* [PATCH v1 01/11] arm64: MAINTAINERS: hisi: Add hisilicon SoC PMU support
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 02/11] dt-bindings: hisi: Add Hisilicon HiP05/06/07 Sysctrl and Djtag dts bindings Anurup M
` (9 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
Add support for Hisilicon SoC hardware event counters
for HIP05/06/07 chip versions.
Signed-off-by: Anurup M <anurup.m@huawei.com>
---
MAINTAINERS | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index b224caa..839abc8 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -5725,6 +5725,16 @@ S: Maintained
F: drivers/net/ethernet/hisilicon/
F: Documentation/devicetree/bindings/net/hisilicon*.txt
+HISILICON SOC PMU
+M: Anurup M <anurup.m@huawei.com>
+W: http://www.hisilicon.com
+S: Supported
+F: drivers/perf/hisilicon/*
+F: drivers/soc/hisilicon/djtag.c
+F: include/linux/soc/hisilicon/djtag.h
+F: Documentation/perf/hisi-pmu.txt
+F: Documentation/devicetree/bindings/arm/hisilicon/pmu.txt
+
HISILICON ROCE DRIVER
M: Lijun Ou <oulijun@huawei.com>
M: Wei Hu(Xavier) <xavier.huwei@huawei.com>
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 02/11] dt-bindings: hisi: Add Hisilicon HiP05/06/07 Sysctrl and Djtag dts bindings
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
2016-11-02 15:42 ` [PATCH v1 01/11] arm64: MAINTAINERS: hisi: Add hisilicon SoC PMU support Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver Anurup M
` (8 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
From: Tan Xiaojun <tanxiaojun@huawei.com>
1) Add Hisilicon HiP05/06/07 CPU and ALGSUB system controller dts
bindings.
2) Add Hisilicon Djtag dts binding.
Signed-off-by: Tan Xiaojun <tanxiaojun@huawei.com>
Signed-off-by: Anurup M <anurup.m@huawei.com>
---
.../bindings/arm/hisilicon/hisilicon.txt | 82 ++++++++++++++++++++++
1 file changed, 82 insertions(+)
diff --git a/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt b/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt
index 7df79a7..341cbb9 100644
--- a/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt
+++ b/Documentation/devicetree/bindings/arm/hisilicon/hisilicon.txt
@@ -270,3 +270,85 @@ Required Properties:
[1]: bootwrapper size
[2]: relocation physical address
[3]: relocation size
+
+-----------------------------------------------------------------------
+The Hisilicon Djtag in CPU die is an independent component which connects with
+some other components in the SoC by Debug Bus. This driver can be configured
+to access the registers of connecting components (like L3 cache, l3 cache PMU
+ etc.) during real time debugging by sysctrl. These components appear as child
+nodes of djtag.
+
+The Hip05/06/07 CPU system controller(sysctrl) support to manage some important
+components (such as clock, reset, soft reset, secure debugger, etc.).
+The CPU sysctrl registers in hip05/06/07 doesnot use syscon but will be mapped
+by djtag driver for use by connecting components.
+
+Hisilicon HiP05 CPU system controller
+Required properties:
+ - compatible : "hisilicon,hip05-cpu-djtag-v1"
+ - reg : Register address and size
+
+Hisilicon HiP06 djtag for CPU sysctrl
+Required properties:
+- compatible : "hisilicon,hip06-sysctrl", "syscon", "simple-mfd";
+- reg : Register address and size
+- djtag :
+ - compatible : "hisilicon,hip06-cpu-djtag-v1"
+ - reg : Register address and size
+
+Hisilicon HiP07 djtag for CPU sysctrl
+Required properties:
+ - compatible : "hisilicon,hip07-cpu-djtag-v2"
+ - reg : Register address and size
+
+Example:
+ /* for Hisilicon HiP05 djtag for CPU sysctrl */
+ djtag0: djtag@80010000 {
+ compatible = "hisilicon,hip05-cpu-djtag-v1";
+ reg = <0x0 0x80010000 0x0 0x10000>;
+
+ /* For L3 cache PMU */
+ pmul3c0 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x02>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
+ };
+
+-----------------------------------------------------------------------
+The Hisilicon HiP05/06/07 ALGSUB system controller(sysctrl) is in IO die
+of SoC. It has a similar function as the Hisilicon HiP05/06/07 CPU system
+controller in CPU die and it manage different components, like RSA, etc.
+The Hisilicon Djtag in IO die has a similar function as in CPU die and maps
+the sysctrl registers for use by connecting components.
+All connecting components shall appear as child nodes of djtag.
+
+Hisilicon HiP05 djtag for ALGSUB sysctrl
+Required properties:
+ - compatible : "hisilicon,hip05-io-djtag-v1"
+ - reg : Register address and size
+
+Hisilicon HiP06 djtag for ALGSUB sysctrl
+Required properties:
+ - compatible : "hisilicon,hip06-io-djtag-v2"
+ - reg : Register address and size
+
+Hisilicon HiP07 djtag for ALGSUB sysctrl
+Required properties:
+ - compatible : "hisilicon,hip07-io-djtag-v2"
+ - reg : Register address and size
+
+Example:
+ /* for Hisilicon HiP05 djtag for alg sysctrl */
+ djtag0: djtag@d0000000 {
+ compatible = "hisilicon,hip05-io-djtag-v1";
+ reg = <0x0 0xd0000000 0x0 0x10000>;
+ };
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
2016-11-02 15:42 ` [PATCH v1 01/11] arm64: MAINTAINERS: hisi: Add hisilicon SoC PMU support Anurup M
2016-11-02 15:42 ` [PATCH v1 02/11] dt-bindings: hisi: Add Hisilicon HiP05/06/07 Sysctrl and Djtag dts bindings Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-03 1:59 ` kbuild test robot
2016-11-07 13:26 ` Arnd Bergmann
2016-11-02 15:42 ` [PATCH v1 04/11] Documentation: perf: hisi: Documentation for HIP05/06/07 PMU event counting Anurup M
` (7 subsequent siblings)
10 siblings, 2 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
From: Tan Xiaojun <tanxiaojun@huawei.com>
The Hisilicon Djtag is an independent component which connects
with some other components in the SoC by Debug Bus. This driver
can be configured to access the registers of connecting components
(like L3 cache) during real time debugging.
Signed-off-by: Tan Xiaojun <tanxiaojun@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Anurup M <anurup.m@huawei.com>
---
drivers/soc/Kconfig | 1 +
drivers/soc/Makefile | 1 +
drivers/soc/hisilicon/Kconfig | 12 +
drivers/soc/hisilicon/Makefile | 1 +
drivers/soc/hisilicon/djtag.c | 639 ++++++++++++++++++++++++++++++++++++
include/linux/soc/hisilicon/djtag.h | 38 +++
6 files changed, 692 insertions(+)
create mode 100644 drivers/soc/hisilicon/Kconfig
create mode 100644 drivers/soc/hisilicon/Makefile
create mode 100644 drivers/soc/hisilicon/djtag.c
create mode 100644 include/linux/soc/hisilicon/djtag.h
diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig
index e6e90e8..89ecd42 100644
--- a/drivers/soc/Kconfig
+++ b/drivers/soc/Kconfig
@@ -3,6 +3,7 @@ menu "SOC (System On Chip) specific Drivers"
source "drivers/soc/bcm/Kconfig"
source "drivers/soc/fsl/qbman/Kconfig"
source "drivers/soc/fsl/qe/Kconfig"
+source "drivers/soc/hisilicon/Kconfig"
source "drivers/soc/mediatek/Kconfig"
source "drivers/soc/qcom/Kconfig"
source "drivers/soc/rockchip/Kconfig"
diff --git a/drivers/soc/Makefile b/drivers/soc/Makefile
index 50c23d0..ce2beb5 100644
--- a/drivers/soc/Makefile
+++ b/drivers/soc/Makefile
@@ -6,6 +6,7 @@ obj-y += bcm/
obj-$(CONFIG_ARCH_DOVE) += dove/
obj-$(CONFIG_MACH_DOVE) += dove/
obj-y += fsl/
+obj-$(CONFIG_ARCH_HISI) += hisilicon/
obj-$(CONFIG_ARCH_MEDIATEK) += mediatek/
obj-$(CONFIG_ARCH_QCOM) += qcom/
obj-$(CONFIG_ARCH_RENESAS) += renesas/
diff --git a/drivers/soc/hisilicon/Kconfig b/drivers/soc/hisilicon/Kconfig
new file mode 100644
index 0000000..6dd4ba0
--- /dev/null
+++ b/drivers/soc/hisilicon/Kconfig
@@ -0,0 +1,12 @@
+#
+# Hisilicon SoC drivers
+#
+config HISI_DJTAG
+ bool "Hisilicon Djtag Support"
+ depends on ARCH_HISI || COMPILE_TEST
+ help
+ Say y here to enable the Hisilicon Djtag support. It is
+ an independent component which connects with some other
+ components in the SoC by Debug Bus. This driver can be
+ configured to access the registers of connecting
+ components during real time debugging.
diff --git a/drivers/soc/hisilicon/Makefile b/drivers/soc/hisilicon/Makefile
new file mode 100644
index 0000000..35a7b4b
--- /dev/null
+++ b/drivers/soc/hisilicon/Makefile
@@ -0,0 +1 @@
+obj-$(CONFIG_HISI_DJTAG) += djtag.o
diff --git a/drivers/soc/hisilicon/djtag.c b/drivers/soc/hisilicon/djtag.c
new file mode 100644
index 0000000..a87c8b6
--- /dev/null
+++ b/drivers/soc/hisilicon/djtag.c
@@ -0,0 +1,639 @@
+/*
+ * Driver for Hisilicon Djtag r/w via System Controller.
+ *
+ * Copyright (C) 2016 Hisilicon Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitops.h>
+#include <linux/init.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+
+#include <asm-generic/delay.h>
+#include <linux/soc/hisilicon/djtag.h>
+
+#define SC_DJTAG_TIMEOUT 100000 /* 100ms */
+
+/* for djtag v1 */
+#define SC_DJTAG_MSTR_EN 0x6800
+#define DJTAG_NOR_CFG BIT(1) /* accelerate R,W */
+#define DJTAG_MSTR_EN BIT(0)
+#define SC_DJTAG_MSTR_START_EN 0x6804
+#define DJTAG_MSTR_START_EN 0x1
+#define SC_DJTAG_DEBUG_MODULE_SEL 0x680c
+#define SC_DJTAG_MSTR_WR 0x6810
+#define DJTAG_MSTR_W 0x1
+#define DJTAG_MSTR_R 0x0
+#define SC_DJTAG_CHAIN_UNIT_CFG_EN 0x6814
+#define CHAIN_UNIT_CFG_EN 0xFFFF
+#define SC_DJTAG_MSTR_ADDR 0x6818
+#define SC_DJTAG_MSTR_DATA 0x681c
+#define SC_DJTAG_RD_DATA_BASE 0xe800
+
+/* for djtag v2 */
+#define SC_DJTAG_SEC_ACC_EN_EX 0xd800
+#define DJTAG_SEC_ACC_EN_EX 0x1
+#define SC_DJTAG_MSTR_CFG_EX 0xd818
+#define DJTAG_MSTR_RW_SHIFT_EX 29
+#define DJTAG_MSTR_RD_EX (0x0 << DJTAG_MSTR_RW_SHIFT_EX)
+#define DJTAG_MSTR_WR_EX (0x1 << DJTAG_MSTR_RW_SHIFT_EX)
+#define DEBUG_MODULE_SEL_SHIFT_EX 16
+#define CHAIN_UNIT_CFG_EN_EX 0xFFFF
+#define SC_DJTAG_MSTR_ADDR_EX 0xd810
+#define SC_DJTAG_MSTR_DATA_EX 0xd814
+#define SC_DJTAG_MSTR_START_EN_EX 0xd81c
+#define DJTAG_MSTR_START_EN_EX 0x1
+#define SC_DJTAG_RD_DATA_BASE_EX 0xe800
+#define SC_DJTAG_OP_ST_EX 0xe828
+#define DJTAG_OP_DONE_EX BIT(8)
+
+#define DJTAG_PREFIX "hisi-djtag-dev-"
+
+DEFINE_IDR(djtag_hosts_idr);
+
+struct hisi_djtag_host {
+ spinlock_t lock;
+ int id;
+ struct device dev;
+ struct list_head client_list;
+ void __iomem *sysctl_reg_map;
+ struct device_node *of_node;
+ int (*djtag_readwrite)(void __iomem *regs_base, u32 offset,
+ u32 mod_sel, u32 mod_mask, bool is_w,
+ u32 wval, int chain_id, u32 *rval);
+};
+
+#define to_hisi_djtag_client(d) container_of(d, struct hisi_djtag_client, dev)
+#define to_hisi_djtag_driver(d) container_of(d, struct hisi_djtag_driver, \
+ driver)
+#define MODULE_PREFIX "hisi_djtag:"
+
+static void djtag_read32_relaxed(void __iomem *regs_base, u32 off, u32 *value)
+{
+ void __iomem *reg_addr = regs_base + off;
+
+ *value = readl_relaxed(reg_addr);
+}
+
+static void djtag_write32(void __iomem *regs_base, u32 off, u32 val)
+{
+ void __iomem *reg_addr = regs_base + off;
+
+ writel(val, reg_addr);
+}
+
+/*
+ * djtag_readwrite_v1/v2: djtag read/write interface
+ * @reg_base: djtag register base address
+ * @offset: register's offset
+ * @mod_sel: module selection
+ * @mod_mask: mask to select specific modules for write
+ * @is_w: write -> true, read -> false
+ * @wval: value to register for write
+ * @chain_id: which sub module for read
+ * @rval: value in register for read
+ *
+ * Return non-zero if error, else return 0.
+ */
+static int djtag_readwrite_v1(void __iomem *regs_base, u32 offset, u32 mod_sel,
+ u32 mod_mask, bool is_w, u32 wval, int chain_id, u32 *rval)
+{
+ u32 rd;
+ int timeout = SC_DJTAG_TIMEOUT;
+
+ if (!(mod_mask & CHAIN_UNIT_CFG_EN)) {
+ pr_warn("djtag: do nothing.\n");
+ return 0;
+ }
+
+ /* djtag mster enable & accelerate R,W */
+ djtag_write32(regs_base, SC_DJTAG_MSTR_EN,
+ DJTAG_NOR_CFG | DJTAG_MSTR_EN);
+
+ /* select module */
+ djtag_write32(regs_base, SC_DJTAG_DEBUG_MODULE_SEL, mod_sel);
+ djtag_write32(regs_base, SC_DJTAG_CHAIN_UNIT_CFG_EN,
+ mod_mask & CHAIN_UNIT_CFG_EN);
+
+ if (is_w) {
+ djtag_write32(regs_base, SC_DJTAG_MSTR_WR, DJTAG_MSTR_W);
+ djtag_write32(regs_base, SC_DJTAG_MSTR_DATA, wval);
+ } else
+ djtag_write32(regs_base, SC_DJTAG_MSTR_WR, DJTAG_MSTR_R);
+
+ /* address offset */
+ djtag_write32(regs_base, SC_DJTAG_MSTR_ADDR, offset);
+
+ /* start to write to djtag register */
+ djtag_write32(regs_base, SC_DJTAG_MSTR_START_EN, DJTAG_MSTR_START_EN);
+
+ /* ensure the djtag operation is done */
+ do {
+ djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN, &rd);
+ if (!(rd & DJTAG_MSTR_EN))
+ break;
+
+ udelay(1);
+ } while (timeout--);
+
+ if (timeout < 0) {
+ pr_err("djtag: %s timeout!\n", is_w ? "write" : "read");
+ return -EBUSY;
+ }
+
+ if (!is_w)
+ djtag_read32_relaxed(regs_base,
+ SC_DJTAG_RD_DATA_BASE + chain_id * 0x4, rval);
+
+ return 0;
+}
+
+static int djtag_readwrite_v2(void __iomem *regs_base, u32 offset, u32 mod_sel,
+ u32 mod_mask, bool is_w, u32 wval, int chain_id, u32 *rval)
+{
+ u32 rd;
+ int timeout = SC_DJTAG_TIMEOUT;
+
+ if (!(mod_mask & CHAIN_UNIT_CFG_EN_EX)) {
+ pr_warn("djtag: do nothing.\n");
+ return 0;
+ }
+
+ /* djtag mster enable */
+ djtag_write32(regs_base, SC_DJTAG_SEC_ACC_EN_EX, DJTAG_SEC_ACC_EN_EX);
+
+ if (is_w) {
+ djtag_write32(regs_base, SC_DJTAG_MSTR_CFG_EX, DJTAG_MSTR_WR_EX
+ | (mod_sel << DEBUG_MODULE_SEL_SHIFT_EX)
+ | (mod_mask & CHAIN_UNIT_CFG_EN_EX));
+ djtag_write32(regs_base, SC_DJTAG_MSTR_DATA_EX, wval);
+ } else
+ djtag_write32(regs_base, SC_DJTAG_MSTR_CFG_EX, DJTAG_MSTR_RD_EX
+ | (mod_sel << DEBUG_MODULE_SEL_SHIFT_EX)
+ | (mod_mask & CHAIN_UNIT_CFG_EN_EX));
+
+ /* address offset */
+ djtag_write32(regs_base, SC_DJTAG_MSTR_ADDR_EX, offset);
+
+ /* start to write to djtag register */
+ djtag_write32(regs_base,
+ SC_DJTAG_MSTR_START_EN_EX, DJTAG_MSTR_START_EN_EX);
+
+ /* ensure the djtag operation is done */
+ do {
+ djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
+
+ if (!(rd & DJTAG_MSTR_START_EN_EX))
+ break;
+
+ udelay(1);
+ } while (timeout--);
+
+ if (timeout < 0)
+ goto timeout;
+
+ timeout = SC_DJTAG_TIMEOUT;
+ do {
+ djtag_read32_relaxed(regs_base, SC_DJTAG_OP_ST_EX, &rd);
+
+ if (rd & DJTAG_OP_DONE_EX)
+ break;
+
+ udelay(1);
+ } while (timeout--);
+
+ if (timeout < 0)
+ goto timeout;
+
+ if (!is_w)
+ djtag_read32_relaxed(regs_base,
+ SC_DJTAG_RD_DATA_BASE_EX + chain_id * 0x4,
+ rval);
+
+ return 0;
+
+timeout:
+ pr_err("djtag: %s timeout!\n", is_w ? "write" : "read");
+ return -EBUSY;
+}
+
+
+/**
+ * djtag_writel - write registers via djtag
+ * @node: djtag node
+ * @offset: register's offset
+ * @mod_sel: module selection
+ * @mod_mask: mask to select specific modules
+ * @val: value to write to register
+ *
+ * If error return errno, otherwise return 0.
+ */
+int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
+ u32 mod_mask, u32 val)
+{
+ void __iomem *reg_map = client->host->sysctl_reg_map;
+ unsigned long flags;
+ int ret = 0;
+
+ spin_lock_irqsave(&client->host->lock, flags);
+ ret = client->host->djtag_readwrite(reg_map, offset, mod_sel, mod_mask,
+ true, val, 0, NULL);
+ if (ret)
+ pr_err("djtag_writel: error! ret=%d\n", ret);
+ spin_unlock_irqrestore(&client->host->lock, flags);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(hisi_djtag_writel);
+
+/**
+ * djtag_readl - read registers via djtag
+ * @node: djtag node
+ * @offset: register's offset
+ * @mod_sel: module type selection
+ * @chain_id: chain_id number, mostly is 0
+ * @val: register's value
+ *
+ * If error return errno, otherwise return 0.
