.h>
- */
-#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h"
-#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
-#include "../../../include/linux/westbridge/cyaserr.h"
-#include "../../../include/linux/westbridge/cyasregs.h"
-#include "../../../include/linux/westbridge/cyasdma.h"
-#include "../../../include/linux/westbridge/cyasintr.h"
-
-#define HAL_REV "1.1.0"
-
-/*
- * uncomment to enable 16bit pnand interface
- */
-#define PNAND_16BIT_MODE
-
-/*
- * selects one of 3 versions of pnand_lbd_read()
- * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
- * reads through the gpmc CONTROLLER REGISTERS
- * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
- * same speed as the above
- * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
- */
-#define PNAND_LBD_READ_NO_PFE
-/* #define ENABLE_GPMC_PF_ENGINE */
-/* #define PFE_LBD_READ_V2 */
-
-/*
- * westbrige astoria ISR options to limit number of
- * back to back DMA transfers per ISR interrupt
- */
-#define MAX_DRQ_LOOPS_IN_ISR 4
-
-/*
- * debug prints enabling
- *#define DBGPRN_ENABLED
- *#define DBGPRN_DMA_SETUP_RD
- *#define DBGPRN_DMA_SETUP_WR
- */
-
-
-/*
- * For performance reasons, we handle storage endpoint transfers upto 4 KB
- * within the HAL itself.
- */
- #define CYASSTORAGE_WRITE_EP_NUM (4)
- #define CYASSTORAGE_READ_EP_NUM (8)
-
-/*
- * size of DMA packet HAL can accept from Storage API
- * HAL will fragment it into smaller chunks that the P port can accept
- */
-#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
-
-/*
- * P port MAX DMA packet size according to interface/ep configurartion
- */
-#define HAL_DMA_PKT_SZ 512
-
-#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
- ((ep) == 6) || ((ep) == 8))
-
-/*
- * persistant, stores current GPMC interface cfg mode
- */
-static uint8_t pnand_16bit;
-
-/*
- * keep processing new WB DRQ in ISR untill all handled (performance feature)
- */
-#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
-
-
-/*
- * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
- */
-#define CASDO 0x05
-#define CASDI 0x85
-#define RDPAGE_B1 0x00
-#define RDPAGE_B2 0x30
-#define PGMPAGE_B1 0x80
-#define PGMPAGE_B2 0x10
-
-/*
- * The type of DMA operation, per endpoint
- */
-typedef enum cy_as_hal_dma_type {
- cy_as_hal_read,
- cy_as_hal_write,
- cy_as_hal_none
-} cy_as_hal_dma_type ;
-
-
-/*
- * SG list halpers defined in scaterlist.h
-#define sg_is_chain(sg) ((sg)->page_link & 0x01)
-#define sg_is_last(sg) ((sg)->page_link & 0x02)
-#define sg_chain_ptr(sg) \
- ((struct scatterlist *) ((sg)->page_link & ~0x03))
-*/
-typedef struct cy_as_hal_endpoint_dma {
- cy_bool buffer_valid ;
- uint8_t *data_p ;
- uint32_t size ;
- /*
- * sg_list_enabled - if true use, r/w DMA transfers use sg list,
- * FALSE use pointer to a buffer
- * sg_p - pointer to the owner's sg list, of there is such
- * (like blockdriver)
- * dma_xfer_sz - size of the next dma xfer on P port
- * seg_xfer_cnt - counts xfered bytes for in current sg_list
- * memory segment
- * req_xfer_cnt - total number of bytes transfered so far in
- * current request
- * req_length - total request length
- */
- bool sg_list_enabled;
- struct scatterlist *sg_p ;
- uint16_t dma_xfer_sz;
- uint32_t seg_xfer_cnt;
- uint16_t req_xfer_cnt;
- uint16_t req_length;
- cy_as_hal_dma_type type ;
- cy_bool pending ;
-} cy_as_hal_endpoint_dma ;
-
-/*
- * The list of OMAP devices (should be one)
- */
-static cy_as_omap_dev_kernel *m_omap_list_p;
-
-/*
- * The callback to call after DMA operations are complete
- */
-static cy_as_hal_dma_complete_callback callback;
-
-/*
- * Pending data size for the endpoints
- */
-static cy_as_hal_endpoint_dma end_points[16] ;
-
-/*
- * Forward declaration
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
-
-static uint16_t intr_sequence_num;
-static uint8_t intr__enable;
-spinlock_t int_lock ;
-
-static u32 iomux_vma;
-static u32 csa_phy;
-
-/*
- * gpmc I/O registers VMA
- */
-static u32 gpmc_base ;
-
-/*
- * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
- */
-static u32 gpmc_data_vma;
-static u32 ndata_reg_vma;
-static u32 ncmd_reg_vma;
-static u32 naddr_reg_vma;
-
-/*
- * fwd declarations
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
-static inline u16 __attribute__((always_inline))
- ast_p_nand_casdo_read(u8 reg_addr8);
-static inline void __attribute__((always_inline))
- ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
-
-/*
- * prints given number of omap registers
- */
-static void cy_as_hal_print_omap_regs(char *name_prefix,
- u8 name_base, u32 virt_base, u16 count)
-{
- u32 reg_val, reg_addr;
- u16 i;
- cy_as_hal_print_message(KERN_INFO "\n");
- for (i = 0; i < count; i++) {
-
- reg_addr = virt_base + (i*4);
- /* use virtual addresses here*/
- reg_val = __raw_readl(reg_addr);
- cy_as_hal_print_message(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
- name_prefix, name_base+i,
- reg_addr, reg_val);
- }
-}
-
-/*
- * setMUX function for a pad + additional pad flags
- */
-static u16 omap_cfg_reg_L(u32 pad_func_index)
-{
- static u8 sanity_check = 1;
-
- u32 reg_vma;
- u16 cur_val, wr_val, rdback_val;
-
- /*
- * do sanity check on the omap_mux_pin_cfg[] table
- */
- cy_as_hal_print_message(KERN_INFO" OMAP pins user_pad cfg ");
- if (sanity_check) {
- if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
- (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
-
- cy_as_hal_print_message(KERN_INFO
- "table is good.\n");
- } else {
- cy_as_hal_print_message(KERN_WARNING
- "table is bad, fix it");
- }
- /*
- * do it only once
- */
- sanity_check = 0;
- }
-
- /*
- * get virtual address to the PADCNF_REG
- */
- reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
-
- /*
- * add additional USER PU/PD/EN flags
- */
- wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
- cur_val = IORD16(reg_vma);
-
- /*
- * PADCFG regs 16 bit long, packed into 32 bit regs,
- * can also be accessed as u16
- */
- IOWR16(reg_vma, wr_val);
- rdback_val = IORD16(reg_vma);
-
- /*
- * in case if the caller wants to save the old value
- */
- return wr_val;
-}
-
-#define BLKSZ_4K 0x1000
-
-/*
- * switch GPMC DATA bus mode
- */
-void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
-{
- uint32_t tmp32;
-
- /*
- * disable gpmc CS4 operation 1st
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
- GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * GPMC NAND data bus can be 8 or 16 bit wide
- */
- if (dbus16_enabled) {
- DBGPRN("enabling 16 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2) |
- GPMC_CONFIG1_DEVICESIZE_16)
- );
- } else {
- DBGPRN(KERN_INFO "enabling 8 bit bus\n");
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2))
- );
- }
-
- /*
- * re-enable astoria CS operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- *remember the state
- */
- pnand_16bit = dbus16_enabled;
-}
-
-static int cy_as_hal_gpmc_init(void)
-{
- u32 tmp32;
- int err;
- struct gpmc_timings timings;
- /*
- * get GPMC i/o registers base(already been i/o mapped
- * in kernel, no need for separate i/o remap)
- */
- gpmc_base = phys_to_virt(OMAP34XX_GPMC_BASE);
- DBGPRN(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
- gpmc_base, __raw_readl(gpmc_base)
- );
-
- /*
- * these are globals are full VMAs of the gpmc_base above
- */
- ncmd_reg_vma = GPMC_VMA(GPMC_NAND_CMD);
- naddr_reg_vma = GPMC_VMA(GPMC_NAND_ADDR);
- ndata_reg_vma = GPMC_VMA(GPMC_NAND_DATA);
-
- /*
- * request GPMC CS for ASTORIA request
- */
- if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
- cy_as_hal_print_message(KERN_ERR "error failed to request"
- "ncs4 for ASTORIA\n");
- return -1;
- } else {
- DBGPRN(KERN_INFO "got phy_addr:%x for "
- "GPMC CS%d GPMC_CFGREG7[CS4]\n",
- csa_phy, AST_GPMC_CS);
- }
-
- /*
- * request VM region for 4K addr space for chip select 4 phy address
- * technically we don't need it for NAND devices, but do it anyway
- * so that data read/write bus cycle can be triggered by reading
- * or writing this mem region
- */
- if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
- err = -EBUSY;
- cy_as_hal_print_message(KERN_ERR "error MEM region "
- "request for phy_addr:%x failed\n",
- csa_phy);
- goto out_free_cs;
- }
-
- /*
- * REMAP mem region associated with our CS
- */
- gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
- if (!gpmc_data_vma) {
- err = -ENOMEM;
- cy_as_hal_print_message(KERN_ERR "error- ioremap()"
- "for phy_addr:%x failed", csa_phy);
-
- goto out_release_mem_region;
- }
- cy_as_hal_print_message(KERN_INFO "ioremap(%x) returned vma=%x\n",
- csa_phy, gpmc_data_vma);
-
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- memset(&timings, 0, sizeof(timings));
-
- /* cs timing */
- timings.cs_on = WB_GPMC_CS_t_o_n;
- timings.cs_wr_off = WB_GPMC_BUSCYC_t;
- timings.cs_rd_off = WB_GPMC_BUSCYC_t;
-
- /* adv timing */
- timings.adv_on = WB_GPMC_ADV_t_o_n;
- timings.adv_rd_off = WB_GPMC_BUSCYC_t;
- timings.adv_wr_off = WB_GPMC_BUSCYC_t;
-
- /* oe timing */
- timings.oe_on = WB_GPMC_OE_t_o_n;
- timings.oe_off = WB_GPMC_OE_t_o_f_f;
- timings.access = WB_GPMC_RD_t_a_c_c;
- timings.rd_cycle = WB_GPMC_BUSCYC_t;
-
- /* we timing */
- timings.we_on = WB_GPMC_WE_t_o_n;
- timings.we_off = WB_GPMC_WE_t_o_f_f;
- timings.wr_access = WB_GPMC_WR_t_a_c_c;
- timings.wr_cycle = WB_GPMC_BUSCYC_t;
-
- timings.page_burst_access = WB_GPMC_BUSCYC_t;
- timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
- gpmc_cs_set_timings(AST_GPMC_CS, &timings);
-
- cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
- GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
-
- /*
- * DISABLE cs4, NOTE GPMC REG7 is already configured
- * at this point by gpmc_cs_request
- */
- tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
- ~GPMC_CONFIG7_CSVALID;
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
-
- /*
- * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (AS_CS_MASK | AS_CS_BADDR));
-
- /*
- * by default configure GPMC into 8 bit mode
- * (to match astoria default mode)
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
- (GPMC_CONFIG1_DEVICETYPE(2) |
- GPMC_CONFIG1_WAIT_PIN_SEL(2)));
-
- /*
- * ENABLE astoria cs operation on GPMC
- */
- gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
- (tmp32 | GPMC_CONFIG7_CSVALID));
-
- /*
- * No method currently exists to write this register through GPMC APIs
- * need to change WAIT2 polarity
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
- IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
-
- tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
- cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
-
- return 0;
-
-out_release_mem_region:
- release_mem_region(csa_phy, BLKSZ_4K);
-
-out_free_cs:
- gpmc_cs_free(AST_GPMC_CS);
-
- return err;
-}
-
-/*
- * west bridge astoria ISR (Interrupt handler)
- */
-static irqreturn_t cy_astoria_int_handler(int irq,
- void *dev_id, struct pt_regs *regs)
-{
- cy_as_omap_dev_kernel *dev_p;
- uint16_t read_val = 0 ;
- uint16_t mask_val = 0 ;
-
- /*
- * debug stuff, counts number of loops per one intr trigger
- */
- uint16_t drq_loop_cnt = 0;
- uint8_t irq_pin;
- /*
- * flags to watch
- */
- const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
- CY_AS_MEM_P0_INTR_REG_MBINT |
- CY_AS_MEM_P0_INTR_REG_PMINT |
- CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
-
- /*
- * sample IRQ pin level (just for statistics)
- */
- irq_pin = __gpio_get_value(AST_INT);
-
- /*
- * this one just for debugging
- */
- intr_sequence_num++ ;
-
- /*
- * astoria device handle
- */
- dev_p = dev_id;
-
- /*
- * read Astoria intr register
- */
- read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) ;
-
- /*
- * save current mask value
- */
- mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG) ;
-
- DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
- intr_sequence_num, read_val);
-
- /*
- * Disable WB interrupt signal generation while we are in ISR
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
- /*
- * this is a DRQ Interrupt
- */
- if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
-
- do {
- /*
- * handle DRQ interrupt
- */
- drq_loop_cnt++;
-
- cy_handle_d_r_q_interrupt(dev_p) ;
-
- /*
- * spending to much time in ISR may impact
- * average system performance
- */
- if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
- break;
-
- /*
- * Keep processing if there is another DRQ int flag
- */
- } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INTR_REG) &
- CY_AS_MEM_P0_INTR_REG_DRQINT);
- }
-
- if (read_val & sentinel)
- cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
-
- DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
- "int_pin:%d DRQ_jobs:%d\n",
- intr_sequence_num,
- mask_val,
- irq_pin,
- drq_loop_cnt);
-
- /*
- * re-enable WB hw interrupts
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
-
- return IRQ_HANDLED ;
-}
-
-static int cy_as_hal_configure_interrupts(void *dev_p)
-{
- int result;
- int irq_pin = AST_INT;
-
- set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
-
- /*
- * for shared IRQS must provide non NULL device ptr
- * othervise the int won't register
- * */
- result = request_irq(OMAP_GPIO_IRQ(irq_pin),
- (irq_handler_t)cy_astoria_int_handler,
- IRQF_SHARED, "AST_INT#", dev_p);
-
- if (result == 0) {
- /*
- * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
- * assigned to this interrupt
- * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
- * of AST_INT GPIO IRQ_ENABLE FLAG
- */
- cy_as_hal_print_message(KERN_INFO"AST_INT omap_pin:"
- "%d assigned IRQ #%d IRQEN1=%d\n",
- irq_pin,
- OMAP_GPIO_IRQ(irq_pin),
- OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
- );
- } else {
- cy_as_hal_print_message("cyasomaphal: interrupt "
- "failed to register\n");
- gpio_free(irq_pin);
- cy_as_hal_print_message(KERN_WARNING
- "ASTORIA: can't get assigned IRQ"
- "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
- }
-
- return result;
-}
-
-/*
- * initialize OMAP pads/pins to user defined functions
- */
-static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- /*
- * browse through the table an dinitiaze the pins
- */
- u32 in_level = 0;
- u16 tmp16, mux_val;
-
- while (pad_cfg_tab->name != NULL) {
-
- if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
-
- pad_cfg_tab->valid = 1;
- mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
-
- /*
- * always set drv level before changing out direction
- */
- __gpio_set_value(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /*
- * "0" - OUT, "1", input omap_set_gpio_direction
- * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
- */
- if (pad_cfg_tab->dir)
- gpio_direction_input(pad_cfg_tab->pin_num);
- else
- gpio_direction_output(pad_cfg_tab->pin_num,
- pad_cfg_tab->drv);
-
- /* sample the pin */
- in_level = __gpio_get_value(pad_cfg_tab->pin_num);
-
- cy_as_hal_print_message(KERN_INFO "configured %s to "
- "OMAP pad_%d, DIR=%d "
- "DOUT=%d, DIN=%d\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num,
- pad_cfg_tab->dir,
- pad_cfg_tab->drv,
- in_level
- );
- } else {
- /*
- * get the pad_mux value to check on the pin_function
- */
- cy_as_hal_print_message(KERN_INFO "couldn't cfg pin %d"
- "for signal %s, its already taken\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- }
-
- tmp16 = *(u16 *)PADCFG_VMA
- (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
-
- cy_as_hal_print_message(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
- "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
- pad_cfg_tab->pin_num, tmp16,
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
- OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
- );
-
- /*
- * next pad_cfg deriptor
- */
- pad_cfg_tab++;
- }
-
- cy_as_hal_print_message(KERN_INFO"pads configured\n");
-}
-
-
-/*
- * release gpios taken by the module
- */
-static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
-{
- while (pad_cfg_tab->name != NULL) {
-
- if (pad_cfg_tab->valid) {
- gpio_free(pad_cfg_tab->pin_num);
- pad_cfg_tab->valid = 0;
- cy_as_hal_print_message(KERN_INFO "GPIO_%d "
- "released from %s\n",
- pad_cfg_tab->pin_num,
- pad_cfg_tab->name);
- } else {
- cy_as_hal_print_message(KERN_INFO "no release "
- "for %s, GPIO_%d, wasn't acquired\n",
- pad_cfg_tab->name,
- pad_cfg_tab->pin_num);
- }
- pad_cfg_tab++;
- }
-}
-
-void cy_as_hal_config_c_s_mux(void)
-{
- /*
- * FORCE the GPMC CS4 pin (it is in use by the zoom system)
- */
- omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
-}
-EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
-
-/*
- * inits all omap h/w
- */
-uint32_t cy_as_hal_processor_hw_init(void)
-{
- int i, err;
-
- cy_as_hal_print_message(KERN_INFO "init OMAP3430 hw...\n");
-
- iomux_vma = (u32)ioremap_nocache(
- (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
- cy_as_hal_print_message(KERN_INFO "PADCONF_VMA=%x val=%x\n",
- iomux_vma, IORD32(iomux_vma));
-
- /*
- * remap gpio banks
- */
- for (i = 0; i < 6; i++) {
- gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
- gpio_vma_tab[i].phy_addr,
- gpio_vma_tab[i].size);
-
- cy_as_hal_print_message(KERN_INFO "%s virt_addr=%x\n",
- gpio_vma_tab[i].name,
- (u32)gpio_vma_tab[i].virt_addr);
- };
-
- /*
- * force OMAP_GPIO_126 to rleased state,
- * will be configured to drive reset
- */
- gpio_free(AST_RESET);
-
- /*
- *same thing with AStoria CS pin
- */
- gpio_free(AST_CS);
-
- /*
- * initialize all the OMAP pads connected to astoria
- */
- cy_as_hal_init_user_pads(user_pad_cfg);
-
- err = cy_as_hal_gpmc_init();
- if (err < 0)
- cy_as_hal_print_message(KERN_INFO"gpmc init failed:%d", err);
-
- cy_as_hal_config_c_s_mux();
-
- return gpmc_data_vma;
-}
-EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
-
-void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
-{
- /*
- * free omap hw resources
- */
- if (gpmc_data_vma != 0)
- iounmap((void *)gpmc_data_vma);
-
- if (csa_phy != 0)
- release_mem_region(csa_phy, BLKSZ_4K);
-
- gpmc_cs_free(AST_GPMC_CS);
-
- free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
-
- cy_as_hal_release_user_pads(user_pad_cfg);
-}
-
-/*
- * These are the functions that are not part of the
- * HAL layer, but are required to be called for this HAL
- */
-
-/*
- * Called On AstDevice LKM exit
- */
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
-
- /*
- * TODO: Need to disable WB interrupt handlere 1st
- */
- if (0 == dev_p)
- return 1 ;
-
- cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
- if (dev_p->m_sig != CY_AS_OMAP_KERNEL_HAL_SIG) {
- cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
- pgm, __func__) ;
- return 1 ;
- }
-
- /*
- * disable interrupt
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
-
-#if 0
- if (dev_p->thread_flag == 0) {
- dev_p->thread_flag = 1 ;
- wait_for_completion(&dev_p->thread_complete) ;
- cy_as_hal_print_message("cyasomaphal:"
- "done cleaning thread\n");
- cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
- }
-#endif
-
- cy_as_hal_omap_hardware_deinit(dev_p);
-
- /*
- * Rearrange the list
- */
- if (m_omap_list_p == dev_p)
- m_omap_list_p = dev_p->m_next_p ;
-
- cy_as_hal_free(dev_p) ;
-
- cy_as_hal_print_message(KERN_INFO"OMAP_kernel_hal stopped\n");
- return 0;
-}
-
-int omap_start_intr(cy_as_hal_device_tag tag)
-{
- cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
- int ret = 0 ;
- const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
- CY_AS_MEM_P0_INTR_REG_MBINT ;
-
- /*
- * register for interrupts
- */
- ret = cy_as_hal_configure_interrupts(dev_p) ;
-
- /*
- * enable only MBox & DRQ interrupts for now
- */
- cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_INT_MASK_REG, mask) ;
-
- return 1 ;
-}
-
-/*
- * Below are the functions that communicate with the WestBridge device.
- * These are system dependent and must be defined by the HAL layer
- * for a given system.
