* [PATCH 0/5] x86: remove idle notifier @ 2016-11-18 6:23 Len Brown 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown 2016-11-18 8:17 ` [PATCH 0/5] x86: remove idle notifier Ingo Molnar 0 siblings, 2 replies; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel The return from idle path is latency sensitive, so the less code and fewer data accesses, the better. Remove the un-maintained i7300_idle driver, the only user of the x86 idle-notifier, and then remove the notifier itself. ^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH 1/5] i7300_idle: remove this driver 2016-11-18 6:23 [PATCH 0/5] x86: remove idle notifier Len Brown @ 2016-11-18 6:23 ` Len Brown 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown ` (4 more replies) 2016-11-18 8:17 ` [PATCH 0/5] x86: remove idle notifier Ingo Molnar 1 sibling, 5 replies; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel, Len Brown From: Len Brown <len.brown@intel.com> In preparation for removing the idle_notifier, remove its only user, the i7300_idle driver. i7300_idle was deployed in 2008 to reduce idle memory power on systems using the i7300 chipset. The driver worked by throttling the fully-buffered DIMMs during idle periods using the IOAT DMA engine. The driver ran only on the i7300 chip-set, and no other hardware has used this mechanism. The driver no longer has a maintainer. Removing this driver will increase idle power on i7300 systems when they run the new kernel without the driver. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> --- MAINTAINERS | 6 - drivers/dma/ioat/registers.h | 2 - drivers/idle/Kconfig | 17 -- drivers/idle/Makefile | 1 - drivers/idle/i7300_idle.c | 612 ------------------------------------------- 5 files changed, 638 deletions(-) delete mode 100644 drivers/idle/i7300_idle.c diff --git a/MAINTAINERS b/MAINTAINERS index 2a58eeac9452..3cdccf5b64f3 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -6087,12 +6087,6 @@ S: Maintained F: Documentation/cdrom/ide-cd F: drivers/ide/ide-cd* -IDLE-I7300 -M: Andy Henroid <andrew.d.henroid@intel.com> -L: linux-pm@vger.kernel.org -S: Supported -F: drivers/idle/i7300_idle.c - IEEE 802.15.4 SUBSYSTEM M: Alexander Aring <aar@pengutronix.de> L: linux-wpan@vger.kernel.org diff --git a/drivers/dma/ioat/registers.h b/drivers/dma/ioat/registers.h index 48fa4cf9f64a..2f3bbc88ff2a 100644 --- a/drivers/dma/ioat/registers.h +++ b/drivers/dma/ioat/registers.h @@ -106,8 +106,6 @@ #define IOAT_DMA_COMP_V1 0x0001 /* Compatibility with DMA version 1 */ #define IOAT_DMA_COMP_V2 0x0002 /* Compatibility with DMA version 2 */ -/* IOAT1 define left for i7300_idle driver to not fail compiling */ -#define IOAT1_CHANSTS_OFFSET 0x04 #define IOAT_CHANSTS_OFFSET 0x08 /* 64-bit Channel Status Register */ #define IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR (~0x3fULL) #define IOAT_CHANSTS_SOFT_ERR 0x10ULL diff --git a/drivers/idle/Kconfig b/drivers/idle/Kconfig index 4732dfc15447..55bcf803841e 100644 --- a/drivers/idle/Kconfig +++ b/drivers/idle/Kconfig @@ -8,20 +8,3 @@ config INTEL_IDLE native Intel hardware idle features. The acpi_idle driver can be configured at the same time, in order to handle processors intel_idle does not support. - -menu "Memory power savings" -depends on X86_64 - -config I7300_IDLE_IOAT_CHANNEL - bool - -config I7300_IDLE - tristate "Intel chipset idle memory power saving driver" - select I7300_IDLE_IOAT_CHANNEL - help - Enable memory power savings when idle with certain Intel server - chipsets. The chipset must have I/O AT support, such as the - Intel 7300. The power savings depends on the type and quantity of - DRAM devices. - -endmenu diff --git a/drivers/idle/Makefile b/drivers/idle/Makefile index 23d295cf10f2..0007111d73e9 100644 --- a/drivers/idle/Makefile +++ b/drivers/idle/Makefile @@ -1,3 +1,2 @@ -obj-$(CONFIG_I7300_IDLE) += i7300_idle.o obj-$(CONFIG_INTEL_IDLE) += intel_idle.o diff --git a/drivers/idle/i7300_idle.c b/drivers/idle/i7300_idle.c deleted file mode 100644 index ffeebc7e9f1c..000000000000 --- a/drivers/idle/i7300_idle.c +++ /dev/null @@ -1,612 +0,0 @@ -/* - * (C) Copyright 2008 Intel Corporation - * Authors: - * Andy Henroid <andrew.d.henroid@intel.com> - * Venkatesh Pallipadi <venkatesh.pallipadi@intel.com> - */ - -/* - * Save DIMM power on Intel 7300-based platforms when all CPUs/cores - * are idle, using the DIMM thermal throttling capability. - * - * This driver depends on the Intel integrated DMA controller (I/O AT). - * If the driver for I/O AT (drivers/dma/ioatdma*) is also enabled, - * this driver should work cooperatively. - */ - -/* #define DEBUG */ - -#include <linux/module.h> -#include <linux/pci.h> -#include <linux/gfp.h> -#include <linux/sched.h> -#include <linux/notifier.h> -#include <linux/cpumask.h> -#include <linux/ktime.h> -#include <linux/delay.h> -#include <linux/debugfs.h> -#include <linux/stop_machine.h> -#include <linux/i7300_idle.h> - -#include <asm/idle.h> - -#include "../dma/ioat/hw.h" -#include "../dma/ioat/registers.h" - -#define I7300_IDLE_DRIVER_VERSION "1.55" -#define I7300_PRINT "i7300_idle:" - -#define MAX_STOP_RETRIES 10 - -static int debug; -module_param_named(debug, debug, uint, 0644); -MODULE_PARM_DESC(debug, "Enable debug printks in this driver"); - -static int forceload; -module_param_named(forceload, forceload, uint, 0644); -MODULE_PARM_DESC(debug, "Enable driver testing on unvalidated i5000"); - -#define dprintk(fmt, arg...) \ - do { if (debug) printk(KERN_INFO I7300_PRINT fmt, ##arg); } while (0) - -/* - * Value to set THRTLOW to when initiating throttling - * 0 = No throttling - * 1 = Throttle when > 4 activations per eval window (Maximum throttling) - * 2 = Throttle when > 8 activations - * 168 = Throttle when > 672 activations (Minimum throttling) - */ -#define MAX_THROTTLE_LOW_LIMIT 168 -static uint throttle_low_limit = 1; -module_param_named(throttle_low_limit, throttle_low_limit, uint, 0644); -MODULE_PARM_DESC(throttle_low_limit, - "Value for THRTLOWLM activation field " - "(0 = disable throttle, 1 = Max throttle, 168 = Min throttle)"); - -/* - * simple invocation and duration statistics - */ -static unsigned long total_starts; -static unsigned long total_us; - -#ifdef DEBUG -static unsigned long past_skip; -#endif - -static struct pci_dev *fbd_dev; - -static raw_spinlock_t i7300_idle_lock; -static int i7300_idle_active; - -static u8 i7300_idle_thrtctl_saved; -static u8 i7300_idle_thrtlow_saved; -static u32 i7300_idle_mc_saved; - -static cpumask_var_t idle_cpumask; -static ktime_t start_ktime; -static unsigned long avg_idle_us; - -static struct dentry *debugfs_dir; - -/* Begin: I/O AT Helper routines */ - -#define IOAT_CHANBASE(ioat_ctl, chan) (ioat_ctl + 0x80 + 0x80 * chan) -/* Snoop control (disable snoops when coherency is not important) */ -#define IOAT_DESC_SADDR_SNP_CTL (1UL << 1) -#define IOAT_DESC_DADDR_SNP_CTL (1UL << 2) - -static struct pci_dev *ioat_dev; -static struct ioat_dma_descriptor *ioat_desc; /* I/O AT desc & data (1 page) */ -static unsigned long ioat_desc_phys; -static u8 *ioat_iomap; /* I/O AT memory-mapped control regs (aka CB_BAR) */ -static u8 *ioat_chanbase; - -/* Start I/O AT memory copy */ -static int i7300_idle_ioat_start(void) -{ - u32 err; - /* Clear error (due to circular descriptor pointer) */ - err = readl(ioat_chanbase + IOAT_CHANERR_OFFSET); - if (err) - writel(err, ioat_chanbase + IOAT_CHANERR_OFFSET); - - writeb(IOAT_CHANCMD_START, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - return 0; -} - -/* Stop I/O AT memory copy */ -static void i7300_idle_ioat_stop(void) -{ - int i; - u64 sts; - - for (i = 0; i < MAX_STOP_RETRIES; i++) { - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - udelay(10); - - sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (sts != IOAT_CHANSTS_ACTIVE) - break; - - } - - if (i == MAX_STOP_RETRIES) { - dprintk("failed to stop I/O AT after %d retries\n", - MAX_STOP_RETRIES); - } -} - -/* Test I/O AT by copying 1024 byte from 2k to 1k */ -static int __init i7300_idle_ioat_selftest(u8 *ctl, - struct ioat_dma_descriptor *desc, unsigned long desc_phys) -{ - u64 chan_sts; - - memset(desc, 0, 2048); - memset((u8 *) desc + 2048, 0xab, 1024); - - desc[0].size = 1024; - desc[0].ctl = 0; - desc[0].src_addr = desc_phys + 2048; - desc[0].dst_addr = desc_phys + 1024; - desc[0].next = 0; - - writeb(IOAT_CHANCMD_RESET, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - writeb(IOAT_CHANCMD_START, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - udelay(1000); - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (chan_sts != IOAT_CHANSTS_DONE) { - /* Not complete, reset the channel */ - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - return -1; - } - - if (*(u32 *) ((u8 *) desc + 3068) != 0xabababab || - *(u32 *) ((u8 *) desc + 2044) != 0xabababab) { - dprintk("Data values src 0x%x, dest 0x%x, memset 0x%x\n", - *(u32 *) ((u8 *) desc + 2048), - *(u32 *) ((u8 *) desc + 1024), - *(u32 *) ((u8 *) desc + 3072)); - return -1; - } - return 0; -} - -static struct device dummy_dma_dev = { - .init_name = "fallback device", - .coherent_dma_mask = DMA_BIT_MASK(64), - .dma_mask = &dummy_dma_dev.coherent_dma_mask, -}; - -/* Setup and initialize I/O AT */ -/* This driver needs I/O AT as the throttling takes effect only when there is - * some memory activity. We use I/O AT to set up a dummy copy, while all CPUs - * go idle and memory is throttled. - */ -static int __init i7300_idle_ioat_init(void) -{ - u8 ver, chan_count, ioat_chan; - u16 chan_ctl; - - ioat_iomap = (u8 *) ioremap_nocache(pci_resource_start(ioat_dev, 0), - pci_resource_len(ioat_dev, 0)); - - if (!ioat_iomap) { - printk(KERN_ERR I7300_PRINT "failed to map I/O AT registers\n"); - goto err_ret; - } - - ver = readb(ioat_iomap + IOAT_VER_OFFSET); - if (ver != IOAT_VER_1_2) { - printk(KERN_ERR I7300_PRINT "unknown I/O AT version (%u.