+ */
+int hisi_djtag_readl(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
+ int chain_id, u32 *val)
+{
+ void __iomem *reg_map = client->host->sysctl_reg_map;
+ unsigned long flags;
+ int ret = 0;
+
+ spin_lock_irqsave(&client->host->lock, flags);
+ ret = client->host->djtag_readwrite(reg_map, offset, mod_sel,
+ 0xffff, false, 0, chain_id, val);
+ if (ret)
+ pr_err("djtag_readl: error! ret=%d\n", ret);
+ spin_unlock_irqrestore(&client->host->lock, flags);
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(hisi_djtag_readl);
+
+static const struct of_device_id djtag_of_match[] = {
+ /* for hip05(D02) cpu die */
+ { .compatible = "hisilicon,hip05-cpu-djtag-v1",
+ .data = (void *)djtag_readwrite_v1 },
+ /* for hip05(D02) io die */
+ { .compatible = "hisilicon,hip05-io-djtag-v1",
+ .data = (void *)djtag_readwrite_v1 },
+ /* for hip06(D03) cpu die */
+ { .compatible = "hisilicon,hip06-cpu-djtag-v1",
+ .data = (void *)djtag_readwrite_v1 },
+ /* for hip06(D03) io die */
+ { .compatible = "hisilicon,hip06-io-djtag-v2",
+ .data = (void *)djtag_readwrite_v2 },
+ /* for hip07(D05) cpu die */
+ { .compatible = "hisilicon,hip07-cpu-djtag-v2",
+ .data = (void *)djtag_readwrite_v2 },
+ /* for hip07(D05) io die */
+ { .compatible = "hisilicon,hip07-io-djtag-v2",
+ .data = (void *)djtag_readwrite_v2 },
+ {},
+};
+
+MODULE_DEVICE_TABLE(of, djtag_of_match);
+
+static ssize_t
+show_modalias(struct device *dev, struct device_attribute *attr, char *buf)
+{
+ struct hisi_djtag_client *client = to_hisi_djtag_client(dev);
+
+ return sprintf(buf, "%s%s\n", MODULE_PREFIX, client->name);
+}
+static DEVICE_ATTR(modalias, 0444, show_modalias, NULL);
+
+static struct attribute *hisi_djtag_dev_attrs[] = {
+ NULL,
+ /* modalias helps coldplug: modprobe $(cat .../modalias) */
+ &dev_attr_modalias.attr,
+ NULL
+};
+ATTRIBUTE_GROUPS(hisi_djtag_dev);
+
+static struct device_type hisi_djtag_client_type = {
+ .groups = hisi_djtag_dev_groups,
+};
+
+struct hisi_djtag_client *hisi_djtag_verify_client(struct device *dev)
+{
+ return (dev->type == &hisi_djtag_client_type)
+ ? to_hisi_djtag_client(dev)
+ : NULL;
+}
+
+static int hisi_djtag_device_probe(struct device *dev)
+{
+ struct hisi_djtag_driver *driver;
+ struct hisi_djtag_client *client;
+ int rc;
+
+ client = hisi_djtag_verify_client(dev);
+ if (!client) {
+ dev_err(dev, "could not find client\n");
+ return -ENODEV;
+ }
+
+ driver = to_hisi_djtag_driver(dev->driver);
+ if (!driver) {
+ dev_err(dev, "could not find driver\n");
+ return -ENODEV;
+ }
+
+ rc = driver->probe(client);
+ if (rc < 0) {
+ dev_err(dev, "client probe failed\n");
+ return rc;
+ }
+
+ return 0;
+}
+
+static int hisi_djtag_device_remove(struct device *dev)
+{
+ struct hisi_djtag_driver *driver;
+ struct hisi_djtag_client *client;
+ int rc;
+
+ client = hisi_djtag_verify_client(dev);
+ if (!client) {
+ dev_err(dev, "could not find client\n");
+ return -ENODEV;
+ }
+
+ driver = to_hisi_djtag_driver(dev->driver);
+ if (!driver) {
+ dev_err(dev, "could not find driver\n");
+ return -ENODEV;
+ }
+
+ rc = driver->remove(client);
+ if (rc < 0) {
+ dev_err(dev, "client probe failed\n");
+ return rc;
+ }
+
+ return 0;
+}
+
+static int hisi_djtag_device_match(struct device *dev,
+ struct device_driver *drv)
+{
+ const struct of_device_id *p;
+ struct hisi_djtag_client *client = hisi_djtag_verify_client(dev);
+
+ if (!client)
+ return false;
+
+ if (of_driver_match_device(dev, drv))
+ return true;
+
+ p = of_match_device(drv->of_match_table, dev);
+ if (!p)
+ return false;
+
+ return true;
+}
+
+struct bus_type hisi_djtag_bus = {
+ .name = "hisi-djtag",
+ .match = hisi_djtag_device_match,
+ .probe = hisi_djtag_device_probe,
+ .remove = hisi_djtag_device_remove,
+};
+
+struct hisi_djtag_client *hisi_djtag_new_device(struct hisi_djtag_host *host,
+ struct device_node *node)
+{
+ struct hisi_djtag_client *client;
+ int status;
+
+ client = kzalloc(sizeof(*client), GFP_KERNEL);
+ if (!client)
+ return NULL;
+
+ client->host = host;
+
+ client->dev.parent = &client->host->dev;
+ client->dev.bus = &hisi_djtag_bus;
+ client->dev.type = &hisi_djtag_client_type;
+ client->dev.of_node = node;
+ snprintf(client->name, DJTAG_CLIENT_NAME_LEN, "%s%s",
+ DJTAG_PREFIX, node->name);
+ dev_set_name(&client->dev, "%s", client->name);
+
+ status = device_register(&client->dev);
+ if (status < 0) {
+ pr_err("error adding new device, status=%d\n", status);
+ kfree(client);
+ return NULL;
+ }
+
+ return client;
+}
+
+static struct hisi_djtag_client *hisi_djtag_of_register_device(
+ struct hisi_djtag_host *host,
+ struct device_node *node)
+{
+ struct hisi_djtag_client *client;
+
+ client = hisi_djtag_new_device(host, node);
+ if (client == NULL) {
+ dev_err(&host->dev, "error registering device %s\n",
+ node->full_name);
+ of_node_put(node);
+ return ERR_PTR(-EINVAL);
+ }
+
+ return client;
+}
+
+static void djtag_register_devices(struct hisi_djtag_host *host)
+{
+ struct device_node *node;
+ struct hisi_djtag_client *client;
+
+ if (!host->of_node)
+ return;
+
+ for_each_available_child_of_node(host->of_node, node) {
+ if (of_node_test_and_set_flag(node, OF_POPULATED))
+ continue;
+ client = hisi_djtag_of_register_device(host, node);
+ list_add(&client->next, &host->client_list);
+ }
+}
+
+static int hisi_djtag_add_host(struct hisi_djtag_host *host)
+{
+ int rc;
+
+ host->dev.bus = &hisi_djtag_bus;
+
+ rc = idr_alloc(&djtag_hosts_idr, host, 0, 0, GFP_KERNEL);
+ if (rc < 0) {
+ dev_err(&host->dev, "No available djtag host ID'!s\n");
+ return rc;
+ }
+ host->id = rc;
+
+ /* Suffix the unique ID and set djtag hostname */
+ dev_set_name(&host->dev, "djtag-host-%d", host->id);
+
+ rc = device_register(&host->dev);
+ if (rc < 0) {
+ dev_err(&host->dev, "add_host dev register failed, rc=%d\n",
+ rc);
+ idr_remove(&djtag_hosts_idr, host->id);
+ return rc;
+ }
+
+ djtag_register_devices(host);
+
+ return 0;
+}
+
+static int djtag_host_probe(struct platform_device *pdev)
+{
+ struct device *dev = &pdev->dev;
+ struct hisi_djtag_host *host;
+ const struct of_device_id *of_id;
+ struct resource *res;
+ int rc;
+
+ of_id = of_match_device(djtag_of_match, dev);
+ if (!of_id)
+ return -EINVAL;
+
+ host = kzalloc(sizeof(*host), GFP_KERNEL);
+ if (!host)
+ return -ENOMEM;
+
+ host->of_node = dev->of_node;
+ host->djtag_readwrite = of_id->data;
+ spin_lock_init(&host->lock);
+
+ INIT_LIST_HEAD(&host->client_list);
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ if (!res) {
+ dev_err(&pdev->dev, "No reg resorces!\n");
+ kfree(host);
+ return -EINVAL;
+ }
+
+ if (!resource_size(res)) {
+ dev_err(&pdev->dev, "Zero reg entry!\n");
+ kfree(host);
+ return -EINVAL;
+ }
+
+ host->sysctl_reg_map = devm_ioremap_resource(dev, res);
+ if (IS_ERR(host->sysctl_reg_map)) {
+ dev_warn(dev, "Unable to map sysctl registers.\n");
+ kfree(host);
+ return -EINVAL;
+ }
+
+ platform_set_drvdata(pdev, host);
+
+ rc = hisi_djtag_add_host(host);
+ if (rc) {
+ dev_err(dev, "add host failed, rc=%d\n", rc);
+ kfree(host);
+ return rc;
+ }
+
+ return 0;
+}
+
+static int djtag_host_remove(struct platform_device *pdev)
+{
+ struct hisi_djtag_host *host;
+ struct hisi_djtag_client *client, *tmp;
+ struct list_head *client_list;
+
+ host = platform_get_drvdata(pdev);
+ client_list = &host->client_list;
+
+ list_for_each_entry_safe(client, tmp, client_list, next) {
+ list_del(&client->next);
+ kfree(client);
+ }
+
+ device_unregister(&host->dev);
+ idr_remove(&djtag_hosts_idr, host->id);
+ kfree(host);
+
+ return 0;
+}
+
+static struct platform_driver djtag_dev_driver = {
+ .driver = {
+ .name = "hisi-djtag",
+ .of_match_table = djtag_of_match,
+ },
+ .probe = djtag_host_probe,
+ .remove = djtag_host_remove,
+};
+module_platform_driver(djtag_dev_driver);
+
+int hisi_djtag_register_driver(struct module *owner,
+ struct hisi_djtag_driver *driver)
+{
+ int rc;
+
+ driver->driver.owner = owner;
+ driver->driver.bus = &hisi_djtag_bus;
+
+ rc = driver_register(&driver->driver);
+ if (rc < 0)
+ pr_err("%s register failed, rc=%d\n", __func__, rc);
+
+ return rc;
+}
+
+void hisi_djtag_unregister_driver(struct hisi_djtag_driver *driver)
+{
+ driver->driver.bus = &hisi_djtag_bus;
+ driver_unregister(&driver->driver);
+}
+
+static int __init hisi_djtag_init(void)
+{
+ int rc;
+
+ rc = bus_register(&hisi_djtag_bus);
+ if (rc) {
+ pr_err("hisi djtag init failed, rc=%d\n", rc);
+ return rc;
+ }
+
+ return 0;
+}
+module_init(hisi_djtag_init);
+
+static void __exit hisi_djtag_exit(void)
+{
+ bus_unregister(&hisi_djtag_bus);
+}
+module_exit(hisi_djtag_exit);
+
+MODULE_DESCRIPTION("Hisilicon djtag driver");
+MODULE_LICENSE("GPL");
+MODULE_VERSION("1.0");
diff --git a/include/linux/soc/hisilicon/djtag.h b/include/linux/soc/hisilicon/djtag.h
new file mode 100644
index 0000000..aae3bca
--- /dev/null
+++ b/include/linux/soc/hisilicon/djtag.h
@@ -0,0 +1,38 @@
+/*
+ * Driver for Hisilicon djtag r/w via System Controller.
+ *
+ * Copyright (C) 2016-2017 Hisilicon Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __HISI_DJTAG_H
+#define __HISI_DJTAG_H
+
+#define DJTAG_CLIENT_NAME_LEN 24
+
+struct hisi_djtag_client {
+ struct hisi_djtag_host *host;
+ struct list_head next;
+ char name[DJTAG_CLIENT_NAME_LEN];
+ struct device dev;
+};
+
+struct hisi_djtag_driver {
+ struct device_driver driver;
+ int (*probe)(struct hisi_djtag_client *);
+ int (*remove)(struct hisi_djtag_client *);
+};
+
+extern struct bus_type hisi_djtag_bus;
+
+int hisi_djtag_register_driver(struct module *owner,
+ struct hisi_djtag_driver *driver);
+void hisi_djtag_unregister_driver(struct hisi_djtag_driver *driver);
+int hisi_djtag_readl(struct hisi_djtag_client *client, u32 offset,
+ u32 mod_sel, int chain_id, u32 *val);
+int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset,
+ u32 mod_sel, u32 mod_mask, u32 val);
+#endif /* __HISI_DJTAG_H */
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-02 15:42 ` [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver Anurup M
@ 2016-11-03 1:59 ` kbuild test robot
2016-11-07 13:26 ` Arnd Bergmann
1 sibling, 0 replies; 33+ messages in thread
From: kbuild test robot @ 2016-11-03 1:59 UTC (permalink / raw)
To: Anurup M
Cc: kbuild-all, anurup.m, linux-arm-kernel, linux-kernel,
will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
[-- Attachment #1: Type: text/plain, Size: 1455 bytes --]
Hi Tan,
[auto build test ERROR on linus/master]
[also build test ERROR on v4.9-rc3 next-20161028]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
[Suggest to use git(>=2.9.0) format-patch --base=<commit> (or --base=auto for convenience) to record what (public, well-known) commit your patch series was built on]
[Check https://git-scm.com/docs/git-format-patch for more information]
url: https://github.com/0day-ci/linux/commits/Anurup-M/perf-arm64-Support-for-Hisilicon-SoC-Hardware-event-counters/20161102-235009
config: arm-allmodconfig (attached as .config)
compiler: arm-linux-gnueabi-gcc (Debian 6.1.1-9) 6.1.1 20160705
reproduce:
wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
make.cross ARCH=arm
All errors (new ones prefixed by >>):
drivers/built-in.o: In function `djtag_readwrite_v2':
>> binder.c:(.text+0x10cd64): undefined reference to `__const_udelay'
binder.c:(.text+0x10cd98): undefined reference to `__const_udelay'
drivers/built-in.o: In function `djtag_readwrite_v1':
binder.c:(.text+0x10cf10): undefined reference to `__const_udelay'
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 59446 bytes --]
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-02 15:42 ` [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver Anurup M
2016-11-03 1:59 ` kbuild test robot
@ 2016-11-07 13:26 ` Arnd Bergmann
2016-11-07 14:15 ` John Garry
2016-11-08 7:02 ` Tan Xiaojun
1 sibling, 2 replies; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-07 13:26 UTC (permalink / raw)
To: linux-arm-kernel
Cc: Anurup M, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, john.garry, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
> From: Tan Xiaojun <tanxiaojun@huawei.com>
>
> The Hisilicon Djtag is an independent component which connects
> with some other components in the SoC by Debug Bus. This driver
> can be configured to access the registers of connecting components
> (like L3 cache) during real time debugging.
The formatting of the text seems odd, please remove the leading spaces.
> drivers/soc/Kconfig | 1 +
> drivers/soc/Makefile | 1 +
> drivers/soc/hisilicon/Kconfig | 12 +
> drivers/soc/hisilicon/Makefile | 1 +
> drivers/soc/hisilicon/djtag.c | 639 ++++++++++++++++++++++++++++++++++++
> include/linux/soc/hisilicon/djtag.h | 38 +++
Do you expect other drivers to be added that reference this interface?
If not, or if you are unsure, just put all of it under drivers/perf
so we don't introduce a global API that has only one user.
> +
> +#include <linux/bitops.h>
> +#include <linux/init.h>
> +#include <linux/list.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/of_address.h>
> +#include <linux/of_device.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +#include <linux/spinlock.h>
> +
> +#include <asm-generic/delay.h>
Never include files from asm-generic directly except from
an architecture specific asm/*.h header file.
> +DEFINE_IDR(djtag_hosts_idr);
make this static
> +static void djtag_read32_relaxed(void __iomem *regs_base, u32 off, u32 *value)
> +{
> + void __iomem *reg_addr = regs_base + off;
> +
> + *value = readl_relaxed(reg_addr);
> +}
> +
> +static void djtag_write32(void __iomem *regs_base, u32 off, u32 val)
> +{
> + void __iomem *reg_addr = regs_base + off;
> +
> + writel(val, reg_addr);
> +}
This looks like an odd combination of interfaces.
Why can the reads be "relaxed" when the writes can not?
Generally speaking, I'd advise to always use non-relaxed accessors
unless there is a strong performance reason, and in that case there
should be a comment explaining the use at each of the callers
of a relaxed accessor.
> + /* ensure the djtag operation is done */
> + do {
> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
> +
> + if (!(rd & DJTAG_MSTR_START_EN_EX))
> + break;
> +
> + udelay(1);
> + } while (timeout--);
This one is obviously not performance critical at all, so use a non-relaxed
accessor. Same for the other two in this function.
Are these functions ever called from atomic context? If yes, please document
from what context they can be called, otherwise please consider changing
the udelay calls into sleeping waits.
> +int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
> + u32 mod_mask, u32 val)
> +{
> + void __iomem *reg_map = client->host->sysctl_reg_map;
> + unsigned long flags;
> + int ret = 0;
> +
> + spin_lock_irqsave(&client->host->lock, flags);
> + ret = client->host->djtag_readwrite(reg_map, offset, mod_sel, mod_mask,
> + true, val, 0, NULL);
> + if (ret)
> + pr_err("djtag_writel: error! ret=%d\n", ret);
> + spin_unlock_irqrestore(&client->host->lock, flags);
> +
> + return ret;
> +}
> +EXPORT_SYMBOL_GPL(hisi_djtag_writel);
That would of course imply changing the spinlock to a mutex here as well.
> +static const struct of_device_id djtag_of_match[] = {
> + /* for hip05(D02) cpu die */
> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
> + .data = (void *)djtag_readwrite_v1 },
> + /* for hip05(D02) io die */
> + { .compatible = "hisilicon,hip05-io-djtag-v1",
> + .data = (void *)djtag_readwrite_v1 },
> + /* for hip06(D03) cpu die */
> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
> + .data = (void *)djtag_readwrite_v1 },
> + /* for hip06(D03) io die */
> + { .compatible = "hisilicon,hip06-io-djtag-v2",
> + .data = (void *)djtag_readwrite_v2 },
> + /* for hip07(D05) cpu die */
> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
> + .data = (void *)djtag_readwrite_v2 },
> + /* for hip07(D05) io die */
> + { .compatible = "hisilicon,hip07-io-djtag-v2",
> + .data = (void *)djtag_readwrite_v2 },
> + {},
> +};
If these are backwards compatible, just mark them as compatible in DT,
e.g. hip06 can use
compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
so you can tell the difference if you need to, but the driver only has to
list the oldest one here.
What is the difference between the cpu and io djtag interfaces?
I think you can also drop the '(void *)'.
> +static void djtag_register_devices(struct hisi_djtag_host *host)
> +{
> + struct device_node *node;
> + struct hisi_djtag_client *client;
> +
> + if (!host->of_node)
> + return;
> +
> + for_each_available_child_of_node(host->of_node, node) {
> + if (of_node_test_and_set_flag(node, OF_POPULATED))
> + continue;
> + client = hisi_djtag_of_register_device(host, node);
> + list_add(&client->next, &host->client_list);
> + }
> +}
Can you explain your thoughts behind creating a new bus type
and adding the child devices manually rather than using
platform_device structures with of_platform_populate()?
Do you expect to see other implementations of this bus type
with incompatible bus drivers?
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-07 13:26 ` Arnd Bergmann
@ 2016-11-07 14:15 ` John Garry
2016-11-07 20:08 ` Arnd Bergmann
2016-11-08 7:02 ` Tan Xiaojun
1 sibling, 1 reply; 33+ messages in thread
From: John Garry @ 2016-11-07 14:15 UTC (permalink / raw)
To: Arnd Bergmann, linux-arm-kernel
Cc: Anurup M, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, will.deacon, linuxarm, xuwei5, zhangshaokun,
sanil.kumar, tanxiaojun, shiju.jose
On 07/11/2016 13:26, Arnd Bergmann wrote:
> On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
>> From: Tan Xiaojun <tanxiaojun@huawei.com>
>>
>> The Hisilicon Djtag is an independent component which connects
>> with some other components in the SoC by Debug Bus. This driver
>> can be configured to access the registers of connecting components
>> (like L3 cache) during real time debugging.
>
> The formatting of the text seems odd, please remove the leading spaces.
>
My only response is at the bottom (I will not snip the code for early
referencing convenience).
Anurup/Tan can respond to the rest.
>> drivers/soc/Kconfig | 1 +
>> drivers/soc/Makefile | 1 +
>> drivers/soc/hisilicon/Kconfig | 12 +
>> drivers/soc/hisilicon/Makefile | 1 +
>> drivers/soc/hisilicon/djtag.c | 639 ++++++++++++++++++++++++++++++++++++
>> include/linux/soc/hisilicon/djtag.h | 38 +++
>
> Do you expect other drivers to be added that reference this interface?
> If not, or if you are unsure, just put all of it under drivers/perf
> so we don't introduce a global API that has only one user.
>
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/init.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/of_address.h>
>> +#include <linux/of_device.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/slab.h>
>> +#include <linux/spinlock.h>
>> +
>> +#include <asm-generic/delay.h>
>
> Never include files from asm-generic directly except from
> an architecture specific asm/*.h header file.