- */
-
-/*
- * GPMC NAND command+addr write phase
- */
-static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
-{
- /*
- * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
- */
- u32 tmpa32 = ((row_addr << 16) | col_addr);
- u8 RA2 = (u8)(row_addr >> 16);
-
- if (!pnand_16bit) {
- /*
- * GPMC PNAND 8bit BUS
- */
- /*
- * CMD1
- */
- IOWR8(ncmd_reg_vma, cmdb1);
-
- /*
- *pnand bus: <CA0,CA1,RA0,RA1>
- */
- IOWR32(naddr_reg_vma, tmpa32);
-
- /*
- * <RA2> , always zero
- */
- IOWR8(naddr_reg_vma, RA2);
-
- } else {
- /*
- * GPMC PNAND 16bit BUS , in 16 bit mode CMD
- * and ADDR sent on [d7..d0]
- */
- uint8_t CA0, CA1, RA0, RA1;
- CA0 = tmpa32 & 0x000000ff;
- CA1 = (tmpa32 >> 8) & 0x000000ff;
- RA0 = (tmpa32 >> 16) & 0x000000ff;
- RA1 = (tmpa32 >> 24) & 0x000000ff;
-
- /*
- * can't use 32 bit writes here omap will not serialize
- * them to lower half in16 bit mode
- */
-
- /*
- *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
- */
- IOWR8(ncmd_reg_vma, cmdb1);
- IOWR8(naddr_reg_vma, CA0);
- IOWR8(naddr_reg_vma, CA1);
- IOWR8(naddr_reg_vma, RA0);
- IOWR8(naddr_reg_vma, RA1);
- IOWR8(naddr_reg_vma, RA2);
- }
-}
-
-/*
- * spin until r/b goes high
- */
-inline int wait_rn_b_high(void)
-{
- u32 w_spins = 0;
-
- /*
- * TODO: note R/b may go low here, need to spin until high
- * while (omap_get_gpio_datain(AST_RnB) == 0) {
- * w_spins++;
- * }
- * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- *
- * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
- * w_spins++;
- * }
- * printk("<1>RnB=0!:%d\n",w_spins);
- * }
- */
- return w_spins;
-}
-
-#ifdef ENABLE_GPMC_PF_ENGINE
-/* #define PFE_READ_DEBUG
- * PNAND block read with OMAP PFE enabled
- * status: Not tested, NW, broken , etc
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint8_t bytes_in_fifo;
-
- /* debug vars*/
-#ifdef PFE_READ_DEBUG
- uint32_t loop_limit;
- uint16_t bytes_read = 0;
-#endif
-
- /*
- * configure the prefetch engine
- */
- uint32_t tmp32;
- uint32_t pfe_status;
-
- /*
- * DISABLE GPMC CS4 operation 1st, this is
- * in case engine is be already disabled
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
-
-#ifdef PFE_READ_DEBUG
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
- if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
- printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
- tmp32, GPMC_PREFETCH_CONFIG1_VAL);
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
- }
-
- /*
- *sanity check 2
- */
- tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
- if (tmp32 != (count))
- printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
- "!= VAL written:%d\n", tmp32, count);
-#endif
-
- /*
- * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
- * on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
-
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * start the prefetch engine
- */
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- while (1) {
- /*
- * GPMC PFE service loop
- */
- do {
- /*
- * spin until PFE fetched some
- * PNAND bus words in the FIFO
- */
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
- bytes_in_fifo = (pfe_status >> 24) & 0x7f;
- } while (bytes_in_fifo == 0);
-
- /* whole 32 bit words in fifo */
- w32cnt = bytes_in_fifo >> 2;
-
-#if 0
- /*
- *NOTE: FIFO_PTR indicates number of NAND bus words bytes
- * already received in the FIFO and available to be read
- * by DMA or MPU whether COUNTVAL indicates number of BUS
- * words yet to be read from PNAND bus words
- */
- printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
- pfe_status, bytes_in_fifo,
- (pfe_status & 0x3fff), w32cnt);
-#endif
-
- while (w32cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- if ((pfe_status & 0x3fff) == 0) {
- /*
- * PFE acc angine done, there still may be data leftover
- * in the FIFO re-read FIFO BYTE counter (check for
- * leftovers from 32 bit read accesses above)
- */
- bytes_in_fifo = (IORD32(
- GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- /*
- * NOTE we may still have one word left in the fifo
- * read it out
- */
- ptr8 = ptr32;
- switch (bytes_in_fifo) {
-
- case 0:
- /*
- * nothing to do we already read the
- * FIFO out with 32 bit accesses
- */
- break;
- case 1:
- /*
- * this only possible
- * for 8 bit pNAND only
- */
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 2:
- /*
- * this one can occur in either modes
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- break;
-
- case 3:
- /*
- * this only possible for 8 bit pNAND only
- */
- *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
- ptr8 += 2;
- *ptr8 = IORD8(gpmc_data_vma);
- break;
-
- case 4:
- /*
- * shouldn't happen, but has been seen
- * in 8 bit mode
- */
- *ptr32 = IORD32(gpmc_data_vma);
- break;
-
- default:
- printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
- bytes_in_fifo);
- break;
- }
- /*
- * read is completed, get out of the while(1) loop
- */
- break;
- }
- }
-}
-#endif
-
-#ifdef PFE_LBD_READ_V2
-/*
- * PFE engine assisted reads with the 64 byte blocks
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint8_t rd_cnt;
- uint32_t *ptr32;
- uint8_t *ptr8;
- uint16_t reminder;
- uint32_t pfe_status;
-
- /*
- * ISSUE PNAND CMD+ADDR
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * setup PFE block
- * count - OMAP number of bytes to access on pnand bus
- */
-
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
- IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
-
- ptr32 = buff;
-
- do {
- pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
- rd_cnt = pfe_status >> (24+2);
-
- while (rd_cnt--)
- *ptr32++ = IORD32(gpmc_data_vma);
-
- } while (pfe_status & 0x3fff);
-
- /*
- * read out the leftover
- */
- ptr8 = ptr32;
- rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
-
- while (rd_cnt--)
- *ptr8++ = IORD8(gpmc_data_vma);
-}
-#endif
-
-#ifdef PNAND_LBD_READ_NO_PFE
-/*
- * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
- * the original working code, works at max speed for 8 bit xfers
- * for 16 bit the bus diagram has gaps
- */
-static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint32_t *ptr32;
- uint16_t *ptr16;
- uint16_t remainder;
-
- DBGPRN("<1> %s(): NO_PFE\n", __func__);
-
- ptr32 = buff;
- /* number of whole 32 bit words in the transfer */
- w32cnt = count >> 2;
-
- /* remainder, in bytes(0..3) */
- remainder = count & 03;
-
- /*
- * note gpmc puts 32b words on the bus least sig. byte 1st
- */
- nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
- IOWR8(ncmd_reg_vma, RDPAGE_B2);
-
- /*
- * read data by 32 bit chunks
- */
- while (w32cnt--)
- *ptr32++ = IORD32(ndata_reg_vma);
-
- /*
- * now do the remainder(it can be 0, 1, 2 or 3)
- * same code for both 8 & 16 bit bus
- * do 1 or 2 MORE words
- */
- ptr16 = (uint16_t *)ptr32;
-
- switch (remainder) {
- case 1:
- /* read one 16 bit word
- * IN 8 BIT WE NEED TO READ even number of bytes
- */
- case 2:
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- *ptr16++ = IORD16(ndata_reg_vma);
- *ptr16 = IORD16(ndata_reg_vma);
- break;
- default:
- /*
- * remainder is 0
- */
- break;
- }
-}
-#endif
-
-/*
- * uses LBD mode to write N bytes into astoria
- * Status: Working, however there are 150ns idle
- * timeafter every 2 (16 bit or 4(8 bit) bus cycles
- */
-static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
-{
- uint16_t w32cnt;
- uint16_t remainder;
- uint8_t *ptr8;
- uint16_t *ptr16;
- uint32_t *ptr32;
-
- remainder = count & 03;
- w32cnt = count >> 2;
- ptr32 = buff;
- ptr8 = buff;
-
- /*
- * send: CMDB1, CA0,CA1,RA0,RA1,RA2
- */
- nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
-
- /*
- * blast the data out in 32bit chunks
- */
- while (w32cnt--)
- IOWR32(ndata_reg_vma, *ptr32++);
-
- /*
- * do the reminder if there is one
- * same handling for both 8 & 16 bit pnand: mode
- */
- ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
-
- switch (remainder) {
- case 1:
- /*
- * read one 16 bit word
- */
- case 2:
- IOWR16(ndata_reg_vma, *ptr16);
- break;
-
- case 3:
- /*
- * for 3 bytes read 2 16 bit words
- */
- IOWR16(ndata_reg_vma, *ptr16++);
- IOWR16(ndata_reg_vma, *ptr16);
- break;
- default:
- /*
- * reminder is 0
- */
- break;
- }
- /*
- * finally issue a PGM cmd
- */
- IOWR8(ncmd_reg_vma, PGMPAGE_B2);
-}
-
-/*
- * write Astoria register
- */
-static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
-{
- unsigned long flags;
- u16 addr16;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t rdreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&rdreg_usage_cnt) != 0) {
- cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
- "* cy_as_hal_write_register usage:%d\n",
- atomic_read(&rdreg_usage_cnt));
- }
-
- atomic_inc(&rdreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- */
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
-
- /*
- * this should be sent on the 16 bit bus
- */
- IOWR16(ndata_reg_vma, data);
- } else {
- /*
- * 8 bit nand mode GPMC will automatically
- * seriallize 16bit or 32 bit writes into
- * 8 bit onesto the lower 8 bit in LE order
- */
- addr16 = 0x0c00 | reg_addr8;
-
- /*
- * CMD1, CA1, CA2,
- */
- IOWR8(ncmd_reg_vma, 0x85);
- IOWR16(naddr_reg_vma, addr16);
- IOWR16(ndata_reg_vma, data);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&rdreg_usage_cnt);
- local_irq_restore(flags);
-}
-
-
-/*
- * read astoria register via pNAND interface
- */
-static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
-{
- u16 data;
- u16 addr16;
- unsigned long flags;
- /*
- * throw an error if called from multiple threads
- */
- static atomic_t wrreg_usage_cnt = { 0 };
-
- /*
- * disable interrupts
- */
- local_irq_save(flags);
-
- if (atomic_read(&wrreg_usage_cnt) != 0) {
- /*
- * if it gets here ( from other threads), this function needs
- * need spin_lock_irq save() protection
- */
- cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
- "cy_as_hal_write_register usage:%d\n",
- atomic_read(&wrreg_usage_cnt));
- }
- atomic_inc(&wrreg_usage_cnt);
-
- /*
- * 2 flavors of GPMC -> PNAND access
- */
- if (pnand_16bit) {
- /*
- * 16 BIT gpmc NAND mode
- * CMD1, CA1, CA2,
- */
-
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR8(naddr_reg_vma, reg_addr8);
- IOWR8(naddr_reg_vma, 0x0c);
- IOWR8(ncmd_reg_vma, 0x00E0);
-
- udelay(1);
-
- /*
- * much faster through the gPMC Register space
- */
- data = IORD16(ndata_reg_vma);
- } else {
- /*
- * 8 BIT gpmc NAND mode
- * CMD1, CA1, CA2, CMD2
- */
- addr16 = 0x0c00 | reg_addr8;
- IOWR8(ncmd_reg_vma, 0x05);
- IOWR16(naddr_reg_vma, addr16);
- IOWR8(ncmd_reg_vma, 0xE0);
- udelay(1);
- data = IORD16(ndata_reg_vma);
- }
-
- /*
- * re-enable interrupts
- */
- atomic_dec(&wrreg_usage_cnt);
- local_irq_restore(flags);
-
- return data;
-}
-
-
-/*
- * This function must be defined to write a register within the WestBridge
- * device. The addr value is the address of the register to write with
- * respect to the base address of the WestBridge device.
- */
-void cy_as_hal_write_register(
- cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data)
-{
- ast_p_nand_casdi_write((u8)addr, data);
-}
-
-/*
- * This function must be defined to read a register from the WestBridge
- * device. The addr value is the address of the register to read with
- * respect to the base address of the WestBridge device.
- */
-uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
-{
- uint16_t data = 0 ;
-
- /*
- * READ ASTORIA REGISTER USING CASDO
- */
- data = ast_p_nand_casdo_read((u8)addr);
-
- return data ;
-}
-
-/*
- * preps Ep pointers & data counters for next packet
- * (fragment of the request) xfer returns true if
- * there is a next transfer, and false if all bytes in
- * current request have been xfered
- */
-static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
-{
-
- if (!end_points[ep].sg_list_enabled) {
- /*
- * no further transfers for non storage EPs
- * (like EP2 during firmware download, done
- * in 64 byte chunks)
- */
- if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
- DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
- __func__, end_points[ep].req_length, ep);
-
- /*
- * no more transfers, we are done with the request
- */
- return false;
- }
-
- /*
- * calculate size of the next DMA xfer, corner
- * case for non-storage EPs where transfer size
- * is not egual N * HAL_DMA_PKT_SZ xfers
- */
- if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
- >= HAL_DMA_PKT_SZ) {
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
- } else {
- /*
- * that would be the last chunk less
- * than P-port max size
- */
- end_points[ep].dma_xfer_sz = end_points[ep].req_length -
- end_points[ep].req_xfer_cnt;
- }
-
- return true;
- }
-
- /*
- * for SG_list assisted dma xfers
- * are we done with current SG ?
- */
- if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
- /*
- * was it the Last SG segment on the list ?
- */
- if (sg_is_last(end_points[ep].sg_p)) {
- DBGPRN("<1> %s: EP:%d completed,"
- "%d bytes xfered\n",
- __func__,
- ep,
- end_points[ep].req_xfer_cnt
- );
-
- return false;
- } else {
- /*
- * There are more SG segments in current
- * request's sg list setup new segment
- */
-
- end_points[ep].seg_xfer_cnt = 0;
- end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
- /* set data pointer for next DMA sg transfer*/
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- DBGPRN("<1> %s new SG:_va:%p\n\n",
- __func__, end_points[ep].data_p);
- }
-
- }
-
- /*
- * for sg list xfers it will always be 512 or 1024
- */
- end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
-
- /*
- * next transfer is required
- */
-
- return true;
-}
-
-/*
- * Astoria DMA read request, APP_CPU reads from WB ep buffer
- */
-static void cy_service_e_p_dma_read_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- uint16_t v, size;
- void *dptr;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
-
- /*
- * get the XFER size frtom WB eP DMA REGISTER
- */
- v = cy_as_hal_read_register(tag, ep_dma_reg);
-
- /*
- * amount of data in EP buff in bytes
- */
- size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
-
- /*
- * memory pointer for this DMA packet xfer (sub_segment)
- */
- dptr = end_points[ep].data_p;
-
- DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
- ep,
- size,
- intr_sequence_num,
- dptr
- );
-
- cy_as_hal_assert(size != 0);
-
- if (size) {
- /*
- * the actual WB-->OMAP memory "soft" DMA xfer
- */
- p_nand_lbd_read(col_addr, row_addr, size, dptr);
- }
-
- /*
- * clear DMAVALID bit indicating that the data has been read
- */
- cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
-
- /*
- * pre-advance data pointer (if it's outside sg
- * list it will be reset anyway
- */
- end_points[ep].data_p += size;
-
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * we have more data to read in this request,
- * setup next dma packet due tell WB how much
- * data we are going to xfer next
- */
- v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, ep_dma_reg, v);
- } else {
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
- end_points[ep].req_xfer_cnt);
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * omap_cpu needs to transfer data to ASTORIA EP buffer
- */
-static void cy_service_e_p_dma_write_request(
- cy_as_omap_dev_kernel *dev_p, uint8_t ep)
-{
- uint16_t addr;
- uint16_t v = 0;
- uint32_t size;
- uint16_t col_addr = 0x0000;
- uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
- void *dptr;
-
- cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
- /*
- * note: size here its the size of the dma transfer could be
- * anything > 0 && < P_PORT packet size
- */
- size = end_points[ep].dma_xfer_sz ;
- dptr = end_points[ep].data_p ;
-
- /*
- * perform the soft DMA transfer, soft in this case
- */
- if (size)
- p_nand_lbd_write(col_addr, row_addr, size, dptr);
-
- end_points[ep].seg_xfer_cnt += size;
- end_points[ep].req_xfer_cnt += size;
- /*
- * pre-advance data pointer
- * (if it's outside sg list it will be reset anyway)
- */
- end_points[ep].data_p += size;
-
- /*
- * now clear DMAVAL bit to indicate we are done
- * transferring data and that the data can now be
- * sent via USB to the USB host, sent to storage,
- * or used internally.
- */
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
- cy_as_hal_write_register(tag, addr, size) ;
-
- /*
- * finally, tell the USB subsystem that the
- * data is gone and we can accept the
- * next request if one exists.
- */
- if (prep_for_next_xfer(tag, ep)) {
- /*
- * There is more data to go. Re-init the WestBridge DMA side
- */
- v = end_points[ep].dma_xfer_sz |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v) ;
- } else {
-
- end_points[ep].pending = cy_false ;
- end_points[ep].type = cy_as_hal_none ;
- end_points[ep].buffer_valid = cy_false ;
-
- /*
- * notify the API that we are done with rq on this EP
- */
- if (callback) {
- /*
- * this callback will wake up the process that might be
- * sleeping on the EP which data is being transferred
- */
- callback(tag, ep,
- end_points[ep].req_xfer_cnt,
- CY_AS_ERROR_SUCCESS);
- }
- }
-}
-
-/*
- * HANDLE DRQINT from Astoria (called in AS_Intr context
- */
-static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
-{
- uint16_t v ;
- static uint8_t service_ep = 2 ;
-
- /*
- * We've got DRQ INT, read DRQ STATUS Register */
- v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
- CY_AS_MEM_P0_DRQ) ;
-
- if (v == 0) {
-#ifndef WESTBRIDGE_NDEBUG
- cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
-#endif
- return;
- }
-
- /*
- * Now, pick a given DMA request to handle, for now, we just
- * go round robin. Each bit position in the service_mask
- * represents an endpoint from EP2 to EP15. We rotate through
- * each of the endpoints to find one that needs to be serviced.
- */
- while ((v & (1 << service_ep)) == 0) {
-
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
- if (end_points[service_ep].type == cy_as_hal_write) {
- /*
- * handle DMA WRITE REQUEST: app_cpu will
- * write data into astoria EP buffer
- */
- cy_service_e_p_dma_write_request(dev_p, service_ep) ;
- } else if (end_points[service_ep].type == cy_as_hal_read) {
- /*
- * handle DMA READ REQUEST: cpu will
- * read EP buffer from Astoria
- */
- cy_service_e_p_dma_read_request(dev_p, service_ep) ;
- }
-#ifndef WESTBRIDGE_NDEBUG
- else
- cy_as_hal_print_message("cyashalomap:interrupt,"
- " w/o pending DMA job,"
- "-check DRQ_MASK logic\n") ;
-#endif
-
- /*
- * Now bump the EP ahead, so other endpoints get
- * a shot before the one we just serviced
- */
- if (end_points[service_ep].type == cy_as_hal_none) {
- if (service_ep == 15)
- service_ep = 2 ;
- else
- service_ep++ ;
- }
-
-}
-
-void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
-{
- DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
- if (end_points[ep].pending)
- cy_as_hal_write_register(tag,
- CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
-
- end_points[ep].buffer_valid = cy_false ;
- end_points[ep].type = cy_as_hal_none;
-}
-
-/*
- * enables/disables SG list assisted DMA xfers for the given EP
- * sg_list assisted XFERS can use physical addresses of mem pages in case if the
- * xfer is performed by a h/w DMA controller rather then the CPU on P port
- */
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
-{
- end_points[ep].sg_list_enabled = sg_xfer_enabled;
- DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
- ep, end_points[ep].sg_list_enabled);
-}
-EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
-
-/*
- * This function must be defined to transfer a block of data to
- * the WestBridge device. This function can use the burst write
- * (DMA) capabilities of WestBridge to do this, or it can just copy
- * the data using writes.
- */
-void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr = 0 ;
- uint16_t v = 0;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem. Another DMA request
- * is coming down before the previous one has completed.
- */
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_write ;
- end_points[ep].pending = cy_true;
-
- /*
- * total length of the request
- */
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * smaller xfers for non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- /*
- * check the EP transfer mode uses sg_list rather then a memory buffer
- * block devices pass it to the HAL, so the hAL could get to the real
- * physical address for each segment and set up a DMA controller
- * hardware ( if there is one)
- */
- if (end_points[ep].sg_list_enabled) {
- /*
- * buf - pointer to the SG list
- * data_p - data pointer to the 1st DMA segment
- * seg_xfer_cnt - keeps track of N of bytes sent in current
- * sg_list segment
- * req_xfer_cnt - keeps track of the total N of bytes
- * transferred for the request
- */
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].req_xfer_cnt = 0;
-
-#ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
- "req_sz:%d, maxsz:%d\n",
- __func__,
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
-#endif
-
- } else {
- /*
- * setup XFER for non sg_list assisted EPs
- */
-
- #ifdef DBGPRN_DMA_SETUP_WR
- DBGPRN("<1>%s non storage or sz < 512:"
- "EP:%d, sz:%d\n", __func__, ep, size);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * will keep track No of bytes xferred for the request
- */
- end_points[ep].req_xfer_cnt = 0;
- }
-
- /*
- * Tell WB we are ready to send data on the given endpoint
- */
- v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
- | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- cy_as_hal_write_register(tag, addr, v) ;
-}
-
-/*
- * This function must be defined to transfer a block of data from
- * the WestBridge device. This function can use the burst read
- * (DMA) capabilities of WestBridge to do this, or it can just
- * copy the data using reads.
- */
-void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf,
- uint32_t size, uint16_t maxsize)
-{
- uint32_t addr ;
- uint16_t v ;
-
- /*
- * Note: "size" is the actual request size
- * "maxsize" - is the P port fragment size
- * No EP0 or EP1 traffic should get here
- */
- cy_as_hal_assert(ep != 0 && ep != 1) ;
-
- /*
- * If this asserts, we have an ordering problem.
- * Another DMA request is coming down before the
- * previous one has completed. we should not get
- * new requests if current is still in process
- */
-
- cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
-
- end_points[ep].buffer_valid = cy_true ;
- end_points[ep].type = cy_as_hal_read ;
- end_points[ep].pending = cy_true;
- end_points[ep].req_xfer_cnt = 0;
- end_points[ep].req_length = size;
-
- if (size >= maxsize) {
- /*
- * set xfer size for very 1st DMA xfer operation
- * port max packet size ( typically 512 or 1024)
- */
- end_points[ep].dma_xfer_sz = maxsize;
- } else {
- /*
- * so that we could handle small xfers on in case
- * of non-storage EPs
- */
- end_points[ep].dma_xfer_sz = size;
- }
-
- addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
-
- if (end_points[ep].sg_list_enabled) {
- /*
- * Handle sg-list assisted EPs
- * seg_xfer_cnt - keeps track of N of sent packets
- * buf - pointer to the SG list
- * data_p - data pointer for the 1st DMA segment
- */
- end_points[ep].seg_xfer_cnt = 0 ;
- end_points[ep].sg_p = buf;
- end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
-
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
- "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
- ep,
- buf,
- end_points[ep].data_p,
- size,
- maxsize);
- #endif
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- } else {
- /*
- * Non sg list EP passed void *buf rather then scatterlist *sg
- */
- #ifdef DBGPRN_DMA_SETUP_RD
- DBGPRN("%s:non-sg_list EP:%d,"
- "RQ_sz:%d, maxsz:%d\n",
- __func__, ep, size, maxsize);
- #endif
-
- end_points[ep].sg_p = NULL;
-
- /*
- * must be a VMA of a membuf in kernel space
- */
- end_points[ep].data_p = buf;
-
- /*
- * Program the EP DMA register for Storage endpoints only.
- */
- if (is_storage_e_p(ep)) {
- v = (end_points[ep].dma_xfer_sz &
- CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
- CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
- cy_as_hal_write_register(tag, addr, v);
- }
- }
-}
-
-/*
- * This function must be defined to allow the WB API to
- * register a callback function that is called when a
- * DMA transfer is complete.
- */
-void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb)
-{
- DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
- __func__, (uint32_t)cb);
- callback = cb ;
-}
-
-/*
- * This function must be defined to return the maximum size of
- * DMA request that can be handled on the given endpoint. The
- * return value should be the maximum size in bytes that the DMA
- * module can handle.
- */
-uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep)
-{
- /*
- * Storage reads and writes are always done in 512 byte blocks.
- * So, we do the count handling within the HAL, and save on
- * some of the data transfer delay.
- */
- if ((ep == CYASSTORAGE_READ_EP_NUM) ||
- (ep == CYASSTORAGE_WRITE_EP_NUM)) {
- /* max DMA request size HAL can handle by itself */
- return CYASSTORAGE_MAX_XFER_SIZE;
- } else {
- /*
- * For the USB - Processor endpoints, the maximum transfer
- * size depends on the speed of USB operation. So, we use
- * the following constant to indicate to the API that
- * splitting of the data into chunks less that or equal to
- * the max transfer size should be handled internally.