%u)\n", - ver >> 4, ver & 0xf); - goto err_unmap; - } - - chan_count = readb(ioat_iomap + IOAT_CHANCNT_OFFSET); - if (!chan_count) { - printk(KERN_ERR I7300_PRINT "unexpected # of I/O AT channels " - "(%u)\n", - chan_count); - goto err_unmap; - } - - ioat_chan = chan_count - 1; - ioat_chanbase = IOAT_CHANBASE(ioat_iomap, ioat_chan); - - chan_ctl = readw(ioat_chanbase + IOAT_CHANCTRL_OFFSET); - if (chan_ctl & IOAT_CHANCTRL_CHANNEL_IN_USE) { - printk(KERN_ERR I7300_PRINT "channel %d in use\n", ioat_chan); - goto err_unmap; - } - - writew(IOAT_CHANCTRL_CHANNEL_IN_USE, - ioat_chanbase + IOAT_CHANCTRL_OFFSET); - - ioat_desc = (struct ioat_dma_descriptor *)dma_alloc_coherent( - &dummy_dma_dev, 4096, - (dma_addr_t *)&ioat_desc_phys, GFP_KERNEL); - if (!ioat_desc) { - printk(KERN_ERR I7300_PRINT "failed to allocate I/O AT desc\n"); - goto err_mark_unused; - } - - writel(ioat_desc_phys & 0xffffffffUL, - ioat_chanbase + IOAT1_CHAINADDR_OFFSET_LOW); - writel(ioat_desc_phys >> 32, - ioat_chanbase + IOAT1_CHAINADDR_OFFSET_HIGH); - - if (i7300_idle_ioat_selftest(ioat_iomap, ioat_desc, ioat_desc_phys)) { - printk(KERN_ERR I7300_PRINT "I/O AT self-test failed\n"); - goto err_free; - } - - /* Setup circular I/O AT descriptor chain */ - ioat_desc[0].ctl = IOAT_DESC_SADDR_SNP_CTL | IOAT_DESC_DADDR_SNP_CTL; - ioat_desc[0].src_addr = ioat_desc_phys + 2048; - ioat_desc[0].dst_addr = ioat_desc_phys + 3072; - ioat_desc[0].size = 128; - ioat_desc[0].next = ioat_desc_phys + sizeof(struct ioat_dma_descriptor); - - ioat_desc[1].ctl = ioat_desc[0].ctl; - ioat_desc[1].src_addr = ioat_desc[0].src_addr; - ioat_desc[1].dst_addr = ioat_desc[0].dst_addr; - ioat_desc[1].size = ioat_desc[0].size; - ioat_desc[1].next = ioat_desc_phys; - - return 0; - -err_free: - dma_free_coherent(&dummy_dma_dev, 4096, (void *)ioat_desc, 0); -err_mark_unused: - writew(0, ioat_chanbase + IOAT_CHANCTRL_OFFSET); -err_unmap: - iounmap(ioat_iomap); -err_ret: - return -ENODEV; -} - -/* Cleanup I/O AT */ -static void __exit i7300_idle_ioat_exit(void) -{ - int i; - u64 chan_sts; - - i7300_idle_ioat_stop(); - - /* Wait for a while for the channel to halt before releasing */ - for (i = 0; i < MAX_STOP_RETRIES; i++) { - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (chan_sts != IOAT_CHANSTS_ACTIVE) { - writew(0, ioat_chanbase + IOAT_CHANCTRL_OFFSET); - break; - } - udelay(1000); - } - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - /* - * We tried to reset multiple times. If IO A/T channel is still active - * flag an error and return without cleanup. Memory leak is better - * than random corruption in that extreme error situation. - */ - if (chan_sts == IOAT_CHANSTS_ACTIVE) { - printk(KERN_ERR I7300_PRINT "Unable to stop IO A/T channels." - " Not freeing resources\n"); - return; - } - - dma_free_coherent(&dummy_dma_dev, 4096, (void *)ioat_desc, 0); - iounmap(ioat_iomap); -} - -/* End: I/O AT Helper routines */ - -#define DIMM_THRTLOW 0x64 -#define DIMM_THRTCTL 0x67 -#define DIMM_THRTCTL_THRMHUNT (1UL << 0) -#define DIMM_MC 0x40 -#define DIMM_GTW_MODE (1UL << 17) -#define DIMM_GBLACT 0x60 - -/* - * Keep track of an exponential-decaying average of recent idle durations. - * The latest duration gets DURATION_WEIGHT_PCT percentage weight - * in this average, with the old average getting the remaining weight. - * - * High weights emphasize recent history, low weights include long history. - */ -#define DURATION_WEIGHT_PCT 55 - -/* - * When the decaying average of recent durations or the predicted duration - * of the next timer interrupt is shorter than duration_threshold, the - * driver will decline to throttle. - */ -#define DURATION_THRESHOLD_US 100 - - -/* Store DIMM thermal throttle configuration */ -static int i7300_idle_thrt_save(void) -{ - u32 new_mc_val; - u8 gblactlm; - - pci_read_config_byte(fbd_dev, DIMM_THRTCTL, &i7300_idle_thrtctl_saved); - pci_read_config_byte(fbd_dev, DIMM_THRTLOW, &i7300_idle_thrtlow_saved); - pci_read_config_dword(fbd_dev, DIMM_MC, &i7300_idle_mc_saved); - /* - * Make sure we have Global Throttling Window Mode set to have a - * "short" window. This (mostly) works around an issue where - * throttling persists until the end of the global throttling window - * size. On the tested system, this was resulting in a maximum of - * 64 ms to exit throttling (average 32 ms). The actual numbers - * depends on system frequencies. Setting the short window reduces - * this by a factor of 4096. - * - * We will only do this only if the system is set for - * unlimited-activations while in open-loop throttling (i.e., when - * Global Activation Throttle Limit is zero). - */ - pci_read_config_byte(fbd_dev, DIMM_GBLACT, &gblactlm); - dprintk("thrtctl_saved = 0x%02x, thrtlow_saved = 0x%02x\n", - i7300_idle_thrtctl_saved, - i7300_idle_thrtlow_saved); - dprintk("mc_saved = 0x%08x, gblactlm = 0x%02x\n", - i7300_idle_mc_saved, - gblactlm); - if (gblactlm == 0) { - new_mc_val = i7300_idle_mc_saved | DIMM_GTW_MODE; - pci_write_config_dword(fbd_dev, DIMM_MC, new_mc_val); - return 0; - } else { - dprintk("could not set GTW_MODE = 1 (OLTT enabled)\n"); - return -ENODEV; - } -} - -/* Restore DIMM thermal throttle configuration */ -static void i7300_idle_thrt_restore(void) -{ - pci_write_config_dword(fbd_dev, DIMM_MC, i7300_idle_mc_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, i7300_idle_thrtlow_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, i7300_idle_thrtctl_saved); -} - -/* Enable DIMM thermal throttling */ -static void i7300_idle_start(void) -{ - u8 new_ctl; - u8 limit; - - new_ctl = i7300_idle_thrtctl_saved & ~DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); - - limit = throttle_low_limit; - if (unlikely(limit > MAX_THROTTLE_LOW_LIMIT)) - limit = MAX_THROTTLE_LOW_LIMIT; - - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, limit); - - new_ctl = i7300_idle_thrtctl_saved | DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); -} - -/* Disable DIMM thermal throttling */ -static void i7300_idle_stop(void) -{ - u8 new_ctl; - u8 got_ctl; - - new_ctl = i7300_idle_thrtctl_saved & ~DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); - - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, i7300_idle_thrtlow_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, i7300_idle_thrtctl_saved); - pci_read_config_byte(fbd_dev, DIMM_THRTCTL, &got_ctl); - WARN_ON_ONCE(got_ctl != i7300_idle_thrtctl_saved); -} - - -/* - * i7300_avg_duration_check() - * return 0 if the decaying average of recent idle durations is - * more than DURATION_THRESHOLD_US - */ -static int i7300_avg_duration_check(void) -{ - if (avg_idle_us >= DURATION_THRESHOLD_US) - return 0; - -#ifdef DEBUG - past_skip++; -#endif - return 1; -} - -/* Idle notifier to look at idle CPUs */ -static int i7300_idle_notifier(struct notifier_block *nb, unsigned long val, - void *data) -{ - unsigned long flags; - ktime_t now_ktime; - static ktime_t idle_begin_time; - static int time_init = 1; - - if (!throttle_low_limit) - return 0; - - if (unlikely(time_init)) { - time_init = 0; - idle_begin_time = ktime_get(); - } - - raw_spin_lock_irqsave(&i7300_idle_lock, flags); - if (val == IDLE_START) { - - cpumask_set_cpu(smp_processor_id(), idle_cpumask); - - if (cpumask_weight(idle_cpumask) != num_online_cpus()) - goto end; - - now_ktime = ktime_get(); - idle_begin_time = now_ktime; - - if (i7300_avg_duration_check()) - goto end; - - i7300_idle_active = 1; - total_starts++; - start_ktime = now_ktime; - - i7300_idle_start(); - i7300_idle_ioat_start(); - - } else if (val == IDLE_END) { - cpumask_clear_cpu(smp_processor_id(), idle_cpumask); - if (cpumask_weight(idle_cpumask) == (num_online_cpus() - 1)) { - /* First CPU coming out of idle */ - u64 idle_duration_us; - - now_ktime = ktime_get(); - - idle_duration_us = ktime_to_us(ktime_sub - (now_ktime, idle_begin_time)); - - avg_idle_us = - ((100 - DURATION_WEIGHT_PCT) * avg_idle_us + - DURATION_WEIGHT_PCT * idle_duration_us) / 100; - - if (i7300_idle_active) { - ktime_t idle_ktime; - - idle_ktime = ktime_sub(now_ktime, start_ktime); - total_us += ktime_to_us(idle_ktime); - - i7300_idle_ioat_stop(); - i7300_idle_stop(); - i7300_idle_active = 0; - } - } - } -end: - raw_spin_unlock_irqrestore(&i7300_idle_lock, flags); - return 0; -} - -static struct notifier_block i7300_idle_nb = { - .notifier_call = i7300_idle_notifier, -}; - -MODULE_DEVICE_TABLE(pci, pci_tbl); - -static ssize_t stats_read_ul(struct file *fp, char __user *ubuf, size_t count, - loff_t *off) -{ - unsigned long *p = fp->private_data; - char buf[32]; - int len; - - len = snprintf(buf, 32, "%lu\n", *p); - return simple_read_from_buffer(ubuf, count, off, buf, len); -} - -static const struct file_operations idle_fops = { - .open = simple_open, - .read = stats_read_ul, - .llseek = default_llseek, -}; - -struct debugfs_file_info { - void *ptr; - char name[32]; - struct dentry *file; -} debugfs_file_list[] = { - {&total_starts, "total_starts", NULL}, - {&total_us, "total_us", NULL}, -#ifdef DEBUG - {&past_skip, "past_skip", NULL}, -#endif - {NULL, "", NULL} - }; - -static int __init i7300_idle_init(void) -{ - raw_spin_lock_init(&i7300_idle_lock); - total_us = 0; - - if (i7300_idle_platform_probe(&fbd_dev, &ioat_dev, forceload)) - return -ENODEV; - - if (i7300_idle_thrt_save()) - return -ENODEV; - - if (i7300_idle_ioat_init()) - return -ENODEV; - - if (!zalloc_cpumask_var(&idle_cpumask, GFP_KERNEL)) - return -ENOMEM; - - debugfs_dir = debugfs_create_dir("i7300_idle", NULL); - if (debugfs_dir) { - int i = 0; - - while (debugfs_file_list[i].ptr != NULL) { - debugfs_file_list[i].file = debugfs_create_file( - debugfs_file_list[i].name, - S_IRUSR, - debugfs_dir, - debugfs_file_list[i].ptr, - &idle_fops); - i++; - } - } - - idle_notifier_register(&i7300_idle_nb); - - printk(KERN_INFO "i7300_idle: loaded v%s\n", I7300_IDLE_DRIVER_VERSION); - return 0; -} - -static void __exit i7300_idle_exit(void) -{ - idle_notifier_unregister(&i7300_idle_nb); - free_cpumask_var(idle_cpumask); - - if (debugfs_dir) { - int i = 0; - - while (debugfs_file_list[i].file != NULL) { - debugfs_remove(debugfs_file_list[i].file); - i++; - } - - debugfs_remove(debugfs_dir); - } - i7300_idle_thrt_restore(); - i7300_idle_ioat_exit(); -} - -module_init(i7300_idle_init); -module_exit(i7300_idle_exit); - -MODULE_AUTHOR("Andy Henroid <andrew.d.henroid@intel.com>"); -MODULE_DESCRIPTION("Intel Chipset DIMM Idle Power Saving Driver v" - I7300_IDLE_DRIVER_VERSION); -MODULE_LICENSE("GPL"); -- 2.11.0.rc1 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [PATCH 2/5] x86: remove idle_notifier 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown @ 2016-11-18 6:23 ` Len Brown 2016-11-18 11:19 ` [tip:x86/idle] x86/idle: Remove idle_notifier tip-bot for Len Brown 2016-12-06 16:54 ` [PATCH 2/5] x86: remove idle_notifier Pavel Machek 2016-11-18 6:23 ` [PATCH 3/5] x86: remove is_idle flag Len Brown ` (3 subsequent siblings) 4 siblings, 2 replies; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel, Len Brown From: Len Brown <len.brown@intel.com> Upon removal of the i7300_idle driver, the idle_notifer is unused. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> --- arch/x86/include/asm/idle.h | 7 ------- arch/x86/kernel/process.c | 15 --------------- 2 files changed, 22 deletions(-) diff --git a/arch/x86/include/asm/idle.h b/arch/x86/include/asm/idle.h index c5d1785373ed..02bab09707f2 100644 --- a/arch/x86/include/asm/idle.h +++ b/arch/x86/include/asm/idle.h @@ -1,13 +1,6 @@ #ifndef _ASM_X86_IDLE_H #define _ASM_X86_IDLE_H -#define IDLE_START 1 -#define IDLE_END 2 - -struct notifier_block; -void idle_notifier_register(struct notifier_block *n); -void idle_notifier_unregister(struct notifier_block *n); - #ifdef CONFIG_X86_64 void enter_idle(void); void exit_idle(void); diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 0888a879120f..f51950715145 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -67,19 +67,6 @@ EXPORT_PER_CPU_SYMBOL(cpu_tss); #ifdef CONFIG_X86_64 static DEFINE_PER_CPU(unsigned char, is_idle); -static ATOMIC_NOTIFIER_HEAD(idle_notifier); - -void idle_notifier_register(struct notifier_block *n) -{ - atomic_notifier_chain_register(&idle_notifier, n); -} -EXPORT_SYMBOL_GPL(idle_notifier_register); - -void idle_notifier_unregister(struct notifier_block *n) -{ - atomic_notifier_chain_unregister(&idle_notifier, n); -} -EXPORT_SYMBOL_GPL(idle_notifier_unregister); #endif /* @@ -255,14 +242,12 @@ static inline void play_dead(void) void enter_idle(void) { this_cpu_write(is_idle, 1); - atomic_notifier_call_chain(&idle_notifier, IDLE_START, NULL); } static void __exit_idle(void) { if (x86_test_and_clear_bit_percpu(0, is_idle) == 0) return; - atomic_notifier_call_chain(&idle_notifier, IDLE_END, NULL); } /* Called from interrupts to signify idle end */ -- 2.11.0.rc1 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [tip:x86/idle] x86/idle: Remove idle_notifier 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown @ 2016-11-18 11:19 ` tip-bot for Len Brown 2016-12-06 16:54 ` [PATCH 2/5] x86: remove idle_notifier Pavel Machek 1 sibling, 0 replies; 18+ messages in thread From: tip-bot for Len Brown @ 2016-11-18 11:19 UTC (permalink / raw) To: linux-tip-commits; +Cc: peterz, hpa, linux-kernel, mingo, tglx, len.brown Commit-ID: 8e7a7ee9ddd5d4275f1194ef2cc243ea86b0fb6f Gitweb: http://git.kernel.org/tip/8e7a7ee9ddd5d4275f1194ef2cc243ea86b0fb6f Author: Len Brown <len.brown@intel.com> AuthorDate: Fri, 18 Nov 2016 01:23:18 -0500 Committer: Thomas Gleixner <tglx@linutronix.de> CommitDate: Fri, 18 Nov 2016 12:07:56 +0100 x86/idle: Remove idle_notifier Upon removal of the i7300_idle driver, the idle_notifer is unused. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/f15385a82ec4bf51f4f06777193d83f03b28cfdd.1479449716.git.len.brown@intel.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- arch/x86/include/asm/idle.h | 7 ------- arch/x86/kernel/process.c | 15 --------------- 2 files changed, 22 deletions(-) diff --git a/arch/x86/include/asm/idle.h b/arch/x86/include/asm/idle.h index c5d1785..02bab09 100644 --- a/arch/x86/include/asm/idle.h +++ b/arch/x86/include/asm/idle.h @@ -1,13 +1,6 @@ #ifndef _ASM_X86_IDLE_H #define _ASM_X86_IDLE_H -#define IDLE_START 1 -#define IDLE_END 2 - -struct notifier_block; -void idle_notifier_register(struct notifier_block *n); -void idle_notifier_unregister(struct notifier_block *n); - #ifdef CONFIG_X86_64 void enter_idle(void); void exit_idle(void); diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 0888a87..f519507 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -67,19 +67,6 @@ EXPORT_PER_CPU_SYMBOL(cpu_tss); #ifdef CONFIG_X86_64 static DEFINE_PER_CPU(unsigned char, is_idle); -static ATOMIC_NOTIFIER_HEAD(idle_notifier); - -void idle_notifier_register(struct notifier_block *n) -{ - atomic_notifier_chain_register(&idle_notifier, n); -} -EXPORT_SYMBOL_GPL(idle_notifier_register); - -void idle_notifier_unregister(struct notifier_block *n) -{ - atomic_notifier_chain_unregister(&idle_notifier, n); -} -EXPORT_SYMBOL_GPL(idle_notifier_unregister); #endif /* @@ -255,14 +242,12 @@ static inline void play_dead(void) void enter_idle(void) { this_cpu_write(is_idle, 1); - atomic_notifier_call_chain(&idle_notifier, IDLE_START, NULL); } static void __exit_idle(void) { if (x86_test_and_clear_bit_percpu(0, is_idle) == 0) return; - atomic_notifier_call_chain(&idle_notifier, IDLE_END, NULL); } /* Called from interrupts to signify idle end */ ^ permalink raw reply related [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown 2016-11-18 11:19 ` [tip:x86/idle] x86/idle: Remove idle_notifier tip-bot for Len Brown @ 2016-12-06 16:54 ` Pavel Machek 2016-12-07 11:46 ` Thomas Gleixner 1 sibling, 1 reply; 18+ messages in thread From: Pavel Machek @ 2016-12-06 16:54 UTC (permalink / raw) To: Len Brown; +Cc: x86, linux-kernel, Len Brown [-- Attachment #1: Type: text/plain, Size: 654 bytes --] Hi! > From: Len Brown <len.brown@intel.com> > > Upon removal of the i7300_idle driver, the idle_notifer is unused. Actually... do we want to keep this this one? There's "cpu is loaded" led trigger, which is implemented on arm but not on x86. I tried to patch process.c, but this would be easier (and does not require new hooks). Does that make sense? Best regards, Pavel > Signed-off-by: Len Brown <len.brown@intel.com> > Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html [-- Attachment #2: Digital signature --] [-- Type: application/pgp-signature, Size: 181 bytes --] ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-12-06 16:54 ` [PATCH 2/5] x86: remove idle_notifier Pavel Machek @ 2016-12-07 11:46 ` Thomas Gleixner 2016-12-07 13:40 ` Pavel Machek 0 