>
>
>> +DEFINE_IDR(djtag_hosts_idr);
>
> make this static
>
>> +static void djtag_read32_relaxed(void __iomem *regs_base, u32 off, u32 *value)
>> +{
>> + void __iomem *reg_addr = regs_base + off;
>> +
>> + *value = readl_relaxed(reg_addr);
>> +}
>> +
>> +static void djtag_write32(void __iomem *regs_base, u32 off, u32 val)
>> +{
>> + void __iomem *reg_addr = regs_base + off;
>> +
>> + writel(val, reg_addr);
>> +}
>
> This looks like an odd combination of interfaces.
> Why can the reads be "relaxed" when the writes can not?
>
> Generally speaking, I'd advise to always use non-relaxed accessors
> unless there is a strong performance reason, and in that case there
> should be a comment explaining the use at each of the callers
> of a relaxed accessor.
>
>> + /* ensure the djtag operation is done */
>> + do {
>> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
>> +
>> + if (!(rd & DJTAG_MSTR_START_EN_EX))
>> + break;
>> +
>> + udelay(1);
>> + } while (timeout--);
>
> This one is obviously not performance critical at all, so use a non-relaxed
> accessor. Same for the other two in this function.
>
> Are these functions ever called from atomic context? If yes, please document
> from what context they can be called, otherwise please consider changing
> the udelay calls into sleeping waits.
>
>> +int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
>> + u32 mod_mask, u32 val)
>> +{
>> + void __iomem *reg_map = client->host->sysctl_reg_map;
>> + unsigned long flags;
>> + int ret = 0;
>> +
>> + spin_lock_irqsave(&client->host->lock, flags);
>> + ret = client->host->djtag_readwrite(reg_map, offset, mod_sel, mod_mask,
>> + true, val, 0, NULL);
>> + if (ret)
>> + pr_err("djtag_writel: error! ret=%d\n", ret);
>> + spin_unlock_irqrestore(&client->host->lock, flags);
>> +
>> + return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(hisi_djtag_writel);
>
> That would of course imply changing the spinlock to a mutex here as well.
>
>> +static const struct of_device_id djtag_of_match[] = {
>> + /* for hip05(D02) cpu die */
>> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip05(D02) io die */
>> + { .compatible = "hisilicon,hip05-io-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip06(D03) cpu die */
>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip06(D03) io die */
>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + /* for hip07(D05) cpu die */
>> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + /* for hip07(D05) io die */
>> + { .compatible = "hisilicon,hip07-io-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + {},
>> +};
>
> If these are backwards compatible, just mark them as compatible in DT,
> e.g. hip06 can use
>
> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
>
> so you can tell the difference if you need to, but the driver only has to
> list the oldest one here.
>
> What is the difference between the cpu and io djtag interfaces?
>
> I think you can also drop the '(void *)'.
>
>> +static void djtag_register_devices(struct hisi_djtag_host *host)
>> +{
>> + struct device_node *node;
>> + struct hisi_djtag_client *client;
>> +
>> + if (!host->of_node)
>> + return;
>> +
>> + for_each_available_child_of_node(host->of_node, node) {
>> + if (of_node_test_and_set_flag(node, OF_POPULATED))
>> + continue;
>> + client = hisi_djtag_of_register_device(host, node);
>> + list_add(&client->next, &host->client_list);
>> + }
>> +}
>
> Can you explain your thoughts behind creating a new bus type
> and adding the child devices manually rather than using
> platform_device structures with of_platform_populate()?
>
> Do you expect to see other implementations of this bus type
> with incompatible bus drivers?
>
> Arnd
>
Hi Arnd,
The new bus type tries to model the djtag in a similar way to I2C/USB
driver arch, where we have a host bus adapter and child devices attached
to the bus. The child devices are bus driver devices and have bus
addresses. We think of the djtag as a separate bus, so we are modelling
it as such.
The bus driver offers a simple host interface for clients to read/write
to the djtag bus: bus accesses are hidden from the client, the host
drives the bus.
> Do you expect to see other implementations of this bus type
> with incompatible bus drivers?
Maybe, not sure.
Cheers,
John
>
> .
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-07 14:15 ` John Garry
@ 2016-11-07 20:08 ` Arnd Bergmann
2016-11-08 11:23 ` John Garry
0 siblings, 1 reply; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-07 20:08 UTC (permalink / raw)
To: John Garry
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On Monday, November 7, 2016 2:15:10 PM CET John Garry wrote:
>
> Hi Arnd,
>
> The new bus type tries to model the djtag in a similar way to I2C/USB
> driver arch, where we have a host bus adapter and child devices attached
> to the bus. The child devices are bus driver devices and have bus
> addresses. We think of the djtag as a separate bus, so we are modelling
> it as such.
>
> The bus driver offers a simple host interface for clients to read/write
> to the djtag bus: bus accesses are hidden from the client, the host
> drives the bus.
Ok, in that case we should probably start out by having a bus specific
DT binding, and separating the description from that of the bus master
device.
I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the master
node, and listing the children by reg property. If the address is not
easily expressed as a single integer, use a larger #address-cells value.
Another option that we have previously used was to actually pretend that
a vendor specific bus is an i2c bus and use the i2c probing infrastructure,
but that only makes sense if the software side closely resembles i2c
(this was the case for Allwinner I think, but I have not looked at
your driver in enough detail to know if it is true here as well).
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread
* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-07 20:08 ` Arnd Bergmann
@ 2016-11-08 11:23 ` John Garry
2016-11-08 11:45 ` Arnd Bergmann
0 siblings, 1 reply; 33+ messages in thread
From: John Garry @ 2016-11-08 11:23 UTC (permalink / raw)
To: Arnd Bergmann
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On 07/11/2016 20:08, Arnd Bergmann wrote:
> On Monday, November 7, 2016 2:15:10 PM CET John Garry wrote:
>>
>> Hi Arnd,
>>
>> The new bus type tries to model the djtag in a similar way to I2C/USB
>> driver arch, where we have a host bus adapter and child devices attached
>> to the bus. The child devices are bus driver devices and have bus
>> addresses. We think of the djtag as a separate bus, so we are modelling
>> it as such.
>>
>> The bus driver offers a simple host interface for clients to read/write
>> to the djtag bus: bus accesses are hidden from the client, the host
>> drives the bus.
>
> Ok, in that case we should probably start out by having a bus specific
> DT binding, and separating the description from that of the bus master
> device.
OK
>
> I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the master
> node, and listing the children by reg property. If the address is not
> easily expressed as a single integer, use a larger #address-cells value.
We already have something equivalent to reg in "module-id" (see patch
02/11), which is the slave device bus address; here's a sample:
+ /* For L3 cache PMU */
+ pmul3c0 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x02>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
FYI, "module-id" is our own internal hw nomenclature.
>
> Another option that we have previously used was to actually pretend that
> a vendor specific bus is an i2c bus and use the i2c probing infrastructure,
> but that only makes sense if the software side closely resembles i2c
> (this was the case for Allwinner I think, but I have not looked at
> your driver in enough detail to know if it is true here as well).
>
OK, let me check this. By chance do you remember the specific AllWinner
driver/hw?
Cheers,
John
> Arnd
>
> .
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 11:23 ` John Garry
@ 2016-11-08 11:45 ` Arnd Bergmann
2016-11-08 13:49 ` John Garry
2016-11-08 15:51 ` Anurup M
0 siblings, 2 replies; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-08 11:45 UTC (permalink / raw)
To: John Garry
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On Tuesday, November 8, 2016 11:23:35 AM CET John Garry wrote:
> On 07/11/2016 20:08, Arnd Bergmann wrote:
> > On Monday, November 7, 2016 2:15:10 PM CET John Garry wrote:
> >>
> >> Hi Arnd,
> >>
> >> The new bus type tries to model the djtag in a similar way to I2C/USB
> >> driver arch, where we have a host bus adapter and child devices attached
> >> to the bus. The child devices are bus driver devices and have bus
> >> addresses. We think of the djtag as a separate bus, so we are modelling
> >> it as such.
> >>
> >> The bus driver offers a simple host interface for clients to read/write
> >> to the djtag bus: bus accesses are hidden from the client, the host
> >> drives the bus.
> >
> > Ok, in that case we should probably start out by having a bus specific
> > DT binding, and separating the description from that of the bus master
> > device.
>
> OK
>
> >
> > I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the master
> > node, and listing the children by reg property. If the address is not
> > easily expressed as a single integer, use a larger #address-cells value.
>
> We already have something equivalent to reg in "module-id" (see patch
> 02/11), which is the slave device bus address; here's a sample:
> + /* For L3 cache PMU */
> + pmul3c0 {
> + compatible = "hisilicon,hisi-pmu-l3c-v1";
> + scl-id = <0x02>;
> + num-events = <0x16>;
> + num-counters = <0x08>;
> + module-id = <0x04>;
> + num-banks = <0x04>;
> + cfgen-map = <0x02 0x04 0x01 0x08>;
> + counter-reg = <0x170>;
> + evctrl-reg = <0x04>;
> + event-en = <0x1000000>;
> + evtype-reg = <0x140>;
> + };
>
> FYI, "module-id" is our own internal hw nomenclature.
Yes, that was my interpretation as well. Please use the standard
"reg" property for this then.
> > Another option that we have previously used was to actually pretend that
> > a vendor specific bus is an i2c bus and use the i2c probing infrastructure,
> > but that only makes sense if the software side closely resembles i2c
> > (this was the case for Allwinner I think, but I have not looked at
> > your driver in enough detail to know if it is true here as well).
> >
>
> OK, let me check this. By chance do you remember the specific AllWinner
> driver/hw?
drivers/i2c/busses/i2c-sun6i-p2wi.c
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 11:45 ` Arnd Bergmann
@ 2016-11-08 13:49 ` John Garry
2016-11-08 15:10 ` Arnd Bergmann
2016-11-08 15:51 ` Anurup M
1 sibling, 1 reply; 33+ messages in thread
From: John Garry @ 2016-11-08 13:49 UTC (permalink / raw)
To: Arnd Bergmann
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On 08/11/2016 11:45, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 11:23:35 AM CET John Garry wrote:
>> On 07/11/2016 20:08, Arnd Bergmann wrote:
>>> On Monday, November 7, 2016 2:15:10 PM CET John Garry wrote:
>>>>
>>>> Hi Arnd,
>>>>
>>>> The new bus type tries to model the djtag in a similar way to I2C/USB
>>>> driver arch, where we have a host bus adapter and child devices attached
>>>> to the bus. The child devices are bus driver devices and have bus
>>>> addresses. We think of the djtag as a separate bus, so we are modelling
>>>> it as such.
>>>>
>>>> The bus driver offers a simple host interface for clients to read/write
>>>> to the djtag bus: bus accesses are hidden from the client, the host
>>>> drives the bus.
>>>
>>> Ok, in that case we should probably start out by having a bus specific
>>> DT binding, and separating the description from that of the bus master
>>> device.
>>
>> OK
>>
>>>
>>> I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the master
>>> node, and listing the children by reg property. If the address is not
>>> easily expressed as a single integer, use a larger #address-cells value.
>>
>> We already have something equivalent to reg in "module-id" (see patch
>> 02/11), which is the slave device bus address; here's a sample:
>> + /* For L3 cache PMU */
>> + pmul3c0 {
>> + compatible = "hisilicon,hisi-pmu-l3c-v1";
>> + scl-id = <0x02>;
>> + num-events = <0x16>;
>> + num-counters = <0x08>;
>> + module-id = <0x04>;
>> + num-banks = <0x04>;
>> + cfgen-map = <0x02 0x04 0x01 0x08>;
>> + counter-reg = <0x170>;
>> + evctrl-reg = <0x04>;
>> + event-en = <0x1000000>;
>> + evtype-reg = <0x140>;
>> + };
>>
>> FYI, "module-id" is our own internal hw nomenclature.
>
> Yes, that was my interpretation as well. Please use the standard
> "reg" property for this then.
>
>>> Another option that we have previously used was to actually pretend that
>>> a vendor specific bus is an i2c bus and use the i2c probing infrastructure,
>>> but that only makes sense if the software side closely resembles i2c
>>> (this was the case for Allwinner I think, but I have not looked at
>>> your driver in enough detail to know if it is true here as well).
>>>
>>
>> OK, let me check this. By chance do you remember the specific AllWinner
>> driver/hw?
>
> drivers/i2c/busses/i2c-sun6i-p2wi.c
Hi Arnd,
Thanks for the reference.
I think the i2c interface doesn't fully satisfy our requirements as we
need more than just a slave bus address when accessing the slave device
(which I think is what i2c uses). We also need to pass "offset" and
"mod_mask" arguments to the djtag adapter to access specific registers
in the slave device.
Cheers,
John
>
> Arnd
>
> .
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 13:49 ` John Garry
@ 2016-11-08 15:10 ` Arnd Bergmann
2016-11-08 15:17 ` John Garry
2016-11-09 10:44 ` Anurup M
0 siblings, 2 replies; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-08 15:10 UTC (permalink / raw)
To: John Garry
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On Tuesday, November 8, 2016 1:49:43 PM CET John Garry wrote:
>
> Hi Arnd,
>
> Thanks for the reference.
>
> I think the i2c interface doesn't fully satisfy our requirements as we
> need more than just a slave bus address when accessing the slave device
> (which I think is what i2c uses). We also need to pass "offset" and
> "mod_mask" arguments to the djtag adapter to access specific registers
> in the slave device.
Ok. Are those values constant per device, or maybe a range? We may want to
include those in the reg property as well then.
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread
* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 15:10 ` Arnd Bergmann
@ 2016-11-08 15:17 ` John Garry
2016-11-09 10:44 ` Anurup M
1 sibling, 0 replies; 33+ messages in thread
From: John Garry @ 2016-11-08 15:17 UTC (permalink / raw)
To: Arnd Bergmann
Cc: linux-arm-kernel, Anurup M, anurup.m, linux-kernel, mark.rutland,
shyju.pv, gabriele.paoloni, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, tanxiaojun, shiju.jose
On 08/11/2016 15:10, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 1:49:43 PM CET John Garry wrote:
>>
>> Hi Arnd,
>>
>> Thanks for the reference.
>>
>> I think the i2c interface doesn't fully satisfy our requirements as we
>> need more than just a slave bus address when accessing the slave device
>> (which I think is what i2c uses). We also need to pass "offset" and
>> "mod_mask" arguments to the djtag adapter to access specific registers
>> in the slave device.
>
> Ok. Are those values constant per device, or maybe a range? We may want to
> include those in the reg property as well then.
>
> Arnd
>
Hi Arnd,
I'm not sure, I'll defer to Tan/Anurup.
Cheers,
John
>
> .
>
^ permalink raw reply [flat|nested] 33+ messages in thread
* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 15:10 ` Arnd Bergmann
2016-11-08 15:17 ` John Garry
@ 2016-11-09 10:44 ` Anurup M
1 sibling, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-09 10:44 UTC (permalink / raw)
To: Arnd Bergmann, John Garry
Cc: linux-arm-kernel, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, will.deacon, linuxarm, xuwei5, zhangshaokun,
sanil.kumar, tanxiaojun, shiju.jose
On Tuesday 08 November 2016 08:40 PM, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 1:49:43 PM CET John Garry wrote:
>> Hi Arnd,
>>
>> Thanks for the reference.
>>
>> I think the i2c interface doesn't fully satisfy our requirements as we
>> need more than just a slave bus address when accessing the slave device
>> (which I think is what i2c uses). We also need to pass "offset" and
>> "mod_mask" arguments to the djtag adapter to access specific registers
>> in the slave device.
> Ok. Are those values constant per device, or maybe a range? We may want to
> include those in the reg property as well then.
>
> Arnd
>
Hi Arnd,
The "mod_mask" is to select the sub-module within a module. This
parameter is
used for djtag write operation.
In the case of L3 cache, this will select the L3 cache bank. 0xFFFF
select all banks.
This value will change based on the L3 cache bank to be written to. I
think this value
can be in the driver itself.
For djtag read operation, the "mod_mask" is ignored. instead the input
parameter
"chain_id" is used. this will identify the sub-module or bank.
For djtag-v1, the "chain_id" is different for each L3 cache bank, But in
the case of
djtag-v2 the "chain_id" is fixed and the value is 0 as In djtag-v2 there
is separate
"module-id" for each sub-module.
The "offset" is the register offset and this value is a range for a module.
Thanks,
Anurup
^ permalink raw reply [flat|nested] 33+ messages in thread
* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 11:45 ` Arnd Bergmann
2016-11-08 13:49 ` John Garry
@ 2016-11-08 15:51 ` Anurup M
2016-11-09 9:06 ` John Garry
1 sibling, 1 reply; 33+ messages in thread
From: Anurup M @ 2016-11-08 15:51 UTC (permalink / raw)
To: Arnd Bergmann, John Garry
Cc: linux-arm-kernel, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, will.deacon, linuxarm, xuwei5, zhangshaokun,
sanil.kumar, tanxiaojun, shiju.jose
On Tuesday 08 November 2016 05:15 PM, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 11:23:35 AM CET John Garry wrote:
>> On 07/11/2016 20:08, Arnd Bergmann wrote:
>>> On Monday, November 7, 2016 2:15:10 PM CET John Garry wrote:
>>>> Hi Arnd,
>>>>
>>>> The new bus type tries to model the djtag in a similar way to I2C/USB
>>>> driver arch, where we have a host bus adapter and child devices attached
>>>> to the bus. The child devices are bus driver devices and have bus
>>>> addresses. We think of the djtag as a separate bus, so we are modelling
>>>> it as such.
>>>>
>>>> The bus driver offers a simple host interface for clients to read/write
>>>> to the djtag bus: bus accesses are hidden from the client, the host
>>>> drives the bus.
>>> Ok, in that case we should probably start out by having a bus specific
>>> DT binding, and separating the description from that of the bus master
>>> device.
>> OK
>>
>>> I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the master
>>> node, and listing the children by reg property. If the address is not
>>> easily expressed as a single integer, use a larger #address-cells value.
>> We already have something equivalent to reg in "module-id" (see patch
>> 02/11), which is the slave device bus address; here's a sample:
>> + /* For L3 cache PMU */
>> + pmul3c0 {
>> + compatible = "hisilicon,hisi-pmu-l3c-v1";
>> + scl-id = <0x02>;
>> + num-events = <0x16>;
>> + num-counters = <0x08>;
>> + module-id = <0x04>;
>> + num-banks = <0x04>;
>> + cfgen-map = <0x02 0x04 0x01 0x08>;
>> + counter-reg = <0x170>;
>> + evctrl-reg = <0x04>;
>> + event-en = <0x1000000>;
>> + evtype-reg = <0x140>;
>> + };
>>
>> FYI, "module-id" is our own internal hw nomenclature.
> Yes, that was my interpretation as well. Please use the standard
> "reg" property for this then.
Hi Arnd,
Firstly my apologies for a mistake in the bindings example in ([PATCH
02/11 ..]).
The module-id property is a list as defined in the PMU bindings patch
([PATCH v1 05/11] dt-bindings .. <https://lkml.org/lkml/2016/11/2/323>).
+ djtag0: djtag@0 {
+ compatible = "hisilicon,hip05-cpu-djtag-v1";
+ pmul3c0 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x02>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04 0x04 0x04 0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
The L3 cache in hip05/06/07 chips consist of 4 banks (each bank has PMU
registers).
In hip05/06 all L3 cache banks are identified with same module-id.
module-id = <0x04 0x04 0x04 0x04>;
But in the case hip07 chip(djtag v2), each L3 cache bank has different
module-id
module-id = <0x01 0x02 0x03 0x04>;
So in this case Please share your opinion on how to model it.
Some more detail of L3 cache PMU.
------------------------------------------------
The hip05/06/07 chips consists of a multiple Super CPU cluster (16 CPU
cores). we call it SCCL.
The L3 cache( 4 banks) is shared by all CPU cores in a SCCL.
Each L3 cache bank has PMU registers. We always take the sum of the
counters to show in perf.
Taking individual L3 cache count is not meaningful as there is no
mapping of CPU cores to individual
L3 cache banks.
Please share your suggestion.
Thanks,
Anurup
>>> Another option that we have previously used was to actually pretend that
>>> a vendor specific bus is an i2c bus and use the i2c probing infrastructure,
>>> but that only makes sense if the software side closely resembles i2c
>>> (this was the case for Allwinner I think, but I have not looked at
>>> your driver in enough detail to know if it is true here as well).
>>>
>> OK, let me check this. By chance do you remember the specific AllWinner
>> driver/hw?