- */
-
- /* DEFINED AS 0xffffffff in cyasdma.h */
- return CY_AS_DMA_MAX_SIZE_HW_SIZE;
- }
-}
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the WestBridge device. Generally this is done via a GPIO of
- * some type.
- */
-cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
-{
- /*
- * Not supported as of now.
- */
- return cy_false ;
-}
-
-void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
-{
- cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
- cy_as_hal_print_message("please check the input voltage levels");
- cy_as_hal_print_message("and clock, and restart the system\n") ;
-}
-
-/*
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- */
-
-/*
- * This function is required by the API to allocate memory.
- * This function is expected to work exactly like malloc().
- */
-void *cy_as_hal_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required by the API to free memory allocated
- * with CyAsHalAlloc(). This function is'expected to work exacly
- * like free().
- */
-void cy_as_hal_free(void *mem_p)
-{
- kfree(mem_p) ;
-}
-
-/*
- * Allocator that can be used in interrupt context.
- * We have to ensure that the kmalloc call does not
- * sleep in this case.
- */
-void *cy_as_hal_c_b_alloc(uint32_t cnt)
-{
- void *ret_p ;
-
- ret_p = kmalloc(cnt, GFP_ATOMIC) ;
- return ret_p ;
-}
-
-/*
- * This function is required to set a block of memory to a
- * specific value. This function is expected to work exactly
- * like memset()
- */
-void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
-{
- memset(ptr, value, cnt) ;
-}
-
-/*
- * This function is expected to create a sleep channel.
- * The data structure that represents the sleep channel object
- * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
- * passed as a pointer, and allpocated by the caller
- * (typically as a local var on the stack) "Create" word should read as
- * "SleepOn", this func doesn't actually create anything
- */
-cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- init_waitqueue_head(&channel->wq) ;
- return cy_true ;
-}
-
-/*
- * for this particular HAL it doesn't actually destroy anything
- * since no actual sleep object is created in CreateSleepChannel()
- * sleep channel is given by the pointer in the argument.
- */
-cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
-{
- return cy_true ;
-}
-
-/*
- * platform specific wakeable Sleep implementation
- */
-cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
-{
- wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
- return cy_true ;
-}
-
-/*
- * wakes up the process waiting on the CHANNEL
- */
-cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
-{
- wake_up_interruptible_all(&channel->wq);
- return cy_true ;
-}
-
-uint32_t cy_as_hal_disable_interrupts()
-{
- if (0 == intr__enable)
- ;
-
- intr__enable++ ;
- return 0 ;
-}
-
-void cy_as_hal_enable_interrupts(uint32_t val)
-{
- intr__enable-- ;
- if (0 == intr__enable)
- ;
-}
-
-/*
- * Sleep atleast 150ns, cpu dependent
- */
-void cy_as_hal_sleep150(void)
-{
- uint32_t i, j;
-
- j = 0;
- for (i = 0; i < 1000; i++)
- j += (~i);
-}
-
-void cy_as_hal_sleep(uint32_t ms)
-{
- cy_as_hal_sleep_channel channel;
-
- cy_as_hal_create_sleep_channel(&channel) ;
- cy_as_hal_sleep_on(&channel, ms) ;
- cy_as_hal_destroy_sleep_channel(&channel) ;
-}
-
-cy_bool cy_as_hal_is_polling()
-{
- return cy_false;
-}
-
-void cy_as_hal_c_b_free(void *ptr)
-{
- cy_as_hal_free(ptr);
-}
-
-/*
- * suppose to reinstate the astoria registers
- * that may be clobbered in sleep mode
- */
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
- (void) is_standby_wakeup;
-}
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
-{
- /* specific to SPI, no implementation required */
- (void) tag;
-}
-
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
-{
- /*
- * we are in asynchronous mode. so no need to handle this
- */
- return true;
-}
-
-/*
- * init OMAP h/w resources
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug)
-{
- cy_as_omap_dev_kernel *dev_p ;
- int i;
- u16 data16[4];
- u8 pncfg_reg;
-
- /*
- * No debug mode support through argument as of now
- */
- (void)debug;
-
- DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
-
- /*
- * Initialize the HAL level endpoint DMA data.
- */
- for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
- end_points[i].data_p = 0 ;
- end_points[i].pending = cy_false ;
- end_points[i].size = 0 ;
- end_points[i].type = cy_as_hal_none ;
- end_points[i].sg_list_enabled = cy_false;
-
- /*
- * by default the DMA transfers to/from the E_ps don't
- * use sg_list that implies that the upper devices like
- * blockdevice have to enable it for the E_ps in their
- * initialization code
- */
- }
-
- /*
- * allocate memory for OMAP HAL
- */
- dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
- sizeof(cy_as_omap_dev_kernel)) ;
- if (dev_p == 0) {
- cy_as_hal_print_message("out of memory allocating OMAP"
- "device structure\n") ;
- return 0 ;
- }
-
- dev_p->m_sig = CY_AS_OMAP_KERNEL_HAL_SIG;
-
- /*
- * initialize OMAP hardware and StartOMAPKernelall gpio pins
- */
- dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
-
- /*
- * Now perform a hard reset of the device to have
- * the new settings take effect
- */
- __gpio_set_value(AST_WAKEUP, 1);
-
- /*
- * do Astoria h/w reset
- */
- DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
-
- /*
- * NEGATIVE PULSE on RST pin
- */
- __gpio_set_value(AST_RESET, 0);
- mdelay(1);
- __gpio_set_value(AST_RESET, 1);
- mdelay(50);
-
- /*
- * note AFTER reset PNAND interface is 8 bit mode
- * so if gpmc Is configured in 8 bit mode upper half will be FF
- */
- pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
-#ifdef PNAND_16BIT_MODE
-
- /*
- * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
- */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
-
- /*
- * now in order to continue to talk to astoria
- * sw OMAP GPMC into 16 bit mode as well
- */
- cy_as_hal_gpmc_enable_16bit_bus(cy_true);
-#else
- /* Astoria and GPMC are already in 8 bit mode, jsut initialize PNAND_CFG */
- ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
-#endif
-
- /*
- * NOTE: if you want to capture bus activity on the LA,
- * don't use printks in between the activities you want to capture.
- * prinks may take milliseconds, and the data of interest
- * will fall outside the LA capture window/buffer
- */
- data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
- data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
-
- if (data16[0] != 0xA200) {
- /*
- * astoria device is not found
- */
- printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
- printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
- data16[0], data16[0]);
- goto bus_acc_error;
- }
-
- cy_as_hal_print_message(KERN_INFO" register access CASDO test:"
- "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
- "PNAND_CFG after RST:%4.4x\n "
- "CY_AS_MEM_PNAND_CFG"
- "after cfg_wr:%4.4x\n\n",
- data16[0], pncfg_reg, data16[1]);
-
- dev_p->thread_flag = 1 ;
- spin_lock_init(&int_lock) ;
- dev_p->m_next_p = m_omap_list_p ;
-
- m_omap_list_p = dev_p ;
- *tag = dev_p;
-
- cy_as_hal_configure_interrupts((void *)dev_p);
-
- cy_as_hal_print_message(KERN_INFO"OMAP3430__hal started tag:%p"
- ", kernel HZ:%d\n", dev_p, HZ);
-
- /*
- *make processor to storage endpoints SG assisted by default
- */
- cy_as_hal_set_ep_dma_mode(4, true);
- cy_as_hal_set_ep_dma_mode(8, true);
-
- return 1 ;
-
- /*
- * there's been a NAND bus access error or
- * astoria device is not connected
- */
-bus_acc_error:
- /*
- * at this point hal tag hasn't been set yet
- * so the device will not call omap_stop
- */
- cy_as_hal_omap_hardware_deinit(dev_p);
- cy_as_hal_free(dev_p) ;
- return 0;
-}
-
-#else
-/*
- * Some compilers do not like empty C files, so if the OMAP hal is not being
- * compiled, we compile this single function. We do this so that for a
- * given target HAL there are not multiple sources for the HAL functions.
- */
-void my_o_m_a_p_kernel_hal_dummy_function(void)
-{
-}
-
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/mach-omap2/cyashalomap_pnand.c 2010-09-07 11:50:34.000000000 -0700
@@ -0,0 +1,2462 @@
+/* Cypress WestBridge OMAP3430 Kernel Hal source file (cyashalomap_kernel.c)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+
+#include <linux/fs.h>
+#include <linux/ioport.h>
+#include <linux/timer.h>
+#include <linux/gpio.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/scatterlist.h>
+#include <linux/mm.h>
+#include <linux/irq.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+/* include seems broken moving for patch submission
+ * #include <mach/mux.h>
+ * #include <mach/gpmc.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h>
+ * #include <mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h>
+ * #include <linux/westbridge/cyaserr.h>
+ * #include <linux/westbridge/cyasregs.h>
+ * #include <linux/westbridge/cyasdma.h>
+ * #include <linux/westbridge/cyasintr.h>
+ */
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h"
+#include "../plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h"
+#include "../../../include/linux/westbridge/cyaserr.h"
+#include "../../../include/linux/westbridge/cyasregs.h"
+#include "../../../include/linux/westbridge/cyasdma.h"
+#include "../../../include/linux/westbridge/cyasintr.h"
+
+#define HAL_REV "1.1.0"
+
+/*
+ * uncomment to enable 16bit pnand interface
+ */
+#define PNAND_16BIT_MODE
+
+/*
+ * selects one of 3 versions of pnand_lbd_read()
+ * PNAND_LBD_READ_NO_PFE - original 8/16 bit code
+ * reads through the gpmc CONTROLLER REGISTERS
+ * ENABLE_GPMC_PF_ENGINE - USES GPMC PFE FIFO reads, in 8 bit mode,
+ * same speed as the above
+ * PFE_LBD_READ_V2 - slightly diffrenet, performance same as above
+ */
+#define PNAND_LBD_READ_NO_PFE
+/* #define ENABLE_GPMC_PF_ENGINE */
+/* #define PFE_LBD_READ_V2 */
+
+/*
+ * westbrige astoria ISR options to limit number of
+ * back to back DMA transfers per ISR interrupt
+ */
+#define MAX_DRQ_LOOPS_IN_ISR 4
+
+/*
+ * debug prints enabling
+ *#define DBGPRN_ENABLED
+ *#define DBGPRN_DMA_SETUP_RD
+ *#define DBGPRN_DMA_SETUP_WR
+ */
+
+
+/*
+ * For performance reasons, we handle storage endpoint transfers upto 4 KB
+ * within the HAL itself.
+ */
+#define CYASSTORAGE_WRITE_EP_NUM (4)
+#define CYASSTORAGE_READ_EP_NUM (8)
+
+/*
+ * size of DMA packet HAL can accept from Storage API
+ * HAL will fragment it into smaller chunks that the P port can accept
+ */
+#define CYASSTORAGE_MAX_XFER_SIZE (2*32768)
+
+/*
+ * P port MAX DMA packet size according to interface/ep configurartion
+ */
+#define HAL_DMA_PKT_SZ 512
+
+#define is_storage_e_p(ep) (((ep) == 2) || ((ep) == 4) || \
+ ((ep) == 6) || ((ep) == 8))
+
+/*
+ * persistant, stores current GPMC interface cfg mode
+ */
+static uint8_t pnand_16bit;
+
+/*
+ * keep processing new WB DRQ in ISR untill all handled (performance feature)
+ */
+#define PROCESS_MULTIPLE_DRQ_IN_ISR (1)
+
+
+/*
+ * ASTORIA PNAND IF COMMANDS, CASDO - READ, CASDI - WRITE
+ */
+#define CASDO 0x05
+#define CASDI 0x85
+#define RDPAGE_B1 0x00
+#define RDPAGE_B2 0x30
+#define PGMPAGE_B1 0x80
+#define PGMPAGE_B2 0x10
+
+/*
+ * The type of DMA operation, per endpoint
+ */
+typedef enum cy_as_hal_dma_type {
+ cy_as_hal_read,
+ cy_as_hal_write,
+ cy_as_hal_none
+} cy_as_hal_dma_type ;
+
+
+/*
+ * SG list halpers defined in scaterlist.h
+#define sg_is_chain(sg) ((sg)->page_link & 0x01)
+#define sg_is_last(sg) ((sg)->page_link & 0x02)
+#define sg_chain_ptr(sg) \
+ ((struct scatterlist *) ((sg)->page_link & ~0x03))
+*/
+typedef struct cy_as_hal_endpoint_dma {
+ cy_bool buffer_valid ;
+ uint8_t *data_p ;
+ uint32_t size ;
+ /*
+ * sg_list_enabled - if true use, r/w DMA transfers use sg list,
+ * FALSE use pointer to a buffer
+ * sg_p - pointer to the owner's sg list, of there is such
+ * (like blockdriver)
+ * dma_xfer_sz - size of the next dma xfer on P port
+ * seg_xfer_cnt - counts xfered bytes for in current sg_list
+ * memory segment
+ * req_xfer_cnt - total number of bytes transfered so far in
+ * current request
+ * req_length - total request length
+ */
+ bool sg_list_enabled;
+ struct scatterlist *sg_p ;
+ uint16_t dma_xfer_sz;
+ uint32_t seg_xfer_cnt;
+ uint16_t req_xfer_cnt;
+ uint16_t req_length;
+ cy_as_hal_dma_type type ;
+ cy_bool pending ;
+} cy_as_hal_endpoint_dma ;
+
+/*
+ * The list of OMAP devices (should be one)
+ */
+static cy_as_omap_dev_kernel *m_omap_list_p;
+
+/*
+ * The callback to call after DMA operations are complete
+ */
+static cy_as_hal_dma_complete_callback callback;
+
+/*
+ * Pending data size for the endpoints
+ */
+static cy_as_hal_endpoint_dma end_points[16] ;
+
+/*
+ * Forward declaration
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p);
+
+static uint16_t intr_sequence_num;
+static uint8_t intr__enable;
+spinlock_t int_lock ;
+
+static u32 iomux_vma;
+static u32 csa_phy;
+
+/*
+ * gpmc I/O registers VMA
+ */
+static u32 gpmc_base ;
+
+/*
+ * gpmc data VMA associated with CS4 (ASTORIA CS on GPMC)
+ */
+static u32 gpmc_data_vma;
+static u32 ndata_reg_vma;
+static u32 ncmd_reg_vma;
+static u32 naddr_reg_vma;
+
+/*
+ * fwd declarations
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff);
+static inline u16 __attribute__((always_inline))
+ ast_p_nand_casdo_read(u8 reg_addr8);
+static inline void __attribute__((always_inline))
+ ast_p_nand_casdi_write(u8 reg_addr8, u16 data);
+
+/*
+ * prints given number of omap registers
+ */
+static void cy_as_hal_print_omap_regs(char *name_prefix,
+ u8 name_base, u32 virt_base, u16 count)
+{
+ u32 reg_val, reg_addr;
+ u16 i;
+ printk(KERN_INFO "\n");
+ for (i = 0; i < count; i++) {
+
+ reg_addr = virt_base + (i*4);
+ /* use virtual addresses here*/
+ reg_val = __raw_readl(reg_addr);
+ printk(KERN_INFO "%s_%d[%8.8x]=%8.8x\n",
+ name_prefix, name_base+i,
+ reg_addr, reg_val);
+ }
+}
+
+/*
+ * setMUX function for a pad + additional pad flags
+ */
+static u16 omap_cfg_reg_L(u32 pad_func_index)
+{
+ static u8 sanity_check = 1;
+
+ u32 reg_vma;
+ u16 cur_val, wr_val, rdback_val;
+
+ /*
+ * do sanity check on the omap_mux_pin_cfg[] table
+ */
+ printk(KERN_INFO " OMAP pins user_pad cfg ");
+ if (sanity_check) {
+ if ((omap_mux_pin_cfg[END_OF_TABLE].name[0] == 'E') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[1] == 'N') &&
+ (omap_mux_pin_cfg[END_OF_TABLE].name[2] == 'D')) {
+
+ printk(KERN_INFO
+ "table is good.\n");
+ } else {
+ cy_as_hal_print_message(KERN_WARNING
+ "table is bad, fix it");
+ }
+ /*
+ * do it only once
+ */
+ sanity_check = 0;
+ }
+
+ /*
+ * get virtual address to the PADCNF_REG
+ */
+ reg_vma = (u32)iomux_vma + omap_mux_pin_cfg[pad_func_index].offset;
+
+ /*
+ * add additional USER PU/PD/EN flags
+ */
+ wr_val = omap_mux_pin_cfg[pad_func_index].mux_val;
+ cur_val = IORD16(reg_vma);
+
+ /*
+ * PADCFG regs 16 bit long, packed into 32 bit regs,
+ * can also be accessed as u16
+ */
+ IOWR16(reg_vma, wr_val);
+ rdback_val = IORD16(reg_vma);
+
+ /*
+ * in case if the caller wants to save the old value
+ */
+ return wr_val;
+}
+
+#define BLKSZ_4K 0x1000
+
+/*
+ * switch GPMC DATA bus mode
+ */
+void cy_as_hal_gpmc_enable_16bit_bus(bool dbus16_enabled)
+{
+ uint32_t tmp32;
+
+ /*
+ * disable gpmc CS4 operation 1st
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS,
+ GPMC_CS_CONFIG7) & ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * GPMC NAND data bus can be 8 or 16 bit wide
+ */
+ if (dbus16_enabled) {
+ DBGPRN("enabling 16 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2) |
+ GPMC_CONFIG1_DEVICESIZE_16)
+ );
+ } else {
+ DBGPRN(KERN_INFO "enabling 8 bit bus\n");
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2))
+ );
+ }
+
+ /*
+ * re-enable astoria CS operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ *remember the state
+ */
+ pnand_16bit = dbus16_enabled;
+}
+
+static int cy_as_hal_gpmc_init(void)
+{
+ u32 tmp32;
+ int err;
+ struct gpmc_timings timings;
+ /*
+ * get GPMC i/o registers base(already been i/o mapped
+ * in kernel, no need for separate i/o remap)
+ */
+ printk(KERN_INFO "%s: mapping phys_to_virt\n", __func__);
+
+ gpmc_base = (u32)ioremap_nocache(OMAP34XX_GPMC_BASE, BLKSZ_4K);
+ printk(KERN_INFO "kernel has gpmc_base=%x , val@ the base=%x",
+ gpmc_base, __raw_readl(gpmc_base)
+ );
+
+ /*
+ * these are globals are full VMAs of the gpmc_base above
+ */
+ ncmd_reg_vma = GPMC_VMA(AST_GPMC_NAND_CMD);
+ naddr_reg_vma = GPMC_VMA(AST_GPMC_NAND_ADDR);
+ ndata_reg_vma = GPMC_VMA(AST_GPMC_NAND_DATA);
+
+ printk(KERN_INFO "%s: calling gpmc_cs_request\n", __func__);
+ /*
+ * request GPMC CS for ASTORIA request
+ */
+ if (gpmc_cs_request(AST_GPMC_CS, SZ_16M, (void *)&csa_phy) < 0) {
+ cy_as_hal_print_message(KERN_ERR "error failed to request"
+ "ncs4 for ASTORIA\n");
+ return -1;
+ } else {
+ printk(KERN_INFO "got phy_addr:%x for "
+ "GPMC CS%d GPMC_CFGREG7[CS4]\n",
+ csa_phy, AST_GPMC_CS);
+ }
+
+ printk(KERN_INFO "%s: calling request_mem_region\n", __func__);
+ /*
+ * request VM region for 4K addr space for chip select 4 phy address
+ * technically we don't need it for NAND devices, but do it anyway
+ * so that data read/write bus cycle can be triggered by reading
+ * or writing this mem region
+ */
+ if (!request_mem_region(csa_phy, BLKSZ_4K, "AST_OMAP_HAL")) {
+ err = -EBUSY;
+ cy_as_hal_print_message(KERN_ERR "error MEM region "
+ "request for phy_addr:%x failed\n",
+ csa_phy);
+ goto out_free_cs;
+ }
+
+ printk(KERN_INFO "%s: calling ioremap_nocache\n", __func__);
+
+ /* REMAP mem region associated with our CS */
+ gpmc_data_vma = (u32)ioremap_nocache(csa_phy, BLKSZ_4K);
+ if (!gpmc_data_vma) {
+ err = -ENOMEM;
+ cy_as_hal_print_message(KERN_ERR "error- ioremap()"
+ "for phy_addr:%x failed", csa_phy);
+
+ goto out_release_mem_region;
+ }
+ printk(KERN_INFO "ioremap(%x) returned vma=%x\n",
+ csa_phy, gpmc_data_vma);
+
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ memset(&timings, 0, sizeof(timings));
+
+ /* cs timing */
+ timings.cs_on = WB_GPMC_CS_t_o_n;
+ timings.cs_wr_off = WB_GPMC_BUSCYC_t;
+ timings.cs_rd_off = WB_GPMC_BUSCYC_t;
+
+ /* adv timing */
+ timings.adv_on = WB_GPMC_ADV_t_o_n;
+ timings.adv_rd_off = WB_GPMC_BUSCYC_t;
+ timings.adv_wr_off = WB_GPMC_BUSCYC_t;
+
+ /* oe timing */
+ timings.oe_on = WB_GPMC_OE_t_o_n;
+ timings.oe_off = WB_GPMC_OE_t_o_f_f;
+ timings.access = WB_GPMC_RD_t_a_c_c;
+ timings.rd_cycle = WB_GPMC_BUSCYC_t;
+
+ /* we timing */
+ timings.we_on = WB_GPMC_WE_t_o_n;
+ timings.we_off = WB_GPMC_WE_t_o_f_f;
+ timings.wr_access = WB_GPMC_WR_t_a_c_c;
+ timings.wr_cycle = WB_GPMC_BUSCYC_t;
+
+ timings.page_burst_access = WB_GPMC_BUSCYC_t;
+ timings.wr_data_mux_bus = WB_GPMC_BUSCYC_t;
+ gpmc_cs_set_timings(AST_GPMC_CS, &timings);
+
+ cy_as_hal_print_omap_regs("GPMC_CONFIG", 1,
+ GPMC_VMA(GPMC_CFG_REG(1, AST_GPMC_CS)), 7);
+
+ /*
+ * DISABLE cs4, NOTE GPMC REG7 is already configured
+ * at this point by gpmc_cs_request
+ */
+ tmp32 = gpmc_cs_read_reg(AST_GPMC_CS, GPMC_CS_CONFIG7) &
+ ~GPMC_CONFIG7_CSVALID;
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7, tmp32);
+
+ /*
+ * PROGRAM chip select Region, (see OMAP3430 TRM PAGE 1088)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (AS_CS_MASK | AS_CS_BADDR));
+
+ /*
+ * by default configure GPMC into 8 bit mode
+ * (to match astoria default mode)
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG1,
+ (GPMC_CONFIG1_DEVICETYPE(2) |
+ GPMC_CONFIG1_WAIT_PIN_SEL(2)));
+
+ /*
+ * ENABLE astoria cs operation on GPMC
+ */
+ gpmc_cs_write_reg(AST_GPMC_CS, GPMC_CS_CONFIG7,
+ (tmp32 | GPMC_CONFIG7_CSVALID));
+
+ /*
+ * No method currently exists to write this register through GPMC APIs
+ * need to change WAIT2 polarity
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ tmp32 = tmp32 | NAND_FORCE_POSTED_WRITE_B | 0x40;
+ IOWR32(GPMC_VMA(GPMC_CONFIG_REG), tmp32);
+
+ tmp32 = IORD32(GPMC_VMA(GPMC_CONFIG_REG));
+ cy_as_hal_print_message("GPMC_CONFIG_REG=0x%x\n", tmp32);
+
+ return 0;
+
+out_release_mem_region:
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+out_free_cs:
+ gpmc_cs_free(AST_GPMC_CS);
+
+ return err;
+}
+
+/*
+ * west bridge astoria ISR (Interrupt handler)
+ */
+static irqreturn_t cy_astoria_int_handler(int irq,
+ void *dev_id, struct pt_regs *regs)
+{
+ cy_as_omap_dev_kernel *dev_p;
+ uint16_t read_val = 0 ;
+ uint16_t mask_val = 0 ;
+
+ /*
+ * debug stuff, counts number of loops per one intr trigger
+ */
+ uint16_t drq_loop_cnt = 0;
+ uint8_t irq_pin;
+ /*
+ * flags to watch
+ */
+ const uint16_t sentinel = (CY_AS_MEM_P0_INTR_REG_MCUINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT |
+ CY_AS_MEM_P0_INTR_REG_PMINT |
+ CY_AS_MEM_P0_INTR_REG_PLLLOCKINT);
+
+ /*
+ * sample IRQ pin level (just for statistics)
+ */
+ irq_pin = __gpio_get_value(AST_INT);
+
+ /*
+ * this one just for debugging
+ */
+ intr_sequence_num++ ;
+
+ /*
+ * astoria device handle
+ */
+ dev_p = dev_id;
+
+ /*
+ * read Astoria intr register
+ */
+ read_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) ;
+
+ /*
+ * save current mask value
+ */
+ mask_val = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG) ;
+
+ DBGPRN("<1>HAL__intr__enter:_seq:%d, P0_INTR_REG:%x\n",
+ intr_sequence_num, read_val);
+
+ /*
+ * Disable WB interrupt signal generation while we are in ISR
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+ /*
+ * this is a DRQ Interrupt
+ */
+ if (read_val & CY_AS_MEM_P0_INTR_REG_DRQINT) {
+
+ do {
+ /*
+ * handle DRQ interrupt
+ */
+ drq_loop_cnt++;
+
+ cy_handle_d_r_q_interrupt(dev_p) ;
+
+ /*
+ * spending to much time in ISR may impact
+ * average system performance
+ */
+ if (drq_loop_cnt >= MAX_DRQ_LOOPS_IN_ISR)
+ break;
+
+ /*
+ * Keep processing if there is another DRQ int flag
+ */
+ } while (cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INTR_REG) &
+ CY_AS_MEM_P0_INTR_REG_DRQINT);
+ }
+
+ if (read_val & sentinel)
+ cy_as_intr_service_interrupt((cy_as_hal_device_tag)dev_p) ;
+
+ DBGPRN("<1>_hal:_intr__exit seq:%d, mask=%4.4x,"
+ "int_pin:%d DRQ_jobs:%d\n",
+ intr_sequence_num,
+ mask_val,
+ irq_pin,
+ drq_loop_cnt);
+
+ /*
+ * re-enable WB hw interrupts
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask_val) ;
+
+ return IRQ_HANDLED ;
+}
+
+static int cy_as_hal_configure_interrupts(void *dev_p)
+{
+ int result;
+ int irq_pin = AST_INT;
+
+ set_irq_type(OMAP_GPIO_IRQ(irq_pin), IRQ_TYPE_LEVEL_LOW);
+
+ /*
+ * for shared IRQS must provide non NULL device ptr
+ * othervise the int won't register
+ * */
+ result = request_irq(OMAP_GPIO_IRQ(irq_pin),
+ (irq_handler_t)cy_astoria_int_handler,
+ IRQF_SHARED, "AST_INT#", dev_p);
+
+ if (result == 0) {
+ /*
+ * OMAP_GPIO_IRQ(irq_pin) - omap logical IRQ number
+ * assigned to this interrupt
+ * OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1) - print status
+ * of AST_INT GPIO IRQ_ENABLE FLAG
+ */
+ printk(KERN_INFO "AST_INT omap_pin:"
+ "%d assigned IRQ #%d IRQEN1=%d\n",
+ irq_pin,
+ OMAP_GPIO_IRQ(irq_pin),
+ OMAP_GPIO_BIT(AST_INT, GPIO_IRQENABLE1)
+ );
+ } else {
+ cy_as_hal_print_message("cyasomaphal: interrupt "
+ "failed to register\n");
+ gpio_free(irq_pin);
+ cy_as_hal_print_message(KERN_WARNING
+ "ASTORIA: can't get assigned IRQ"
+ "%i for INT#\n", OMAP_GPIO_IRQ(irq_pin));
+ }
+
+ return result;
+}
+
+/*
+ * initialize OMAP pads/pins to user defined functions
+ */
+static void cy_as_hal_init_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ /*
+ * browse through the table an dinitiaze the pins
+ */
+ u32 in_level = 0;
+ u16 tmp16, mux_val;
+
+ while (pad_cfg_tab->name != NULL) {
+
+ if (gpio_request(pad_cfg_tab->pin_num, NULL) == 0) {
+
+ pad_cfg_tab->valid = 1;
+ mux_val = omap_cfg_reg_L(pad_cfg_tab->mux_func);
+
+ /*
+ * always set drv level before changing out direction
+ */
+ __gpio_set_value(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /*
+ * "0" - OUT, "1", input omap_set_gpio_direction
+ * (pad_cfg_tab->pin_num, pad_cfg_tab->dir);
+ */
+ if (pad_cfg_tab->dir)
+ gpio_direction_input(pad_cfg_tab->pin_num);
+ else
+ gpio_direction_output(pad_cfg_tab->pin_num,
+ pad_cfg_tab->drv);
+
+ /* sample the pin */
+ in_level = __gpio_get_value(pad_cfg_tab->pin_num);
+
+ printk(KERN_INFO "configured %s to "
+ "OMAP pad_%d, DIR=%d "
+ "DOUT=%d, DIN=%d\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->dir,
+ pad_cfg_tab->drv,
+ in_level
+ );
+ } else {
+ /*
+ * get the pad_mux value to check on the pin_function
+ */
+ printk(KERN_INFO "couldn't cfg pin %d"
+ "for signal %s, its already taken\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ }
+
+ tmp16 = *(u16 *)PADCFG_VMA
+ (omap_mux_pin_cfg[pad_cfg_tab->mux_func].offset);
+
+ printk(KERN_INFO "GPIO_%d(PAD_CFG=%x,OE=%d"
+ "DOUT=%d, DIN=%d IRQEN=%d)\n\n",
+ pad_cfg_tab->pin_num, tmp16,
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_OE),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_OUT),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_DATA_IN),
+ OMAP_GPIO_BIT(pad_cfg_tab->pin_num, GPIO_IRQENABLE1)
+ );
+
+ /*
+ * next pad_cfg deriptor
+ */
+ pad_cfg_tab++;
+ }
+
+ printk(KERN_INFO "pads configured\n");
+}
+
+
+/*
+ * release gpios taken by the module
+ */
+static void cy_as_hal_release_user_pads(user_pad_cfg_t *pad_cfg_tab)
+{
+ while (pad_cfg_tab->name != NULL) {
+
+ if (pad_cfg_tab->valid) {
+ gpio_free(pad_cfg_tab->pin_num);
+ pad_cfg_tab->valid = 0;
+ printk(KERN_INFO "GPIO_%d "
+ "released from %s\n",
+ pad_cfg_tab->pin_num,
+ pad_cfg_tab->name);
+ } else {
+ printk(KERN_INFO "no release "
+ "for %s, GPIO_%d, wasn't acquired\n",
+ pad_cfg_tab->name,
+ pad_cfg_tab->pin_num);
+ }
+ pad_cfg_tab++;
+ }
+}
+
+void cy_as_hal_config_c_s_mux(void)
+{
+ /*
+ * FORCE the GPMC CS4 pin (it is in use by the zoom system)
+ */
+ omap_cfg_reg_L(T8_OMAP3430_GPMC_n_c_s4);
+}
+EXPORT_SYMBOL(cy_as_hal_config_c_s_mux);
+
+/*
+ * inits all omap h/w
+ */
+uint32_t cy_as_hal_processor_hw_init(void)
+{
+ int i, err;
+
+ printk(KERN_INFO "init OMAP3430 hw...\n");
+
+ iomux_vma = (u32)ioremap_nocache(
+ (u32)CTLPADCONF_BASE_ADDR, CTLPADCONF_SIZE);
+ printk(KERN_INFO "PADCONF_VMA=%x val=%x\n",
+ iomux_vma, IORD32(iomux_vma));
+
+ /*
+ * remap gpio banks
+ */
+ for (i = 0; i < 6; i++) {
+ gpio_vma_tab[i].virt_addr = (u32)ioremap_nocache(
+ gpio_vma_tab[i].phy_addr,
+ gpio_vma_tab[i].size);
+
+ printk(KERN_INFO "%s virt_addr=%x\n",
+ gpio_vma_tab[i].name,
+ (u32)gpio_vma_tab[i].virt_addr);
+ };
+
+ /*
+ * force OMAP_GPIO_126 to rleased state,
+ * will be configured to drive reset
+ */
+ gpio_free(AST_RESET);
+
+ /*
+ *same thing with AStoria CS pin
+ */
+ gpio_free(AST_CS);
+
+ /*
+ * initialize all the OMAP pads connected to astoria
+ */
+ cy_as_hal_init_user_pads(user_pad_cfg);
+
+ err = cy_as_hal_gpmc_init();
+ if (err < 0)
+ printk(KERN_INFO "gpmc init failed:%d", err);
+
+ cy_as_hal_config_c_s_mux();
+
+ return gpmc_data_vma;
+}
+EXPORT_SYMBOL(cy_as_hal_processor_hw_init);
+
+void cy_as_hal_omap_hardware_deinit(cy_as_omap_dev_kernel *dev_p)
+{
+ /*
+ * free omap hw resources
+ */
+ if (gpmc_data_vma != 0)
+ iounmap((void *)gpmc_data_vma);
+
+ if (csa_phy != 0)
+ release_mem_region(csa_phy, BLKSZ_4K);
+
+ gpmc_cs_free(AST_GPMC_CS);
+
+ free_irq(OMAP_GPIO_IRQ(AST_INT), dev_p);
+
+ cy_as_hal_release_user_pads(user_pad_cfg);
+}
+
+/*
+ * These are the functions that are not part of the
+ * HAL layer, but are required to be called for this HAL
+ */
+
+/*
+ * Called On AstDevice LKM exit
+ */
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+
+ /*
+ * TODO: Need to disable WB interrupt handlere 1st
+ */
+ if (0 == dev_p)
+ return 1 ;
+
+ cy_as_hal_print_message("<1>_stopping OMAP34xx HAL layer object\n");
+ if (dev_p->m_sig != CY_AS_OMAP_PNAND_HAL_SIG) {
+ cy_as_hal_print_message("<1>%s: %s: bad HAL tag\n",
+ pgm, __func__) ;
+ return 1 ;
+ }
+
+ /*
+ * disable interrupt
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, 0x0000) ;
+
+#if 0
+ if (dev_p->thread_flag == 0) {
+ dev_p->thread_flag = 1 ;
+ wait_for_completion(&dev_p->thread_complete) ;
+ cy_as_hal_print_message("cyasomaphal:"
+ "done cleaning thread\n");
+ cy_as_hal_destroy_sleep_channel(&dev_p->thread_sc) ;
+ }
+#endif
+
+ cy_as_hal_omap_hardware_deinit(dev_p);
+
+ /*
+ * Rearrange the list
+ */
+ if (m_omap_list_p == dev_p)
+ m_omap_list_p = dev_p->m_next_p ;
+
+ cy_as_hal_free(dev_p) ;
+
+ printk(KERN_INFO "OMAP_kernel_hal stopped\n");
+ return 0;
+}
+
+int omap_start_intr(cy_as_hal_device_tag tag)
+{
+ cy_as_omap_dev_kernel *dev_p = (cy_as_omap_dev_kernel *)tag ;
+ int ret = 0 ;
+ const uint16_t mask = CY_AS_MEM_P0_INTR_REG_DRQINT |
+ CY_AS_MEM_P0_INTR_REG_MBINT ;
+
+ /*
+ * register for interrupts
+ */
+ ret = cy_as_hal_configure_interrupts(dev_p) ;
+
+ /*
+ * enable only MBox & DRQ interrupts for now
+ */
+ cy_as_hal_write_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_INT_MASK_REG, mask) ;
+
+ return 1 ;
+}
+
+/*
+ * Below are the functions that communicate with the WestBridge device.
+ * These are system dependent and must be defined by the HAL layer
+ * for a given system.
+ */
+
+/*
+ * GPMC NAND command+addr write phase
+ */
+static inline void nand_cmd_n_addr(u8 cmdb1, u16 col_addr, u32 row_addr)
+{
+ /*
+ * byte order on the bus <cmd> <CA0,CA1,RA0,RA1, RA2>
+ */
+ u32 tmpa32 = ((row_addr << 16) | col_addr);
+ u8 RA2 = (u8)(row_addr >> 16);
+
+ if (!pnand_16bit) {
+ /*
+ * GPMC PNAND 8bit BUS
+ */
+ /*
+ * CMD1
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+
+ /*
+ *pnand bus: <CA0,CA1,RA0,RA1>
+ */
+ IOWR32(naddr_reg_vma, tmpa32);
+
+ /*
+ * <RA2> , always zero
+ */
+ IOWR8(naddr_reg_vma, RA2);
+
+ } else {
+ /*
+ * GPMC PNAND 16bit BUS , in 16 bit mode CMD
+ * and ADDR sent on [d7..d0]
+ */
+ uint8_t CA0, CA1, RA0, RA1;
+ CA0 = tmpa32 & 0x000000ff;
+ CA1 = (tmpa32 >> 8) & 0x000000ff;
+ RA0 = (tmpa32 >> 16) & 0x000000ff;
+ RA1 = (tmpa32 >> 24) & 0x000000ff;
+
+ /*
+ * can't use 32 bit writes here omap will not serialize
+ * them to lower half in16 bit mode
+ */
+
+ /*
+ *pnand bus: <CMD1, CA0,CA1,RA0,RA1, RA2 (always zero)>
+ */
+ IOWR8(ncmd_reg_vma, cmdb1);
+ IOWR8(naddr_reg_vma, CA0);
+ IOWR8(naddr_reg_vma, CA1);
+ IOWR8(naddr_reg_vma, RA0);
+ IOWR8(naddr_reg_vma, RA1);
+ IOWR8(naddr_reg_vma, RA2);
+ }
+}
+
+/*
+ * spin until r/b goes high
+ */
+inline int wait_rn_b_high(void)
+{
+ u32 w_spins = 0;
+
+ /*
+ * TODO: note R/b may go low here, need to spin until high
+ * while (omap_get_gpio_datain(AST_RnB) == 0) {
+ * w_spins++;
+ * }
+ * if (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ *
+ * while (OMAP_GPIO_BIT(AST_RnB, GPIO_DATA_IN) == 0) {
+ * w_spins++;
+ * }
+ * printk("<1>RnB=0!:%d\n",w_spins);
+ * }
+ */
+ return w_spins;
+}
+
+#ifdef ENABLE_GPMC_PF_ENGINE
+/* #define PFE_READ_DEBUG
+ * PNAND block read with OMAP PFE enabled
+ * status: Not tested, NW, broken , etc
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint8_t bytes_in_fifo;
+
+ /* debug vars*/
+#ifdef PFE_READ_DEBUG
+ uint32_t loop_limit;
+ uint16_t bytes_read = 0;
+#endif
+
+ /*
+ * configure the prefetch engine
+ */
+ uint32_t tmp32;
+ uint32_t pfe_status;
+
+ /*
+ * DISABLE GPMC CS4 operation 1st, this is
+ * in case engine is be already disabled
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x0);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+
+#ifdef PFE_READ_DEBUG
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG1));
+ if (tmp32 != GPMC_PREFETCH_CONFIG1_VAL) {
+ printk(KERN_INFO "<1> prefetch is CONFIG1 read val:%8.8x, != VAL written:%8.8x\n",
+ tmp32, GPMC_PREFETCH_CONFIG1_VAL);
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ printk(KERN_INFO "<1> GPMC_PREFETCH_STATUS : %8.8x\n", tmp32);
+ }
+
+ /*
+ *sanity check 2
+ */
+ tmp32 = IORD32(GPMC_VMA(GPMC_PREFETCH_CONFIG2));
+ if (tmp32 != (count))
+ printk(KERN_INFO "<1> GPMC_PREFETCH_CONFIG2 read val:%d, "
+ "!= VAL written:%d\n", tmp32, count);
+#endif
+
+ /*
+ * ISSUE PNAND CMD+ADDR, note gpmc puts 32b words
+ * on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * start the prefetch engine
+ */
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ while (1) {
+ /*
+ * GPMC PFE service loop
+ */
+ do {
+ /*
+ * spin until PFE fetched some
+ * PNAND bus words in the FIFO
+ */
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS));
+ bytes_in_fifo = (pfe_status >> 24) & 0x7f;
+ } while (bytes_in_fifo == 0);
+
+ /* whole 32 bit words in fifo */
+ w32cnt = bytes_in_fifo >> 2;
+
+#if 0
+ /*
+ *NOTE: FIFO_PTR indicates number of NAND bus words bytes
+ * already received in the FIFO and available to be read
+ * by DMA or MPU whether COUNTVAL indicates number of BUS
+ * words yet to be read from PNAND bus words
+ */
+ printk(KERN_ERR "<1> got PF_STATUS:%8.8x FIFO_PTR:%d, COUNTVAL:%d, w32cnt:%d\n",
+ pfe_status, bytes_in_fifo,
+ (pfe_status & 0x3fff), w32cnt);
+#endif
+
+ while (w32cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ if ((pfe_status & 0x3fff) == 0) {
+ /*
+ * PFE acc angine done, there still may be data leftover
+ * in the FIFO re-read FIFO BYTE counter (check for
+ * leftovers from 32 bit read accesses above)
+ */
+ bytes_in_fifo = (IORD32(
+ GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ /*
+ * NOTE we may still have one word left in the fifo
+ * read it out
+ */
+ ptr8 = ptr32;
+ switch (bytes_in_fifo) {
+
+ case 0:
+ /*
+ * nothing to do we already read the
+ * FIFO out with 32 bit accesses
+ */
+ break;
+ case 1:
+ /*
+ * this only possible
+ * for 8 bit pNAND only
+ */
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 2:
+ /*
+ * this one can occur in either modes
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ break;
+
+ case 3:
+ /*
+ * this only possible for 8 bit pNAND only
+ */
+ *(uint16_t *)ptr8 = IORD16(gpmc_data_vma);
+ ptr8 += 2;
+ *ptr8 = IORD8(gpmc_data_vma);
+ break;
+
+ case 4:
+ /*
+ * shouldn't happen, but has been seen
+ * in 8 bit mode
+ */
+ *ptr32 = IORD32(gpmc_data_vma);
+ break;
+
+ default:
+ printk(KERN_ERR"<1>_error: PFE FIFO bytes leftover is not read:%d\n",
+ bytes_in_fifo);
+ break;
+ }
+ /*
+ * read is completed, get out of the while(1) loop
+ */
+ break;
+ }
+ }
+}
+#endif
+
+#ifdef PFE_LBD_READ_V2
+/*
+ * PFE engine assisted reads with the 64 byte blocks
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint8_t rd_cnt;
+ uint32_t *ptr32;
+ uint8_t *ptr8;
+ uint16_t reminder;
+ uint32_t pfe_status;
+
+ /*
+ * ISSUE PNAND CMD+ADDR
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * setup PFE block
+ * count - OMAP number of bytes to access on pnand bus
+ */
+
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG1), GPMC_PREFETCH_CONFIG1_VAL);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONFIG2), count);
+ IOWR32(GPMC_VMA(GPMC_PREFETCH_CONTROL), 0x1);
+
+ ptr32 = buff;
+
+ do {
+ pfe_status = IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) ;
+ rd_cnt = pfe_status >> (24+2);
+
+ while (rd_cnt--)
+ *ptr32++ = IORD32(gpmc_data_vma);
+
+ } while (pfe_status & 0x3fff);
+
+ /*
+ * read out the leftover
+ */
+ ptr8 = ptr32;
+ rd_cnt = (IORD32(GPMC_VMA(GPMC_PREFETCH_STATUS)) >> 24) & 0x7f;
+
+ while (rd_cnt--)
+ *ptr8++ = IORD8(gpmc_data_vma);
+}
+#endif
+
+#ifdef PNAND_LBD_READ_NO_PFE
+/*
+ * Endpoint buffer read w/o OMAP GPMC Prefetch Engine
+ * the original working code, works at max speed for 8 bit xfers
+ * for 16 bit the bus diagram has gaps
+ */
+static void p_nand_lbd_read(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint32_t *ptr32;
+ uint16_t *ptr16;
+ uint16_t remainder;
+
+ DBGPRN("<1> %s(): NO_PFE\n", __func__);
+
+ ptr32 = buff;
+ /* number of whole 32 bit words in the transfer */
+ w32cnt = count >> 2;
+
+ /* remainder, in bytes(0..3) */
+ remainder = count & 03;
+
+ /*
+ * note gpmc puts 32b words on the bus least sig. byte 1st
+ */
+ nand_cmd_n_addr(RDPAGE_B1, col_addr, row_addr);
+ IOWR8(ncmd_reg_vma, RDPAGE_B2);
+
+ /*
+ * read data by 32 bit chunks
+ */
+ while (w32cnt--)
+ *ptr32++ = IORD32(ndata_reg_vma);
+
+ /*
+ * now do the remainder(it can be 0, 1, 2 or 3)
+ * same code for both 8 & 16 bit bus
+ * do 1 or 2 MORE words
+ */
+ ptr16 = (uint16_t *)ptr32;
+
+ switch (remainder) {
+ case 1:
+ /* read one 16 bit word
+ * IN 8 BIT WE NEED TO READ even number of bytes
+ */
+ case 2:
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ *ptr16++ = IORD16(ndata_reg_vma);
+ *ptr16 = IORD16(ndata_reg_vma);
+ break;
+ default:
+ /*
+ * remainder is 0
+ */
+ break;
+ }
+}
+#endif
+
+/*
+ * uses LBD mode to write N bytes into astoria
+ * Status: Working, however there are 150ns idle
+ * timeafter every 2 (16 bit or 4(8 bit) bus cycles
+ */
+static void p_nand_lbd_write(u16 col_addr, u32 row_addr, u16 count, void *buff)
+{
+ uint16_t w32cnt;
+ uint16_t remainder;
+ uint8_t *ptr8;
+ uint16_t *ptr16;
+ uint32_t *ptr32;
+
+ remainder = count & 03;
+ w32cnt = count >> 2;
+ ptr32 = buff;
+ ptr8 = buff;
+
+ /*
+ * send: CMDB1, CA0,CA1,RA0,RA1,RA2
+ */
+ nand_cmd_n_addr(PGMPAGE_B1, col_addr, row_addr);
+
+ /*
+ * blast the data out in 32bit chunks
+ */
+ while (w32cnt--)
+ IOWR32(ndata_reg_vma, *ptr32++);
+
+ /*
+ * do the reminder if there is one
+ * same handling for both 8 & 16 bit pnand: mode
+ */
+ ptr16 = (uint16_t *)ptr32; /* do 1 or 2 words */
+
+ switch (remainder) {
+ case 1:
+ /*
+ * read one 16 bit word
+ */
+ case 2:
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+
+ case 3:
+ /*
+ * for 3 bytes read 2 16 bit words
+ */
+ IOWR16(ndata_reg_vma, *ptr16++);
+ IOWR16(ndata_reg_vma, *ptr16);
+ break;
+ default:
+ /*
+ * reminder is 0
+ */
+ break;
+ }
+ /*
+ * finally issue a PGM cmd
+ */
+ IOWR8(ncmd_reg_vma, PGMPAGE_B2);
+}
+
+/*
+ * write Astoria register
+ */
+static inline void ast_p_nand_casdi_write(u8 reg_addr8, u16 data)
+{
+ unsigned long flags;
+ u16 addr16;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t rdreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&rdreg_usage_cnt) != 0) {
+ cy_as_hal_print_message(KERN_ERR "cy_as_omap_hal:"
+ "* cy_as_hal_write_register usage:%d\n",
+ atomic_read(&rdreg_usage_cnt));
+ }
+
+ atomic_inc(&rdreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ */
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+
+ /*
+ * this should be sent on the 16 bit bus
+ */
+ IOWR16(ndata_reg_vma, data);
+ } else {
+ /*
+ * 8 bit nand mode GPMC will automatically
+ * seriallize 16bit or 32 bit writes into
+ * 8 bit onesto the lower 8 bit in LE order
+ */
+ addr16 = 0x0c00 | reg_addr8;
+
+ /*
+ * CMD1, CA1, CA2,
+ */
+ IOWR8(ncmd_reg_vma, 0x85);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR16(ndata_reg_vma, data);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&rdreg_usage_cnt);
+ local_irq_restore(flags);
+}
+
+
+/*
+ * read astoria register via pNAND interface
+ */
+static inline u16 ast_p_nand_casdo_read(u8 reg_addr8)
+{
+ u16 data;
+ u16 addr16;
+ unsigned long flags;
+ /*
+ * throw an error if called from multiple threads
+ */
+ static atomic_t wrreg_usage_cnt = { 0 };
+
+ /*
+ * disable interrupts
+ */
+ local_irq_save(flags);
+
+ if (atomic_read(&wrreg_usage_cnt) != 0) {
+ /*
+ * if it gets here ( from other threads), this function needs
+ * need spin_lock_irq save() protection
+ */
+ cy_as_hal_print_message(KERN_ERR"cy_as_omap_hal: "
+ "cy_as_hal_write_register usage:%d\n",
+ atomic_read(&wrreg_usage_cnt));
+ }
+ atomic_inc(&wrreg_usage_cnt);
+
+ /*
+ * 2 flavors of GPMC -> PNAND access
+ */
+ if (pnand_16bit) {
+ /*
+ * 16 BIT gpmc NAND mode
+ * CMD1, CA1, CA2,
+ */
+
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR8(naddr_reg_vma, reg_addr8);
+ IOWR8(naddr_reg_vma, 0x0c);
+ IOWR8(ncmd_reg_vma, 0x00E0);
+
+ udelay(1);
+
+ /*
+ * much faster through the gPMC Register space
+ */
+ data = IORD16(ndata_reg_vma);
+ } else {
+ /*
+ * 8 BIT gpmc NAND mode
+ * CMD1, CA1, CA2, CMD2
+ */
+ addr16 = 0x0c00 | reg_addr8;
+ IOWR8(ncmd_reg_vma, 0x05);
+ IOWR16(naddr_reg_vma, addr16);
+ IOWR8(ncmd_reg_vma, 0xE0);
+ udelay(1);
+ data = IORD16(ndata_reg_vma);
+ }
+
+ /*
+ * re-enable interrupts
+ */
+ atomic_dec(&wrreg_usage_cnt);
+ local_irq_restore(flags);
+
+ return data;
+}
+
+
+/*
+ * This function must be defined to write a register within the WestBridge
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the WestBridge device.