siblings, 1 reply; 18+ messages in thread From: Thomas Gleixner @ 2016-12-07 11:46 UTC (permalink / raw) To: Pavel Machek; +Cc: Len Brown, x86, linux-kernel, Len Brown On Tue, 6 Dec 2016, Pavel Machek wrote: > > From: Len Brown <len.brown@intel.com> > > > > Upon removal of the i7300_idle driver, the idle_notifer is unused. > > Actually... do we want to keep this this one? There's "cpu is loaded" > led trigger, which is implemented on arm but not on x86. > > I tried to patch process.c, but this would be easier (and does not > require new hooks). > > Does that make sense? No. Notifiers are crap. I assume this LED thing is just for entertainment purposes as I cannot figure out why it would be useful. ARM has it does not qualify as useful. Thanks, tglx ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-12-07 11:46 ` Thomas Gleixner @ 2016-12-07 13:40 ` Pavel Machek 2016-12-08 9:18 ` Thomas Gleixner 0 siblings, 1 reply; 18+ messages in thread From: Pavel Machek @ 2016-12-07 13:40 UTC (permalink / raw) To: Thomas Gleixner; +Cc: Len Brown, x86, linux-kernel, Len Brown [-- Attachment #1: Type: text/plain, Size: 1028 bytes --] On Wed 2016-12-07 12:46:12, Thomas Gleixner wrote: > On Tue, 6 Dec 2016, Pavel Machek wrote: > > > From: Len Brown <len.brown@intel.com> > > > > > > Upon removal of the i7300_idle driver, the idle_notifer is unused. > > > > Actually... do we want to keep this this one? There's "cpu is loaded" > > led trigger, which is implemented on arm but not on x86. > > > > I tried to patch process.c, but this would be easier (and does not > > require new hooks). > > > > Does that make sense? > > No. Notifiers are crap. > > I assume this LED thing is just for entertainment purposes as I cannot > figure out why it would be useful. ARM has it does not qualify as useful. Well, on low-speed systems, that led is actually quite important. You know if your keypress was registered, and CPU is just slow, or if you need to press it again. Best regards, Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html [-- Attachment #2: Digital signature --] [-- Type: application/pgp-signature, Size: 181 bytes --] ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-12-07 13:40 ` Pavel Machek @ 2016-12-08 9:18 ` Thomas Gleixner 2016-12-08 22:05 ` Pavel Machek 0 siblings, 1 reply; 18+ messages in thread From: Thomas Gleixner @ 2016-12-08 9:18 UTC (permalink / raw) To: Pavel Machek; +Cc: Len Brown, x86, linux-kernel, Len Brown On Wed, 7 Dec 2016, Pavel Machek wrote: > On Wed 2016-12-07 12:46:12, Thomas Gleixner wrote: > > On Tue, 6 Dec 2016, Pavel Machek wrote: > > > > From: Len Brown <len.brown@intel.com> > > > > > > > > Upon removal of the i7300_idle driver, the idle_notifer is unused. > > > > > > Actually... do we want to keep this this one? There's "cpu is loaded" > > > led trigger, which is implemented on arm but not on x86. > > > > > > I tried to patch process.c, but this would be easier (and does not > > > require new hooks). > > > > > > Does that make sense? > > > > No. Notifiers are crap. > > > > I assume this LED thing is just for entertainment purposes as I cannot > > figure out why it would be useful. ARM has it does not qualify as useful. > > Well, on low-speed systems, that led is actually quite important. You > know if your keypress was registered, and CPU is just slow, or if you > need to press it again. I'm dealing with low-speed systems for 20 years now and that LED was never important for me, quite the contrary, it's annoying to have the extra work in the idle wakeup path which causes extra pointless latency. If you can't figure out your keypress lag without that LED then feel free to patch your own kernel, but stop trying to impose that nonsense on everyone. Thanks, tglx ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-12-08 9:18 ` Thomas Gleixner @ 2016-12-08 22:05 ` Pavel Machek 2016-12-08 22:53 ` Thomas Gleixner 0 siblings, 1 reply; 18+ messages in thread From: Pavel Machek @ 2016-12-08 22:05 UTC (permalink / raw) To: Thomas Gleixner; +Cc: Len Brown, x86, linux-kernel, Len Brown [-- Attachment #1: Type: text/plain, Size: 2009 bytes --] On Thu 2016-12-08 10:18:13, Thomas Gleixner wrote: > On Wed, 7 Dec 2016, Pavel Machek wrote: > > On Wed 2016-12-07 12:46:12, Thomas Gleixner wrote: > > > On Tue, 6 Dec 2016, Pavel Machek wrote: > > > > > From: Len Brown <len.brown@intel.com> > > > > > > > > > > Upon removal of the i7300_idle driver, the idle_notifer is unused. > > > > > > > > Actually... do we want to keep this this one? There's "cpu is loaded" > > > > led trigger, which is implemented on arm but not on x86. > > > > > > > > I tried to patch process.c, but this would be easier (and does not > > > > require new hooks). > > > > > > > > Does that make sense? > > > > > > No. Notifiers are crap. > > > > > > I assume this LED thing is just for entertainment purposes as I cannot > > > figure out why it would be useful. ARM has it does not qualify as useful. > > > > Well, on low-speed systems, that led is actually quite important. You > > know if your keypress was registered, and CPU is just slow, or if you > > need to press it again. > > I'm dealing with low-speed systems for 20 years now and that LED was never > important for me, quite the contrary, it's annoying to have the extra work > in the idle wakeup path which causes extra pointless latency. If you can't > figure out your keypress lag without that LED then feel free to patch your > own kernel, but stop trying to impose that nonsense on everyone. The kernel already has the hooks, and they did not seem to bother anyone. Arm already has the functionality, and it is useful. You may not care, but so what. Leds are broken on x86, plain and simple. Perhaps that should not be your decision? Feel free to patch it out of your kernel. Or feel free to argue that it needs to be removed from arm. But having unneccessary differences between architectures is just ugly. Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html [-- Attachment #2: Digital signature --] [-- Type: application/pgp-signature, Size: 181 bytes --] ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/5] x86: remove idle_notifier 2016-12-08 22:05 ` Pavel Machek @ 2016-12-08 22:53 ` Thomas Gleixner 0 siblings, 0 replies; 18+ messages in thread From: Thomas Gleixner @ 2016-12-08 22:53 UTC (permalink / raw) To: Pavel Machek; +Cc: Len Brown, x86, LKML, Len Brown, Peter Zijlstra On Thu, 8 Dec 2016, Pavel Machek wrote: > On Thu 2016-12-08 10:18:13, Thomas Gleixner wrote: > > I'm dealing with low-speed systems for 20 years now and that LED was never > > important for me, quite the contrary, it's annoying to have the extra work > > in the idle wakeup path which causes extra pointless latency. If you can't > > figure out your keypress lag without that LED then feel free to patch your > > own kernel, but stop trying to impose that nonsense on everyone. > > The kernel already has the hooks, and they did not seem to bother > anyone. The have bothered a lot of people up to the point where we could remove them, simply because they are pointless ballast and overhead in the wake from idle path. > Arm already has the functionality, and it is useful. You may > not care, but so what. Leds are broken on x86, plain and simple. BlinkenLEDs for idle were never available on x86 to begin with, so they can't be broken. Aisde of that the vast majority of x86 systems simply do not have LEDs which are accessible from that context. So just to support a very questionable use case with a very limited usefulness you want to impose extra code into a code path which is already known to be too heavy weight and people working on it to reduce the overhead. > Perhaps that should not be your decision? Feel free to complain to Linus. My maintainer decision stands. > Feel free to patch it out of your kernel. Or feel free to argue that it > needs to be removed from arm. But having unneccessary differences between > architectures is just ugly. Following that argumentation would require to add this to the core idle code, so _ALL_ architetures have access to this, but that got turned down by the core maintainers a more than a year ago for the very same reason. Thanks, tglx ^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH 3/5] x86: remove is_idle flag 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown @ 2016-11-18 6:23 ` Len Brown 2016-11-18 11:20 ` [tip:x86/idle] x86/idle: Remove " tip-bot for Len Brown 2016-11-18 6:23 ` [PATCH 4/5] x86: remove x86_test_and_clear_bit_percpu() Len Brown ` (2 subsequent siblings) 4 siblings, 1 reply; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel, Len Brown From: Len Brown <len.brown@intel.com> Upon removal of the idle_notifier, all accesses to the "is_idle" flag serve no purpose. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> --- arch/x86/kernel/process.