> drivers/i2c/busses/i2c-sun6i-p2wi.c
>
> Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 15:51 ` Anurup M
@ 2016-11-09 9:06 ` John Garry
2016-11-11 10:35 ` Anurup M
0 siblings, 1 reply; 33+ messages in thread
From: John Garry @ 2016-11-09 9:06 UTC (permalink / raw)
To: Anurup M, Arnd Bergmann
Cc: linux-arm-kernel, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, will.deacon, linuxarm, xuwei5, zhangshaokun,
sanil.kumar, tanxiaojun, shiju.jose
>>>> I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the
>>>> master
>>>> node, and listing the children by reg property. If the address is not
>>>> easily expressed as a single integer, use a larger #address-cells
>>>> value.
>>> We already have something equivalent to reg in "module-id" (see patch
>>> 02/11), which is the slave device bus address; here's a sample:
>>> + /* For L3 cache PMU */
>>> + pmul3c0 {
>>> + compatible = "hisilicon,hisi-pmu-l3c-v1";
>>> + scl-id = <0x02>;
>>> + num-events = <0x16>;
>>> + num-counters = <0x08>;
>>> + module-id = <0x04>;
>>> + num-banks = <0x04>;
>>> + cfgen-map = <0x02 0x04 0x01 0x08>;
>>> + counter-reg = <0x170>;
>>> + evctrl-reg = <0x04>;
>>> + event-en = <0x1000000>;
>>> + evtype-reg = <0x140>;
>>> + };
>>>
>>> FYI, "module-id" is our own internal hw nomenclature.
>> Yes, that was my interpretation as well. Please use the standard
>> "reg" property for this then.
> Hi Arnd,
>
> Firstly my apologies for a mistake in the bindings example in ([PATCH
> 02/11 ..]).
> The module-id property is a list as defined in the PMU bindings patch
> ([PATCH v1 05/11] dt-bindings .. <https://lkml.org/lkml/2016/11/2/323>).
>
> + djtag0: djtag@0 {
> + compatible = "hisilicon,hip05-cpu-djtag-v1";
> + pmul3c0 {
> + compatible = "hisilicon,hisi-pmu-l3c-v1";
> + scl-id = <0x02>;
> + num-events = <0x16>;
> + num-counters = <0x08>;
> + module-id = <0x04 0x04 0x04 0x04>;
> + num-banks = <0x04>;
> + cfgen-map = <0x02 0x04 0x01 0x08>;
> + counter-reg = <0x170>;
> + evctrl-reg = <0x04>;
> + event-en = <0x1000000>;
> + evtype-reg = <0x140>;
> + };
>
>
> The L3 cache in hip05/06/07 chips consist of 4 banks (each bank has PMU
> registers).
>
> In hip05/06 all L3 cache banks are identified with same module-id.
> module-id = <0x04 0x04 0x04 0x04>;
>
> But in the case hip07 chip(djtag v2), each L3 cache bank has different
> module-id
> module-id = <0x01 0x02 0x03 0x04>;
>
> So in this case Please share your opinion on how to model it.
>
My suggestion is to have a single PMU per module, whether that is 4
banks or 1 bank per module, as this makes the driver simpler.
I think you mentioned that a separate PMU per bank does not make much
sense, and you would rather treat all banks as a single bank and
aggregrate their perf statstics under a single PMU: Can you just use a
script in userspace which can do this aggregration work if you have
separate PMUs?
Maybe perf guys have a view on this also.
John
> Some more detail of L3 cache PMU.
> ------------------------------------------------
> The hip05/06/07 chips consists of a multiple Super CPU cluster (16 CPU
> cores). we call it SCCL.
> The L3 cache( 4 banks) is shared by all CPU cores in a SCCL.
> Each L3 cache bank has PMU registers. We always take the sum of the
> counters to show in perf.
> Taking individual L3 cache count is not meaningful as there is no
> mapping of CPU cores to individual
> L3 cache banks.
>
> Please share your suggestion.
>
> Thanks,
> Anurup
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-09 9:06 ` John Garry
@ 2016-11-11 10:35 ` Anurup M
0 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-11 10:35 UTC (permalink / raw)
To: John Garry, Arnd Bergmann
Cc: linux-arm-kernel, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, will.deacon, linuxarm, xuwei5, zhangshaokun,
sanil.kumar, tanxiaojun, shiju.jose
On Wednesday 09 November 2016 02:36 PM, John Garry wrote:
>
>>>>> I'd suggest requiring #address-cells=<1> and #size-cells=<0> in the
>>>>> master
>>>>> node, and listing the children by reg property. If the address is not
>>>>> easily expressed as a single integer, use a larger #address-cells
>>>>> value.
>>>> We already have something equivalent to reg in "module-id" (see patch
>>>> 02/11), which is the slave device bus address; here's a sample:
>>>> + /* For L3 cache PMU */
>>>> + pmul3c0 {
>>>> + compatible = "hisilicon,hisi-pmu-l3c-v1";
>>>> + scl-id = <0x02>;
>>>> + num-events = <0x16>;
>>>> + num-counters = <0x08>;
>>>> + module-id = <0x04>;
>>>> + num-banks = <0x04>;
>>>> + cfgen-map = <0x02 0x04 0x01 0x08>;
>>>> + counter-reg = <0x170>;
>>>> + evctrl-reg = <0x04>;
>>>> + event-en = <0x1000000>;
>>>> + evtype-reg = <0x140>;
>>>> + };
>>>>
>>>> FYI, "module-id" is our own internal hw nomenclature.
>>> Yes, that was my interpretation as well. Please use the standard
>>> "reg" property for this then.
>> Hi Arnd,
>>
>> Firstly my apologies for a mistake in the bindings example in ([PATCH
>> 02/11 ..]).
>> The module-id property is a list as defined in the PMU bindings patch
>> ([PATCH v1 05/11] dt-bindings .. <https://lkml.org/lkml/2016/11/2/323>).
>>
>> + djtag0: djtag@0 {
>> + compatible = "hisilicon,hip05-cpu-djtag-v1";
>> + pmul3c0 {
>> + compatible = "hisilicon,hisi-pmu-l3c-v1";
>> + scl-id = <0x02>;
>> + num-events = <0x16>;
>> + num-counters = <0x08>;
>> + module-id = <0x04 0x04 0x04 0x04>;
>> + num-banks = <0x04>;
>> + cfgen-map = <0x02 0x04 0x01 0x08>;
>> + counter-reg = <0x170>;
>> + evctrl-reg = <0x04>;
>> + event-en = <0x1000000>;
>> + evtype-reg = <0x140>;
>> + };
>>
>>
>> The L3 cache in hip05/06/07 chips consist of 4 banks (each bank has PMU
>> registers).
>>
>> In hip05/06 all L3 cache banks are identified with same module-id.
>> module-id = <0x04 0x04 0x04 0x04>;
>>
>> But in the case hip07 chip(djtag v2), each L3 cache bank has different
>> module-id
>> module-id = <0x01 0x02 0x03 0x04>;
>>
>> So in this case Please share your opinion on how to model it.
>>
>
> My suggestion is to have a single PMU per module, whether that is 4
> banks or 1 bank per module, as this makes the driver simpler.
>
> I think you mentioned that a separate PMU per bank does not make much
> sense, and you would rather treat all banks as a single bank and
> aggregrate their perf statstics under a single PMU: Can you just use a
> script in userspace which can do this aggregration work if you have
> separate PMUs?
Hi John,
Mark also suggest the same view.
I have some concerns or doubts in having separate PMU for each L3 cache
bank.
We can discuss it in the same thread [[RESEND PATCH v1 07/11]]
<http://www.spinics.net/lists/arm-kernel/msg541938.html>].
Thanks,
Anurup
>
> Maybe perf guys have a view on this also.
>
> John
>
>> Some more detail of L3 cache PMU.
>> ------------------------------------------------
>> The hip05/06/07 chips consists of a multiple Super CPU cluster (16 CPU
>> cores). we call it SCCL.
>> The L3 cache( 4 banks) is shared by all CPU cores in a SCCL.
>> Each L3 cache bank has PMU registers. We always take the sum of the
>> counters to show in perf.
>> Taking individual L3 cache count is not meaningful as there is no
>> mapping of CPU cores to individual
>> L3 cache banks.
>>
>> Please share your suggestion.
>>
>> Thanks,
>> Anurup
>
^ permalink raw reply [flat|nested] 33+ messages in thread
* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-07 13:26 ` Arnd Bergmann
2016-11-07 14:15 ` John Garry
@ 2016-11-08 7:02 ` Tan Xiaojun
2016-11-08 7:38 ` Anurup M
1 sibling, 1 reply; 33+ messages in thread
From: Tan Xiaojun @ 2016-11-08 7:02 UTC (permalink / raw)
To: Arnd Bergmann, linux-arm-kernel
Cc: Anurup M, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, john.garry, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, shiju.jose
On 2016/11/7 21:26, Arnd Bergmann wrote:
> On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
>> From: Tan Xiaojun <tanxiaojun@huawei.com>
>>
>> The Hisilicon Djtag is an independent component which connects
>> with some other components in the SoC by Debug Bus. This driver
>> can be configured to access the registers of connecting components
>> (like L3 cache) during real time debugging.
>
> The formatting of the text seems odd, please remove the leading spaces.
>
Sorry for that. We will fix it.
>> drivers/soc/Kconfig | 1 +
>> drivers/soc/Makefile | 1 +
>> drivers/soc/hisilicon/Kconfig | 12 +
>> drivers/soc/hisilicon/Makefile | 1 +
>> drivers/soc/hisilicon/djtag.c | 639 ++++++++++++++++++++++++++++++++++++
>> include/linux/soc/hisilicon/djtag.h | 38 +++
>
> Do you expect other drivers to be added that reference this interface?
> If not, or if you are unsure, just put all of it under drivers/perf
> so we don't introduce a global API that has only one user.
>
OK. For now, this suggestion sounds good.
>> +
>> +#include <linux/bitops.h>
>> +#include <linux/init.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/of_address.h>
>> +#include <linux/of_device.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/slab.h>
>> +#include <linux/spinlock.h>
>> +
>> +#include <asm-generic/delay.h>
>
> Never include files from asm-generic directly except from
> an architecture specific asm/*.h header file.
>
>
OK. Sorry for that.
>> +DEFINE_IDR(djtag_hosts_idr);
>
> make this static
>
OK.
>> +static void djtag_read32_relaxed(void __iomem *regs_base, u32 off, u32 *value)
>> +{
>> + void __iomem *reg_addr = regs_base + off;
>> +
>> + *value = readl_relaxed(reg_addr);
>> +}
>> +
>> +static void djtag_write32(void __iomem *regs_base, u32 off, u32 val)
>> +{
>> + void __iomem *reg_addr = regs_base + off;
>> +
>> + writel(val, reg_addr);
>> +}
>
> This looks like an odd combination of interfaces.
> Why can the reads be "relaxed" when the writes can not?
>
> Generally speaking, I'd advise to always use non-relaxed accessors
> unless there is a strong performance reason, and in that case there
> should be a comment explaining the use at each of the callers
> of a relaxed accessor.
>
Yes, it is our mistake.
>> + /* ensure the djtag operation is done */
>> + do {
>> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
>> +
>> + if (!(rd & DJTAG_MSTR_START_EN_EX))
>> + break;
>> +
>> + udelay(1);
>> + } while (timeout--);
>
> This one is obviously not performance critical at all, so use a non-relaxed
> accessor. Same for the other two in this function.
>
> Are these functions ever called from atomic context? If yes, please document
> from what context they can be called, otherwise please consider changing
> the udelay calls into sleeping waits.
>
Yes, this is not reentrant.
>> +int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
>> + u32 mod_mask, u32 val)
>> +{
>> + void __iomem *reg_map = client->host->sysctl_reg_map;
>> + unsigned long flags;
>> + int ret = 0;
>> +
>> + spin_lock_irqsave(&client->host->lock, flags);
>> + ret = client->host->djtag_readwrite(reg_map, offset, mod_sel, mod_mask,
>> + true, val, 0, NULL);
>> + if (ret)
>> + pr_err("djtag_writel: error! ret=%d\n", ret);
>> + spin_unlock_irqrestore(&client->host->lock, flags);
>> +
>> + return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(hisi_djtag_writel);
>
> That would of course imply changing the spinlock to a mutex here as well.
>
>> +static const struct of_device_id djtag_of_match[] = {
>> + /* for hip05(D02) cpu die */
>> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip05(D02) io die */
>> + { .compatible = "hisilicon,hip05-io-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip06(D03) cpu die */
>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip06(D03) io die */
>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + /* for hip07(D05) cpu die */
>> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + /* for hip07(D05) io die */
>> + { .compatible = "hisilicon,hip07-io-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>> + {},
>> +};
>
> If these are backwards compatible, just mark them as compatible in DT,
> e.g. hip06 can use
>
> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
>
> so you can tell the difference if you need to, but the driver only has to
> list the oldest one here.
>
> What is the difference between the cpu and io djtag interfaces?
>
> I think you can also drop the '(void *)'.
>
OK. We will consider it.
Thanks.
Xiaojun.
>> +static void djtag_register_devices(struct hisi_djtag_host *host)
>> +{
>> + struct device_node *node;
>> + struct hisi_djtag_client *client;
>> +
>> + if (!host->of_node)
>> + return;
>> +
>> + for_each_available_child_of_node(host->of_node, node) {
>> + if (of_node_test_and_set_flag(node, OF_POPULATED))
>> + continue;
>> + client = hisi_djtag_of_register_device(host, node);
>> + list_add(&client->next, &host->client_list);
>> + }
>> +}
>
> Can you explain your thoughts behind creating a new bus type
> and adding the child devices manually rather than using
> platform_device structures with of_platform_populate()?
>
> Do you expect to see other implementations of this bus type
> with incompatible bus drivers?
>
> Arnd
>
>
> .
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 7:02 ` Tan Xiaojun
@ 2016-11-08 7:38 ` Anurup M
2016-11-08 11:43 ` Arnd Bergmann
0 siblings, 1 reply; 33+ messages in thread
From: Anurup M @ 2016-11-08 7:38 UTC (permalink / raw)
To: Arnd Bergmann, linux-arm-kernel
Cc: Tan Xiaojun, anurup.m, linux-kernel, mark.rutland, shyju.pv,
gabriele.paoloni, john.garry, will.deacon, linuxarm, xuwei5,
zhangshaokun, sanil.kumar, shiju.jose
On Tuesday 08 November 2016 12:32 PM, Tan Xiaojun wrote:
> On 2016/11/7 21:26, Arnd Bergmann wrote:
>> On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
>>> From: Tan Xiaojun <tanxiaojun@huawei.com>
>>>
>>> The Hisilicon Djtag is an independent component which connects
>>> with some other components in the SoC by Debug Bus. This driver
>>> can be configured to access the registers of connecting components
>>> (like L3 cache) during real time debugging.
>> The formatting of the text seems odd, please remove the leading spaces.
>>
> Sorry for that. We will fix it.
>
>>> drivers/soc/Kconfig | 1 +
>>> drivers/soc/Makefile | 1 +
>>> drivers/soc/hisilicon/Kconfig | 12 +
>>> drivers/soc/hisilicon/Makefile | 1 +
>>> drivers/soc/hisilicon/djtag.c | 639 ++++++++++++++++++++++++++++++++++++
>>> include/linux/soc/hisilicon/djtag.h | 38 +++
>> Do you expect other drivers to be added that reference this interface?
>> If not, or if you are unsure, just put all of it under drivers/perf
>> so we don't introduce a global API that has only one user.
>>
> OK. For now, this suggestion sounds good.
>
>>> +
>>> +#include <linux/bitops.h>
>>> +#include <linux/init.h>
>>> +#include <linux/list.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of.h>
>>> +#include <linux/of_address.h>
>>> +#include <linux/of_device.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/slab.h>
>>> +#include <linux/spinlock.h>
>>> +
>>> +#include <asm-generic/delay.h>
>> Never include files from asm-generic directly except from
>> an architecture specific asm/*.h header file.
>>
>>
> OK. Sorry for that.
>
>>> +DEFINE_IDR(djtag_hosts_idr);
>> make this static
>>
> OK.
>
>>> +static void djtag_read32_relaxed(void __iomem *regs_base, u32 off, u32 *value)
>>> +{
>>> + void __iomem *reg_addr = regs_base + off;
>>> +
>>> + *value = readl_relaxed(reg_addr);
>>> +}
>>> +
>>> +static void djtag_write32(void __iomem *regs_base, u32 off, u32 val)
>>> +{
>>> + void __iomem *reg_addr = regs_base + off;
>>> +
>>> + writel(val, reg_addr);
>>> +}
>> This looks like an odd combination of interfaces.
>> Why can the reads be "relaxed" when the writes can not?
>>
>> Generally speaking, I'd advise to always use non-relaxed accessors
>> unless there is a strong performance reason, and in that case there
>> should be a comment explaining the use at each of the callers
>> of a relaxed accessor.
>>
> Yes, it is our mistake.
Shall use non-relaxed version and make it consistent.
>>> + /* ensure the djtag operation is done */
>>> + do {
>>> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
>>> +
>>> + if (!(rd & DJTAG_MSTR_START_EN_EX))
>>> + break;
>>> +
>>> + udelay(1);
>>> + } while (timeout--);
>> This one is obviously not performance critical at all, so use a non-relaxed
>> accessor. Same for the other two in this function.
>>
>> Are these functions ever called from atomic context? If yes, please document
>> from what context they can be called, otherwise please consider changing
>> the udelay calls into sleeping waits.
>>
> Yes, this is not reentrant.
The read/write functions shall also be called from irq handler (for
handling counter overflow).
So need to use udelay calls. Shall Document it in v2.
>>> +int hisi_djtag_writel(struct hisi_djtag_client *client, u32 offset, u32 mod_sel,
>>> + u32 mod_mask, u32 val)
>>> +{
>>> + void __iomem *reg_map = client->host->sysctl_reg_map;
>>> + unsigned long flags;
>>> + int ret = 0;
>>> +
>>> + spin_lock_irqsave(&client->host->lock, flags);
>>> + ret = client->host->djtag_readwrite(reg_map, offset, mod_sel, mod_mask,
>>> + true, val, 0, NULL);
>>> + if (ret)
>>> + pr_err("djtag_writel: error! ret=%d\n", ret);
>>> + spin_unlock_irqrestore(&client->host->lock, flags);
>>> +
>>> + return ret;
>>> +}
>>> +EXPORT_SYMBOL_GPL(hisi_djtag_writel);
>> That would of course imply changing the spinlock to a mutex here as well.
>>
>>> +static const struct of_device_id djtag_of_match[] = {
>>> + /* for hip05(D02) cpu die */
>>> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
>>> + .data = (void *)djtag_readwrite_v1 },
>>> + /* for hip05(D02) io die */
>>> + { .compatible = "hisilicon,hip05-io-djtag-v1",
>>> + .data = (void *)djtag_readwrite_v1 },
>>> + /* for hip06(D03) cpu die */
>>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
>>> + .data = (void *)djtag_readwrite_v1 },
>>> + /* for hip06(D03) io die */
>>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
>>> + .data = (void *)djtag_readwrite_v2 },
>>> + /* for hip07(D05) cpu die */
>>> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
>>> + .data = (void *)djtag_readwrite_v2 },
>>> + /* for hip07(D05) io die */
>>> + { .compatible = "hisilicon,hip07-io-djtag-v2",
>>> + .data = (void *)djtag_readwrite_v2 },
>>> + {},
>>> +};
>> If these are backwards compatible, just mark them as compatible in DT,
>> e.g. hip06 can use
>>
>> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
>>
>> so you can tell the difference if you need to, but the driver only has to
>> list the oldest one here.
>>
>> What is the difference between the cpu and io djtag interfaces?
On some chips like hip06, the djtag version is different for IO die.
Thanks,
Anurup
>>
>> I think you can also drop the '(void *)'.
>>
> OK. We will consider it.
>
> Thanks.
> Xiaojun.
>
>>> +static void djtag_register_devices(struct hisi_djtag_host *host)
>>> +{
>>> + struct device_node *node;
>>> + struct hisi_djtag_client *client;
>>> +
>>> + if (!host->of_node)
>>> + return;
>>> +
>>> + for_each_available_child_of_node(host->of_node, node) {
>>> + if (of_node_test_and_set_flag(node, OF_POPULATED))
>>> + continue;
>>> + client = hisi_djtag_of_register_device(host, node);
>>> + list_add(&client->next, &host->client_list);
>>> + }
>>> +}
>> Can you explain your thoughts behind creating a new bus type
>> and adding the child devices manually rather than using
>> platform_device structures with of_platform_populate()?