+ */
+void cy_as_hal_write_register(
+ cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data)
+{
+ ast_p_nand_casdi_write((u8)addr, data);
+}
+
+/*
+ * This function must be defined to read a register from the WestBridge
+ * device. The addr value is the address of the register to read with
+ * respect to the base address of the WestBridge device.
+ */
+uint16_t cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr)
+{
+ uint16_t data = 0 ;
+
+ /*
+ * READ ASTORIA REGISTER USING CASDO
+ */
+ data = ast_p_nand_casdo_read((u8)addr);
+
+ return data ;
+}
+
+/*
+ * preps Ep pointers & data counters for next packet
+ * (fragment of the request) xfer returns true if
+ * there is a next transfer, and false if all bytes in
+ * current request have been xfered
+ */
+static inline bool prep_for_next_xfer(cy_as_hal_device_tag tag, uint8_t ep)
+{
+
+ if (!end_points[ep].sg_list_enabled) {
+ /*
+ * no further transfers for non storage EPs
+ * (like EP2 during firmware download, done
+ * in 64 byte chunks)
+ */
+ if (end_points[ep].req_xfer_cnt >= end_points[ep].req_length) {
+ DBGPRN("<1> %s():RQ sz:%d non-_sg EP:%d completed\n",
+ __func__, end_points[ep].req_length, ep);
+
+ /*
+ * no more transfers, we are done with the request
+ */
+ return false;
+ }
+ /* this is not defined currently*/
+ #ifdef WESTBRIDGE_PNAND_HAL_PREP_CHANGE
+ /*
+ * calculate size of the next DMA xfer, corner
+ * case for non-storage EPs where transfer size
+ * is not egual N * HAL_DMA_PKT_SZ xfers
+ */
+ if ((end_points[ep].req_length - end_points[ep].req_xfer_cnt)
+ >= HAL_DMA_PKT_SZ) {
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+ } else {
+ /*
+ * that would be the last chunk less
+ * than P-port max size
+ */
+ end_points[ep].dma_xfer_sz = end_points[ep].req_length -
+ end_points[ep].req_xfer_cnt;
+ }
+ #endif
+ return true;
+ }
+
+ /*
+ * for SG_list assisted dma xfers
+ * are we done with current SG ?
+ */
+ if (end_points[ep].seg_xfer_cnt == end_points[ep].sg_p->length) {
+ /*
+ * was it the Last SG segment on the list ?
+ */
+ if (sg_is_last(end_points[ep].sg_p)) {
+ DBGPRN("<1> %s: EP:%d completed,"
+ "%d bytes xfered\n",
+ __func__,
+ ep,
+ end_points[ep].req_xfer_cnt
+ );
+
+ return false;
+ } else {
+ /*
+ * There are more SG segments in current
+ * request's sg list setup new segment
+ */
+
+ end_points[ep].seg_xfer_cnt = 0;
+ end_points[ep].sg_p = sg_next(end_points[ep].sg_p);
+ /* set data pointer for next DMA sg transfer*/
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ DBGPRN("<1> %s new SG:_va:%p\n\n",
+ __func__, end_points[ep].data_p);
+ }
+
+ }
+
+ /*
+ * for sg list xfers it will always be 512 or 1024
+ */
+ end_points[ep].dma_xfer_sz = HAL_DMA_PKT_SZ;
+
+ /*
+ * next transfer is required
+ */
+
+ return true;
+}
+
+/*
+ * Astoria DMA read request, APP_CPU reads from WB ep buffer
+ */
+static void cy_service_e_p_dma_read_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ uint16_t v, size;
+ void *dptr;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ uint16_t ep_dma_reg = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2;
+
+ /*
+ * get the XFER size frtom WB eP DMA REGISTER
+ */
+ v = cy_as_hal_read_register(tag, ep_dma_reg);
+
+ /*
+ * amount of data in EP buff in bytes
+ */
+ size = v & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK;
+
+ /*
+ * memory pointer for this DMA packet xfer (sub_segment)
+ */
+ dptr = end_points[ep].data_p;
+
+ DBGPRN("<1>HAL:_svc_dma_read on EP_%d sz:%d, intr_seq:%d, dptr:%p\n",
+ ep,
+ size,
+ intr_sequence_num,
+ dptr
+ );
+
+ cy_as_hal_assert(size != 0);
+
+ if (size) {
+ /*
+ * the actual WB-->OMAP memory "soft" DMA xfer
+ */
+ p_nand_lbd_read(col_addr, row_addr, size, dptr);
+ }
+
+ /*
+ * clear DMAVALID bit indicating that the data has been read
+ */
+ cy_as_hal_write_register(tag, ep_dma_reg, 0) ;
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+
+ /*
+ * pre-advance data pointer (if it's outside sg
+ * list it will be reset anyway
+ */
+ end_points[ep].data_p += size;
+
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * we have more data to read in this request,
+ * setup next dma packet due tell WB how much
+ * data we are going to xfer next
+ */
+ v = end_points[ep].dma_xfer_sz/*HAL_DMA_PKT_SZ*/ |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, ep_dma_reg, v);
+ } else {
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ DBGPRN("<1>trigg rd_dma completion cb: xfer_sz:%d\n",
+ end_points[ep].req_xfer_cnt);
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * omap_cpu needs to transfer data to ASTORIA EP buffer
+ */
+static void cy_service_e_p_dma_write_request(
+ cy_as_omap_dev_kernel *dev_p, uint8_t ep)
+{
+ uint16_t addr;
+ uint16_t v = 0;
+ uint32_t size;
+ uint16_t col_addr = 0x0000;
+ uint32_t row_addr = CYAS_DEV_CALC_EP_ADDR(ep);
+ void *dptr;
+
+ cy_as_hal_device_tag tag = (cy_as_hal_device_tag)dev_p ;
+ /*
+ * note: size here its the size of the dma transfer could be
+ * anything > 0 && < P_PORT packet size
+ */
+ size = end_points[ep].dma_xfer_sz ;
+ dptr = end_points[ep].data_p ;
+
+ /*
+ * perform the soft DMA transfer, soft in this case
+ */
+ if (size)
+ p_nand_lbd_write(col_addr, row_addr, size, dptr);
+
+ end_points[ep].seg_xfer_cnt += size;
+ end_points[ep].req_xfer_cnt += size;
+ /*
+ * pre-advance data pointer
+ * (if it's outside sg list it will be reset anyway)
+ */
+ end_points[ep].data_p += size;
+
+ /*
+ * now clear DMAVAL bit to indicate we are done
+ * transferring data and that the data can now be
+ * sent via USB to the USB host, sent to storage,
+ * or used internally.
+ */
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+ cy_as_hal_write_register(tag, addr, size) ;
+
+ /*
+ * finally, tell the USB subsystem that the
+ * data is gone and we can accept the
+ * next request if one exists.
+ */
+ if (prep_for_next_xfer(tag, ep)) {
+ /*
+ * There is more data to go. Re-init the WestBridge DMA side
+ */
+ v = end_points[ep].dma_xfer_sz |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v) ;
+ } else {
+
+ end_points[ep].pending = cy_false ;
+ end_points[ep].type = cy_as_hal_none ;
+ end_points[ep].buffer_valid = cy_false ;
+
+ /*
+ * notify the API that we are done with rq on this EP
+ */
+ if (callback) {
+ /*
+ * this callback will wake up the process that might be
+ * sleeping on the EP which data is being transferred
+ */
+ callback(tag, ep,
+ end_points[ep].req_xfer_cnt,
+ CY_AS_ERROR_SUCCESS);
+ }
+ }
+}
+
+/*
+ * HANDLE DRQINT from Astoria (called in AS_Intr context
+ */
+static void cy_handle_d_r_q_interrupt(cy_as_omap_dev_kernel *dev_p)
+{
+ uint16_t v ;
+ static uint8_t service_ep = 2 ;
+
+ /*
+ * We've got DRQ INT, read DRQ STATUS Register */
+ v = cy_as_hal_read_register((cy_as_hal_device_tag)dev_p,
+ CY_AS_MEM_P0_DRQ) ;
+
+ if (v == 0) {
+#ifndef WESTBRIDGE_NDEBUG
+ cy_as_hal_print_message("stray DRQ interrupt detected\n") ;
+#endif
+ return;
+ }
+
+ /*
+ * Now, pick a given DMA request to handle, for now, we just
+ * go round robin. Each bit position in the service_mask
+ * represents an endpoint from EP2 to EP15. We rotate through
+ * each of the endpoints to find one that needs to be serviced.
+ */
+ while ((v & (1 << service_ep)) == 0) {
+
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+ if (end_points[service_ep].type == cy_as_hal_write) {
+ /*
+ * handle DMA WRITE REQUEST: app_cpu will
+ * write data into astoria EP buffer
+ */
+ cy_service_e_p_dma_write_request(dev_p, service_ep) ;
+ } else if (end_points[service_ep].type == cy_as_hal_read) {
+ /*
+ * handle DMA READ REQUEST: cpu will
+ * read EP buffer from Astoria
+ */
+ cy_service_e_p_dma_read_request(dev_p, service_ep) ;
+ }
+#ifndef WESTBRIDGE_NDEBUG
+ else
+ cy_as_hal_print_message("cyashalomap:interrupt,"
+ " w/o pending DMA job,"
+ "-check DRQ_MASK logic\n") ;
+#endif
+
+ /*
+ * Now bump the EP ahead, so other endpoints get
+ * a shot before the one we just serviced
+ */
+ if (end_points[service_ep].type == cy_as_hal_none) {
+ if (service_ep == 15)
+ service_ep = 2 ;
+ else
+ service_ep++ ;
+ }
+
+}
+
+void cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep)
+{
+ DBGPRN("cy_as_hal_dma_cancel_request on ep:%d", ep);
+ if (end_points[ep].pending)
+ cy_as_hal_write_register(tag,
+ CY_AS_MEM_P0_EP2_DMA_REG + ep - 2, 0);
+
+ end_points[ep].buffer_valid = cy_false ;
+ end_points[ep].type = cy_as_hal_none;
+}
+
+/*
+ * enables/disables SG list assisted DMA xfers for the given EP
+ * sg_list assisted XFERS can use physical addresses of mem pages in case if the
+ * xfer is performed by a h/w DMA controller rather then the CPU on P port
+ */
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled)
+{
+ end_points[ep].sg_list_enabled = sg_xfer_enabled;
+ DBGPRN("<1> EP:%d sg_list assisted DMA mode set to = %d\n",
+ ep, end_points[ep].sg_list_enabled);
+}
+EXPORT_SYMBOL(cy_as_hal_set_ep_dma_mode);
+
+/*
+ * This function must be defined to transfer a block of data to
+ * the WestBridge device. This function can use the burst write
+ * (DMA) capabilities of WestBridge to do this, or it can just copy
+ * the data using writes.
+ */
+void cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr = 0 ;
+ uint16_t v = 0;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem. Another DMA request
+ * is coming down before the previous one has completed.
+ */
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false) ;
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_write ;
+ end_points[ep].pending = cy_true;
+
+ /*
+ * total length of the request
+ */
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * smaller xfers for non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ /*
+ * check the EP transfer mode uses sg_list rather then a memory buffer
+ * block devices pass it to the HAL, so the hAL could get to the real
+ * physical address for each segment and set up a DMA controller
+ * hardware ( if there is one)
+ */
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * buf - pointer to the SG list
+ * data_p - data pointer to the 1st DMA segment
+ * seg_xfer_cnt - keeps track of N of bytes sent in current
+ * sg_list segment
+ * req_xfer_cnt - keeps track of the total N of bytes
+ * transferred for the request
+ */
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].req_xfer_cnt = 0;
+
+#ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("cyasomaphal:%s: EP:%d, buf:%p, buf_va:%p,"
+ "req_sz:%d, maxsz:%d\n",
+ __func__,
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+#endif
+
+ } else {
+ /*
+ * setup XFER for non sg_list assisted EPs
+ */
+
+ #ifdef DBGPRN_DMA_SETUP_WR
+ DBGPRN("<1>%s non storage or sz < 512:"
+ "EP:%d, sz:%d\n", __func__, ep, size);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * will keep track No of bytes xferred for the request
+ */
+ end_points[ep].req_xfer_cnt = 0;
+ }
+
+ /*
+ * Tell WB we are ready to send data on the given endpoint
+ */
+ v = (end_points[ep].dma_xfer_sz & CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK)
+ | CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ cy_as_hal_write_register(tag, addr, v) ;
+}
+
+/*
+ * This function must be defined to transfer a block of data from
+ * the WestBridge device. This function can use the burst read
+ * (DMA) capabilities of WestBridge to do this, or it can just
+ * copy the data using reads.
+ */
+void cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf,
+ uint32_t size, uint16_t maxsize)
+{
+ uint32_t addr ;
+ uint16_t v ;
+
+ /*
+ * Note: "size" is the actual request size
+ * "maxsize" - is the P port fragment size
+ * No EP0 or EP1 traffic should get here
+ */
+ cy_as_hal_assert(ep != 0 && ep != 1) ;
+
+ /*
+ * If this asserts, we have an ordering problem.
+ * Another DMA request is coming down before the
+ * previous one has completed. we should not get
+ * new requests if current is still in process
+ */
+
+ cy_as_hal_assert(end_points[ep].buffer_valid == cy_false);
+
+ end_points[ep].buffer_valid = cy_true ;
+ end_points[ep].type = cy_as_hal_read ;
+ end_points[ep].pending = cy_true;
+ end_points[ep].req_xfer_cnt = 0;
+ end_points[ep].req_length = size;
+
+ if (size >= maxsize) {
+ /*
+ * set xfer size for very 1st DMA xfer operation
+ * port max packet size ( typically 512 or 1024)
+ */
+ end_points[ep].dma_xfer_sz = maxsize;
+ } else {
+ /*
+ * so that we could handle small xfers on in case
+ * of non-storage EPs
+ */
+ end_points[ep].dma_xfer_sz = size;
+ }
+
+ addr = CY_AS_MEM_P0_EP2_DMA_REG + ep - 2 ;
+
+ if (end_points[ep].sg_list_enabled) {
+ /*
+ * Handle sg-list assisted EPs
+ * seg_xfer_cnt - keeps track of N of sent packets
+ * buf - pointer to the SG list
+ * data_p - data pointer for the 1st DMA segment
+ */
+ end_points[ep].seg_xfer_cnt = 0 ;
+ end_points[ep].sg_p = buf;
+ end_points[ep].data_p = sg_virt(end_points[ep].sg_p);
+
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("cyasomaphal:DMA_setup_read sg_list EP:%d, "
+ "buf:%p, buf_va:%p, req_sz:%d, maxsz:%d\n",
+ ep,
+ buf,
+ end_points[ep].data_p,
+ size,
+ maxsize);
+ #endif
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ } else {
+ /*
+ * Non sg list EP passed void *buf rather then scatterlist *sg
+ */
+ #ifdef DBGPRN_DMA_SETUP_RD
+ DBGPRN("%s:non-sg_list EP:%d,"
+ "RQ_sz:%d, maxsz:%d\n",
+ __func__, ep, size, maxsize);
+ #endif
+
+ end_points[ep].sg_p = NULL;
+
+ /*
+ * must be a VMA of a membuf in kernel space
+ */
+ end_points[ep].data_p = buf;
+
+ /*
+ * Program the EP DMA register for Storage endpoints only.
+ */
+ if (is_storage_e_p(ep)) {
+ v = (end_points[ep].dma_xfer_sz &
+ CY_AS_MEM_P0_E_pn_DMA_REG_COUNT_MASK) |
+ CY_AS_MEM_P0_E_pn_DMA_REG_DMAVAL ;
+ cy_as_hal_write_register(tag, addr, v);
+ }
+ }
+}
+
+/*
+ * This function must be defined to allow the WB API to
+ * register a callback function that is called when a
+ * DMA transfer is complete.
+ */
+void cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb)
+{
+ DBGPRN("<1>\n%s: WB API has registered a dma_complete callback:%x\n",
+ __func__, (uint32_t)cb);
+ callback = cb ;
+}
+
+/*
+ * This function must be defined to return the maximum size of
+ * DMA request that can be handled on the given endpoint. The
+ * return value should be the maximum size in bytes that the DMA
+ * module can handle.
+ */
+uint32_t cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep)
+{
+ /*
+ * Storage reads and writes are always done in 512 byte blocks.
+ * So, we do the count handling within the HAL, and save on
+ * some of the data transfer delay.
+ */
+ if ((ep == CYASSTORAGE_READ_EP_NUM) ||
+ (ep == CYASSTORAGE_WRITE_EP_NUM)) {
+ /* max DMA request size HAL can handle by itself */
+ return CYASSTORAGE_MAX_XFER_SIZE;
+ } else {
+ /*
+ * For the USB - Processor endpoints, the maximum transfer
+ * size depends on the speed of USB operation. So, we use
+ * the following constant to indicate to the API that
+ * splitting of the data into chunks less that or equal to
+ * the max transfer size should be handled internally.
+ */
+
+ /* DEFINED AS 0xffffffff in cyasdma.h */
+ return CY_AS_DMA_MAX_SIZE_HW_SIZE;
+ }
+}
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the WestBridge device. Generally this is done via a GPIO of
+ * some type.
+ */
+cy_bool cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state)
+{
+ if (state) {
+ cy_as_hal_gpmc_enable_16bit_bus(cy_false);
+ __gpio_set_value(AST_WAKEUP, 1);
+ } else {
+ __gpio_set_value(AST_WAKEUP, 0);
+ }
+ return cy_true;
+}
+
+void cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag)
+{
+ cy_as_hal_print_message("error: astoria PLL lock is lost\n") ;
+ cy_as_hal_print_message("please check the input voltage levels");
+ cy_as_hal_print_message("and clock, and restart the system\n") ;
+}
+
+/*
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ */
+
+/*
+ * This function is required by the API to allocate memory.
+ * This function is expected to work exactly like malloc().
+ */
+void *cy_as_hal_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required by the API to free memory allocated
+ * with CyAsHalAlloc(). This function is'expected to work exacly
+ * like free().
+ */
+void cy_as_hal_free(void *mem_p)
+{
+ kfree(mem_p) ;
+}
+
+/*
+ * Allocator that can be used in interrupt context.
+ * We have to ensure that the kmalloc call does not
+ * sleep in this case.
+ */
+void *cy_as_hal_c_b_alloc(uint32_t cnt)
+{
+ void *ret_p ;
+
+ ret_p = kmalloc(cnt, GFP_ATOMIC) ;
+ return ret_p ;
+}
+
+/*
+ * This function is required to set a block of memory to a
+ * specific value. This function is expected to work exactly
+ * like memset()
+ */
+void cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt)
+{
+ memset(ptr, value, cnt) ;
+}
+
+/*
+ * This function is expected to create a sleep channel.