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index f51950715145..d8e9d794e114 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -65,10 +65,6 @@ __visible DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, cpu_tss) = { }; EXPORT_PER_CPU_SYMBOL(cpu_tss); -#ifdef CONFIG_X86_64 -static DEFINE_PER_CPU(unsigned char, is_idle); -#endif - /* * this gets called so that we can store lazy state into memory and copy the * current task into the new thread. @@ -241,13 +237,10 @@ static inline void play_dead(void) #ifdef CONFIG_X86_64 void enter_idle(void) { - this_cpu_write(is_idle, 1); } static void __exit_idle(void) { - if (x86_test_and_clear_bit_percpu(0, is_idle) == 0) - return; } /* Called from interrupts to signify idle end */ -- 2.11.0.rc1 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [tip:x86/idle] x86/idle: Remove is_idle flag 2016-11-18 6:23 ` [PATCH 3/5] x86: remove is_idle flag Len Brown @ 2016-11-18 11:20 ` tip-bot for Len Brown 0 siblings, 0 replies; 18+ messages in thread From: tip-bot for Len Brown @ 2016-11-18 11:20 UTC (permalink / raw) To: linux-tip-commits; +Cc: hpa, linux-kernel, mingo, len.brown, tglx, peterz Commit-ID: f08b5fe2d4eeb0a8a6e0e7e71928cf2c7b1b791d Gitweb: http://git.kernel.org/tip/f08b5fe2d4eeb0a8a6e0e7e71928cf2c7b1b791d Author: Len Brown <len.brown@intel.com> AuthorDate: Fri, 18 Nov 2016 01:23:19 -0500 Committer: Thomas Gleixner <tglx@linutronix.de> CommitDate: Fri, 18 Nov 2016 12:07:57 +0100 x86/idle: Remove is_idle flag Upon removal of the idle_notifier, all accesses to the "is_idle" flag serve no purpose. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/e4a24197cf9c227fcd1ca2df09999eaec9052f49.1479449716.git.len.brown@intel.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- arch/x86/kernel/process.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index f519507..d8e9d79 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -65,10 +65,6 @@ __visible DEFINE_PER_CPU_SHARED_ALIGNED(struct tss_struct, cpu_tss) = { }; EXPORT_PER_CPU_SYMBOL(cpu_tss); -#ifdef CONFIG_X86_64 -static DEFINE_PER_CPU(unsigned char, is_idle); -#endif - /* * this gets called so that we can store lazy state into memory and copy the * current task into the new thread. @@ -241,13 +237,10 @@ static inline void play_dead(void) #ifdef CONFIG_X86_64 void enter_idle(void) { - this_cpu_write(is_idle, 1); } static void __exit_idle(void) { - if (x86_test_and_clear_bit_percpu(0, is_idle) == 0) - return; } /* Called from interrupts to signify idle end */ ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [PATCH 4/5] x86: remove x86_test_and_clear_bit_percpu() 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown 2016-11-18 6:23 ` [PATCH 3/5] x86: remove is_idle flag Len Brown @ 2016-11-18 6:23 ` Len Brown 2016-11-18 11:20 ` [tip:x86/idle] x86: Remove x86_test_and_clear_bit_percpu() tip-bot for Len Brown 2016-11-18 6:23 ` [PATCH 5/5] x86: remove enter_idle(), exit_idle() Len Brown 2016-11-18 11:19 ` [tip:x86/idle] i7300_idle: Remove this driver tip-bot for Len Brown 4 siblings, 1 reply; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel, Len Brown From: Len Brown <len.brown@intel.com> Upon removal of the "is_idle" flag, x86_test_and_clear_bit_percpu() is no longer used. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> --- arch/x86/include/asm/percpu.h | 11 ----------- 1 file changed, 11 deletions(-) diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h index 84f58de08c2b..9fa03604b2b3 100644 --- a/arch/x86/include/asm/percpu.h +++ b/arch/x86/include/asm/percpu.h @@ -507,17 +507,6 @@ do { \ #endif -/* This is not atomic against other CPUs -- CPU preemption needs to be off */ -#define x86_test_and_clear_bit_percpu(bit, var) \ -({ \ - bool old__; \ - asm volatile("btr %2,"__percpu_arg(1)"\n\t" \ - CC_SET(c) \ - : CC_OUT(c) (old__), "+m" (var) \ - : "dIr" (bit)); \ - old__; \ -}) - static __always_inline bool x86_this_cpu_constant_test_bit(unsigned int nr, const unsigned long __percpu *addr) { -- 2.11.0.rc1 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [tip:x86/idle] x86: Remove x86_test_and_clear_bit_percpu() 2016-11-18 6:23 ` [PATCH 4/5] x86: remove x86_test_and_clear_bit_percpu() Len Brown @ 2016-11-18 11:20 ` tip-bot for Len Brown 0 siblings, 0 replies; 18+ messages in thread From: tip-bot for Len Brown @ 2016-11-18 11:20 UTC (permalink / raw) To: linux-tip-commits; +Cc: tglx, peterz, hpa, linux-kernel, mingo, len.brown Commit-ID: 9694be731dc81fec5ad59e863fa6538878ac496a Gitweb: http://git.kernel.org/tip/9694be731dc81fec5ad59e863fa6538878ac496a Author: Len Brown <len.brown@intel.com> AuthorDate: Fri, 18 Nov 2016 01:23:20 -0500 Committer: Thomas Gleixner <tglx@linutronix.de> CommitDate: Fri, 18 Nov 2016 12:07:57 +0100 x86: Remove x86_test_and_clear_bit_percpu() Upon removal of the "is_idle" flag, x86_test_and_clear_bit_percpu() is no longer used. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/b334ae6819507e3dfc0a4b33ed974714d067eb4a.1479449716.git.len.brown@intel.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- arch/x86/include/asm/percpu.h | 11 ----------- 1 file changed, 11 deletions(-) diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h index 84f58de..9fa0360 100644 --- a/arch/x86/include/asm/percpu.h +++ b/arch/x86/include/asm/percpu.h @@ -507,17 +507,6 @@ do { \ #endif -/* This is not atomic against other CPUs -- CPU preemption needs to be off */ -#define x86_test_and_clear_bit_percpu(bit, var) \ -({ \ - bool old__; \ - asm volatile("btr %2,"__percpu_arg(1)"\n\t" \ - CC_SET(c) \ - : CC_OUT(c) (old__), "+m" (var) \ - : "dIr" (bit)); \ - old__; \ -}) - static __always_inline bool x86_this_cpu_constant_test_bit(unsigned int nr, const unsigned long __percpu *addr) { ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [PATCH 5/5] x86: remove enter_idle(), exit_idle() 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown ` (2 preceding siblings ...) 2016-11-18 6:23 ` [PATCH 4/5] x86: remove x86_test_and_clear_bit_percpu() Len Brown @ 2016-11-18 6:23 ` Len Brown 2016-11-18 11:21 ` [tip:x86/idle] x86/idle: Remove " tip-bot for Len Brown 2016-11-18 11:19 ` [tip:x86/idle] i7300_idle: Remove this driver tip-bot for Len Brown 4 siblings, 1 reply; 18+ messages in thread From: Len Brown @ 2016-11-18 6:23 UTC (permalink / raw) To: x86; +Cc: linux-kernel, Len Brown From: Len Brown <len.brown@intel.com> Upon removal of the is_idle flag, these routines became NOPs. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> --- arch/x86/include/asm/apic.h | 1 - arch/x86/include/asm/idle.h | 9 --------- arch/x86/kernel/kvm.c | 2 -- arch/x86/kernel/process.c | 25 ------------------------- drivers/xen/events/events_base.c | 1 - 5 files changed, 38 deletions(-) diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index f5aaf6c83222..5731274bfdba 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -639,7 +639,6 @@ extern void irq_exit(void); static inline void entering_irq(void) { irq_enter(); - exit_idle(); } static inline void entering_ack_irq(void) diff --git a/arch/x86/include/asm/idle.h b/arch/x86/include/asm/idle.h index 02bab09707f2..dcebb1c634f1 100644 --- a/arch/x86/include/asm/idle.h +++ b/arch/x86/include/asm/idle.h @@ -1,15 +1,6 @@ #ifndef _ASM_X86_IDLE_H #define _ASM_X86_IDLE_H -#ifdef CONFIG_X86_64 -void enter_idle(void); -void exit_idle(void); -#else /* !CONFIG_X86_64 */ -static inline void enter_idle(void) { } -static inline void exit_idle(void) { } -static inline void __exit_idle(void) { } -#endif /* CONFIG_X86_64 */ - void amd_e400_remove_cpu(int cpu); #endif /* _ASM_X86_IDLE_H */ diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c index edbbfc854e39..093f550f372d 100644 --- a/arch/x86/kernel/kvm.c +++ b/arch/x86/kernel/kvm.c @@ -267,13 +267,11 @@ do_async_page_fault(struct pt_regs *regs, unsigned long error_code) case KVM_PV_REASON_PAGE_NOT_PRESENT: /* page is swapped out by the host. */ prev_state = exception_enter(); - exit_idle(); kvm_async_pf_task_wait((u32)read_cr2()); exception_exit(prev_state); break; case KVM_PV_REASON_PAGE_READY: rcu_irq_enter(); - exit_idle(); kvm_async_pf_task_wake((u32)read_cr2()); rcu_irq_exit(); break; diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index d8e9d794e114..ee023919e476 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -234,34 +234,9 @@ static inline void play_dead(void) } #endif -#ifdef CONFIG_X86_64 -void enter_idle(void) -{ -} - -static void __exit_idle(void) -{ -} - -/* Called from interrupts to signify idle end */ -void exit_idle(void) -{ - /* idle loop has pid 0 */ - if (current->pid) - return; - __exit_idle(); -} -#endif - void arch_cpu_idle_enter(void) { local_touch_nmi(); - enter_idle(); -} - -void arch_cpu_idle_exit(void) -{ - __exit_idle(); } void arch_cpu_idle_dead(void) diff --git a/drivers/xen/events/events_base.c b/drivers/xen/events/events_base.c index 9ecfcdcdd6d6..9ad622ab05dc 100644 --- a/drivers/xen/events/events_base.c +++ b/drivers/xen/events/events_base.c @@ -1256,7 +1256,6 @@ void xen_evtchn_do_upcall(struct pt_regs *regs) irq_enter(); #ifdef CONFIG_X86 - exit_idle(); inc_irq_stat(irq_hv_callback_count); #endif -- 2.11.0.rc1 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [tip:x86/idle] x86/idle: Remove enter_idle(), exit_idle() 2016-11-18 6:23 ` [PATCH 5/5] x86: remove enter_idle(), exit_idle() Len Brown @ 2016-11-18 11:21 ` tip-bot for Len Brown 0 siblings, 0 replies; 18+ messages in thread From: tip-bot for Len Brown @ 2016-11-18 11:21 UTC (permalink / raw) To: linux-tip-commits; +Cc: hpa, tglx, linux-kernel, len.brown, peterz, mingo Commit-ID: 7a3e686e1bb57c34f73d3f19d620fe29035a6c99 Gitweb: http://git.kernel.org/tip/7a3e686e1bb57c34f73d3f19d620fe29035a6c99 Author: Len Brown <len.brown@intel.com> AuthorDate: Fri, 18 Nov 2016 01:23:21 -0500 Committer: Thomas Gleixner <tglx@linutronix.de> CommitDate: Fri, 18 Nov 2016 12:07:57 +0100 x86/idle: Remove enter_idle(), exit_idle() Upon removal of the is_idle flag, these routines became NOPs. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/822f2c22cc5890f7b8ea0eeec60277eb44505b4e.1479449716.git.len.brown@intel.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- arch/x86/include/asm/apic.h | 1 - arch/x86/include/asm/idle.h | 9 --------- arch/x86/kernel/kvm.c | 2 -- arch/x86/kernel/process.c | 25 ------------------------- drivers/xen/events/events_base.c | 1 - 5 files changed, 38 deletions(-) diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index f5aaf6c..5731274 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -639,7 +639,6 @@ extern void irq_exit(void); static inline void entering_irq(void) { irq_enter(); - exit_idle(); } static inline void entering_ack_irq(void) diff --git a/arch/x86/include/asm/idle.h b/arch/x86/include/asm/idle.h index 02bab09..dcebb1c 100644 --- a/arch/x86/include/asm/idle.h +++ b/arch/x86/include/asm/idle.h @@ -1,15 +1,6 @@ #ifndef _ASM_X86_IDLE_H #define _ASM_X86_IDLE_H -#ifdef CONFIG_X86_64 -void enter_idle(void); -void exit_idle(void); -#else /* !CONFIG_X86_64 */ -static inline void enter_idle(void) { } -static inline void exit_idle(void) { } -static inline void __exit_idle(void) { } -#endif /* CONFIG_X86_64 */ - void amd_e400_remove_cpu(int cpu); #endif /* _ASM_X86_IDLE_H */ diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c index edbbfc8..093f550 100644 --- a/arch/x86/kernel/kvm.c +++ b/arch/x86/kernel/kvm.c @@ -267,13 +267,11 @@ do_async_page_fault(struct pt_regs *regs, unsigned long error_code) case KVM_PV_REASON_PAGE_NOT_PRESENT: /* page is swapped out by the host. */ prev_state = exception_enter(); - exit_idle(); kvm_async_pf_task_wait((u32)read_cr2()); exception_exit(prev_state); break; case KVM_PV_REASON_PAGE_READY: rcu_irq_enter(); - exit_idle(); kvm_async_pf_task_wake((u32)read_cr2()); rcu_irq_exit(); break; diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index d8e9d79..ee023919 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -234,34 +234,9 @@ static inline void play_dead(void) } #endif -#ifdef CONFIG_X86_64 -void enter_idle(void) -{ -} - -static void __exit_idle(void) -{ -} - -/* Called from interrupts to signify idle end */ -void exit_idle(void) -{ - /* idle loop has pid 0 */ - if (current->pid) - return; - __exit_idle(); -} -#endif - void arch_cpu_idle_enter(void) { local_touch_nmi(); - enter_idle(); -} - -void arch_cpu_idle_exit(void) -{ - __exit_idle(); } void arch_cpu_idle_dead(void) diff --git a/drivers/xen/events/events_base.c b/drivers/xen/events/events_base.c index 9ecfcdc..9ad622a 100644 --- a/drivers/xen/events/events_base.c +++ b/drivers/xen/events/events_base.c @@ -1256,7 +1256,6 @@ void xen_evtchn_do_upcall(struct pt_regs *regs) irq_enter(); #ifdef CONFIG_X86 - exit_idle(); inc_irq_stat(irq_hv_callback_count); #endif ^ permalink raw reply related [flat|nested] 18+ messages in thread
* [tip:x86/idle] i7300_idle: Remove this driver 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown ` (3 preceding siblings ...) 2016-11-18 6:23 ` [PATCH 5/5] x86: remove enter_idle(), exit_idle() Len Brown @ 2016-11-18 11:19 ` tip-bot for Len Brown 4 siblings, 0 replies; 18+ messages in thread From: tip-bot for Len Brown @ 2016-11-18 11:19 UTC (permalink / raw) To: linux-tip-commits; +Cc: mingo, len.brown, linux-kernel, tglx, peterz, hpa Commit-ID: b65ce83f2a607ad478c5492812f5f218e8d57a6b Gitweb: http://git.kernel.org/tip/b65ce83f2a607ad478c5492812f5f218e8d57a6b Author: Len Brown <len.brown@intel.com> AuthorDate: Fri, 18 Nov 2016 01:23:17 -0500 Committer: Thomas Gleixner <tglx@linutronix.de> CommitDate: Fri, 18 Nov 2016 12:07:56 +0100 i7300_idle: Remove this driver In preparation for removing the idle_notifier, remove its only user, the i7300_idle driver. i7300_idle was deployed in 2008 to reduce idle memory power on systems using the i7300 chipset. The driver worked by throttling the fully-buffered DIMMs during idle periods using the IOAT DMA engine. The driver ran only on the i7300 chip-set, and no other hardware has used this mechanism. The driver no longer has a maintainer. Removing this driver will increase idle power on i7300 systems when they run the new kernel without the driver. Signed-off-by: Len Brown <len.brown@intel.com> Acked-by: Ingo Molnar <mingo@kernel.org> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/ad6a044e57cc75f44cc8621abe846e58f7882243.1479449716.git.len.brown@intel.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- MAINTAINERS | 6 - drivers/dma/ioat/registers.h | 2 - drivers/idle/Kconfig | 17 -- drivers/idle/Makefile | 1 - drivers/idle/i7300_idle.c | 612 ------------------------------------------- 5 files changed, 638 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 851b89b..80db200 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -6087,12 +6087,6 @@ S: Maintained F: Documentation/cdrom/ide-cd F: drivers/ide/ide-cd* -IDLE-I7300 -M: Andy Henroid <andrew.d.henroid@intel.com> -L: linux-pm@vger.kernel.org -S: Supported -F: drivers/idle/i7300_idle.c - IEEE 802.15.4 SUBSYSTEM M: Alexander Aring <aar@pengutronix.de> L: linux-wpan@vger.kernel.org diff --git a/drivers/dma/ioat/registers.h b/drivers/dma/ioat/registers.h index 48fa4cf..2f3bbc8 100644 --- a/drivers/dma/ioat/registers.h +++ b/drivers/dma/ioat/registers.h @@ -106,8 +106,6 @@ #define IOAT_DMA_COMP_V1 0x0001 /* Compatibility with DMA version 1 */ #define IOAT_DMA_COMP_V2 0x0002 /* Compatibility with DMA version 2 */ -/* IOAT1 define left for i7300_idle driver to not fail compiling */ -#define IOAT1_CHANSTS_OFFSET 0x04 #define IOAT_CHANSTS_OFFSET 0x08 /* 64-bit Channel Status Register */ #define IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR (~0x3fULL) #define IOAT_CHANSTS_SOFT_ERR 0x10ULL diff --git a/drivers/idle/Kconfig b/drivers/idle/Kconfig index 4732dfc..55bcf80 100644 --- a/drivers/idle/Kconfig +++ b/drivers/idle/Kconfig @@ -8,20 +8,3 @@ config INTEL_IDLE native Intel hardware idle features. The acpi_idle driver can be configured at the same time, in order to handle processors intel_idle does not support. - -menu "Memory power savings" -depends on X86_64 - -config I7300_IDLE_IOAT_CHANNEL - bool - -config I7300_IDLE - tristate "Intel chipset idle memory power saving driver" - select I7300_IDLE_IOAT_CHANNEL - help - Enable memory power savings when idle with certain Intel server - chipsets. The chipset must have I/O AT support, such as the - Intel 7300. The power savings depends on the type and quantity of - DRAM devices. - -endmenu diff --git a/drivers/idle/Makefile b/drivers/idle/Makefile index 23d295c..0007111 100644 --- a/drivers/idle/Makefile +++ b/drivers/idle/Makefile @@ -1,3 +1,2 @@ -obj-$(CONFIG_I7300_IDLE) += i7300_idle.o obj-$(CONFIG_INTEL_IDLE) += intel_idle.o diff --git a/drivers/idle/i7300_idle.c b/drivers/idle/i7300_idle.c deleted file mode 100644 index ffeebc7..0000000 --- a/drivers/idle/i7300_idle.c +++ /dev/null @@ -1,612 +0,0 @@ -/* - * (C) Copyright 2008 Intel Corporation - * Authors: - * Andy Henroid <andrew.d.henroid@intel.com> - * Venkatesh Pallipadi <venkatesh.pallipadi@intel.com> - */ - -/* - * Save DIMM power on Intel 7300-based platforms when all CPUs/cores - * are idle, using the DIMM thermal throttling capability. - * - * This driver depends on the Intel integrated DMA controller (I/O AT). - * If the driver for I/O AT (drivers/dma/ioatdma*) is also enabled, - * this driver should work cooperatively. - */ - -/* #define DEBUG */ - -#include <linux/module.h> -#include <linux/pci.h> -#include <linux/gfp.h> -#include <linux/sched.h> -#include <linux/notifier.h> -#include <linux/cpumask.h> -#include <linux/ktime.h> -#include <linux/delay.h> -#include <linux/debugfs.h> -#include <linux/stop_machine.h> -#include <linux/i7300_idle.h> - -#include <asm/idle.h> - -#include "../dma/ioat/hw.h" -#include "../dma/ioat/registers.h" - -#define I7300_IDLE_DRIVER_VERSION "1.55" -#define I7300_PRINT "i7300_idle:" - -#define MAX_STOP_RETRIES 10 - -static int debug; -module_param_named(debug, debug, uint, 0644); -MODULE_PARM_DESC(debug, "Enable debug printks in this driver"); - -static int forceload; -module_param_named(forceload, forceload, uint, 0644); -MODULE_PARM_DESC(debug, "Enable driver testing on unvalidated i5000"); - -#define dprintk(fmt, arg...) \ - do { if (debug) printk(KERN_INFO I7300_PRINT fmt, ##arg); } while (0) - -/* - * Value to set THRTLOW to when initiating throttling - * 0 = No throttling - * 1 = Throttle when > 4 activations per eval window (Maximum throttling) - * 2 = Throttle when > 8 activations - * 168 = Throttle when > 672 activations (Minimum throttling) - */ -#define MAX_THROTTLE_LOW_LIMIT 168 -static uint throttle_low_limit = 1; -module_param_named(throttle_low_limit, throttle_low_limit, uint, 0644); -MODULE_PARM_DESC(throttle_low_limit, - "Value for THRTLOWLM activation field " - "(0 = disable throttle, 1 = Max throttle, 168 = Min throttle)"); - -/* - * simple invocation and duration statistics - */ -static unsigned long total_starts; -static unsigned long total_us; - -#ifdef DEBUG -static unsigned long past_skip; -#endif - -static struct pci_dev *fbd_dev; - -static raw_spinlock_t i7300_idle_lock; -static int i7300_idle_active; - -static u8 i7300_idle_thrtctl_saved; -static u8 i7300_idle_thrtlow_saved; -static u32 i7300_idle_mc_saved; - -static cpumask_var_t idle_cpumask; -static ktime_t start_ktime; -static unsigned long avg_idle_us; - -static struct dentry *debugfs_dir; - -/* Begin: I/O AT Helper routines */ - -#define IOAT_CHANBASE(ioat_ctl, chan) (ioat_ctl + 0x80 + 0x80 * chan) -/* Snoop control (disable snoops when coherency is not important) */ -#define IOAT_DESC_SADDR_SNP_CTL (1UL << 1) -#define IOAT_DESC_DADDR_SNP_CTL (1UL << 2) - -static struct pci_dev *ioat_dev; -static struct ioat_dma_descriptor *ioat_desc; /* I/O AT desc & data (1 page) */ -static unsigned long ioat_desc_phys; -static u8 *ioat_iomap; /* I/O AT memory-mapped control regs (aka CB_BAR) */ -static u8 *ioat_chanbase; - -/* Start I/O AT memory copy */ -static int i7300_idle_ioat_start(void) -{ - u32 err; - /* Clear error (due to circular descriptor pointer) */ - err = readl(ioat_chanbase + IOAT_CHANERR_OFFSET); - if (err) - writel(err, ioat_chanbase + IOAT_CHANERR_OFFSET); - - writeb(IOAT_CHANCMD_START, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - return 0; -} - -/* Stop I/O AT memory copy */ -static void i7300_idle_ioat_stop(void) -{ - int i; - u64 sts; - - for (i = 0; i < MAX_STOP_RETRIES; i++) { - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - udelay(10); - - sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (sts != IOAT_CHANSTS_ACTIVE) - break; - - } - - if (i == MAX_STOP_RETRIES) { - dprintk("failed to stop I/O AT after %d retries\n", - MAX_STOP_RETRIES); - } -} - -/* Test I/O AT by copying 1024 byte from 2k to 1k */ -static int __init i7300_idle_ioat_selftest(u8 *ctl, - struct ioat_dma_descriptor *desc, unsigned long desc_phys) -{ - u64 chan_sts; - - memset(desc, 0, 2048); - memset((u8 *) desc + 2048, 0xab, 1024); - - desc[0].size = 1024; - desc[0].ctl = 0; - desc[0].src_addr = desc_phys + 2048; - desc[0].dst_addr = desc_phys + 1024; - desc[0].next = 0; - - writeb(IOAT_CHANCMD_RESET, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - writeb(IOAT_CHANCMD_START, ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - udelay(1000); - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (chan_sts != IOAT_CHANSTS_DONE) { - /* Not complete, reset the channel */ - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - return -1; - } - - if (*(u32 *) ((u8 *) desc + 3068) != 0xabababab || - *(u32 *) ((u8 *) desc + 2044) != 0xabababab) { - dprintk("Data values src 0x%x, dest 0x%x, memset 0x%x\n", - *(u32 *) ((u8 *) desc + 2048), - *(u32 *) ((u8 *) desc + 1024), - *(u32 *) ((u8 *) desc + 3072)); - return -1; - } - return 0; -} - -static struct device dummy_dma_dev = { - .init_name = "fallback device", - .coherent_dma_mask = DMA_BIT_MASK(64), - .dma_mask = &dummy_dma_dev.coherent_dma_mask, -}; - -/* Setup and initialize I/O AT */ -/* This driver needs I/O AT as the throttling takes effect only when there is - * some memory activity. We use I/O AT to set up a dummy copy, while all CPUs - * go idle and memory is throttled. - */ -static int __init i7300_idle_ioat_init(void) -{ - u8 ver, chan_count, ioat_chan; - u16 chan_ctl; - - ioat_iomap = (u8 *) ioremap_nocache(pci_resource_start(ioat_dev, 0), - pci_resource_len(ioat_dev, 0)); - - if (!ioat_iomap) { - printk(KERN_ERR I7300_PRINT "failed to map I/O AT registers\n"); - goto err_ret; - } - - ver = readb(ioat_iomap + IOAT_VER_OFFSET); - if (ver != IOAT_VER_1_2) { - printk(KERN_ERR I7300_PRINT "unknown I/O AT version (%u.%u)\n", - ver >> 4, ver & 0xf); - goto err_unmap; - } - - chan_count = readb(ioat_iomap + IOAT_CHANCNT_OFFSET); - if (!chan_count) { - printk(KERN_ERR I7300_PRINT "unexpected # of I/O AT channels " - "(%u)\n", - chan_count); - goto err_unmap; - } - - ioat_chan = chan_count - 1; - ioat_chanbase = IOAT_CHANBASE(ioat_iomap, ioat_chan); - - chan_ctl = readw(ioat_chanbase + IOAT_CHANCTRL_OFFSET); - if (chan_ctl & IOAT_CHANCTRL_CHANNEL_IN_USE) { - printk(KERN_ERR I7300_PRINT "channel %d in use\n", ioat_chan); - goto err_unmap; - } - - writew(IOAT_CHANCTRL_CHANNEL_IN_USE, - ioat_chanbase + IOAT_CHANCTRL_OFFSET); - - ioat_desc = (struct ioat_dma_descriptor *)dma_alloc_coherent( - &dummy_dma_dev, 4096, - (dma_addr_t *)&ioat_desc_phys, GFP_KERNEL); - if (!ioat_desc) { - printk(KERN_ERR I7300_PRINT "failed to allocate I/O AT desc\n"); - goto err_mark_unused; - } - - writel(ioat_desc_phys & 0xffffffffUL, - ioat_chanbase + IOAT1_CHAINADDR_OFFSET_LOW); - writel(ioat_desc_phys >> 32, - ioat_chanbase + IOAT1_CHAINADDR_OFFSET_HIGH); - - if (i7300_idle_ioat_selftest(ioat_iomap, ioat_desc, ioat_desc_phys)) { - printk(KERN_ERR I7300_PRINT "I/O AT self-test failed\n"); - goto err_free; - } - - /* Setup circular I/O AT descriptor chain */ - ioat_desc[0].ctl = IOAT_DESC_SADDR_SNP_CTL | IOAT_DESC_DADDR_SNP_CTL; - ioat_desc[0].src_addr = ioat_desc_phys + 2048; - ioat_desc[0].dst_addr = ioat_desc_phys + 3072; - ioat_desc[0].size = 128; - ioat_desc[0].next = ioat_desc_phys + sizeof(struct ioat_dma_descriptor); - - ioat_desc[1].ctl = ioat_desc[0].ctl; - ioat_desc[1].src_addr = ioat_desc[0].src_addr; - ioat_desc[1].dst_addr = ioat_desc[0].dst_addr; - ioat_desc[1].size = ioat_desc[0].size; - ioat_desc[1].next = ioat_desc_phys; - - return 0; - -err_free: - dma_free_coherent(&dummy_dma_dev, 4096, (void *)ioat_desc, 0); -err_mark_unused: - writew(0, ioat_chanbase + IOAT_CHANCTRL_OFFSET); -err_unmap: - iounmap(ioat_iomap); -err_ret: - return -ENODEV; -} - -/* Cleanup I/O AT */ -static void __exit i7300_idle_ioat_exit(void) -{ - int i; - u64 chan_sts; - - i7300_idle_ioat_stop(); - - /* Wait for a while for the channel to halt before releasing */ - for (i = 0; i < MAX_STOP_RETRIES; i++) { - writeb(IOAT_CHANCMD_RESET, - ioat_chanbase + IOAT1_CHANCMD_OFFSET); - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - if (chan_sts != IOAT_CHANSTS_ACTIVE) { - writew(0, ioat_chanbase + IOAT_CHANCTRL_OFFSET); - break; - } - udelay(1000); - } - - chan_sts = readq(ioat_chanbase + IOAT1_CHANSTS_OFFSET) & - IOAT_CHANSTS_STATUS; - - /* - * We tried to reset multiple times. If IO A/T channel is still active - * flag an error and return without cleanup. Memory leak is better - * than random corruption in that extreme error situation. - */ - if (chan_sts == IOAT_CHANSTS_ACTIVE) { - printk(KERN_ERR I7300_PRINT "Unable to stop IO A/T channels." - " Not freeing resources\n"); - return; - } - - dma_free_coherent(&dummy_dma_dev, 4096, (void *)ioat_desc, 0); - iounmap(ioat_iomap); -} - -/* End: I/O AT Helper routines */ - -#define DIMM_THRTLOW 0x64 -#define DIMM_THRTCTL 0x67 -#define DIMM_THRTCTL_THRMHUNT (1UL << 0) -#define DIMM_MC 0x40 -#define DIMM_GTW_MODE (1UL << 17) -#define DIMM_GBLACT 0x60 - -/* - * Keep track of an exponential-decaying average of recent idle durations. - * The latest duration gets DURATION_WEIGHT_PCT percentage weight - * in this average, with the old average getting the remaining weight. - * - * High weights emphasize recent history, low weights include long history. - */ -#define DURATION_WEIGHT_PCT 55 - -/* - * When the decaying average of recent durations or the predicted duration - * of the next timer interrupt is shorter than duration_threshold, the - * driver will decline to throttle. - */ -#define DURATION_THRESHOLD_US 100 - - -/* Store DIMM thermal throttle configuration */ -static int i7300_idle_thrt_save(void) -{ - u32 new_mc_val; - u8 