>>
>> Do you expect to see other implementations of this bus type
>> with incompatible bus drivers?
>>
>> Arnd
>>
>>
>> .
>>
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 7:38 ` Anurup M
@ 2016-11-08 11:43 ` Arnd Bergmann
2016-11-08 13:46 ` Anurup M
0 siblings, 1 reply; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-08 11:43 UTC (permalink / raw)
To: Anurup M
Cc: linux-arm-kernel, Tan Xiaojun, anurup.m, linux-kernel,
mark.rutland, shyju.pv, gabriele.paoloni, john.garry,
will.deacon, linuxarm, xuwei5, zhangshaokun, sanil.kumar,
shiju.jose
On Tuesday, November 8, 2016 1:08:31 PM CET Anurup M wrote:
> On Tuesday 08 November 2016 12:32 PM, Tan Xiaojun wrote:
> > On 2016/11/7 21:26, Arnd Bergmann wrote:
> >> On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
> >>> From: Tan Xiaojun <tanxiaojun@huawei.com>
> >>> + /* ensure the djtag operation is done */
> >>> + do {
> >>> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
> >>> +
> >>> + if (!(rd & DJTAG_MSTR_START_EN_EX))
> >>> + break;
> >>> +
> >>> + udelay(1);
> >>> + } while (timeout--);
> >> This one is obviously not performance critical at all, so use a non-relaxed
> >> accessor. Same for the other two in this function.
> >>
> >> Are these functions ever called from atomic context? If yes, please document
> >> from what context they can be called, otherwise please consider changing
> >> the udelay calls into sleeping waits.
> >>
> > Yes, this is not reentrant.
> The read/write functions shall also be called from irq handler (for
> handling counter overflow).
> So need to use udelay calls. Shall Document it in v2.
Ok.
> >>> +static const struct of_device_id djtag_of_match[] = {
> >>> + /* for hip05(D02) cpu die */
> >>> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
> >>> + .data = (void *)djtag_readwrite_v1 },
> >>> + /* for hip05(D02) io die */
> >>> + { .compatible = "hisilicon,hip05-io-djtag-v1",
> >>> + .data = (void *)djtag_readwrite_v1 },
> >>> + /* for hip06(D03) cpu die */
> >>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
> >>> + .data = (void *)djtag_readwrite_v1 },
> >>> + /* for hip06(D03) io die */
> >>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
> >>> + .data = (void *)djtag_readwrite_v2 },
> >>> + /* for hip07(D05) cpu die */
> >>> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
> >>> + .data = (void *)djtag_readwrite_v2 },
> >>> + /* for hip07(D05) io die */
> >>> + { .compatible = "hisilicon,hip07-io-djtag-v2",
> >>> + .data = (void *)djtag_readwrite_v2 },
> >>> + {},
> >>> +};
> >>
> >> If these are backwards compatible, just mark them as compatible in DT,
> >> e.g. hip06 can use
> >>
> >> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
> >>
> >> so you can tell the difference if you need to, but the driver only has to
> >> list the oldest one here.
> >>
> >> What is the difference between the cpu and io djtag interfaces?
> On some chips like hip06, the djtag version is different for IO die.
In what way? The driver doesn't seem to care about the difference.
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 11:43 ` Arnd Bergmann
@ 2016-11-08 13:46 ` Anurup M
2016-11-08 15:08 ` Arnd Bergmann
0 siblings, 1 reply; 33+ messages in thread
From: Anurup M @ 2016-11-08 13:46 UTC (permalink / raw)
To: Arnd Bergmann
Cc: linux-arm-kernel, Tan Xiaojun, anurup.m, linux-kernel,
mark.rutland, shyju.pv, gabriele.paoloni, john.garry,
will.deacon, linuxarm, xuwei5, zhangshaokun, sanil.kumar,
shiju.jose
On Tuesday 08 November 2016 05:13 PM, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 1:08:31 PM CET Anurup M wrote:
>
>> On Tuesday 08 November 2016 12:32 PM, Tan Xiaojun wrote:
>>> On 2016/11/7 21:26, Arnd Bergmann wrote:
>>>> On Wednesday, November 2, 2016 11:42:46 AM CET Anurup M wrote:
>>>>> From: Tan Xiaojun <tanxiaojun@huawei.com>
>>>>> + /* ensure the djtag operation is done */
>>>>> + do {
>>>>> + djtag_read32_relaxed(regs_base, SC_DJTAG_MSTR_START_EN_EX, &rd);
>>>>> +
>>>>> + if (!(rd & DJTAG_MSTR_START_EN_EX))
>>>>> + break;
>>>>> +
>>>>> + udelay(1);
>>>>> + } while (timeout--);
>>>> This one is obviously not performance critical at all, so use a non-relaxed
>>>> accessor. Same for the other two in this function.
>>>>
>>>> Are these functions ever called from atomic context? If yes, please document
>>>> from what context they can be called, otherwise please consider changing
>>>> the udelay calls into sleeping waits.
>>>>
>>> Yes, this is not reentrant.
>> The read/write functions shall also be called from irq handler (for
>> handling counter overflow).
>> So need to use udelay calls. Shall Document it in v2.
> Ok.
>
>>>>> +static const struct of_device_id djtag_of_match[] = {
>>>>> + /* for hip05(D02) cpu die */
>>>>> + { .compatible = "hisilicon,hip05-cpu-djtag-v1",
>>>>> + .data = (void *)djtag_readwrite_v1 },
>>>>> + /* for hip05(D02) io die */
>>>>> + { .compatible = "hisilicon,hip05-io-djtag-v1",
>>>>> + .data = (void *)djtag_readwrite_v1 },
>>>>> + /* for hip06(D03) cpu die */
>>>>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
>>>>> + .data = (void *)djtag_readwrite_v1 },
>>>>> + /* for hip06(D03) io die */
>>>>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
>>>>> + .data = (void *)djtag_readwrite_v2 },
>>>>> + /* for hip07(D05) cpu die */
>>>>> + { .compatible = "hisilicon,hip07-cpu-djtag-v2",
>>>>> + .data = (void *)djtag_readwrite_v2 },
>>>>> + /* for hip07(D05) io die */
>>>>> + { .compatible = "hisilicon,hip07-io-djtag-v2",
>>>>> + .data = (void *)djtag_readwrite_v2 },
>>>>> + {},
>>>>> +};
>>>> If these are backwards compatible, just mark them as compatible in DT,
>>>> e.g. hip06 can use
>>>>
>>>> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
>>>>
>>>> so you can tell the difference if you need to, but the driver only has to
>>>> list the oldest one here.
>>>>
>>>> What is the difference between the cpu and io djtag interfaces?
>> On some chips like hip06, the djtag version is different for IO die.
> In what way? The driver doesn't seem to care about the difference.
There is a difference in djtag version of CPU and IO die (in some chips).
For ex: in hip06 djtag for CPU is v1 and for IO is v2.
so they use different readwrite handlers djtag_readwrite_(v1/2).
+ /* for hip06(D03) cpu die */
+ { .compatible = "hisilicon,hip06-cpu-djtag-v1",
+ .data = (void *)djtag_readwrite_v1 },
+ /* for hip06(D03) io die */
+ { .compatible = "hisilicon,hip06-io-djtag-v2",
+ .data = (void *)djtag_readwrite_v2 },
For the same djtag version, there is no difference in handling in the
driver.
Thanks,
Anurup
> Arnd
>
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 13:46 ` Anurup M
@ 2016-11-08 15:08 ` Arnd Bergmann
2016-11-09 4:28 ` Anurup M
0 siblings, 1 reply; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-08 15:08 UTC (permalink / raw)
To: Anurup M
Cc: linux-arm-kernel, Tan Xiaojun, anurup.m, linux-kernel,
mark.rutland, shyju.pv, gabriele.paoloni, john.garry,
will.deacon, linuxarm, xuwei5, zhangshaokun, sanil.kumar,
shiju.jose
On Tuesday, November 8, 2016 7:16:30 PM CET Anurup M wrote:
> >>>> If these are backwards compatible, just mark them as compatible in DT,
> >>>> e.g. hip06 can use
> >>>>
> >>>> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
> >>>>
> >>>> so you can tell the difference if you need to, but the driver only has to
> >>>> list the oldest one here.
> >>>>
> >>>> What is the difference between the cpu and io djtag interfaces?
> >> On some chips like hip06, the djtag version is different for IO die.
> > In what way? The driver doesn't seem to care about the difference.
> There is a difference in djtag version of CPU and IO die (in some chips).
> For ex: in hip06 djtag for CPU is v1 and for IO is v2.
> so they use different readwrite handlers djtag_readwrite_(v1/2).
>
> + /* for hip06(D03) cpu die */
> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
> + .data = (void *)djtag_readwrite_v1 },
> + /* for hip06(D03) io die */
> + { .compatible = "hisilicon,hip06-io-djtag-v2",
> + .data = (void *)djtag_readwrite_v2 },
>
>
> For the same djtag version, there is no difference in handling in the
> driver.
Right, but my point was about the compatibility with the older chips
using the same IP block, marking the ones as compatible that actually
use the same interface.
I also see that the compatible strings have the version included in
them, and you can probably drop them by requiring them only in the
fallback:
compatible = "hisilicon,hip05-cpu-djtag", "hisilicon,djtag-v1";
compatible = "hisilicon,hip05-io-djtag", "hisilicon,djtag-v1";
compatible = "hisilicon,hip06-cpu-djtag", "hisilicon,djtag-v1";
compatible = "hisilicon,hip06-io-djtag", "hisilicon,djtag-v2";
compatible = "hisilicon,hip07-cpu-djtag", "hisilicon,djtag-v2";
compatible = "hisilicon,hip07-io-djtag", "hisilicon,djtag-v2";
We want to have the first entry be as specific as possible, but
the last (second) entry is the one that can be used by the driver
for matching. When a future hip08/hip09/... chip uses an existing
interface, you then don't have to update the driver.
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-08 15:08 ` Arnd Bergmann
@ 2016-11-09 4:28 ` Anurup M
2016-11-09 21:40 ` Arnd Bergmann
0 siblings, 1 reply; 33+ messages in thread
From: Anurup M @ 2016-11-09 4:28 UTC (permalink / raw)
To: Arnd Bergmann
Cc: linux-arm-kernel, Tan Xiaojun, anurup.m, linux-kernel,
mark.rutland, shyju.pv, gabriele.paoloni, john.garry,
will.deacon, linuxarm, xuwei5, zhangshaokun, sanil.kumar,
shiju.jose
On Tuesday 08 November 2016 08:38 PM, Arnd Bergmann wrote:
> On Tuesday, November 8, 2016 7:16:30 PM CET Anurup M wrote:
>>>>>> If these are backwards compatible, just mark them as compatible in DT,
>>>>>> e.g. hip06 can use
>>>>>>
>>>>>> compatible = "hisilicon,hip06-cpu-djtag-v1", "hisilicon,hip05-cpu-djtag-v1";
>>>>>>
>>>>>> so you can tell the difference if you need to, but the driver only has to
>>>>>> list the oldest one here.
>>>>>>
>>>>>> What is the difference between the cpu and io djtag interfaces?
>>>> On some chips like hip06, the djtag version is different for IO die.
>>> In what way? The driver doesn't seem to care about the difference.
>> There is a difference in djtag version of CPU and IO die (in some chips).
>> For ex: in hip06 djtag for CPU is v1 and for IO is v2.
>> so they use different readwrite handlers djtag_readwrite_(v1/2).
>>
>> + /* for hip06(D03) cpu die */
>> + { .compatible = "hisilicon,hip06-cpu-djtag-v1",
>> + .data = (void *)djtag_readwrite_v1 },
>> + /* for hip06(D03) io die */
>> + { .compatible = "hisilicon,hip06-io-djtag-v2",
>> + .data = (void *)djtag_readwrite_v2 },
>>
>>
>> For the same djtag version, there is no difference in handling in the
>> driver.
> Right, but my point was about the compatibility with the older chips
> using the same IP block, marking the ones as compatible that actually
> use the same interface.
>
> I also see that the compatible strings have the version included in
> them, and you can probably drop them by requiring them only in the
> fallback:
>
> compatible = "hisilicon,hip05-cpu-djtag", "hisilicon,djtag-v1";
> compatible = "hisilicon,hip05-io-djtag", "hisilicon,djtag-v1";
> compatible = "hisilicon,hip06-cpu-djtag", "hisilicon,djtag-v1";
> compatible = "hisilicon,hip06-io-djtag", "hisilicon,djtag-v2";
> compatible = "hisilicon,hip07-cpu-djtag", "hisilicon,djtag-v2";
> compatible = "hisilicon,hip07-io-djtag", "hisilicon,djtag-v2";
>
> We want to have the first entry be as specific as possible, but
> the last (second) entry is the one that can be used by the driver
> for matching. When a future hip08/hip09/... chip uses an existing
> interface, you then don't have to update the driver.
Thanks. I had a similar thought on this. So as I have the version string
in the
second entry "-v(1/2)".
I can use it in driver for matching. So i think I will change it as below.
Please correct me if my understanding is wrong.
static const struct of_device_id djtag_of_match[] = {
- /* for hip05(D02) cpu die */
- { .compatible = "hisilicon,hip05-cpu-djtag-v1",
+ /* for hisi djtag-v1 cpu die */
+ { .compatible = "hisilicon,hisi-cpu-djtag-v1",
.data = djtag_readwrite_v1 },
- /* for hip05(D02) io die */
- { .compatible = "hisilicon,hip05-io-djtag-v1",
+ /* for hisi djtag-v1 io die */
+ { .compatible = "hisilicon,hisi-io-djtag-v1",
.data = djtag_readwrite_v1 },
- /* for hip06(D03) cpu die */
- { .compatible = "hisilicon,hip06-cpu-djtag-v1",
- .data = djtag_readwrite_v1 },
- /* for hip06(D03) io die */
- { .compatible = "hisilicon,hip06-io-djtag-v2",
- .data = djtag_readwrite_v2 },
- /* for hip07(D05) cpu die */
- { .compatible = "hisilicon,hip07-cpu-djtag-v2",
+ /* for hisi djtag-v2 cpu die */
+ { .compatible = "hisilicon,hisi-cpu-djtag-v2",
.data = djtag_readwrite_v2 },
- /* for hip07(D05) io die */
- { .compatible = "hisilicon,hip07-io-djtag-v2",
+ /* for hisi djtag-v2 io die */
+ { .compatible = "hisilicon,hisi-io-djtag-v2",
.data = (djtag_readwrite_v2 },
{},
};
Thanks,
Anurup
> Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-09 4:28 ` Anurup M
@ 2016-11-09 21:40 ` Arnd Bergmann
2016-11-11 10:23 ` Anurup M
0 siblings, 1 reply; 33+ messages in thread
From: Arnd Bergmann @ 2016-11-09 21:40 UTC (permalink / raw)
To: linux-arm-kernel
Cc: Anurup M, mark.rutland, zhangshaokun, gabriele.paoloni,
john.garry, will.deacon, linux-kernel, xuwei5, linuxarm,
shyju.pv, sanil.kumar, shiju.jose, Tan Xiaojun, anurup.m
On Wednesday, November 9, 2016 9:58:38 AM CET Anurup M wrote:
>
> > I also see that the compatible strings have the version included in
> > them, and you can probably drop them by requiring them only in the
> > fallback:
> >
> > compatible = "hisilicon,hip05-cpu-djtag", "hisilicon,djtag-v1";
> > compatible = "hisilicon,hip05-io-djtag", "hisilicon,djtag-v1";
> > compatible = "hisilicon,hip06-cpu-djtag", "hisilicon,djtag-v1";
> > compatible = "hisilicon,hip06-io-djtag", "hisilicon,djtag-v2";
> > compatible = "hisilicon,hip07-cpu-djtag", "hisilicon,djtag-v2";
> > compatible = "hisilicon,hip07-io-djtag", "hisilicon,djtag-v2";
> >
> > We want to have the first entry be as specific as possible, but
> > the last (second) entry is the one that can be used by the driver
> > for matching. When a future hip08/hip09/... chip uses an existing
> > interface, you then don't have to update the driver.
> Thanks. I had a similar thought on this. So as I have the version string
> in the
> second entry "-v(1/2)".
> I can use it in driver for matching. So i think I will change it as below.
> Please correct me if my understanding is wrong.
>
> static const struct of_device_id djtag_of_match[] = {
> - /* for hip05(D02) cpu die */
> - { .compatible = "hisilicon,hip05-cpu-djtag-v1",
> + /* for hisi djtag-v1 cpu die */
> + { .compatible = "hisilicon,hisi-cpu-djtag-v1",
> .data = djtag_readwrite_v1 },
> - /* for hip05(D02) io die */
> - { .compatible = "hisilicon,hip05-io-djtag-v1",
> + /* for hisi djtag-v1 io die */
> + { .compatible = "hisilicon,hisi-io-djtag-v1",
>From the code it looks like "hisilicon,hisi-io-djtag-v1" and
"hisilicon,hisi-cpu-djtag-v1" have the same register-level interface,
so we just need one compatible string for them to match the driver.
Arnd
^ permalink raw reply [flat|nested] 33+ messages in thread* Re: [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver
2016-11-09 21:40 ` Arnd Bergmann
@ 2016-11-11 10:23 ` Anurup M
0 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-11 10:23 UTC (permalink / raw)
To: Arnd Bergmann, linux-arm-kernel
Cc: mark.rutland, zhangshaokun, gabriele.paoloni, john.garry,
will.deacon, linux-kernel, xuwei5, linuxarm, shyju.pv,
sanil.kumar, shiju.jose, Tan Xiaojun, anurup.m
On Thursday 10 November 2016 03:10 AM, Arnd Bergmann wrote:
> On Wednesday, November 9, 2016 9:58:38 AM CET Anurup M wrote:
>>> I also see that the compatible strings have the version included in
>>> them, and you can probably drop them by requiring them only in the
>>> fallback:
>>>
>>> compatible = "hisilicon,hip05-cpu-djtag", "hisilicon,djtag-v1";
>>> compatible = "hisilicon,hip05-io-djtag", "hisilicon,djtag-v1";
>>> compatible = "hisilicon,hip06-cpu-djtag", "hisilicon,djtag-v1";
>>> compatible = "hisilicon,hip06-io-djtag", "hisilicon,djtag-v2";
>>> compatible = "hisilicon,hip07-cpu-djtag", "hisilicon,djtag-v2";
>>> compatible = "hisilicon,hip07-io-djtag", "hisilicon,djtag-v2";
>>>
>>> We want to have the first entry be as specific as possible, but
>>> the last (second) entry is the one that can be used by the driver
>>> for matching. When a future hip08/hip09/... chip uses an existing
>>> interface, you then don't have to update the driver.
>> Thanks. I had a similar thought on this. So as I have the version string
>> in the
>> second entry "-v(1/2)".
>> I can use it in driver for matching. So i think I will change it as below.
>> Please correct me if my understanding is wrong.
>>
>> static const struct of_device_id djtag_of_match[] = {
>> - /* for hip05(D02) cpu die */
>> - { .compatible = "hisilicon,hip05-cpu-djtag-v1",
>> + /* for hisi djtag-v1 cpu die */
>> + { .compatible = "hisilicon,hisi-cpu-djtag-v1",
>> .data = djtag_readwrite_v1 },
>> - /* for hip05(D02) io die */
>> - { .compatible = "hisilicon,hip05-io-djtag-v1",
>> + /* for hisi djtag-v1 io die */
>> + { .compatible = "hisilicon,hisi-io-djtag-v1",
> From the code it looks like "hisilicon,hisi-io-djtag-v1" and
> "hisilicon,hisi-cpu-djtag-v1" have the same register-level interface,
> so we just need one compatible string for them to match the driver.
>
> Arnd
The djtag versions in CPU die and IO die can be different in the same chip.
For example in hip06, the CPU die has djtag-v1 whereas IO die has djtag-v2.
So I think it need two different compatible string
for hip06 chip CPU DIE "hisilicon,hisi-cpu-djtag-v1"
for hip06 chip IO DIE "hisilicon,hisi-io-djtag-v2"
Thanks,
Anurup
^ permalink raw reply [flat|nested] 33+ messages in thread
* [PATCH v1 04/11] Documentation: perf: hisi: Documentation for HIP05/06/07 PMU event counting.