+ * The data structure that represents the sleep channel object
+ * sleep channel (which is Linux "wait_queue_head_t wq" for this paticular HAL)
+ * passed as a pointer, and allpocated by the caller
+ * (typically as a local var on the stack) "Create" word should read as
+ * "SleepOn", this func doesn't actually create anything
+ */
+cy_bool cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ init_waitqueue_head(&channel->wq) ;
+ return cy_true ;
+}
+
+/*
+ * for this particular HAL it doesn't actually destroy anything
+ * since no actual sleep object is created in CreateSleepChannel()
+ * sleep channel is given by the pointer in the argument.
+ */
+cy_bool cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel)
+{
+ return cy_true ;
+}
+
+/*
+ * platform specific wakeable Sleep implementation
+ */
+cy_bool cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms)
+{
+ wait_event_interruptible_timeout(channel->wq, 0, ((ms * HZ)/1000)) ;
+ return cy_true ;
+}
+
+/*
+ * wakes up the process waiting on the CHANNEL
+ */
+cy_bool cy_as_hal_wake(cy_as_hal_sleep_channel *channel)
+{
+ wake_up_interruptible_all(&channel->wq);
+ return cy_true ;
+}
+
+uint32_t cy_as_hal_disable_interrupts()
+{
+ if (0 == intr__enable)
+ ;
+
+ intr__enable++ ;
+ return 0 ;
+}
+
+void cy_as_hal_enable_interrupts(uint32_t val)
+{
+ intr__enable-- ;
+ if (0 == intr__enable)
+ ;
+}
+
+/*
+ * Sleep atleast 150ns, cpu dependent
+ */
+void cy_as_hal_sleep150(void)
+{
+ uint32_t i, j;
+
+ j = 0;
+ for (i = 0; i < 1000; i++)
+ j += (~i);
+}
+
+void cy_as_hal_sleep(uint32_t ms)
+{
+ cy_as_hal_sleep_channel channel;
+
+ cy_as_hal_create_sleep_channel(&channel) ;
+ cy_as_hal_sleep_on(&channel, ms) ;
+ cy_as_hal_destroy_sleep_channel(&channel) ;
+}
+
+cy_bool cy_as_hal_is_polling()
+{
+ return cy_false;
+}
+
+void cy_as_hal_c_b_free(void *ptr)
+{
+ cy_as_hal_free(ptr);
+}
+
+/*
+ * suppose to reinstate the astoria registers
+ * that may be clobbered in sleep mode
+ */
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup)
+{
+ (void) tag;
+ (void) is_standby_wakeup;
+
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+
+}
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag)
+{
+ /* specific to SPI, no implementation required */
+ (void) tag;
+}
+
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag)
+{
+ /*
+ * we are in asynchronous mode. so no need to handle this
+ */
+ return true;
+}
+
+/*
+ * init OMAP h/w resources
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug)
+{
+ cy_as_omap_dev_kernel *dev_p ;
+ int i;
+ u16 data16[4];
+ u8 pncfg_reg;
+
+ /*
+ * No debug mode support through argument as of now
+ */
+ (void)debug;
+
+ DBGPRN(KERN_INFO"starting OMAP34xx HAL...\n");
+
+ /*
+ * Initialize the HAL level endpoint DMA data.
+ */
+ for (i = 0 ; i < sizeof(end_points)/sizeof(end_points[0]) ; i++) {
+ end_points[i].data_p = 0 ;
+ end_points[i].pending = cy_false ;
+ end_points[i].size = 0 ;
+ end_points[i].type = cy_as_hal_none ;
+ end_points[i].sg_list_enabled = cy_false;
+
+ /*
+ * by default the DMA transfers to/from the E_ps don't
+ * use sg_list that implies that the upper devices like
+ * blockdevice have to enable it for the E_ps in their
+ * initialization code
+ */
+ }
+
+ /*
+ * allocate memory for OMAP HAL
+ */
+ dev_p = (cy_as_omap_dev_kernel *)cy_as_hal_alloc(
+ sizeof(cy_as_omap_dev_kernel)) ;
+ if (dev_p == 0) {
+ cy_as_hal_print_message("out of memory allocating OMAP"
+ "device structure\n") ;
+ return 0 ;
+ }
+
+ dev_p->m_sig = CY_AS_OMAP_PNAND_HAL_SIG;
+
+ /*
+ * initialize OMAP hardware and StartOMAPKernelall gpio pins
+ */
+ dev_p->m_addr_base = (void *)cy_as_hal_processor_hw_init();
+
+ /*
+ * Now perform a hard reset of the device to have
+ * the new settings take effect
+ */
+ __gpio_set_value(AST_WAKEUP, 1);
+
+ /*
+ * do Astoria h/w reset
+ */
+ DBGPRN(KERN_INFO"-_-_pulse -> westbridge RST pin\n");
+
+ /*
+ * NEGATIVE PULSE on RST pin
+ */
+ __gpio_set_value(AST_RESET, 0);
+ mdelay(1);
+ __gpio_set_value(AST_RESET, 1);
+ mdelay(50);
+
+ /*
+ * note AFTER reset PNAND interface is 8 bit mode
+ * so if gpmc Is configured in 8 bit mode upper half will be FF
+ */
+ pncfg_reg = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+#ifdef PNAND_16BIT_MODE
+
+ /*
+ * switch to 16 bit mode, force NON-LNA LBD mode, 3 RA addr bytes
+ */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0001);
+
+ /*
+ * now in order to continue to talk to astoria
+ * sw OMAP GPMC into 16 bit mode as well
+ */
+ cy_as_hal_gpmc_enable_16bit_bus(cy_true);
+#else
+ /* Astoria and GPMC are already in 8 bit mode,
+ * just initialize PNAND_CFG */
+ ast_p_nand_casdi_write(CY_AS_MEM_PNAND_CFG, 0x0000);
+#endif
+
+ /*
+ * NOTE: if you want to capture bus activity on the LA,
+ * don't use printks in between the activities you want to capture.
+ * prinks may take milliseconds, and the data of interest
+ * will fall outside the LA capture window/buffer
+ */
+ data16[0] = ast_p_nand_casdo_read(CY_AS_MEM_CM_WB_CFG_ID);
+ data16[1] = ast_p_nand_casdo_read(CY_AS_MEM_PNAND_CFG);
+
+ if (data16[0] != 0xA200) {
+ /*
+ * astoria device is not found
+ */
+ printk(KERN_ERR "ERROR: astoria device is not found, CY_AS_MEM_CM_WB_CFG_ID ");
+ printk(KERN_ERR "read returned:%4.4X: CY_AS_MEM_PNAND_CFG:%4.4x !\n",
+ data16[0], data16[0]);
+ goto bus_acc_error;
+ }
+
+ printk(KERN_INFO " register access CASDO test:"
+ "\n CY_AS_MEM_CM_WB_CFG_ID:%4.4x\n"
+ "PNAND_CFG after RST:%4.4x\n "
+ "CY_AS_MEM_PNAND_CFG"
+ "after cfg_wr:%4.4x\n\n",
+ data16[0], pncfg_reg, data16[1]);
+
+ dev_p->thread_flag = 1 ;
+ spin_lock_init(&int_lock) ;
+ dev_p->m_next_p = m_omap_list_p ;
+
+ m_omap_list_p = dev_p ;
+ *tag = dev_p;
+
+ cy_as_hal_configure_interrupts((void *)dev_p);
+
+ printk(KERN_INFO "OMAP3430__hal started tag:%p"
+ ", kernel HZ:%d\n", dev_p, HZ);
+
+ /*
+ *make processor to storage endpoints SG assisted by default
+ */
+ cy_as_hal_set_ep_dma_mode(4, true);
+ cy_as_hal_set_ep_dma_mode(8, true);
+
+ return 1 ;
+
+ /*
+ * there's been a NAND bus access error or
+ * astoria device is not connected
+ */
+bus_acc_error:
+ /*
+ * at this point hal tag hasn't been set yet
+ * so the device will not call omap_stop
+ */
+ cy_as_hal_omap_hardware_deinit(dev_p);
+ cy_as_hal_free(dev_p) ;
+ return 0;
+}
+
+#else
+/*
+ * Some compilers do not like empty C files, so if the OMAP hal is not being
+ * compiled, we compile this single function. We do this so that for a
+ * given target HAL there are not multiple sources for the HAL functions.
+ */
+void my_o_m_a_p_kernel_hal_dummy_function(void)
+{
+}
+
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h 2010-09-07 11:03:12.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_cram.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+/* Debug printing enabled */
+
+#define DBGPRN(...) printk(__VA_ARGS__)
+#define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+/** NO DEBUG PRINTING **/
+#define DBGPRN(...)
+#define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_cram_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_cram_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasmemmap.h 2010-09-07 11:05:48.000000000 -0700
@@ -0,0 +1,564 @@
+/*
+OMAP3430 ZOOM MDK astoria interface defs(cyasmemmap.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+#include <mach/mux.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+#include <linux/../../arch/arm/plat-omap/include/plat/mux.h>
+
+#ifndef _INCLUDED_CYASMEMMAP_H_
+#define _INCLUDED_CYASMEMMAP_H_
+
+/* defines copied from OMAP kernel branch */
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
+#define OMAP34XX_MUX_MODE0 0
+#define OMAP34XX_MUX_MODE4 4
+#define OMAP3_INPUT_EN (1 << 8)
+#define OMAP34XX_PIN_INPUT_PULLUP (OMAP2_PULL_ENA | OMAP3_INPUT_EN \
+ | OMAP2_PULL_UP)
+
+/*
+ * for OMAP3430 <-> astoria : ADmux mode, 8 bit data path
+ * WB Signal- OMAP3430 signal COMMENTS
+ * --------------------------- --------------------
+ * CS_L -GPMC_nCS4_GPIO_53 ZOOM I SOM board
+ * signal: up_nCS_A_EXT
+ * AD[7:0]-upD[7:0] buffered on the
+ * transposer board
+ * GPMC_ADDR
+ * [A8:A1]->upD[7:0]
+ * INT# -GPMC_nWP_GPIO_62
+ * DACK -N/C not conected
+ * WAKEUP-GPIO_167
+ * RESET-GPIO_126
+ * R/B -GPMC_WAIT2_GPIO_64
+ * -------------------------------------------
+ * The address range for nCS1B is 0x06000000 - 0x07FF FFFF.
+*/
+
+/*
+ *OMAP_ZOOM LEDS
+ */
+#define LED_0 156
+#define LED_1 128
+#define LED_2 64
+#define LED_3 60
+
+#define HIGH 1
+#define LOW 1
+
+/*
+ *omap GPIO number
+ */
+#define AST_WAKEUP 167
+#define AST_RESET 126
+#define AST__rn_b 64
+
+/*
+ * NOTE THIS PIN IS USED AS WP for OMAP NAND
+ */
+#define AST_INT 62
+
+/* N4_OMAP3430_GPMC_A1
+ M4_OMAP3430_GPMC_A2
+ L4_OMAP3430_GPMC_A3
+ K4_OMAP3430_GPMC_A4
+ T3_OMAP3430_GPMC_A5
+ R3_OMAP3430_GPMC_A6
+ N3_OMAP3430_GPMC_A7
+ M3_OMAP3430_GPMC_A8
+ * as an I/O, it is actually controlled by GPMC
+ */
+#define AST_CS 55
+
+
+/*
+ *GPMC prefetch engine
+ */
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG1 0x01E0
+
+ /*32 bytes for 16 bit pnand mode*/
+ #define PFE_THRESHOLD 31
+
+ /*
+ * bit fields
+ * PF_ACCESSMODE - 0 - read mode, 1 - write mode
+ * PF_DMAMODE - 0 - default only intr line signal will be generated
+ * PF_SYNCHROMODE - default 0 - engin will start access as soon as
+ * ctrl re STARTENGINE is set
+ * PF_WAITPINSEL - FOR synchro mode selects WAIT pin whch edge
+ * will be monitored
+ * PF_EN_ENGINE - 1- ENABLES ENGINE, but it needs to be started after
+ * that C ctrl reg bit 0
+ * PF_FIFO_THRESHOLD - FIFO threshhold in number of BUS(8 or 16) words
+ * PF_WEIGHTED_PRIO - NUM of cycles granted to PFE if RND_ROBIN
+ * prioritization is enabled
+ * PF_ROUND_ROBIN - if enabled, gives priority to other CS, but
+ * reserves NUM of cycles for PFE's turn
+ * PF_ENGIN_CS_SEL - GPMC CS assotiated with PFE function
+ */
+ #define PF_ACCESSMODE (0 << 0)
+ #define PF_DMAMODE (0 << 2)
+ #define PF_SYNCHROMODE (0 << 3)
+ #define PF_WAITPINSEL (0x0 << 4)
+ #define PF_EN_ENGINE (1 << 7)
+ #define PF_FIFO_THRESHOLD (PFE_THRESHOLD << 8)
+ #define PF_WEIGHTED_PRIO (0x0 << 16)
+ #define PF_ROUND_ROBIN (0 << 23)
+ #define PF_ENGIN_CS_SEL (AST_GPMC_CS << 24)
+ #define PF_EN_OPTIM_ACC (0 << 27)
+ #define PF_CYCLEOPTIM (0x0 << 28)
+
+#define GPMC_PREFETCH_CONFIG1_VAL (PF_ACCESSMODE | \
+ PF_DMAMODE | PF_SYNCHROMODE | \
+ PF_WAITPINSEL | PF_EN_ENGINE | \
+ PF_FIFO_THRESHOLD | PF_FIFO_THRESHOLD | \
+ PF_WEIGHTED_PRIO | PF_ROUND_ROBIN | \
+ PF_ENGIN_CS_SEL | PF_EN_OPTIM_ACC | \
+ PF_CYCLEOPTIM)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONFIG2 0x01E4
+ /*
+ * bit fields
+ * 14 bit field NOTE this counts is also
+ * is in number of BUS(8 or 16) words
+ */
+ #define PF_TRANSFERCOUNT (0x000)
+
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_CONTROL 0x01EC
+ /*
+ * bit fields , ONLY BIT 0 is implemented
+ * PFWE engin must be programmed with this bit = 0
+ */
+ #define PFPW_STARTENGINE (1 << 0)
+
+/* register and its bit fields */
+#define GPMC_PREFETCH_STATUS 0x01F0
+
+ /* */
+ #define PFE_FIFO_THRESHOLD (1 << 16)
+
+/*
+ * GPMC posted write/prefetch engine end
+ */
+
+
+/*
+ * chip select number on GPMC ( 0..7 )
+ */
+#define AST_GPMC_CS 4
+
+/*
+ * not connected
+ */
+#define AST_DACK 00
+
+
+/*
+ * Physical address above the NAND flash
+ * we use CS For mapping in OMAP3430 RAM space use 0x0600 0000
+ */
+#define CYAS_DEV_BASE_ADDR (0x20000000)
+#define CYAS_DEV_MAX_ADDR (0xFF)
+#define CYAS_DEV_ADDR_RANGE (CYAS_DEV_MAX_ADDR << 1)
+
+/* in CRAM or PSRAM mode OMAP A1..An wires-> Astoria, there is no A0 line */
+#define CYAS_DEV_CALC_ADDR(cyas_addr) (cyas_addr << 1)
+#define CYAS_DEV_CALC_EP_ADDR(ep) (ep << 1)
+
+/*
+ *OMAP3430 i/o access macros
+ */
+#define IORD32(addr) (*(volatile u32 *)(addr))
+#define IOWR32(addr, val) (*(volatile u32 *)(addr) = val)
+
+#define IORD16(addr) (*(volatile u16 *)(addr))
+#define IOWR16(addr, val) (*(volatile u16 *)(addr) = val)
+
+#define IORD8(addr) (*(volatile u8 *)(addr))
+#define IOWR8(addr, val) (*(volatile u8 *)(addr) = val)
+
+/*
+ * local defines for accessing to OMAP GPIO ***
+ */
+#define CTLPADCONF_BASE_ADDR 0x48002000
+#define CTLPADCONF_SIZE 0x1000
+
+#define GPIO1_BASE_ADDR 0x48310000
+#define GPIO2_BASE_ADDR 0x49050000
+#define GPIO3_BASE_ADDR 0x49052000
+#define GPIO4_BASE_ADDR 0x49054000
+#define GPIO5_BASE_ADDR 0x49056000
+#define GPIO6_BASE_ADDR 0x49058000
+#define GPIO_SPACE_SIZE 0x1000
+
+
+/*
+ * OMAP3430 GPMC timing for pNAND interface
+ */
+#define GPMC_BASE 0x6E000000
+#define GPMC_REGION_SIZE 0x1000
+#define GPMC_CONFIG_REG (0x50)
+
+/*
+ * bit 0 in the GPMC_CONFIG_REG
+ */
+#define NAND_FORCE_POSTED_WRITE_B 1
+
+/*
+ * WAIT2STATUS, must be (1 << 10)
+ */
+#define AS_WAIT_PIN_MASK (1 << 10)
+
+/*
+ * GPMC_CONFIG(reg number [1..7] [for chip sel CS[0..7])
+ */
+#define GPMC_CFG_REG(N, CS) ((0x60 + (4*(N-1))) + (0x30*CS))
+
+/*
+ * we get "gpmc_base" from kernel
+ */
+#define GPMC_VMA(offset) (gpmc_base + offset)
+
+/*
+ * GPMC CS space VMA start address
+ */
+#define GPMC_CS_VMA(offset) (gpmc_data_vma + offset)
+
+/*
+ * PAD_CFG mux space VMA
+ */
+#define PADCFG_VMA(offset) (iomux_vma + offset)
+
+/*
+ * CONFIG1: by default, sngle access, async r/w RD_MULTIPLE[30]
+ * WR_MULTIPLE[28]; GPMC_FCL_DIV[1:0]
+ */
+#define GPMC_FCLK_DIV ((0) << 0)
+
+/*
+ * ADDITIONAL DIVIDER FOR ALL TIMING PARAMS
+ */
+#define TIME_GRAN_SCALE ((0) << 4)
+
+/*
+ * for use by gpmc_set_timings api, measured in ns, not clocks
+ */
+#define WB_GPMC_BUSCYC_t (7 * 6)
+#define WB_GPMC_CS_t_on (0)
+#define WB_GPMC_ADV_t_on (1 * 6)
+#define WB_GPMC_ADV_t_off (3 * 6)
+#define WB_GPMC_OE_t_on (2 * 6)
+#define WB_GPMC_OE_t_off (6 * 6)
+#define WB_GPMC_WE_t_on (2 * 6)
+#define WB_GPMC_WE_t_off (6 * 6)
+#define WB_GPMC_RDS_ADJ (1 * 6)
+#define WB_GPMC_RD_t_a_c_c (WB_GPMC_OE_t_off + WB_GPMC_RDS_ADJ)
+#define WB_GPMC_WR_t_a_c_c (WB_GPMC_BUSCYC_t)
+
+#define DIR_OUT 0
+#define DIR_INP 1
+#define DRV_HI 1
+#define DRV_LO 0
+
+/*
+ * GPMC_CONFIG7[cs] register bit fields3
+ * AS_CS_MASK - 3 bit mask for A26,A25,A24,
+ * AS_CS_BADDR - 6 BIT VALUE A29 ...A24
+ * CSVALID_B - CSVALID bit on GPMC_CONFIG7[cs] register
+ */
+#define AS_CS_MASK (0X7 << 8)
+#define AS_CS_BADDR 0x02
+#define CSVALID_B (1 << 6)
+
+/*
+ * DEFINE OMAP34XX GPIO OFFSETS (should have been defined in kernel /arch
+ * these are offsets from the BASE_ADDRESS of the GPIO BLOCK
+ */
+#define GPIO_REVISION 0x000
+#define GPIO_SYSCONFIG 0x010
+#define GPIO_SYSSTATUS1 0x014
+#define GPIO_IRQSTATUS1 0x0183
+#define GPIO_IRQENABLE1 0x01C
+#define GPIO_IRQSTATUS2 0x028
+#define GPIO_CTRL 0x030
+#define GPIO_OE 0x034
+#define GPIO_DATA_IN 0x038
+#define GPIO_DATA_OUT 0x03C
+#define GPIO_LEVELDETECT0 0x040
+#define GPIO_LEVELDETECT1 0x044
+#define GPIO_RISINGDETECT 0x048
+#define GPIO_FALLINGDETECT 0x04c
+#define GPIO_CLEAR_DATAOUT 0x090
+#define GPIO_SET_DATAOUT 0x094
+
+typedef struct {
+ char *name;
+ u32 phy_addr;
+ u32 virt_addr;
+ u32 size;
+} io2vma_tab_t;
+
+/*
+ * GPIO phy to translation VMA table
+ */
+static io2vma_tab_t gpio_vma_tab[6] = {
+ {"GPIO1_BASE_ADDR", GPIO1_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO2_BASE_ADDR", GPIO2_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO3_BASE_ADDR", GPIO3_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO4_BASE_ADDR", GPIO4_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO5_BASE_ADDR", GPIO5_BASE_ADDR , 0 , GPIO_SPACE_SIZE},
+ {"GPIO6_BASE_ADDR", GPIO6_BASE_ADDR , 0 , GPIO_SPACE_SIZE}
+};
+/*
+ * name - USER signal name assigned to the pin ( for printks)
+ * mux_func - enum index NAME for the pad_cfg function
+ * pin_num - pin_number if mux_func is GPIO, if not a GPIO it is -1
+ * mux_ptr - pointer to the corresponding pad_cfg_reg
+ * (used for pad release )
+ * mux_save - preserve here original PAD_CNF value for this
+ * pin (used for pad release)
+ * dir - if GPIO: 0 - OUT , 1 - IN
+ * dir_save - save original pin direction
+ * drv - initial drive level "0" or "1"
+ * drv_save - save original pin drive level
+ * valid - 1 if successfuly configured
+*/
+typedef struct {
+ char *name;
+ u32 mux_func;
+ int pin_num;
+ u16 *mux_ptr;
+ u16 mux_save;
+ u8 dir;
+ u8 dir_save;
+ u8 drv;
+ u8 drv_save;
+ u8 valid;
+} user_pad_cfg_t;
+
+/*
+ * need to ensure that enums are in sync with the
+ * omap_mux_pin_cfg table, these enums designate
+ * functions that OMAP pads can be configured to
+ */
+enum {
+ B23_OMAP3430_GPIO_167,
+ D23_OMAP3430_GPIO_126,
+ H1_OMAP3430_GPIO_62,
+ H1_OMAP3430_GPMC_n_w_p,
+ T8_OMAP3430_GPMC_n_c_s4,
+ T8_OMAP3430_GPIO_55,
+ R25_OMAP3430_GPIO_156,
+ R27_OMAP3430_GPIO_128,
+ K8_OMAP3430_GPIO_64,
+ K8_GPMC_WAIT2,
+ G3_OMAP3430_GPIO_60,
+ G3_OMAP3430_n_b_e0_CLE,
+ C6_GPMC_WAIT3,
+ J1_OMAP3430_GPIO_61,
+ C6_OMAP3430_GPIO_65,
+ N4_OMAP3430_GPMC_A1,
+ M4_OMAP3430_GPMC_A2,
+ L4_OMAP3430_GPMC_A3,
+ K4_OMAP3430_GPMC_A4,
+ T3_OMAP3430_GPMC_A5,
+ R3_OMAP3430_GPMC_A6,
+ N3_OMAP3430_GPMC_A7,
+ M3_OMAP3430_GPMC_A8,
+ END_OF_TABLE
+};
+
+/*
+ * number of GPIOS we plan to grab
+ */
+#define GPIO_SLOTS 8
+
+/*
+ * user_pads_init() reads(and saves) from/to this table
+ * used in conjunction with omap_3430_mux_t table in .h file
+ * because the way it's done in the kernel code
+ * TODO: implement restore of the the original cfg and i/o regs
+ */
+
+static user_pad_cfg_t user_pad_cfg[] = {
+ /*
+ * name,pad_func,pin_num, mux_ptr, mux_sav, dir,
+ * dir_sav, drv, drv_save, valid
+ */
+ {"AST_WAKEUP", B23_OMAP3430_GPIO_167, 167, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST_RESET", D23_OMAP3430_GPIO_126, 126, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"AST__rn_b", K8_GPMC_WAIT2, 64, NULL, 0,
+ DIR_INP, 0, 0, 0, 0},
+ {"AST_INTR", H1_OMAP3430_GPIO_62, 62, NULL, 0,
+ DIR_INP, 0, DRV_HI, 0, 0},
+ {"AST_CS", T8_OMAP3430_GPMC_n_c_s4, 55, NULL, 0,
+ DIR_OUT, 0, DRV_HI, 0, 0},
+ {"LED_0", R25_OMAP3430_GPIO_156, 156, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"LED_1", R27_OMAP3430_GPIO_128, 128, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ {"AST_CLE", G3_OMAP3430_n_b_e0_CLE , 60, NULL, 0,
+ DIR_OUT, 0, DRV_LO, 0, 0},
+ /*
+ * Z terminator, must always be present
+ * for sanity check, don't remove
+ */
+ {NULL}
+};
+
+#define GPIO_BANK(pin) (pin >> 5)
+#define REG_WIDTH 32
+#define GPIO_REG_VMA(pin_num, offset) \
+ (gpio_vma_tab[GPIO_BANK(pin_num)].virt_addr + offset)
+
+/*
+ * OMAP GPIO_REG 32 BIT MASK for a bit or
+ * flag in gpio_No[0..191] apply it to a 32 b3it
+ * location to set clear or check on a corresponding
+ * gpio bit or flag
+ */
+#define GPIO_REG_MASK(pin_num) (1 << \
+ (pin_num - (GPIO_BANK(pin_num) * REG_WIDTH)))
+
+/*
+ * OMAP GPIO registers bitwise access macros
+ */
+
+#define OMAP_GPIO_BIT(pin_num, reg) \
+ ((*((u32 *)GPIO_REG_VMA(pin_num, reg)) \
+ & GPIO_REG_MASK(pin_num)) ? 1 : 0)
+
+#define RD_OMAP_GPIO_BIT(pin_num, v) OMAP_GPIO_BIT(pin_num, reg)
+
+/*
+ *these are superfast set/clr bitbang macro, 48ns cyc tyme
+ */