gblactlm; - - pci_read_config_byte(fbd_dev, DIMM_THRTCTL, &i7300_idle_thrtctl_saved); - pci_read_config_byte(fbd_dev, DIMM_THRTLOW, &i7300_idle_thrtlow_saved); - pci_read_config_dword(fbd_dev, DIMM_MC, &i7300_idle_mc_saved); - /* - * Make sure we have Global Throttling Window Mode set to have a - * "short" window. This (mostly) works around an issue where - * throttling persists until the end of the global throttling window - * size. On the tested system, this was resulting in a maximum of - * 64 ms to exit throttling (average 32 ms). The actual numbers - * depends on system frequencies. Setting the short window reduces - * this by a factor of 4096. - * - * We will only do this only if the system is set for - * unlimited-activations while in open-loop throttling (i.e., when - * Global Activation Throttle Limit is zero). - */ - pci_read_config_byte(fbd_dev, DIMM_GBLACT, &gblactlm); - dprintk("thrtctl_saved = 0x%02x, thrtlow_saved = 0x%02x\n", - i7300_idle_thrtctl_saved, - i7300_idle_thrtlow_saved); - dprintk("mc_saved = 0x%08x, gblactlm = 0x%02x\n", - i7300_idle_mc_saved, - gblactlm); - if (gblactlm == 0) { - new_mc_val = i7300_idle_mc_saved | DIMM_GTW_MODE; - pci_write_config_dword(fbd_dev, DIMM_MC, new_mc_val); - return 0; - } else { - dprintk("could not set GTW_MODE = 1 (OLTT enabled)\n"); - return -ENODEV; - } -} - -/* Restore DIMM thermal throttle configuration */ -static void i7300_idle_thrt_restore(void) -{ - pci_write_config_dword(fbd_dev, DIMM_MC, i7300_idle_mc_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, i7300_idle_thrtlow_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, i7300_idle_thrtctl_saved); -} - -/* Enable DIMM thermal throttling */ -static void i7300_idle_start(void) -{ - u8 new_ctl; - u8 limit; - - new_ctl = i7300_idle_thrtctl_saved & ~DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); - - limit = throttle_low_limit; - if (unlikely(limit > MAX_THROTTLE_LOW_LIMIT)) - limit = MAX_THROTTLE_LOW_LIMIT; - - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, limit); - - new_ctl = i7300_idle_thrtctl_saved | DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); -} - -/* Disable DIMM thermal throttling */ -static void i7300_idle_stop(void) -{ - u8 new_ctl; - u8 got_ctl; - - new_ctl = i7300_idle_thrtctl_saved & ~DIMM_THRTCTL_THRMHUNT; - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, new_ctl); - - pci_write_config_byte(fbd_dev, DIMM_THRTLOW, i7300_idle_thrtlow_saved); - pci_write_config_byte(fbd_dev, DIMM_THRTCTL, i7300_idle_thrtctl_saved); - pci_read_config_byte(fbd_dev, DIMM_THRTCTL, &got_ctl); - WARN_ON_ONCE(got_ctl != i7300_idle_thrtctl_saved); -} - - -/* - * i7300_avg_duration_check() - * return 0 if the decaying average of recent idle durations is - * more than DURATION_THRESHOLD_US - */ -static int i7300_avg_duration_check(void) -{ - if (avg_idle_us >= DURATION_THRESHOLD_US) - return 0; - -#ifdef DEBUG - past_skip++; -#endif - return 1; -} - -/* Idle notifier to look at idle CPUs */ -static int i7300_idle_notifier(struct notifier_block *nb, unsigned long val, - void *data) -{ - unsigned long flags; - ktime_t now_ktime; - static ktime_t idle_begin_time; - static int time_init = 1; - - if (!throttle_low_limit) - return 0; - - if (unlikely(time_init)) { - time_init = 0; - idle_begin_time = ktime_get(); - } - - raw_spin_lock_irqsave(&i7300_idle_lock, flags); - if (val == IDLE_START) { - - cpumask_set_cpu(smp_processor_id(), idle_cpumask); - - if (cpumask_weight(idle_cpumask) != num_online_cpus()) - goto end; - - now_ktime = ktime_get(); - idle_begin_time = now_ktime; - - if (i7300_avg_duration_check()) - goto end; - - i7300_idle_active = 1; - total_starts++; - start_ktime = now_ktime; - - i7300_idle_start(); - i7300_idle_ioat_start(); - - } else if (val == IDLE_END) { - cpumask_clear_cpu(smp_processor_id(), idle_cpumask); - if (cpumask_weight(idle_cpumask) == (num_online_cpus() - 1)) { - /* First CPU coming out of idle */ - u64 idle_duration_us; - - now_ktime = ktime_get(); - - idle_duration_us = ktime_to_us(ktime_sub - (now_ktime, idle_begin_time)); - - avg_idle_us = - ((100 - DURATION_WEIGHT_PCT) * avg_idle_us + - DURATION_WEIGHT_PCT * idle_duration_us) / 100; - - if (i7300_idle_active) { - ktime_t idle_ktime; - - idle_ktime = ktime_sub(now_ktime, start_ktime); - total_us += ktime_to_us(idle_ktime); - - i7300_idle_ioat_stop(); - i7300_idle_stop(); - i7300_idle_active = 0; - } - } - } -end: - raw_spin_unlock_irqrestore(&i7300_idle_lock, flags); - return 0; -} - -static struct notifier_block i7300_idle_nb = { - .notifier_call = i7300_idle_notifier, -}; - -MODULE_DEVICE_TABLE(pci, pci_tbl); - -static ssize_t stats_read_ul(struct file *fp, char __user *ubuf, size_t count, - loff_t *off) -{ - unsigned long *p = fp->private_data; - char buf[32]; - int len; - - len = snprintf(buf, 32, "%lu\n", *p); - return simple_read_from_buffer(ubuf, count, off, buf, len); -} - -static const struct file_operations idle_fops = { - .open = simple_open, - .read = stats_read_ul, - .llseek = default_llseek, -}; - -struct debugfs_file_info { - void *ptr; - char name[32]; - struct dentry *file; -} debugfs_file_list[] = { - {&total_starts, "total_starts", NULL}, - {&total_us, "total_us", NULL}, -#ifdef DEBUG - {&past_skip, "past_skip", NULL}, -#endif - {NULL, "", NULL} - }; - -static int __init i7300_idle_init(void) -{ - raw_spin_lock_init(&i7300_idle_lock); - total_us = 0; - - if (i7300_idle_platform_probe(&fbd_dev, &ioat_dev, forceload)) - return -ENODEV; - - if (i7300_idle_thrt_save()) - return -ENODEV; - - if (i7300_idle_ioat_init()) - return -ENODEV; - - if (!zalloc_cpumask_var(&idle_cpumask, GFP_KERNEL)) - return -ENOMEM; - - debugfs_dir = debugfs_create_dir("i7300_idle", NULL); - if (debugfs_dir) { - int i = 0; - - while (debugfs_file_list[i].ptr != NULL) { - debugfs_file_list[i].file = debugfs_create_file( - debugfs_file_list[i].name, - S_IRUSR, - debugfs_dir, - debugfs_file_list[i].ptr, - &idle_fops); - i++; - } - } - - idle_notifier_register(&i7300_idle_nb); - - printk(KERN_INFO "i7300_idle: loaded v%s\n", I7300_IDLE_DRIVER_VERSION); - return 0; -} - -static void __exit i7300_idle_exit(void) -{ - idle_notifier_unregister(&i7300_idle_nb); - free_cpumask_var(idle_cpumask); - - if (debugfs_dir) { - int i = 0; - - while (debugfs_file_list[i].file != NULL) { - debugfs_remove(debugfs_file_list[i].file); - i++; - } - - debugfs_remove(debugfs_dir); - } - i7300_idle_thrt_restore(); - i7300_idle_ioat_exit(); -} - -module_init(i7300_idle_init); -module_exit(i7300_idle_exit); - -MODULE_AUTHOR("Andy Henroid <andrew.d.henroid@intel.com>"); -MODULE_DESCRIPTION("Intel Chipset DIMM Idle Power Saving Driver v" - I7300_IDLE_DRIVER_VERSION); -MODULE_LICENSE("GPL"); ^ permalink raw reply related [flat|nested] 18+ messages in thread
* Re: [PATCH 0/5] x86: remove idle notifier 2016-11-18 6:23 [PATCH 0/5] x86: remove idle notifier Len Brown 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown @ 2016-11-18 8:17 ` Ingo Molnar 1 sibling, 0 replies; 18+ messages in thread From: Ingo Molnar @ 2016-11-18 8:17 UTC (permalink / raw) To: Len Brown; +Cc: x86, linux-kernel, Thomas Gleixner * Len Brown <lenb@kernel.org> wrote: > The return from idle path is latency sensitive, > so the less code and fewer data accesses, the better. > > Remove the un-maintained i7300_idle driver, > the only user of the x86 idle-notifier, > and then remove the notifier itself. Acked-by: Ingo Molnar <mingo@kernel.org> Thanks, Ingo ^ permalink raw reply [flat|nested] 18+ messages in thread
end of thread, other threads:[~2016-12-08 22:56 UTC | newest] Thread overview: 18+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2016-11-18 6:23 [PATCH 0/5] x86: remove idle notifier Len Brown 2016-11-18 6:23 ` [PATCH 1/5] i7300_idle: remove this driver Len Brown 2016-11-18 6:23 ` [PATCH 2/5] x86: remove idle_notifier Len Brown 2016-11-18 11:19 ` [tip:x86/idle] x86/idle: Remove idle_notifier tip-bot for Len Brown 2016-12-06 16:54 ` [PATCH 2/5] x86: remove idle_notifier Pavel Machek 2016-12-07 11:46 ` Thomas Gleixner 2016-12-07 13:40 ` Pavel Machek 2016-12-08 9:18 ` Thomas Gleixner 2016-12-08 22:05 ` Pavel Machek 2016-12-08 22:53 ` Thomas Gleixner 2016-11-18 6:23 ` [PATCH 3/5] x86: remove is_idle flag Len Brown 2016-11-18 11:20 ` [tip:x86/idle] x86/idle: Remove " tip-bot for Len Brown 2016-11-18 6:23 ` [PATCH 4/5] x86: remove x86_test_and_clear_bit_percpu() Len Brown 2016-11-18 11:20 ` [tip:x86/idle] x86: Remove x86_test_and_clear_bit_percpu() tip-bot for Len Brown 2016-11-18 6:23 ` [PATCH 5/5] x86: remove enter_idle(), exit_idle() Len Brown 2016-11-18 11:21 ` [tip:x86/idle] x86/idle: Remove " tip-bot for Len Brown 2016-11-18 11:19 ` [tip:x86/idle] i7300_idle: Remove this driver tip-bot for Len Brown 2016-11-18 8:17 ` [PATCH 0/5] x86: remove idle notifier Ingo Molnar
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).