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (2 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 03/11] drivers: soc: hisi: Add support for Hisilicon Djtag driver Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 05/11] dt-bindings: perf: hisi: Add Devicetree bindings for Hisilicon SoC PMU Anurup M
` (6 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
Documentation for perf usage and Hisilicon SoC PMU uncore events.
The Hisilicon SOC has event counters for hardware modules like
L3 cache, Miscellaneous node, DDR cntroller etc. These events are
all uncore.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
---
Documentation/perf/hisi-pmu.txt | 80 +++++++++++++++++++++++++++++++++++++++++
1 file changed, 80 insertions(+)
create mode 100644 Documentation/perf/hisi-pmu.txt
diff --git a/Documentation/perf/hisi-pmu.txt b/Documentation/perf/hisi-pmu.txt
new file mode 100644
index 0000000..670a9df
--- /dev/null
+++ b/Documentation/perf/hisi-pmu.txt
@@ -0,0 +1,80 @@
+Hisilicon SoC PMU (Performance Monitoring Unit)
+================================================
+The Hisilicon SoC hip05/06/07 chips consist of varous independent system
+device PMU's such as L3 cache(L3C), Miscellaneous Nodes(MN) and DDR
+controllers. These PMU devices are independent and have hardware logic to
+gather statistics and performance information.
+
+Hip0x chips are encapsulated by multiple CPU and IO die's. The CPU die is
+called as Super CPU cluster (SCCL) which includes 16 cpu-cores. Every SCCL
+is further grouped as CPU clusters (CCL) which includes 4 cpu-cores each.
+Each SCCL has 1 L3 cache and 1 MN units.
+
+The L3 cache is shared by all CPU cores in a CPU die. The L3C has four banks
+(or instances). Each bank or instance of L3C has Eight 32-bit counter
+registers. The hip05/06 chip L3 cache has 22 statistics events. The hip07
+chip has 66 statistics events. These events are very useful for debugging.
+
+The MN module is also shared by all CPU cores in a CPU die. It receives
+barriers and DVM(Distributed Virtual Memory) messages from cpu or smmu, and
+perform the required actions and return response messages. These events are
+very useful for debugging. The MN has total 9 statistics events and support
+four 32-bit counter registers in hip05/06/07 chips.
+
+The DDR conroller supports various statistics events. Every SCCL has fot 2
+DDR channels and hence 2 DDR controllers. The Hip05/06/07 has support for a
+total of 13 statistics events.
+
+There is no memory mapping for L3 cache and MN registers. It can be accessed
+by using the Hisilicon djtag interface. The Djtag in a SCCL is an independent
+module which connects with some modules in the SoC by Debug Bus.
+
+Hisilicon SoC (hip05/06/07) PMU driver
+--------------------------------------
+The hip0x PMU driver shall register perf PMU drivers like L3 cache, MN, DDRC
+etc.
+Separate PMU shall be registered for L3 cache and MN for each Super CPU
+cluster.
+For DRR controller separate PMU shall be registered for each channel in a
+Super CPU cluster.
+
+The available events and configuration options shall be described in the sysfs.
+The "perf list" shall list the available events from sysfs.
+eg. hisi_l3c2/read_allocate/ [kernel PMU event]
+
+The Super Cluster ID will be the number suffix to PMU name
+e.g. hisi_l3c2. Here Super cluster ID is 2 and so hisi_l3c2/read_allocate
+is the event for read_allocate of SCCL #2.
+
+For DDRC the channel number will be suffix at the end.
+eg: hisi_ddrc2_0/flux_read/. Here Super cluster ID is 2 and the channel number
+is 0 for the event flux_read.
+
+The event code is represented by 12 bits.
+ i) event 0-11
+ The event code will be represented using the LSB 12 bits.
+
+The driver also provides a "cpumask" sysfs attribute, which shows the CPU core
+ID used to count the uncore PMU event.
+
+Example usage of perf:
+$# perf list
+hisi_l3c2/read_hit/ [kernel PMU event]
+hisi_l3c2/write_hit/ [kernel PMU event]
+------------------------------------------
+hisi_l3c1/read_hit/ [kernel PMU event]
+hisi_l3c1/write_hit/ [kernel PMU event]
+------------------------------------------
+hisi_mn2/read_req/ [kernel PMU event]
+hisi_mn2/write_req/ [kernel PMU event]
+------------------------------------------
+hisi_ddrc2_0/flux_read/ [kernel PMU event]
+------------------------------------------
+
+$# perf stat -a -e hisi_l3c2/read_allocate/ sleep 5
+
+The current driver doesnot support sampling. so "perf record" is unsupported.
+Also attach to a task is unsupported as the events are all uncore.
+
+Note: Please contact the maintainer for a complete list of events supported for
+the PMU devices in the SoC and its information if needed.
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 05/11] dt-bindings: perf: hisi: Add Devicetree bindings for Hisilicon SoC PMU
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (3 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 04/11] Documentation: perf: hisi: Documentation for HIP05/06/07 PMU event counting Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 06/11] perf: hisi: Update Kconfig for Hisilicon PMU support Anurup M
` (5 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1) Device tree bindings for Hisilicon SoC PMU.
2) Add example for Hisilicon L3 cache, MN and DDRC PMU.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
---
.../devicetree/bindings/arm/hisilicon/pmu.txt | 127 +++++++++++++++++++++
1 file changed, 127 insertions(+)
create mode 100644 Documentation/devicetree/bindings/arm/hisilicon/pmu.txt
diff --git a/Documentation/devicetree/bindings/arm/hisilicon/pmu.txt b/Documentation/devicetree/bindings/arm/hisilicon/pmu.txt
new file mode 100644
index 0000000..e7b35e0
--- /dev/null
+++ b/Documentation/devicetree/bindings/arm/hisilicon/pmu.txt
@@ -0,0 +1,127 @@
+Hisilicon SoC hip05/06/07 ARMv8 PMU
+===================================
+
+The Hisilicon SoC chips like hip05/06/07 etc. consist of varous independent
+system device PMU's such as L3 cache (L3C), Miscellaneous Nodes(MN) and DDR
+comtroller. These PMU devices are independent and have hardware logic to
+gather statistics and performance information.
+
+HiSilicon SoC chip is encapsulated by multiple CPU and IO die's. The CPU die
+is called as Super CPU cluster (SCCL) which includes 16 cpu-cores. Every SCCL
+is further grouped as CPU clusters (CCL) which includes 4 cpu-cores each.
+e.g. In the case of hip05/06/07, each SCCL has 1 L3 cache and 1 MN PMU device.
+
+The Hisilicon SoC PMU DT node bindigs for uncore PMU devices are as below.
+For PMU devices like L3 cache. MN etc. which are accessed using the djtag,
+the parent node will be the djtag node of the corresponding CPU die(SCCL).
+
+For uncore PMU devices there are some common required properties as detailed
+below.
+
+Required properties:
+ - compatible : This field contain two values. The first value is
+ always "hisilicon" and second value is the Module type as shown
+ in below examples:
+ (a) "hisilicon,hisi-pmu-l3c-v1" for Hisilicon SoC L3C PMU
+ device (Version 1)
+ (b) "hisilicon,hisi-pmu-mn-v1" for Hisilicon SoC MN PMU
+ device (Version 1)
+ (c) "hisilicon,hisi-pmu-ddrc-v1" for Hisilicon SoC DDRC PMU
+ device (Version 1)
+ The hip05/06/07 chips have v1 hardware for L3C, MN and DDRC.
+
+ - scl-id : The Super Cluster ID. This can be the ID of the CPU die
+ or IO die in the chip.
+
+ - num-events : No of events supported by this PMU device.
+
+ - num-counters : No of hardware counters available for counting.
+
+L3 cache
+--------
+The L3 cache is dedicated for each SCCL and hence there are separate DT nodes
+for L3 cache for each SCCL. For L3 cache PMU the additional required properties
+are
+ - counter-reg : Counter register offset.
+
+ - evtype-reg : Event select register offset.
+
+ - evctrl-reg : Event counting control(LAUCTRL) register offset.
+
+ - event-en : Event enable value.
+
+ - module-id : Module ID to input for djtag. This property is an array of
+ module_id for each L3 cache banks.
+
+ - num-banks : Number of banks or instances of the device.
+
+ - cfgen-map : Config enable array to select the bank.
+
+Miscellaneous Node
+-------------------
+The MN is dedicated for each SCCL and hence there are separate DT nodes for MN
+for each SCCL. For MN PMU the additional required properties are
+ - counter-reg : Counter register offset.
+
+ - evtype-reg : Event select register offset.
+
+ - evctrl-reg : Event counting control register offset.
+
+ - module-id : Module ID to input for djtag. As MN doesnot have multiple banks
+ this property is a single value.
+
+ - cfgen-map : Config enable to select the bank. For MN it is a single value
+
+ - event-en : Event enable value.
+
+Example:
+
+ djtag0: djtag@0 {
+ compatible = "hisilicon,hip05-cpu-djtag-v1";
+ pmul3c0 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x02>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04 0x04 0x04 0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
+
+ pmumn0 {
+ compatible = "hisilicon,hisi-pmu-mn-v1";
+ scl-id = <0x02>;
+ num-events = <0x09>;
+ num-counters = <0x04>;
+ module-id = <0x0b>;
+ cfgen-map = <0x01>;
+ counter-reg = <0x30>;
+ evctrl-reg = <0x40>;
+ event-en = <0x01>;
+ evtype-reg = <0x48>;
+ };
+ };
+
+DDR controller
+--------------
+Each SCCL in Hip05/06/07 chips have 2 DDR channels and hence 2 DDR controllers.
+There are separate DT nodes for each DDR channel.
+For DDRC PMU the additional required properties are
+
+ - ch-id : DDRC Channel ID.
+ - reg : Register base address and range for the DDRC channel.
+
+Example:
+ /* DDRC for CPU die scl #2 Channel #1 for hip05 */
+ pmu_sccl0_ddrc1: pmu_ddrc1@80358000 {
+ compatible = "hisilicon,hisi-pmu-ddrc-v1";
+ scl-id = <0x02>;
+ ch-id = <0x1>;
+ num-events = <0x0D>;
+ num-counters = <0x04>;
+ reg = <0x80358000 0x10000>; /* TOTEMC DDRC1 */
+ };
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 06/11] perf: hisi: Update Kconfig for Hisilicon PMU support
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (4 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 05/11] dt-bindings: perf: hisi: Add Devicetree bindings for Hisilicon SoC PMU Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 07/11] perf: hisi: Add support for Hisilicon SoC event counters Anurup M
` (4 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1. Update Kconfig for Hip05/06/07 PMU support.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
---
drivers/perf/Kconfig | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index 4d5c5f9..da8dd97 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -19,4 +19,13 @@ config XGENE_PMU
help
Say y if you want to use APM X-Gene SoC performance monitors.
+config HISI_PMU
+ bool "Enable hardware event counter support for HiSilicon SoC"
+ depends on HW_PERF_EVENTS && ARM64
+ depends on HISI_DJTAG
+ help
+ Enable hardware event counter support for hardware event counters
+ in Hisilicon hip05/06/07 SoC. The hardware modules like L3C, MN and
+ DDRC have hardware events and counters.
+
endmenu
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 07/11] perf: hisi: Add support for Hisilicon SoC event counters
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (5 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 06/11] perf: hisi: Update Kconfig for Hisilicon PMU support Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 08/11] perf: hisi: Add sysfs attributes for L3 cache(L3C) PMU Anurup M
` (3 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1. Hip05/06/07 uncore PMU to support different hardware
event counters.
2. Hisilicon PMU shall use the DJTAG hardware interface
to access hardware event counters and configuration
register.
3. Routines to initialize and setup PMU.
4. Routines to enable/disable/add/del/start/stop hardware
event counting.
5. Add support to count L3 cache hardware events.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
---
drivers/perf/Makefile | 1 +
drivers/perf/hisilicon/Makefile | 1 +
drivers/perf/hisilicon/hisi_uncore_l3c.c | 571 +++++++++++++++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_l3c.h | 67 ++++
drivers/perf/hisilicon/hisi_uncore_pmu.c | 331 ++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_pmu.h | 108 ++++++
6 files changed, 1079 insertions(+)
create mode 100644 drivers/perf/hisilicon/Makefile
create mode 100644 drivers/perf/hisilicon/hisi_uncore_l3c.c
create mode 100644 drivers/perf/hisilicon/hisi_uncore_l3c.h
create mode 100644 drivers/perf/hisilicon/hisi_uncore_pmu.c
create mode 100644 drivers/perf/hisilicon/hisi_uncore_pmu.h
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index b116e98..061f229 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -1,2 +1,3 @@
obj-$(CONFIG_ARM_PMU) += arm_pmu.o
obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o
+obj-$(CONFIG_HISI_PMU) += hisilicon/
diff --git a/drivers/perf/hisilicon/Makefile b/drivers/perf/hisilicon/Makefile
new file mode 100644
index 0000000..e1766cf
--- /dev/null
+++ b/drivers/perf/hisilicon/Makefile
@@ -0,0 +1 @@
+obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c.o
diff --git a/drivers/perf/hisilicon/hisi_uncore_l3c.c b/drivers/perf/hisilicon/hisi_uncore_l3c.c
new file mode 100644
index 0000000..f78f7b2
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_l3c.c
@@ -0,0 +1,571 @@
+/*
+ * HiSilicon SoC L3C Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitmap.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/perf_event.h>
+#include "hisi_uncore_l3c.h"
+
+static inline int hisi_l3c_counter_valid(int idx)
+{
+ return (idx >= HISI_IDX_L3C_COUNTER0 &&
+ idx <= HISI_IDX_L3C_COUNTER_MAX);
+}
+
+static u32 hisi_read_l3c_counter(struct hisi_l3c_data *l3c_hwmod_data,
+ int cntr_idx, int bank_idx)
+{
+ struct hisi_djtag_client *client = l3c_hwmod_data->client;
+ u32 module_id = l3c_hwmod_data->l3c_hwcfg.module_id[bank_idx];
+ u32 cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[bank_idx];
+ u32 reg_offset, value;
+
+ reg_offset = l3c_hwmod_data->l3c_hwcfg.counter_reg0_off +
+ (cntr_idx * 4);
+
+ hisi_djtag_readreg(module_id, cfg_en, reg_offset, client, &value);
+
+ return value;
+}
+
+static u64 hisi_l3c_event_update(struct perf_event *event,
+ struct hw_perf_event *hwc, int idx)
+{
+ struct hisi_pmu *pl3c_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u64 delta, prev_raw_count, total_raw_count = 0, avg_raw_count = 0;
+ u32 num_banks = l3c_hwmod_data->l3c_hwcfg.num_banks;
+ int i;
+
+ if (!hisi_l3c_counter_valid(idx)) {
+ dev_err(pl3c_pmu->dev, "Unsupported event index:%d!\n", idx);
+ return 0;
+ }
+
+ /* Check if the L3C data is initialized for this SCCL */
+ if (!l3c_hwmod_data->client) {
+ dev_err(pl3c_pmu->dev, "SCL=%d not initialized!\n",
+ pl3c_pmu->scl_id);
+ return 0;
+ }
+
+ do {
+ /* Get count from individual L3C banks and sum them up */
+ for (i = 0; i < num_banks; i++) {
+ total_raw_count += hisi_read_l3c_counter(l3c_hwmod_data,
+ idx, i);
+ }
+ prev_raw_count = local64_read(&hwc->prev_count);
+
+ /*
+ * As prev_raw_count is updated with average value of
+ * L3 cache banks, we multiply it by no of banks and
+ * compute the delta
+ */
+ delta = (total_raw_count - (prev_raw_count * num_banks)) &
+ HISI_MAX_PERIOD;
+
+ local64_add(delta, &event->count);
+
+ /*
+ * Divide by num of banks to get average count and
+ * update prev_count with this value
+ */
+ avg_raw_count = total_raw_count / num_banks;
+ } while (local64_cmpxchg(
+ &hwc->prev_count, prev_raw_count, avg_raw_count) !=
+ prev_raw_count);
+
+ return total_raw_count;
+}
+
+static void hisi_set_l3c_evtype(struct hisi_pmu *pl3c_pmu, int idx, u32 val)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u32 reg_offset = l3c_hwmod_data->l3c_hwcfg.evtype_reg0_off;
+ u32 event_value, value = 0;
+ u32 cfg_en, module_id;
+ int i;
+
+ event_value = (val -
+ HISI_HWEVENT_L3C_READ_ALLOCATE);
+
+ /* Select the appropriate Event select register */
+ if (idx > 3)
+ reg_offset += 4;
+
+ /* Value to write to event type register */
+ val = event_value << (8 * idx);
+
+ /* Find the djtag Identifier of the Unit */
+ client = l3c_hwmod_data->client;
+
+ /*
+ * Set the event in L3C_EVENT_TYPEx Register
+ * for all L3C banks
+ */
+ for (i = 0; i < l3c_hwmod_data->l3c_hwcfg.num_banks; i++) {