+#define OMAP_SET_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_SET_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+#define OMAP_CLR_GPIO(pin_num) \
+ (*(u32 *)GPIO_REG_VMA(pin_num, GPIO_CLEAR_DATAOUT) \
+ = GPIO_REG_MASK(pin_num))
+
+#define WR_OMAP_GPIO_BIT(pin_num, v) \
+ (v ? (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_SET_DATAOUT) = GPIO_REG_MASK(pin_num)) \
+ : (*(u32 *)GPIO_REG_VMA(pin_num, \
+ GPIO_CLEAR_DATAOUT) = GPIO_REG_MASK(pin_num)))
+
+/*
+ * Note this pin cfg mimicks similar implementation
+ * in linux kernel, which unfortunately doesn't allow
+ * us to dynamically insert new custom GPIO mux
+ * configurations all REG definitions used in this
+ * applications. to add a new pad_cfg function, insert
+ * a new ENUM and new pin_cfg entry in omap_mux_pin_cfg[]
+ * table below3
+ *
+ * offset - note this is a word offset since the
+ * SCM regs are 16 bit packed in one 32 bit word
+ * mux_val - just enough to describe pins used
+ */
+typedef struct {
+ char *name;
+ u16 offset;
+ u16 mux_val;
+} omap_3430_mux_t;
+
+/*
+ * "OUTIN" is configuration when DATA reg drives the
+ * pin but the level at the pin can be sensed
+ */
+#define PAD_AS_OUTIN (OMAP34XX_MUX_MODE4 | \
+ OMAP34XX_PIN_OUTPUT | OMAP34XX_PIN_INPUT)
+
+omap_3430_mux_t omap_mux_pin_cfg[] = {
+ /*
+ * B23_OMAP3430_GPIO_167 - GPIO func to PA3D 167 WB wakeup
+ * D23_OMAP3430_GPIO_126 - drive GPIO_126 ( AST RESET)
+ * H1_OMAP3430_GPIO_62 - need a pullup on this pin
+ * H1_OMAP3430_GPMC_n_w_p - GPMC NAND CTRL n_w_p out
+ * T8_OMAP3430_GPMC_n_c_s4" - T8 is controlled b_y GPMC NAND ctrl
+ * R25_OMAP3430_GPIO_156 - OMAPZOOM drive LED_0
+ * R27_OMAP3430_GPIO_128 - OMAPZOOM drive LED_1
+ * K8_OMAP3430_GPIO_64 - OMAPZOOM drive LED_2
+ * K8_GPMC_WAIT2 - GPMC WAIT2 function on PAD K8
+ * G3_OMAP3430_GPIO_60 - OMAPZOOM drive LED_3
+ * G3_OMAP3430_n_b_e0_CLE -GPMC NAND ctrl CLE signal
+ */
+
+ {"B23_OMAP3430_GPIO_167", 0x0130, (OMAP34XX_MUX_MODE4)},
+ {"D23_OMAP3430_GPIO_126", 0x0132, (OMAP34XX_MUX_MODE4)},
+ {"H1_OMAP3430_GPIO_62", 0x00CA, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP) },
+ {"H1_OMAP3430_GPMC_n_w_p", 0x00CA, (OMAP34XX_MUX_MODE0)},
+ {"T8_OMAP3430_GPMC_n_c_s4", 0x00B6, (OMAP34XX_MUX_MODE0) },
+ {"T8_OMAP3430_GPIO_55", 0x00B6, (OMAP34XX_MUX_MODE4) },
+ {"R25_OMAP3430_GPIO_156", 0x018C, (OMAP34XX_MUX_MODE4) },
+ {"R27_OMAP3430_GPIO_128", 0x0154, (OMAP34XX_MUX_MODE4) },
+ {"K8_OMAP3430_GPIO_64", 0x00d0, (OMAP34XX_MUX_MODE4) },
+ {"K8_GPMC_WAIT2", 0x00d0, (OMAP34XX_MUX_MODE0) },
+ {"G3_OMAP3430_GPIO_60", 0x00C6, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"G3_OMAP3430_n_b_e0_CLE", 0x00C6, (OMAP34XX_MUX_MODE0)},
+ {"C6_GPMC_WAIT3", 0x00d2, (OMAP34XX_MUX_MODE0)},
+ {"C6_OMAP3430_GPIO_65", 0x00d2, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN)},
+ {"J1_OMAP3430_GPIO_61", 0x00C8, (OMAP34XX_MUX_MODE4 |
+ OMAP3_INPUT_EN | OMAP34XX_PIN_INPUT_PULLUP)},
+ {"N4_OMAP3430_GPMC_A1", 0x007A, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M4_OMAP3430_GPMC_A2", 0x007C, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"L4_OMAP3430_GPMC_A3", 0x007E, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"K4_OMAP3430_GPMC_A4", 0x0080, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"T3_OMAP3430_GPMC_A5", 0x0082, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"R3_OMAP3430_GPMC_A6", 0x0084, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"N3_OMAP3430_GPMC_A7", 0x0086, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ {"M3_OMAP3430_GPMC_A8", 0x0088, (OMAP34XX_MUX_MODE0 |
+ OMAP34XX_PIN_INPUT_PULLUP)},
+ /*
+ * don't remove, used for sanity check.
+ */
+ {"END_OF_TABLE"}
+};
+
+
+#endif /* _INCLUDED_CYASMEMMAP_H_ */
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyasomapdev_cram.h 2010-09-07 11:06:34.000000000 -0700
@@ -0,0 +1,73 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_CRAM_HAL_SIG (0x2332)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_phy_addr_base;
+
+ void *m_vma_addr_base;
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,319 +0,0 @@
-/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-/*
- * This file contains the defintion of the hardware abstraction
- * layer on OMAP3430 talking to the West Bridge Astoria device
- */
-
-
-#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
-#define _INCLUDED_CYASHALOMAP_KERNEL_H_
-
-#include <linux/kernel.h>
-#include <linux/module.h>
-#include <linux/wait.h>
-#include <linux/string.h>
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-typedef struct cy_as_hal_sleep_channel_t {
- wait_queue_head_t wq ;
-} cy_as_hal_sleep_channel ;
-
-/* moved to staging location, eventual location
- * considered is here
-#include <mach/westbridge/cyashaldef.h>
-#include <linux/westbridge/cyastypes.h>
-#include <linux/westbridge/cyas_cplus_start.h>
-*/
-#include "../cyashaldef.h"
-#include "../../../../../../../include/linux/westbridge/cyastypes.h"
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#include "cyasomapdev_kernel.h"
-
-/*
- * Below are the data structures that must be defined by the HAL layer
- */
-
-/*
- * The HAL layer must define a TAG for identifying a specific Astoria
- * device in the system. In this case the tag is a void * which is
- * really an OMAP device pointer
- */
-typedef void *cy_as_hal_device_tag ;
-
-
-/* This must be included after the CyAsHalDeviceTag type is defined */
-
-/* moved to staging location, eventual location
- * considered is here
- * #include <linux/westbridge/cyashalcb.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
-/*
- * Below are the functions that communicate with the West Bridge
- * device. These are system dependent and must be defined by
- * the HAL layer for a given system.
- */
-
-/*
- * This function must be defined to write a register within the Antioch
- * device. The addr value is the address of the register to write with
- * respect to the base address of the Antioch device.
- */
-void
-cy_as_hal_write_register(cy_as_hal_device_tag tag,
- uint16_t addr, uint16_t data) ;
-
-/*
- * This function must be defined to read a register from
- * the west bridge device. The addr value is the address of
- * the register to read with respect to the base address
- * of the west bridge device.
- */
-uint16_t
-cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
-
-/*
- * This function must be defined to transfer a block of data
- * to the west bridge device. This function can use the burst write
- * (DMA) capabilities of Antioch to do this, or it can just copy
- * the data using writes.
- */
-void
-cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
- uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to transfer a block of data
- * from the Antioch device. This function can use the burst
- * read (DMA) capabilities of Antioch to do this, or it can
- * just copy the data using reads.
- */
-void
-cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
- void *buf, uint32_t size, uint16_t maxsize) ;
-
-/*
- * This function must be defined to cancel any pending DMA request.
- */
-void
-cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
-
-/*
- * This function must be defined to allow the Antioch API to
- * register a callback function that is called when a DMA transfer
- * is complete.
- */
-void
-cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
- cy_as_hal_dma_complete_callback cb) ;
-
-/*
- * This function must be defined to return the maximum size of DMA
- * request that can be handled on the given endpoint. The return
- * value should be the maximum size in bytes that the DMA module can
- * handle.
- */
-uint32_t
-cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
- cy_as_end_point_number_t ep) ;
-
-/*
- * This function must be defined to set the state of the WAKEUP pin
- * on the Antioch device. Generally this is done via a GPIO of some
- * type.
- */
-cy_bool
-cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
-
-/*
- * This function is called when the Antioch PLL loses lock, because
- * of a problem in the supply voltage or the input clock.
- */
-void
-cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
-
-
-/**********************************************************************
- *
- * Below are the functions that must be defined to provide the basic
- * operating system services required by the API.
- *
-***********************************************************************/
-
-/*
- * This function is required by the API to allocate memory. This function
- * is expected to work exactly like malloc().
- */
-void *
-cy_as_hal_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsHalAlloc(). This function is expected to work exacly like free().
- */
-void
-cy_as_hal_free(void *mem_p) ;
-
-/*
- * This function is required by the API to allocate memory during a
- * callback. This function must be able to provide storage at inturupt
- * time.
- */
-void *
-cy_as_hal_c_b_alloc(uint32_t cnt) ;
-
-/*
- * This function is required by the API to free memory allocated with
- * CyAsCBHalAlloc().
- */
-void
-cy_as_hal_c_b_free(void *ptr) ;
-
-/*
- * This function is required to set a block of memory to a specific
- * value. This function is expected to work exactly like memset()
- */
-void
-cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
-
-/*
- * This function is expected to create a sleep channel. The data
- * structure that represents the sleep channel is given by the
- * pointer in the argument.
- */
-cy_bool
-cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-/*
- * This function is expected to destroy a sleep channel. The data
- * structure that represents the sleep channel is given by
- * the pointer in the argument.
- */
-
-
-cy_bool
-cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
-
-cy_bool
-cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
-
-cy_bool
-cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
-
-uint32_t
-cy_as_hal_disable_interrupts(void) ;
-
-void
-cy_as_hal_enable_interrupts(uint32_t);
-
-void
-cy_as_hal_sleep150(void);
-
-void
-cy_as_hal_sleep(uint32_t ms);
-
-cy_bool
-cy_as_hal_is_polling(void);
-
-void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
- cy_bool is_standby_wakeup);
-
-/*
- * required only in spi mode
- */
-cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
-
-void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
-
-
-#ifndef NDEBUG
-#define cy_as_hal_assert(cond) if (!(cond))\
- printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
-#else
-#define cy_as_hal_assert(cond)
-#endif
-
-#define cy_as_hal_print_message printk
-
-/* removable debug printks */
-#ifndef WESTBRIDGE_NDEBUG
-#define DBG_PRINT_ENABLED
-#endif
-
-/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
-
-
-#ifdef DBG_PRINT_ENABLED
- /* Debug printing enabled */
-
- #define DBGPRN(...) printk(__VA_ARGS__)
- #define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
- current->pid, __func__)
-
-#else
- /** NO DEBUG PRINTING **/
- #define DBGPRN(...)
- #define DBGPRN_FUNC_NAME
-
-#endif
-
-/*
-CyAsMiscSetLogLevel(uint8_t level)
-{
- debug_level = level ;
-}
-
-#ifdef CY_AS_LOG_SUPPORT
-
-void
-cy_as_log_debug_message(int level, const char *str)
-{
- if (level <= debug_level)
- cy_as_hal_print_message("log %d: %s\n", level, str) ;
-}
-*/
-
-
-/*
- * print buffer helper
- */
-void cyashal_prn_buf(void *buf, uint16_t offset, int len);
-
-/*
- * These are the functions that are not part of the HAL layer,
- * but are required to be called for this HAL.
- */
-int start_o_m_a_p_kernel(const char *pgm,
- cy_as_hal_device_tag *tag, cy_bool debug) ;
-int stop_o_m_a_p_kernel(const char *pgm, cy_as_hal_device_tag tag) ;
-int omap_start_intr(cy_as_hal_device_tag tag) ;
-void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
-
-/* moved to staging location
-#include <linux/westbridge/cyas_cplus_end.h>
-*/
-#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
-#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h 2010-09-07 11:56:27.000000000 -0700
@@ -0,0 +1,319 @@
+/* Cypress Antioch HAL for OMAP KERNEL header file (cyashalomapkernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+/*
+ * This file contains the defintion of the hardware abstraction
+ * layer on OMAP3430 talking to the West Bridge Astoria device
+ */
+
+
+#ifndef _INCLUDED_CYASHALOMAP_KERNEL_H_
+#define _INCLUDED_CYASHALOMAP_KERNEL_H_
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/string.h>
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+typedef struct cy_as_hal_sleep_channel_t {
+ wait_queue_head_t wq ;
+} cy_as_hal_sleep_channel ;
+
+/* moved to staging location, eventual location
+ * considered is here
+#include <mach/westbridge/cyashaldef.h>
+#include <linux/westbridge/cyastypes.h>
+#include <linux/westbridge/cyas_cplus_start.h>
+*/
+#include "../cyashaldef.h"
+#include "../../../../../../../include/linux/westbridge/cyastypes.h"
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#include "cyasomapdev_pnand.h"
+
+/*
+ * Below are the data structures that must be defined by the HAL layer
+ */
+
+/*
+ * The HAL layer must define a TAG for identifying a specific Astoria
+ * device in the system. In this case the tag is a void * which is
+ * really an OMAP device pointer
+ */
+typedef void *cy_as_hal_device_tag ;
+
+
+/* This must be included after the CyAsHalDeviceTag type is defined */
+
+/* moved to staging location, eventual location
+ * considered is here
+ * #include <linux/westbridge/cyashalcb.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyashalcb.h"
+/*
+ * Below are the functions that communicate with the West Bridge
+ * device. These are system dependent and must be defined by
+ * the HAL layer for a given system.
+ */
+
+/*
+ * This function must be defined to write a register within the Antioch
+ * device. The addr value is the address of the register to write with
+ * respect to the base address of the Antioch device.
+ */
+void
+cy_as_hal_write_register(cy_as_hal_device_tag tag,
+ uint16_t addr, uint16_t data) ;
+
+/*
+ * This function must be defined to read a register from
+ * the west bridge device. The addr value is the address of
+ * the register to read with respect to the base address
+ * of the west bridge device.
+ */
+uint16_t
+cy_as_hal_read_register(cy_as_hal_device_tag tag, uint16_t addr) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * to the west bridge device. This function can use the burst write
+ * (DMA) capabilities of Antioch to do this, or it can just copy
+ * the data using writes.
+ */
+void
+cy_as_hal_dma_setup_write(cy_as_hal_device_tag tag,
+ uint8_t ep, void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to transfer a block of data
+ * from the Antioch device. This function can use the burst
+ * read (DMA) capabilities of Antioch to do this, or it can
+ * just copy the data using reads.
+ */
+void
+cy_as_hal_dma_setup_read(cy_as_hal_device_tag tag, uint8_t ep,
+ void *buf, uint32_t size, uint16_t maxsize) ;
+
+/*
+ * This function must be defined to cancel any pending DMA request.
+ */
+void
+cy_as_hal_dma_cancel_request(cy_as_hal_device_tag tag, uint8_t ep) ;
+
+/*
+ * This function must be defined to allow the Antioch API to
+ * register a callback function that is called when a DMA transfer
+ * is complete.
+ */
+void
+cy_as_hal_dma_register_callback(cy_as_hal_device_tag tag,
+ cy_as_hal_dma_complete_callback cb) ;
+
+/*
+ * This function must be defined to return the maximum size of DMA
+ * request that can be handled on the given endpoint. The return
+ * value should be the maximum size in bytes that the DMA module can
+ * handle.
+ */
+uint32_t
+cy_as_hal_dma_max_request_size(cy_as_hal_device_tag tag,
+ cy_as_end_point_number_t ep) ;
+
+/*
+ * This function must be defined to set the state of the WAKEUP pin
+ * on the Antioch device. Generally this is done via a GPIO of some
+ * type.
+ */
+cy_bool
+cy_as_hal_set_wakeup_pin(cy_as_hal_device_tag tag, cy_bool state) ;
+
+/*
+ * This function is called when the Antioch PLL loses lock, because
+ * of a problem in the supply voltage or the input clock.
+ */
+void
+cy_as_hal_pll_lock_loss_handler(cy_as_hal_device_tag tag) ;
+
+
+/**********************************************************************
+ *
+ * Below are the functions that must be defined to provide the basic
+ * operating system services required by the API.
+ *
+***********************************************************************/
+
+/*
+ * This function is required by the API to allocate memory. This function
+ * is expected to work exactly like malloc().
+ */
+void *
+cy_as_hal_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsHalAlloc(). This function is expected to work exacly like free().
+ */
+void
+cy_as_hal_free(void *mem_p) ;
+
+/*
+ * This function is required by the API to allocate memory during a
+ * callback. This function must be able to provide storage at inturupt
+ * time.
+ */
+void *
+cy_as_hal_c_b_alloc(uint32_t cnt) ;
+
+/*
+ * This function is required by the API to free memory allocated with
+ * CyAsCBHalAlloc().
+ */
+void
+cy_as_hal_c_b_free(void *ptr) ;
+
+/*
+ * This function is required to set a block of memory to a specific
+ * value. This function is expected to work exactly like memset()
+ */
+void
+cy_as_hal_mem_set(void *ptr, uint8_t value, uint32_t cnt) ;
+
+/*
+ * This function is expected to create a sleep channel. The data
+ * structure that represents the sleep channel is given by the
+ * pointer in the argument.
+ */
+cy_bool
+cy_as_hal_create_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+/*
+ * This function is expected to destroy a sleep channel. The data
+ * structure that represents the sleep channel is given by
+ * the pointer in the argument.
+ */
+
+
+cy_bool
+cy_as_hal_destroy_sleep_channel(cy_as_hal_sleep_channel *channel) ;
+
+cy_bool
+cy_as_hal_sleep_on(cy_as_hal_sleep_channel *channel, uint32_t ms) ;
+
+cy_bool
+cy_as_hal_wake(cy_as_hal_sleep_channel *channel) ;
+
+uint32_t
+cy_as_hal_disable_interrupts(void) ;
+
+void
+cy_as_hal_enable_interrupts(uint32_t);
+
+void
+cy_as_hal_sleep150(void);
+
+void
+cy_as_hal_sleep(uint32_t ms);
+
+cy_bool
+cy_as_hal_is_polling(void);
+
+void cy_as_hal_init_dev_registers(cy_as_hal_device_tag tag,
+ cy_bool is_standby_wakeup);
+
+/*
+ * required only in spi mode
+ */
+cy_bool cy_as_hal_sync_device_clocks(cy_as_hal_device_tag tag);
+
+void cy_as_hal_read_regs_before_standby(cy_as_hal_device_tag tag);
+
+
+#ifndef NDEBUG
+#define cy_as_hal_assert(cond) if (!(cond))\
+ printk(KERN_WARNING"assertion failed at %s:%d\n", __FILE__, __LINE__);
+#else
+#define cy_as_hal_assert(cond)
+#endif
+
+#define cy_as_hal_print_message printk
+
+/* removable debug printks */
+#ifndef WESTBRIDGE_NDEBUG
+#define DBG_PRINT_ENABLED
+#endif
+
+/*#define MBOX_ACCESS_DBG_PRINT_ENABLED*/
+
+
+#ifdef DBG_PRINT_ENABLED
+/* Debug printing enabled */
+
+#define DBGPRN(...) printk(__VA_ARGS__)
+#define DBGPRN_FUNC_NAME printk("<1> %x:_func: %s\n", \
+ current->pid, __func__)
+
+#else
+/* NO DEBUG PRINTING */
+#define DBGPRN(...)