+ module_id = l3c_hwmod_data->l3c_hwcfg.module_id[i];
+ cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[i];
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(0xff << (8 * idx));
+ value |= val;
+
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+ }
+}
+
+static u32 hisi_write_l3c_counter(struct hisi_pmu *pl3c_pmu,
+ struct hw_perf_event *hwc, u32 value)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u32 reg_offset, cfg_en, module_id;
+ int i, ret = 0;
+ int idx = GET_CNTR_IDX(hwc);
+
+ if (!hisi_l3c_counter_valid(idx)) {
+ dev_err(pl3c_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return -EINVAL;
+ }
+
+ reg_offset = l3c_hwmod_data->l3c_hwcfg.counter_reg0_off +
+ (idx * 4);
+
+ client = l3c_hwmod_data->client;
+
+ for (i = 0; i < l3c_hwmod_data->l3c_hwcfg.num_banks; i++) {
+ module_id = l3c_hwmod_data->l3c_hwcfg.module_id[i];
+ cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[i];
+ ret = hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+ if (!ret)
+ ret = value;
+ }
+
+ return ret;
+}
+
+static void hisi_enable_l3c_counter(struct hisi_pmu *pl3c_pmu, int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u32 reg_offset = l3c_hwmod_data->l3c_hwcfg.event_ctrl_reg_off;
+ u32 eventen = l3c_hwmod_data->l3c_hwcfg.event_enable;
+ u32 value, cfg_en, module_id;
+ int i;
+
+ if (!hisi_l3c_counter_valid(idx)) {
+ dev_err(pl3c_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ client = l3c_hwmod_data->client;
+
+ /*
+ * Set the event_bus_en bit in L3C AUCNTRL to enable counting
+ * for all L3C banks
+ */
+ for (i = 0; i < l3c_hwmod_data->l3c_hwcfg.num_banks; i++) {
+ module_id = l3c_hwmod_data->l3c_hwcfg.module_id[i];
+ cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[i];
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value |= eventen;
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+ }
+}
+
+static void hisi_disable_l3c_counter(struct hisi_pmu *pl3c_pmu, int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u32 reg_offset = l3c_hwmod_data->l3c_hwcfg.event_ctrl_reg_off;
+ u32 eventen = l3c_hwmod_data->l3c_hwcfg.event_enable;
+ u32 value, cfg_en, module_id;
+ int i;
+
+ if (!hisi_l3c_counter_valid(idx)) {
+ dev_err(pl3c_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ /* Find the djtag Identifier of the Unit */
+ client = l3c_hwmod_data->client;
+
+ /*
+ * Clear the event_bus_en bit in L3C AUCNTRL if no other
+ * event counting for all L3C banks
+ */
+ for (i = 0; i < l3c_hwmod_data->l3c_hwcfg.num_banks; i++) {
+ module_id = l3c_hwmod_data->l3c_hwcfg.module_id[i];
+ cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[i];
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(eventen);
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+ }
+}
+
+static void hisi_clear_l3c_event_idx(struct hisi_pmu *pl3c_pmu,
+ int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ u32 reg_offset = l3c_hwmod_data->l3c_hwcfg.evtype_reg0_off;
+ void *bitmap_addr;
+ u32 cfg_en, value, module_id;
+ int i;
+
+ if (!hisi_l3c_counter_valid(idx)) {
+ dev_err(pl3c_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ bitmap_addr = l3c_hwmod_data->hisi_l3c_event_used_mask;
+
+ __clear_bit(idx, bitmap_addr);
+
+ /* Clear Counting in L3C event config register */
+ if (idx > 3)
+ reg_offset += 4;
+
+ client = l3c_hwmod_data->client;
+
+ /*
+ * Clear the event in L3C_EVENT_TYPEx Register
+ * for all L3C banks
+ */
+ for (i = 0; i < l3c_hwmod_data->l3c_hwcfg.num_banks; i++) {
+ module_id = l3c_hwmod_data->l3c_hwcfg.module_id[i];
+ cfg_en = l3c_hwmod_data->l3c_hwcfg.bank_cfgen[i];
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(0xff << (8 * idx));
+ value |= (0xff << (8 * idx));
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+ }
+}
+
+static int hisi_l3c_get_event_idx(struct hisi_pmu *pl3c_pmu)
+{
+ struct hisi_l3c_data *l3c_hwmod_data = pl3c_pmu->hwmod_data;
+ int event_idx;
+
+ event_idx =
+ find_first_zero_bit(
+ l3c_hwmod_data->hisi_l3c_event_used_mask,
+ pl3c_pmu->num_counters);
+
+ if (event_idx == HISI_MAX_CFG_L3C_CNTR)
+ return -EAGAIN;
+
+ __set_bit(event_idx,
+ l3c_hwmod_data->hisi_l3c_event_used_mask);
+
+ return event_idx;
+}
+
+static void hisi_free_l3c_data(struct hisi_pmu *pl3c_pmu)
+{
+ kfree(pl3c_pmu->hwmod_data);
+ pl3c_pmu->hwmod_data = NULL;
+}
+
+static int init_hisi_l3c_hwcfg(struct device *dev,
+ struct hisi_l3c_data *pl3c_data)
+{
+ struct hisi_l3c_hwcfg *pl3c_hwcfg = &pl3c_data->l3c_hwcfg;
+ struct device_node *node = dev->of_node;
+ u32 prop_len;
+ int ret;
+
+ if (of_property_read_u32(node, "counter-reg",
+ &pl3c_hwcfg->counter_reg0_off)) {
+ dev_err(dev, "DT:Couldnot read counter-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "evctrl-reg",
+ &pl3c_hwcfg->event_ctrl_reg_off)) {
+ dev_err(dev, "DT:Couldnot read evctrl-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "event-en",
+ &pl3c_hwcfg->event_enable)) {
+ dev_err(dev, "DT:Couldnot read event-en!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "evtype-reg",
+ &pl3c_hwcfg->evtype_reg0_off)) {
+ dev_err(dev, "DT:Couldnot read evtype-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "num-banks",
+ &pl3c_hwcfg->num_banks)) {
+ dev_err(dev, "DT:Couldnot read num-banks!\n");
+ return -EINVAL;
+ }
+
+ prop_len = of_property_count_u32_elems(node, "module-id");
+ if (prop_len != pl3c_hwcfg->num_banks) {
+ dev_err(dev, "DT:module-id entry not valid!\n");
+ return -EINVAL;
+ }
+
+ ret = of_property_read_u32_array(node, "module-id",
+ &pl3c_hwcfg->module_id[0],
+ pl3c_hwcfg->num_banks);
+ if (ret < 0) {
+ dev_err(dev, "DT:Couldnot read module-id!\n");
+ return -EINVAL;
+ }
+
+ prop_len = of_property_count_u32_elems(node, "cfgen-map");
+ if (prop_len != pl3c_hwcfg->num_banks) {
+ dev_err(dev, "DT:cfgen-map entrynot valid!\n");
+ return -EINVAL;
+ }
+
+ ret = of_property_read_u32_array(node, "cfgen-map",
+ &pl3c_hwcfg->bank_cfgen[0],
+ pl3c_hwcfg->num_banks);
+ if (ret < 0) {
+ dev_err(dev, "DT:Couldnot read cfgen-map!\n");
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int init_hisi_l3c_data(struct device *dev,
+ struct hisi_pmu *pl3c_pmu,
+ struct hisi_djtag_client *client)
+{
+ struct hisi_l3c_data *l3c_hwmod_data = NULL;
+ int ret;
+
+ l3c_hwmod_data = kzalloc(sizeof(struct hisi_l3c_data),
+ GFP_KERNEL);
+ if (!l3c_hwmod_data)
+ return -ENOMEM;
+
+ /* Set the djtag Identifier */
+ l3c_hwmod_data->client = client;
+
+ pl3c_pmu->hw_events.events = devm_kcalloc(dev,
+ pl3c_pmu->num_counters,
+ sizeof(*pl3c_pmu->hw_events.events),
+ GFP_KERNEL);
+ if (!pl3c_pmu->hw_events.events) {
+ ret = -ENOMEM;
+ goto fail;
+ }
+
+ raw_spin_lock_init(&pl3c_pmu->hw_events.pmu_lock);
+
+ pl3c_pmu->hwmod_data = l3c_hwmod_data;
+
+ ret = init_hisi_l3c_hwcfg(dev, l3c_hwmod_data);
+ if (ret)
+ goto fail;
+
+ return 0;
+
+fail:
+ hisi_free_l3c_data(pl3c_pmu);
+ return ret;
+}
+
+static struct hisi_uncore_ops hisi_uncore_l3c_ops = {
+ .set_evtype = hisi_set_l3c_evtype,
+ .set_event_period = hisi_pmu_set_event_period,
+ .get_event_idx = hisi_l3c_get_event_idx,
+ .clear_event_idx = hisi_clear_l3c_event_idx,
+ .event_update = hisi_l3c_event_update,
+ .enable_counter = hisi_enable_l3c_counter,
+ .disable_counter = hisi_disable_l3c_counter,
+ .write_counter = hisi_write_l3c_counter,
+};
+
+static int hisi_l3c_pmu_init(struct device *dev,
+ struct hisi_pmu *pl3c_pmu)
+{
+ int ret;
+
+ /* Read common PMU properties */
+ ret = hisi_uncore_common_fwprop_read(dev, pl3c_pmu);
+ if (ret)
+ return ret;
+
+ pl3c_pmu->name = kasprintf(GFP_KERNEL, "hisi_l3c%d",
+ pl3c_pmu->scl_id);
+ pl3c_pmu->ops = &hisi_uncore_l3c_ops;
+ pl3c_pmu->dev = dev;
+
+ /* Pick one core to use for cpumask attributes */
+ cpumask_set_cpu(smp_processor_id(), &pl3c_pmu->cpu);
+
+ return 0;
+}
+
+static int hisi_pmu_l3c_dev_probe(struct hisi_djtag_client *client)
+{
+ struct hisi_pmu *pl3c_pmu = NULL;
+ struct device *dev = &client->dev;
+ int ret;
+
+ pl3c_pmu = hisi_pmu_alloc(dev);
+ if (IS_ERR(pl3c_pmu))
+ return PTR_ERR(pl3c_pmu);
+
+ ret = hisi_l3c_pmu_init(dev, pl3c_pmu);
+ if (ret)
+ return ret;
+
+ ret = init_hisi_l3c_data(dev, pl3c_pmu, client);
+ if (ret)
+ goto fail_init;
+
+ /* Register with perf PMU */
+ pl3c_pmu->pmu = (struct pmu) {
+ .name = pl3c_pmu->name,
+ .task_ctx_nr = perf_invalid_context,
+ .event_init = hisi_uncore_pmu_event_init,
+ .add = hisi_uncore_pmu_add,
+ .del = hisi_uncore_pmu_del,
+ .start = hisi_uncore_pmu_start,
+ .stop = hisi_uncore_pmu_stop,
+ .read = hisi_uncore_pmu_read,
+ };
+
+ ret = hisi_uncore_pmu_setup(pl3c_pmu, pl3c_pmu->name);
+ if (ret) {
+ dev_err(dev, "hisi_uncore_pmu_init FAILED!!\n");
+ goto fail;
+ }
+
+ /* Set the drv data to l3c_pmu */
+ dev_set_drvdata(dev, pl3c_pmu);
+
+ return 0;
+
+fail:
+ hisi_free_l3c_data(pl3c_pmu);
+
+fail_init:
+ dev_err(dev, "%s failed\n", __func__);
+ return ret;
+}
+
+static int hisi_pmu_l3c_dev_remove(struct hisi_djtag_client *client)
+{
+ struct hisi_pmu *pl3c_pmu = NULL;
+ struct device *dev = &client->dev;
+
+ pl3c_pmu = dev_get_drvdata(dev);
+
+ perf_pmu_unregister(&pl3c_pmu->pmu);
+ hisi_free_l3c_data(pl3c_pmu);
+
+ return 0;
+}
+
+static const struct of_device_id l3c_of_match[] = {
+ { .compatible = "hisilicon,hisi-pmu-l3c-v1", },
+ {},
+};
+MODULE_DEVICE_TABLE(of, l3c_of_match);
+
+static struct hisi_djtag_driver hisi_pmu_l3c_driver = {
+ .driver = {
+ .name = "hisi-pmu-l3c",
+ .of_match_table = l3c_of_match,
+ },
+ .probe = hisi_pmu_l3c_dev_probe,
+ .remove = hisi_pmu_l3c_dev_remove,
+};
+
+static int __init hisi_pmu_l3c_init(void)
+{
+ int rc;
+
+ rc = hisi_djtag_register_driver(THIS_MODULE, &hisi_pmu_l3c_driver);
+ if (rc < 0) {
+ pr_err("hisi pmu l3c init failed, rc=%d\n", rc);
+ return rc;
+ }
+
+ return 0;
+}
+module_init(hisi_pmu_l3c_init);
+
+static void __exit hisi_pmu_l3c_exit(void)
+{
+ hisi_djtag_unregister_driver(&hisi_pmu_l3c_driver);
+
+}
+module_exit(hisi_pmu_l3c_exit);
+
+MODULE_DESCRIPTION("HiSilicon SoC HIP0x L3C PMU driver");
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Anurup M");
diff --git a/drivers/perf/hisilicon/hisi_uncore_l3c.h b/drivers/perf/hisilicon/hisi_uncore_l3c.h
new file mode 100644
index 0000000..a4a1777
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_l3c.h
@@ -0,0 +1,67 @@
+/*
+ * HiSilicon SoC L3C Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __HISI_UNCORE_L3C_H__
+#define __HISI_UNCORE_L3C_H__
+
+#include "hisi_uncore_pmu.h"
+
+/*
+ * ARMv8 HiSilicon L3C RAW event types.
+ */
+enum armv8_hisi_l3c_event_types {
+ HISI_HWEVENT_L3C_READ_ALLOCATE = 0x0,
+ HISI_HWEVENT_L3C_WRITE_ALLOCATE = 0x01,
+ HISI_HWEVENT_L3C_READ_NOALLOCATE = 0x02,
+ HISI_HWEVENT_L3C_WRITE_NOALLOCATE = 0x03,
+ HISI_HWEVENT_L3C_READ_HIT = 0x04,
+ HISI_HWEVENT_L3C_WRITE_HIT = 0x05,
+ HISI_HWEVENT_L3C_EVENT_MAX = 0x15,
+};
+
+/*
+ * ARMv8 HiSilicon Hardware counter Index.
+ */
+enum armv8_hisi_l3c_counters {
+ HISI_IDX_L3C_COUNTER0 = 0x0,
+ HISI_IDX_L3C_COUNTER_MAX = 0x7,
+};
+
+#define HISI_MAX_CFG_L3C_CNTR 0x08
+
+struct hisi_l3c_hwcfg {
+ u32 evtype_reg0_off;
+ u32 counter_reg0_off;
+ u32 event_ctrl_reg_off;
+ u32 event_enable;
+ u32 module_id[MAX_BANKS];
+ u32 num_banks;
+ u32 bank_cfgen[MAX_BANKS];
+};
+
+struct hisi_l3c_data {
+ struct hisi_djtag_client *client;
+ DECLARE_BITMAP(hisi_l3c_event_used_mask,
+ HISI_MAX_CFG_L3C_CNTR);
+ struct hisi_l3c_hwcfg l3c_hwcfg;
+};
+
+#endif /* __HISI_UNCORE_L3C_H__ */
diff --git a/drivers/perf/hisilicon/hisi_uncore_pmu.c b/drivers/perf/hisilicon/hisi_uncore_pmu.c
new file mode 100644
index 0000000..8d29fcc
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_pmu.c
@@ -0,0 +1,331 @@
+/*
+ * HiSilicon SoC Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitmap.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/perf_event.h>
+#include "hisi_uncore_pmu.h"
+
+/* djtag read interface - Call djtag driver to access SoC registers */
+int hisi_djtag_readreg(int module_id, int bank, u32 offset,
+ struct hisi_djtag_client *client, u32 *pvalue)
+{
+ int ret;
+ u32 chain_id = 0;
+
+ while (bank != 1) {
+ bank = (bank >> 0x1);
+ chain_id++;
+ }
+
+ ret = hisi_djtag_readl(client, offset, module_id,
+ chain_id, pvalue);
+ if (ret)
+ dev_err(&client->dev, "read failed, ret=%d!\n", ret);
+
+ return ret;
+}
+
+/* djtag write interface - Call djtag driver to access SoC registers */
+int hisi_djtag_writereg(int module_id, int bank,
+ u32 offset, u32 value,
+ struct hisi_djtag_client *client)
+{
+ int ret;
+
+ ret = hisi_djtag_writel(client, offset, module_id,
+ HISI_DJTAG_MOD_MASK, value);
+ if (ret)
+ dev_err(&client->dev, "write failed, ret=%d!\n", ret);
+
+ return ret;
+}
+
+static int pmu_map_event(struct perf_event *event)
+{
+ return (int)(event->attr.config & HISI_EVTYPE_EVENT);
+}
+
+static int
+__hw_perf_event_init(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+ struct device *dev = phisi_pmu->dev;
+ int mapping;
+
+ mapping = pmu_map_event(event);
+ if (mapping < 0) {
+ dev_err(dev, "event %x:%llx not supported\n", event->attr.type,
+ event->attr.config);
+ return mapping;
+ }
+
+ /*
+ * We don't assign an index until we actually place the event onto
+ * hardware. Use -1 to signify that we haven't decided where to put it
+ * yet.
+ */
+ hwc->idx = -1;
+ hwc->config = 0;
+ hwc->event_base = 0;
+
+ /* For HiSilicon SoC L3C update config_base based on event encoding */
+ hwc->config_base = event->attr.config;
+
+ return 0;
+}
+
+int hisi_uncore_pmu_event_init(struct perf_event *event)
+{
+ int err;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+
+ if (event->attr.type != event->pmu->type)
+ return -ENOENT;
+
+ /* we do not support sampling as the counters are all
+ * shared by all CPU cores in a CPU die(SCCL). Also we
+ * donot support attach to a task(per-process mode)
+ */
+ if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
+ return -EOPNOTSUPP;
+
+ /* counters do not have these bits */
+ if (event->attr.exclude_user ||
+ event->attr.exclude_kernel ||
+ event->attr.exclude_host ||
+ event->attr.exclude_guest ||
+ event->attr.exclude_hv ||
+ event->attr.exclude_idle)
+ return -EINVAL;
+
+ if (event->cpu < 0)
+ return -EINVAL;
+
+ event->cpu = cpumask_first(&phisi_pmu->cpu);
+
+ err = __hw_perf_event_init(event);
+
+ return err;
+}
+
+/*
+ * Enable counter and set the counter to count
+ * the event that we're interested in.
+ */
+void hisi_uncore_pmu_enable_event(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+
+ /* Disable the hardware event counting */
+ if (phisi_pmu->ops->disable_counter)
+ phisi_pmu->ops->disable_counter(phisi_pmu, GET_CNTR_IDX(hwc));
+
+ /*
+ * Set event (if destined for Hisilicon SoC counters).
+ */
+ if (phisi_pmu->ops->set_evtype)
+ phisi_pmu->ops->set_evtype(phisi_pmu, GET_CNTR_IDX(hwc),
+ hwc->config_base);
+
+ /* Enable the hardware event counting */
+ if (phisi_pmu->ops->enable_counter)
+ phisi_pmu->ops->enable_counter(phisi_pmu, GET_CNTR_IDX(hwc));
+}
+
+void hisi_pmu_set_event_period(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+
+ /*
+ * The Hisilicon PMU counters have a period of 2^32. To account for the
+ * possiblity of extreme interrupt latency we program for a period of
+ * half that. Hopefully we can handle the interrupt before another 2^31
+ * events occur and the counter overtakes its previous value.
+ */
+ u64 val = 1ULL << 31;
+
+ local64_set(&hwc->prev_count, val);
+
+ /* Write to the hardware event counter */
+ phisi_pmu->ops->write_counter(phisi_pmu, hwc, val);
+}
+
+void hisi_uncore_pmu_start(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_pmu_hw_events *hw_events;
+
+ hw_events = &phisi_pmu->hw_events;
+
+ if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
+ return;
+
+ WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+ hwc->state = 0;
+
+ if (phisi_pmu->ops->set_event_period)
+ phisi_pmu->ops->set_event_period(event);
+
+ if (flags & PERF_EF_RELOAD) {
+ u64 prev_raw_count = local64_read(&hwc->prev_count);
+
+ phisi_pmu->ops->write_counter(phisi_pmu, hwc,
+ (u32)prev_raw_count);
+ }
+
+ hisi_uncore_pmu_enable_event(event);
+ perf_event_update_userpage(event);
+}
+
+void hisi_uncore_pmu_stop(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+
+ if (hwc->state & PERF_HES_UPTODATE)
+ return;
+
+ /*
+ * We always reprogram the counter, so ignore PERF_EF_UPDATE.