+#define DBGPRN_FUNC_NAME
+
+#endif
+
+/*
+CyAsMiscSetLogLevel(uint8_t level)
+{
+ debug_level = level ;
+}
+
+#ifdef CY_AS_LOG_SUPPORT
+
+void
+cy_as_log_debug_message(int level, const char *str)
+{
+ if (level <= debug_level)
+ cy_as_hal_print_message("log %d: %s\n", level, str) ;
+}
+*/
+
+
+/*
+ * print buffer helper
+ */
+void cyashal_prn_buf(void *buf, uint16_t offset, int len);
+
+/*
+ * These are the functions that are not part of the HAL layer,
+ * but are required to be called for this HAL.
+ */
+int cy_as_hal_omap_pnand_start(const char *pgm,
+ cy_as_hal_device_tag *tag, cy_bool debug) ;
+int cy_as_hal_omap_pnand_stop(const char *pgm, cy_as_hal_device_tag tag) ;
+int omap_start_intr(cy_as_hal_device_tag tag) ;
+void cy_as_hal_set_ep_dma_mode(uint8_t ep, bool sg_xfer_enabled);
+
+/* moved to staging location
+#include <linux/westbridge/cyas_cplus_end.h>
+*/
+#include "../../../../../../../include/linux/westbridge/cyas_cplus_start.h"
+#endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasmemmap.h 2010-09-01 16:04:25.000000000 -0700
@@ -31,6 +31,9 @@
#define _INCLUDED_CYASMEMMAP_H_
/* defines copied from OMAP kernel branch */
+
+#define OMAP2_PULL_UP (1 << 4)
+#define OMAP2_PULL_ENA (1 << 3)
#define OMAP34XX_MUX_MODE0 0
#define OMAP34XX_MUX_MODE4 4
#define OMAP3_INPUT_EN (1 << 8)
@@ -250,9 +253,9 @@
/*
*gpmc nand registers for CS4
*/
-#define GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
-#define GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_CMD (0x7c + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_ADDR (0x80 + (0x30*AST_GPMC_CS))
+#define AST_GPMC_NAND_DATA (0x84 + (0x30*AST_GPMC_CS))
#define GPMC_STAT_REG (0x54)
#define GPMC_ERR_TYPE (0x48)
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_kernel.h 1969-12-31 16:00:00.000000000 -0800
@@ -1,72 +0,0 @@
-/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
-## ===========================
-## Copyright (C) 2010 Cypress Semiconductor
-##
-## This program is free software; you can redistribute it and/or
-## modify it under the terms of the GNU General Public License
-## as published by the Free Software Foundation; either version 2
-## of the License, or (at your option) any later version.
-##
-## This program is distributed in the hope that it will be useful,
-## but WITHOUT ANY WARRANTY; without even the implied warranty of
-## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
-## GNU General Public License for more details.
-##
-## You should have received a copy of the GNU General Public License
-## along with this program; if not, write to the Free Software
-## Foundation, Inc., 51 Franklin Street, Fifth Floor,
-## Boston, MA 02110-1301, USA.
-## ===========================
-*/
-
-#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
-#define __CY_AS_OMAP_DEV_KERNEL_H__
-
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/completion.h>
-
-/* include does not seem to work
- * moving for patch submission
-#include <mach/gpmc.h>
-*/
-#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
-
-/*
- * Constants
- */
-#define CY_AS_OMAP_KERNEL_HAL_SIG (0x1441)
-
-
-/*
- * Data structures
- */
-typedef struct cy_as_omap_dev_kernel {
- /* This is the signature for this data structure */
- unsigned int m_sig ;
-
- /* Address base of Antioch Device */
- void *m_addr_base;
-
- /* This is a pointer to the next Antioch device in the system */
- struct cy_as_omap_dev_kernel *m_next_p;
-
- /* This is for thread sync */
- struct completion thread_complete;
-
- /* This is for thread to wait for interrupts */
- cy_as_hal_sleep_channel thread_sc;
-
- /* This is for thread to exit upon StopOmapKernel */
- int thread_flag; /* set 1 to exit */
-
- int dma_ch;
-
- /* This is for dma sync */
- struct completion dma_complete;
-} cy_as_omap_dev_kernel;
-
-#endif
-
-/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 1969-12-31 16:00:00.000000000 -0800
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyasomapdev_pnand.h 2010-08-20 14:36:23.000000000 -0700
@@ -0,0 +1,72 @@
+/* Cypress Antioch OMAP KERNEL file (cyanomapdev_kernel.h)
+## ===========================
+## Copyright (C) 2010 Cypress Semiconductor
+##
+## This program is free software; you can redistribute it and/or
+## modify it under the terms of the GNU General Public License
+## as published by the Free Software Foundation; either version 2
+## of the License, or (at your option) any later version.
+##
+## This program is distributed in the hope that it will be useful,
+## but WITHOUT ANY WARRANTY; without even the implied warranty of
+## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+## GNU General Public License for more details.
+##
+## You should have received a copy of the GNU General Public License
+## along with this program; if not, write to the Free Software
+## Foundation, Inc., 51 Franklin Street, Fifth Floor,
+## Boston, MA 02110-1301, USA.
+## ===========================
+*/
+
+#ifndef __CY_AS_OMAP_DEV_KERNEL_H__
+#define __CY_AS_OMAP_DEV_KERNEL_H__
+
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/completion.h>
+
+/* include does not seem to work
+ * moving for patch submission
+#include <mach/gpmc.h>
+*/
+#include <linux/../../arch/arm/plat-omap/include/plat/gpmc.h>
+
+/*
+ * Constants
+ */
+#define CY_AS_OMAP_PNAND_HAL_SIG (0x1441)
+
+
+/*
+ * Data structures
+ */
+typedef struct cy_as_omap_dev_kernel {
+ /* This is the signature for this data structure */
+ unsigned int m_sig ;
+
+ /* Address base of Antioch Device */
+ void *m_addr_base;
+
+ /* This is a pointer to the next Antioch device in the system */
+ struct cy_as_omap_dev_kernel *m_next_p;
+
+ /* This is for thread sync */
+ struct completion thread_complete;
+
+ /* This is for thread to wait for interrupts */
+ cy_as_hal_sleep_channel thread_sc;
+
+ /* This is for thread to exit upon StopOmapKernel */
+ int thread_flag; /* set 1 to exit */
+
+ int dma_ch;
+
+ /* This is for dma sync */
+ struct completion dma_complete;
+} cy_as_omap_dev_kernel;
+
+#endif
+
+/*[]*/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile
--- linux-next-vanilla/drivers/staging/westbridge/astoria/device/Makefile 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/device/Makefile 2010-08-19 14:30:50.000000000 -0700
@@ -12,12 +12,20 @@ obj-$(CONFIG_WESTBRIDGE_DEVICE_DRIVER)
ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL),y)
#moved for staging compatbility
#cyasdev-y := ../../../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o
-cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_kernel.o cyasdevice.o \
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_pnand.o cyasdevice.o \
../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
../api/src/cyasstorage.o ../api/src/cyasusb.o
else
+
+ifeq ($(CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL),y)
+cyasdev-y := ../arch/arm/mach-omap2/cyashalomap_cram.o cyasdevice.o \
+ ../api/src/cyasdma.o ../api/src/cyasintr.o ../api/src/cyaslep2pep.o \
+ ../api/src/cyaslowlevel.o ../api/src/cyasmisc.o ../api/src/cyasmtp.o \
+ ../api/src/cyasstorage.o ../api/src/cyasusb.o
+else
# should not get here, need to be built with some hal
cyasdev-y := cyasdevice.o
endif
+endif
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h
--- linux-next-vanilla/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/astoria/include/linux/westbridge/cyashal.h 2010-09-07 11:09:06.000000000 -0700
@@ -62,31 +62,48 @@
/***** OMAP5912 Kernel HAL ****/
#ifdef __CY_ASTORIA_OMAP_5912_KERNEL_HAL__
- #ifdef CY_HAL_DEFINED
- #error only one HAL layer can be defined
- #endif
+ #ifdef CY_HAL_DEFINED
+ #error only one HAL layer can be defined
+ #endif
- #define CY_HAL_DEFINED
+ #define CY_HAL_DEFINED
- #include "cyanhalomap_kernel.h"
+ #include "cyanhalomap_kernel.h"
#endif
/***** eof OMAP5912 Kernel HAL ****/
-/***** OMAP3430 Kernel HAL ****/
+/***** OMAP3430 PNAND HAL ****/
#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
- #ifdef CY_HAL_DEFINED
- #error only one HAL layer can be defined
- #endif
+ #ifdef CY_HAL_DEFINED
+ #error only one HAL layer can be defined
+ #endif
- #define CY_HAL_DEFINED
+ #define CY_HAL_DEFINED
/* moved to staging location, eventual implementation
* considered is here
* #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
*/
- #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h"
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_pnand.h"
+
+#endif
+/*****************************/
+
+/***** OMAP3430 CRAM HAL ****/
+#ifdef CONFIG_MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+
+ #ifdef CY_HAL_DEFINED
+ #error only one HAL layer can be defined
+ #endif
+
+ #define CY_HAL_DEFINED
+/* moved to staging location, eventual implementation
+ * considered is here
+ * #include mach/westbridge/westbridge-omap3-pnand-hal/cyashalomap_kernel.h>
+*/
+ #include "../../../arch/arm/plat-omap/include/mach/westbridge/westbridge-omap3-cram-hal/cyashalomap_cram.h"
#endif
/*****************************/
diff -uprN -X linux-next-vanilla/Documentation/dontdiff linux-next-vanilla/drivers/staging/westbridge/Kconfig linux-next-incl-sdk/drivers/staging/westbridge/Kconfig
--- linux-next-vanilla/drivers/staging/westbridge/Kconfig 2010-08-31 19:32:51.000000000 -0700
+++ linux-next-incl-sdk/drivers/staging/westbridge/Kconfig 2010-09-01 15:45:46.000000000 -0700
@@ -4,7 +4,7 @@
menuconfig WESTBRIDGE
tristate "West Bridge support"
- depends on HAS_IOMEM && BROKEN
+ depends on WESTBRIDGE_HAL_SELECTED
help
This selects West Bridge Peripheral controller support.
@@ -12,7 +12,7 @@ menuconfig WESTBRIDGE
menuconfig WESTBRIDGE_ASTORIA
bool "West Bridge Astoria support"
- depends on WESTBRIDGE != n
+ depends on WESTBRIDGE != n && WESTBRIDGE_HAL_SELECTED
help
This option enables support for West Bridge Astoria
@@ -20,11 +20,43 @@ if WESTBRIDGE_ASTORIA
source "drivers/staging/westbridge/astoria/Kconfig"
endif #WESTBRIDGE_ASTORIA
-menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+#menuconfig MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
+# bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
+# depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+# help
+# Include the OMAP3430 HAL for PNAND interface
+
+config WESTBRIDGE_HAL_SELECTED
+ boolean
+
+choice
+ prompt "West Bridge HAL"
+ help
+ West Bridge HAL/processor interface to be used
+
+#
+# HAL Layers
+#
+
+config MACH_OMAP3_WESTBRIDGE_AST_PNAND_HAL
bool "WESTBRIDGE OMAP3430 Astoria PNAND HAL"
- depends on ARCH_OMAP3 && WESTBRIDGE_ASTORIA
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_OMAP3_WESTBRIDGE_AST_CRAM_HAL
+ bool "WESTBRIDGE OMAP3430 Astoria CRAM HAL"
+ depends on ARCH_OMAP3
+ select WESTBRIDGE_HAL_SELECTED
+ help
+ Include the OMAP3430 HAL for PNAND interface
+
+config MACH_NO_WESTBRIDGE
+ bool "no West Bridge HAL selected"
help
- Include the OMAP3430 Linux Based HAL
+ Do not include any HAL layer(de-activates West Bridge option)
+endchoice
config WESTBRIDGE_DEBUG
bool "West Bridge debugging"
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] west bridge, kconfig and hal fixes
2010-09-07 19:22 ` [PATCH] west bridge, kconfig and hal fixes David Cross
@ 2010-09-07 23:57 ` Greg KH
2010-09-08 20:56 ` David Cross
0 siblings, 1 reply; 12+ messages in thread
From: Greg KH @ 2010-09-07 23:57 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Tue, Sep 07, 2010 at 12:22:10PM -0700, David Cross wrote:
First off, what's with the "Re:" of the Subject? What are you
responding to here?
> This patch contains the kconfig changes necessary to fix build errors
> that could come up in the linux-next version. It also includes an
> additional HAL layer for the west bridge CRAM interface.
Again, one patch per change, please break this up.
> The inclusion
> of this interface support did require the reorganization of some of the
> existing code, which is part of the reason for the size of the patch.
> Moving files and directories makes this patch seem larger than it really
> is.
If you use git you can send a patch that properly shows only what
changes even if the file is moved. Care to do that?
> The Kconfig changes are closely related to the inclusion of the CRAM
> HAL layer, and as such this patch is difficult to logically separate.
Why is it necessary?
> The linux-next tree does not seem to have a config for the zoom2, and
> trying to build it for that board seems to make the compilation break.
Why should the driver care about which arch it is built for? It should
build on _all_ arches, right?
> As such, the only thing that I tested was compilation using the two
> different HALs (one of which is added in this patch). Please let me know
> if there are problems or questions with this.
Why do you need a "HAL" at all here?
confused,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
* RE: [PATCH] west bridge, kconfig and hal fixes
2010-09-07 23:57 ` Greg KH
@ 2010-09-08 20:56 ` David Cross
2010-09-09 3:51 ` Greg KH
0 siblings, 1 reply; 12+ messages in thread
From: David Cross @ 2010-09-08 20:56 UTC (permalink / raw)
To: 'Greg KH'; +Cc: linux-kernel
From: Greg KH [mailto:greg@kroah.com]
On Tue, Sep 07, 2010 at 12:22:10PM -0700, David Cross wrote:
> First off, what's with the "Re:" of the Subject? What are you
> responding to here?
I will removed that going forward when submitting patches, it does seem
appropriate on this response, so am leaving it in.
> > This patch contains the kconfig changes necessary to fix build errors
> > that could come up in the linux-next version. It also includes an
> > additional HAL layer for the west bridge CRAM interface.
> Again, one patch per change, please break this up.
Can I break it up as 1) moving PNAND HAL layer + Kconfig changes and 2) CRAM
HAL layer inclusion or is this still too tightly coupled? The issue that I
am having is that I was working on these drivers while waiting on inclusion
in the staging area in order to incorporate all of the feedback I have
gotten so far. As such, I now either have to artificially break up the
patches or start with a clean kernel and start sequentially hand coding them
again. I think this latter might be what you want, but this is really
time-consuming to do at this point.
While I definitely understand the need for this type of review of patches
for changes to the kernel itself, it is hard for me to understand why the
same rigor needs to apply to this driver in the staging area. As you
mentioned, it needs a lot of work to meet the standards for acceptance. I
have done a fair amount of work to move it closer. Do I really need to undo
and repeat all of that in order to clean up "BROKEN" code in the staging
area of the linux-next tree?
> > The inclusion
> > of this interface support did require the reorganization of some of the
> > existing code, which is part of the reason for the size of the patch.
> > Moving files and directories makes this patch seem larger than it really
> > is.
> If you use git you can send a patch that properly shows only what
> changes even if the file is moved. Care to do that?
I have not used git so far, other than to pull from repositories. I probably
could, but it will take me some time to learn. Is this "nice to have" or do
I need to do it to get this patch applied?
> > The Kconfig changes are closely related to the inclusion of the CRAM
> > HAL layer, and as such this patch is difficult to logically separate.
> Why is it necessary?
The HAL layer in general or this specific HAL layer? One of the reasons for
needing memory interface options is the allocation of pins on a given
platform. For example, if CLE was used as a GPIO for some other purpose, it
is nice to have the option to move to an SRAM like interface. Also, I needed
to develop this for unrelated reasons. As such, making minor structural
changes such that new HALs could be easily incorporated in an intuitive
manner seemed to be a plus. I also hoped that it would make the structure
and rationale behind the initial driver structure more intuitive to the
kernel maintainers.
> > The linux-next tree does not seem to have a config for the zoom2, and
> > trying to build it for that board seems to make the compilation break.
> Why should the driver care about which arch it is built for? It should
> build on _all_ arches, right?
It should build on all architectures that have an appropriate HAL
implemented. Otherwise, if you don't have a method for low level
communication with west bridge, why would you want to build the driver?
> > As such, the only thing that I tested was compilation using the two
> > different HALs (one of which is added in this patch). Please let me know
> > if there are problems or questions with this.
> Why do you need a "HAL" at all here?
For low level communication. The interface to the processor is a memory
interface across which mailbox messages are passed. These mailbox messages
makeup a sort of data link layer on top of which higher levels of
abstraction can be built.
Additionally, the HAL defines how you pump data to the device's endpoints,
since this is also done across the memory interface from the processor's
perspective. Other power management items are also included, and the
complete list of functions can be seen in the header files.
Thanks,
david
---------------------------------------------------------------
This message and any attachments may contain Cypress (or its
subsidiaries) confidential information. If it has been received
in error, please advise the sender and immediately delete this
message.
---------------------------------------------------------------
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH] west bridge, kconfig and hal fixes
2010-09-08 20:56 ` David Cross
@ 2010-09-09 3:51 ` Greg KH
0 siblings, 0 replies; 12+ messages in thread
From: Greg KH @ 2010-09-09 3:51 UTC (permalink / raw)
To: David Cross; +Cc: linux-kernel
On Wed, Sep 08, 2010 at 01:56:09PM -0700, David Cross wrote:
>
> From: Greg KH [mailto:greg@kroah.com]
>
> On Tue, Sep 07, 2010 at 12:22:10PM -0700, David Cross wrote:
>
> > First off, what's with the "Re:" of the Subject? What are you
> > responding to here?
> I will removed that going forward when submitting patches, it does seem
> appropriate on this response, so am leaving it in.
Yes, of course it is correct here :)
> > > This patch contains the kconfig changes necessary to fix build errors
> > > that could come up in the linux-next version. It also includes an
> > > additional HAL layer for the west bridge CRAM interface.
>
> > Again, one patch per change, please break this up.
>
> Can I break it up as 1) moving PNAND HAL layer + Kconfig changes and
Wait, look, you had a "and" in there. Why?
Why not just do:
- move PNAND HAL layer
- Kconfig changes
as individual patches.
Each one does only one thing. Again, please try to remember that.
> 2) CRAM HAL layer inclusion or is this still too tightly coupled? The
> issue that I am having is that I was working on these drivers while
> waiting on inclusion in the staging area in order to incorporate all
> of the feedback I have gotten so far. As such, I now either have to
> artificially break up the patches or start with a clean kernel and
> start sequentially hand coding them again. I think this latter might
> be what you want, but this is really time-consuming to do at this
> point.
Well, it shouldn't take all that long of time, as you know what you want
to accomplish. It's also how kernel development works, you usually go
off and do a bunch of work, then step back and figure out how to break
it up into logical chunks to submit it. Everyone does it. After time
you will get used to doing things in logical steps so you don't have to
redo things, but that takes a bit of time.
> While I definitely understand the need for this type of review of patches
> for changes to the kernel itself, it is hard for me to understand why the
> same rigor needs to apply to this driver in the staging area.
There is no difference here for what type of patches I can accept in the
staging area. Staging is for code that doesn't meet the "normal"
acceptance guidelines, but that does not mean that the proper
development process does not also apply here at all. It's a matter of
both cleaning up the code, and cleaning up the developers who are
working on the code at the same time :)
> As you gmentioned, it needs a lot of work to meet the standards for
> acceptance. I have done a fair amount of work to move it closer. Do I
> really need to undo and repeat all of that in order to clean up
> "BROKEN" code in the staging area of the linux-next tree?
It's marked "BROKEN" because it breaks the build. Once that is fixed we
can remove that marking, that's the only thing keeping it that way.
> > > The inclusion
> > > of this interface support did require the reorganization of some of the
> > > existing code, which is part of the reason for the size of the patch.
> > > Moving files and directories makes this patch seem larger than it really
> > > is.
>
> > If you use git you can send a patch that properly shows only what
> > changes even if the file is moved. Care to do that?
>
> I have not used git so far, other than to pull from repositories. I probably
> could, but it will take me some time to learn. Is this "nice to have" or do
> I need to do it to get this patch applied?
What are you using to develop your patches? If you don't use git, you
can use quilt, or some other type of patch management system. But it
might be worth spending a few hours to get up and running on one of them
as this is going to be something you use a lot in the future, and it
will help you get work done much faster and be able to work with the
rest of the community better as well.
Trust me, taking the time to learn git is worth it in the end.
> > > The Kconfig changes are closely related to the inclusion of the CRAM
> > > HAL layer, and as such this patch is difficult to logically separate.
>
> > Why is it necessary?
>
> The HAL layer in general or this specific HAL layer?
Any HAL layer.
> One of the reasons for needing memory interface options is the
> allocation of pins on a given platform. For example, if CLE was used
> as a GPIO for some other purpose, it is nice to have the option to
> move to an SRAM like interface. Also, I needed to develop this for
> unrelated reasons. As such, making minor structural changes such that
> new HALs could be easily incorporated in an intuitive manner seemed to
> be a plus. I also hoped that it would make the structure and rationale
> behind the initial driver structure more intuitive to the kernel
> maintainers.
Note that almost no other driver in the kernel needs such a layer. What
you do is have a "backend" that talks to either GPIO or SRAM or
something else. Then, at runtime, you can pick the one you need/want
depending on the hardware the code is running on, and everything "just
works".
Remember, the goal is to be able to build your driver once for any
platform it might run on. Don't rely on Kconfig options to select the
hardware types, that's not the correct way to do things in the end at
all.
> > > The linux-next tree does not seem to have a config for the zoom2, and
> > > trying to build it for that board seems to make the compilation break.
>
> > Why should the driver care about which arch it is built for? It should
> > build on _all_ arches, right?
>
> It should build on all architectures that have an appropriate HAL
> implemented. Otherwise, if you don't have a method for low level
> communication with west bridge, why would you want to build the driver?
Testing that the build isn't broken? Building one image to run on lots
of different platforms? Distros really need this and there is a large
unification effort for ARM to get this to work for that platform right
now. PPC has done this a lot already as well.
Hope this helps,
greg k-h
^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2010-09-09 5:22 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <5BFACB1451C1459BA8562A5561D0A4E6@stanford.edu>
2010-09-02 0:08 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig fixes David Cross
2010-09-02 1:37 ` Greg KH
2010-09-02 22:43 ` [PATCH] West Bridge Astoria Driver 2.6.35, Kconfig and HAL fixes David Cross
2010-09-05 5:17 ` Greg KH
2010-09-07 19:22 ` [PATCH] west bridge, kconfig and hal fixes David Cross
2010-09-07 23:57 ` Greg KH
2010-09-08 20:56 ` David Cross
2010-09-09 3:51 ` Greg KH
2010-09-02 22:47 ` [PATCH] West Bridge Astoria Driver 2.6.35, cyasgadget fixes and mpage_cleardirty David Cross
2010-09-05 5:20 ` Greg KH
2010-09-02 22:49 ` [PATCH] West Bridge Astoria Driver 2.6.35, minor block and device driver updates David Cross
2010-09-05 5:24 ` Greg KH
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.