+ * See hisi_uncore_pmu_start()
+ */
+ if (phisi_pmu->ops->disable_counter)
+ phisi_pmu->ops->disable_counter(phisi_pmu,
+ GET_CNTR_IDX(hwc));
+
+ WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
+ hwc->state |= PERF_HES_STOPPED;
+ if (hwc->state & PERF_HES_UPTODATE)
+ return;
+
+ /* Read hardware counter and update the Perf counter statistics */
+ phisi_pmu->ops->event_update(event, hwc, GET_CNTR_IDX(hwc));
+ hwc->state |= PERF_HES_UPTODATE;
+}
+
+int hisi_uncore_pmu_add(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_pmu_hw_events *hw_events;
+ int idx;
+
+ hw_events = &phisi_pmu->hw_events;
+
+ hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
+
+ /* If we don't have a free counter then return early. */
+ idx = phisi_pmu->ops->get_event_idx(phisi_pmu);
+ if (idx < 0)
+ return -EAGAIN;
+
+ event->hw.idx = idx;
+ hw_events->events[idx] = event;
+
+ if (flags & PERF_EF_START)
+ hisi_uncore_pmu_start(event, PERF_EF_RELOAD);
+
+ /* Propagate our changes to the userspace mapping. */
+ perf_event_update_userpage(event);
+
+ return 0;
+}
+
+void hisi_uncore_pmu_del(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_pmu_hw_events *hw_events;
+
+ hw_events = &phisi_pmu->hw_events;
+
+ hisi_uncore_pmu_stop(event, PERF_EF_UPDATE);
+
+ phisi_pmu->ops->clear_event_idx(phisi_pmu, GET_CNTR_IDX(hwc));
+ perf_event_update_userpage(event);
+ hw_events->events[GET_CNTR_IDX(hwc)] = NULL;
+}
+
+struct hisi_pmu *hisi_pmu_alloc(struct device *dev)
+{
+ struct hisi_pmu *phisi_pmu;
+
+ phisi_pmu = devm_kzalloc(dev, sizeof(*phisi_pmu), GFP_KERNEL);
+ if (!phisi_pmu)
+ return ERR_PTR(-ENOMEM);
+
+ return phisi_pmu;
+}
+
+void hisi_uncore_pmu_read(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *phisi_pmu = to_hisi_pmu(event->pmu);
+
+ /* Read hardware counter and update the Perf counter statistics */
+ phisi_pmu->ops->event_update(event, hwc, GET_CNTR_IDX(hwc));
+}
+
+int hisi_uncore_common_fwprop_read(struct device *dev,
+ struct hisi_pmu *phisi_pmu)
+{
+ if (device_property_read_u32(dev, "num-events",
+ &phisi_pmu->num_events)) {
+ dev_err(dev, "Cant read num-events from DT!\n");
+ return -EINVAL;
+ }
+
+ if (device_property_read_u32(dev, "num-counters",
+ &phisi_pmu->num_counters)) {
+ dev_err(dev, "Cant read num-counters from DT!\n");
+ return -EINVAL;
+ }
+
+ /* Find the SCL ID */
+ if (device_property_read_u32(dev, "scl-id",
+ &phisi_pmu->scl_id)) {
+ dev_err(dev, "Cant read scl-id!\n");
+ return -EINVAL;
+ }
+
+ if (phisi_pmu->scl_id == 0 ||
+ phisi_pmu->scl_id >= MAX_UNITS) {
+ dev_err(dev, "Invalid SCL=%d!\n",
+ phisi_pmu->scl_id);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+int hisi_uncore_pmu_setup(struct hisi_pmu *phisi_pmu,
+ const char *pmu_name)
+{
+ /* Register the events with perf */
+ return perf_pmu_register(&phisi_pmu->pmu, pmu_name, -1);
+}
diff --git a/drivers/perf/hisilicon/hisi_uncore_pmu.h b/drivers/perf/hisilicon/hisi_uncore_pmu.h
new file mode 100644
index 0000000..b6b16df
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_pmu.h
@@ -0,0 +1,108 @@
+/*
+ * HiSilicon SoC Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __HISI_UNCORE_PMU_H__
+#define __HISI_UNCORE_PMU_H__
+
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/soc/hisilicon/djtag.h>
+#include <linux/types.h>
+#include <asm/local64.h>
+
+#undef pr_fmt
+#define pr_fmt(fmt) "hisi_pmu: " fmt
+
+#define HISI_DJTAG_MOD_MASK (0xFFFF)
+#define HISI_CNTR_SCCL_MASK (0xF00)
+
+#define HISI_EVTYPE_EVENT 0xfff
+#define HISI_MAX_PERIOD ((1LLU << 32) - 1)
+
+#define MAX_BANKS 8
+#define MAX_COUNTERS 30
+#define MAX_UNITS 8
+
+#define GET_CNTR_IDX(hwc) (hwc->idx)
+#define to_hisi_pmu(c) (container_of(c, struct hisi_pmu, pmu))
+
+#define GET_UNIT_IDX(event_code) \
+ (((event_code & HISI_SCCL_MASK) >> \
+ HISI_SCCL_SHIFT) - 1)
+
+struct hisi_pmu;
+
+struct hisi_uncore_ops {
+ void (*set_evtype)(struct hisi_pmu *, int, u32);
+ void (*set_event_period)(struct perf_event *);
+ int (*get_event_idx)(struct hisi_pmu *);
+ void (*clear_event_idx)(struct hisi_pmu *, int);
+ u64 (*event_update)(struct perf_event *,
+ struct hw_perf_event *, int);
+ u32 (*read_counter)(struct hisi_pmu *, int, int);
+ u32 (*write_counter)(struct hisi_pmu *,
+ struct hw_perf_event *, u32);
+ void (*enable_counter)(struct hisi_pmu *, int);
+ void (*disable_counter)(struct hisi_pmu *, int);
+};
+
+struct hisi_pmu_hw_events {
+ struct perf_event **events;
+ raw_spinlock_t pmu_lock;
+};
+
+/* Generic pmu struct for different pmu types */
+struct hisi_pmu {
+ const char *name;
+ struct hisi_pmu_hw_events hw_events;
+ struct hisi_uncore_ops *ops;
+ struct device *dev;
+ void *hwmod_data; /* Hardware module specific data */
+ cpumask_t cpu;
+ struct pmu pmu;
+ u32 scl_id;
+ int num_counters;
+ int num_events;
+ int num_units;
+};
+
+void hisi_uncore_pmu_read(struct perf_event *event);
+void hisi_uncore_pmu_del(struct perf_event *event, int flags);
+int hisi_uncore_pmu_add(struct perf_event *event, int flags);
+void hisi_uncore_pmu_start(struct perf_event *event, int flags);
+void hisi_uncore_pmu_stop(struct perf_event *event, int flags);
+void hisi_pmu_set_event_period(struct perf_event *event);
+void hisi_uncore_pmu_enable_event(struct perf_event *event);
+int hisi_uncore_pmu_setup(struct hisi_pmu *phisi_pmu, const char *pmu_name);
+int hisi_uncore_pmu_event_init(struct perf_event *event);
+int hisi_djtag_readreg(int module_id, int bank, u32 offset,
+ struct hisi_djtag_client *client,
+ u32 *pvalue);
+int hisi_djtag_writereg(int module_id, int bank,
+ u32 offset, u32 value,
+ struct hisi_djtag_client *client);
+struct hisi_pmu *hisi_pmu_alloc(struct device *dev);
+int hisi_uncore_common_fwprop_read(struct device *dev,
+ struct hisi_pmu *phisi_pmu);
+#endif /* __HISI_UNCORE_PMU_H__ */
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 08/11] perf: hisi: Add sysfs attributes for L3 cache(L3C) PMU
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (6 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 07/11] perf: hisi: Add support for Hisilicon SoC event counters Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 09/11] perf: hisi: Miscellanous node(MN) event counting in perf Anurup M
` (2 subsequent siblings)
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1. Add L3 caches events to /sys/devices/hisi_l3c2/events/
The events can be selected as shown in perf list
e.g.: For L3C_READ_ALLOCATE event for Super CPU cluster 2 the
event format is
-e "hisi_l3c2/read_allocate/"
2. Add cpu_mask attribute group for showing the available CPU
for counting.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
---
drivers/perf/hisilicon/hisi_uncore_l3c.c | 57 ++++++++++++++++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_pmu.c | 40 ++++++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_pmu.h | 22 ++++++++++++
3 files changed, 119 insertions(+)
diff --git a/drivers/perf/hisilicon/hisi_uncore_l3c.c b/drivers/perf/hisilicon/hisi_uncore_l3c.c
index f78f7b2..428fba0 100644
--- a/drivers/perf/hisilicon/hisi_uncore_l3c.c
+++ b/drivers/perf/hisilicon/hisi_uncore_l3c.c
@@ -436,6 +436,62 @@ static int init_hisi_l3c_data(struct device *dev,
return ret;
}
+static struct attribute *hisi_l3c_format_attr[] = {
+ HISI_PMU_FORMAT_ATTR(event, "config:0-11"),
+ NULL,
+};
+
+static struct attribute_group hisi_l3c_format_group = {
+ .name = "format",
+ .attrs = hisi_l3c_format_attr,
+};
+
+static struct attribute *hisi_l3c_events_attr[] = {
+ HISI_PMU_EVENT_ATTR_STR(read_allocate,
+ "event=0x0"),
+ HISI_PMU_EVENT_ATTR_STR(write_allocate,
+ "event=0x01"),
+ HISI_PMU_EVENT_ATTR_STR(read_noallocate,
+ "event=0x02"),
+ HISI_PMU_EVENT_ATTR_STR(write_noallocate,
+ "event=0x03"),
+ HISI_PMU_EVENT_ATTR_STR(read_hit, "event=0x04"),
+ HISI_PMU_EVENT_ATTR_STR(write_hit, "event=0x05"),
+ NULL,
+};
+
+static struct attribute_group hisi_l3c_events_group = {
+ .name = "events",
+ .attrs = hisi_l3c_events_attr,
+};
+
+static struct attribute *hisi_l3c_attrs[] = {
+ NULL,
+};
+
+struct attribute_group hisi_l3c_attr_group = {
+ .attrs = hisi_l3c_attrs,
+};
+
+static DEVICE_ATTR(cpumask, 0444, hisi_cpumask_sysfs_show, NULL);
+
+static struct attribute *hisi_l3c_cpumask_attrs[] = {
+ &dev_attr_cpumask.attr,
+ NULL,
+};
+
+static const struct attribute_group hisi_l3c_cpumask_attr_group = {
+ .attrs = hisi_l3c_cpumask_attrs,
+};
+
+static const struct attribute_group *hisi_l3c_pmu_attr_groups[] = {
+ &hisi_l3c_attr_group,
+ &hisi_l3c_format_group,
+ &hisi_l3c_events_group,
+ &hisi_l3c_cpumask_attr_group,
+ NULL,
+};
+
static struct hisi_uncore_ops hisi_uncore_l3c_ops = {
.set_evtype = hisi_set_l3c_evtype,
.set_event_period = hisi_pmu_set_event_period,
@@ -496,6 +552,7 @@ static int hisi_pmu_l3c_dev_probe(struct hisi_djtag_client *client)
.start = hisi_uncore_pmu_start,
.stop = hisi_uncore_pmu_stop,
.read = hisi_uncore_pmu_read,
+ .attr_groups = hisi_l3c_pmu_attr_groups,
};
ret = hisi_uncore_pmu_setup(pl3c_pmu, pl3c_pmu->name);
diff --git a/drivers/perf/hisilicon/hisi_uncore_pmu.c b/drivers/perf/hisilicon/hisi_uncore_pmu.c
index 8d29fcc..d0a911a 100644
--- a/drivers/perf/hisilicon/hisi_uncore_pmu.c
+++ b/drivers/perf/hisilicon/hisi_uncore_pmu.c
@@ -26,6 +26,46 @@
#include <linux/perf_event.h>
#include "hisi_uncore_pmu.h"
+/*
+ * PMU format attributes
+ */
+ssize_t hisi_format_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct dev_ext_attribute *eattr;
+
+ eattr = container_of(attr, struct dev_ext_attribute,
+ attr);
+ return sprintf(buf, "%s\n", (char *) eattr->var);
+}
+
+/*
+ * PMU event attributes
+ */
+ssize_t hisi_event_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct perf_pmu_events_attr *pmu_attr =
+ container_of(attr, struct perf_pmu_events_attr, attr);
+
+ if (pmu_attr->event_str)
+ return sprintf(buf, "%s", pmu_attr->event_str);
+
+ return 0;
+}
+
+/*
+ * sysfs cpumask attributes
+ */
+ssize_t hisi_cpumask_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct pmu *pmu = dev_get_drvdata(dev);
+ struct hisi_pmu *hisi_pmu = to_hisi_pmu(pmu);
+
+ return cpumap_print_to_pagebuf(true, buf, &hisi_pmu->cpu);
+}
+
/* djtag read interface - Call djtag driver to access SoC registers */
int hisi_djtag_readreg(int module_id, int bank, u32 offset,
struct hisi_djtag_client *client, u32 *pvalue)
diff --git a/drivers/perf/hisilicon/hisi_uncore_pmu.h b/drivers/perf/hisilicon/hisi_uncore_pmu.h
index b6b16df..a948752 100644
--- a/drivers/perf/hisilicon/hisi_uncore_pmu.h
+++ b/drivers/perf/hisilicon/hisi_uncore_pmu.h
@@ -51,6 +51,22 @@
(((event_code & HISI_SCCL_MASK) >> \
HISI_SCCL_SHIFT) - 1)
+#define HISI_PMU_FORMAT_ATTR(_name, _config) \
+ (&((struct dev_ext_attribute[]) { \
+ { .attr = __ATTR(_name, 0444, \
+ hisi_format_sysfs_show, NULL), \
+ .var = (void *) _config, \
+ } \
+ })[0].attr.attr)
+
+#define HISI_PMU_EVENT_ATTR_STR(_name, _str) \
+ (&((struct perf_pmu_events_attr[]) { \
+ { .attr = __ATTR(_name, 0444, \
+ hisi_event_sysfs_show, NULL), \
+ .event_str = _str, \
+ } \
+ })[0].attr.attr)
+
struct hisi_pmu;
struct hisi_uncore_ops {
@@ -105,4 +121,10 @@ int hisi_djtag_writereg(int module_id, int bank,
struct hisi_pmu *hisi_pmu_alloc(struct device *dev);
int hisi_uncore_common_fwprop_read(struct device *dev,
struct hisi_pmu *phisi_pmu);
+ssize_t hisi_event_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf);
+ssize_t hisi_format_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf);
+ssize_t hisi_cpumask_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf);
#endif /* __HISI_UNCORE_PMU_H__ */
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 09/11] perf: hisi: Miscellanous node(MN) event counting in perf
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (7 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 08/11] perf: hisi: Add sysfs attributes for L3 cache(L3C) PMU Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 10/11] perf: hisi: Support for Hisilicon DDRC PMU Anurup M
2016-11-02 15:42 ` [PATCH v1 11/11] dts: arm64: hip06: Add Hisilicon SoC PMU support Anurup M
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
From: Shaokun Zhang <zhangshaokun@hisilicon.com>
1. Add support to count MN hardware events.
2. Mn events are listed in sysfs at /sys/devices/hisi_mn2/events/
The events can be selected as shown in perf list
e.g.: For MN_READ_REQUEST event for Super CPU cluster 2 the
event format is
-e "hisi_mn2/read_req/"
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: Anurup M <anurup.m@huawei.com>
---
drivers/perf/hisilicon/Makefile | 2 +-
drivers/perf/hisilicon/hisi_uncore_mn.c | 571 ++++++++++++++++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_mn.h | 68 ++++
3 files changed, 640 insertions(+), 1 deletion(-)
create mode 100644 drivers/perf/hisilicon/hisi_uncore_mn.c
create mode 100644 drivers/perf/hisilicon/hisi_uncore_mn.h
diff --git a/drivers/perf/hisilicon/Makefile b/drivers/perf/hisilicon/Makefile
index e1766cf..8975104 100644
--- a/drivers/perf/hisilicon/Makefile
+++ b/drivers/perf/hisilicon/Makefile
@@ -1 +1 @@
-obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c.o
+obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c.o hisi_uncore_mn.o
diff --git a/drivers/perf/hisilicon/hisi_uncore_mn.c b/drivers/perf/hisilicon/hisi_uncore_mn.c
new file mode 100644
index 0000000..920e346
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_mn.c
@@ -0,0 +1,571 @@
+/*
+ * HiSilicon SoC MN Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Shaokun Zhang <zhangshaokun@hisilicon.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitmap.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/perf_event.h>
+#include "hisi_uncore_mn.h"
+
+static inline int hisi_mn_counter_valid(int idx)
+{
+ return (idx >= HISI_IDX_MN_COUNTER0 &&
+ idx <= HISI_IDX_MN_COUNTER_MAX);
+}
+
+static u32 hisi_read_mn_counter(struct hisi_mn_data *mn_hwmod_data,
+ int idx)
+{
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ struct hisi_djtag_client *client = mn_hwmod_data->client;
+ u32 cfg_en, reg_offset, value;
+
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+ reg_offset = mn_hwmod_data->mn_hwcfg.counter_reg0_off + (idx * 4);
+
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ return value;
+}
+
+static u64 hisi_mn_event_update(struct perf_event *event,
+ struct hw_perf_event *hwc, int idx)
+{
+ struct hisi_pmu *pmn_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_mn_data *mn_hwmod_data;
+ u64 delta, prev_raw_count, new_raw_count = 0;
+ u32 cfg_en;
+
+ if (!hisi_mn_counter_valid(idx)) {
+ dev_err(pmn_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return 0;
+ }
+
+ mn_hwmod_data = pmn_pmu->hwmod_data;
+
+ /* Check if the MN data is initialized for this SCCL */
+ if (!mn_hwmod_data->client) {
+ dev_err(pmn_pmu->dev,
+ "SCL=%d not initialized!\n", pmn_pmu->scl_id);
+ return 0;
+ }
+
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ do {
+ prev_raw_count = local64_read(&hwc->prev_count);
+ new_raw_count =
+ hisi_read_mn_counter(mn_hwmod_data, idx);
+ delta = (new_raw_count - prev_raw_count) &
+ HISI_MAX_PERIOD;
+
+ local64_add(delta, &event->count);
+ } while (local64_cmpxchg(
+ &hwc->prev_count, prev_raw_count, new_raw_count) !=
+ prev_raw_count);
+
+ return new_raw_count;
+}
+
+static void hisi_set_mn_evtype(struct hisi_pmu *pmn_pmu, int idx, u32 val)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ u32 reg_offset = mn_hwmod_data->mn_hwcfg.evtype_reg0_off;
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ u32 cfg_en, event_value, value = 0;
+
+ event_value = (val -
+ HISI_HWEVENT_MN_EO_BARR_REQ);
+
+ /* Value to write to event type register */
+ val = event_value << (8 * idx);
+
+ /* Find the djtag Identifier of the Unit */
+ client = mn_hwmod_data->client;
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ /*
+ * Set the event in MN_EVENT_TYPE Register
+ */
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(0xff << (8 * idx));
+ value |= val;
+
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+}
+
+static u32 hisi_write_mn_counter(struct hisi_pmu *pmn_pmu,
+ struct hw_perf_event *hwc, u32 value)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ u32 cfg_en, reg_offset;
+ int ret;
+ int idx = GET_CNTR_IDX(hwc);
+
+ if (!hisi_mn_counter_valid(idx)) {
+ dev_err(pmn_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return -EINVAL;
+ }
+
+ reg_offset = mn_hwmod_data->mn_hwcfg.counter_reg0_off +
+ (idx * 4);
+ /* Find the djtag Identifier of the Unit */
+ client = mn_hwmod_data->client;
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ ret = hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+
+ return ret;
+}
+
+static void hisi_enable_mn_counter(struct hisi_pmu *pmn_pmu, int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ u32 reg_offset = mn_hwmod_data->mn_hwcfg.event_ctrl_reg_off;
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ u32 event_en = mn_hwmod_data->mn_hwcfg.event_enable;
+ u32 cfg_en, value;
+
+ if (!hisi_mn_counter_valid(idx)) {
+ dev_err(pmn_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ /* Find the djtag Identifier of the Unit */
+ client = mn_hwmod_data->client;
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ /*
+ * Set the event_bus_en bit in MN_EVENT_CTRL to enable counting
+ */
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client,
+ &value);
+
+ value |= event_en;
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+}
+
+static void hisi_disable_mn_counter(struct hisi_pmu *pmn_pmu, int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ u32 reg_offset = mn_hwmod_data->mn_hwcfg.event_ctrl_reg_off;
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ u32 event_en = mn_hwmod_data->mn_hwcfg.event_enable;
+ u32 cfg_en, value;
+
+ if (!hisi_mn_counter_valid(idx)) {
+ dev_err(pmn_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ /* Find the djtag Identifier of the Unit */
+ client = mn_hwmod_data->client;
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ /*
+ * Clear the event_bus_en bit in MN event control
+ */
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(event_en);
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+}
+
+static void hisi_clear_mn_event_idx(struct hisi_pmu *pmn_pmu, int idx)
+{
+ struct hisi_djtag_client *client;
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ u32 reg_offset = mn_hwmod_data->mn_hwcfg.evtype_reg0_off;
+ u32 module_id = mn_hwmod_data->mn_hwcfg.module_id;
+ void *bitmap_addr;
+ u32 cfg_en, value;
+
+ if (!hisi_mn_counter_valid(idx)) {
+ dev_err(pmn_pmu->dev,
+ "Unsupported event index:%d!\n", idx);
+ return;
+ }
+
+ bitmap_addr = mn_hwmod_data->hisi_mn_event_used_mask;
+
+ __clear_bit(idx, bitmap_addr);
+
+ /* Find the djtag Identifier of the Unit */
+ client = mn_hwmod_data->client;
+ cfg_en = mn_hwmod_data->mn_hwcfg.bank_cfgen;
+
+ /*
+ * Clear the event in MN_EVENT_TYPE Register
+ */
+ hisi_djtag_readreg(module_id,
+ cfg_en,
+ reg_offset,
+ client, &value);
+
+ value &= ~(0xff << (8 * idx));
+ value |= (0xff << (8 * idx));
+ hisi_djtag_writereg(module_id,
+ cfg_en,
+ reg_offset,
+ value,
+ client);
+}
+
+static int hisi_mn_get_event_idx(struct hisi_pmu *pmn_pmu)
+{
+ struct hisi_mn_data *mn_hwmod_data = pmn_pmu->hwmod_data;
+ int event_idx;
+
+ event_idx =
+ find_first_zero_bit(
+ mn_hwmod_data->hisi_mn_event_used_mask,
+ HISI_MAX_CFG_MN_CNTR);
+
+ if (event_idx == HISI_MAX_CFG_MN_CNTR)
+ return -EAGAIN;
+
+ __set_bit(event_idx,
+ mn_hwmod_data->hisi_mn_event_used_mask);
+
+ return event_idx;
+}
+
+static void hisi_free_mn_data(struct hisi_pmu *pmn_pmu)
+{
+ kfree(pmn_pmu->hwmod_data);
+ pmn_pmu->hwmod_data = NULL;
+}
+
+static int init_hisi_mn_hwcfg(struct device *dev,
+ struct hisi_mn_data *pmn_data)
+{
+ struct hisi_mn_hwcfg *pmn_hwcfg = &pmn_data->mn_hwcfg;
+ struct device_node *node = dev->of_node;
+
+ if (of_property_read_u32(node, "counter-reg",
+ &pmn_hwcfg->counter_reg0_off)) {
+ dev_err(dev, "DT:Couldnot read counter-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "evctrl-reg",
+ &pmn_hwcfg->event_ctrl_reg_off)) {
+ dev_err(dev, "DT:Couldnot read evctrl-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "event-en",
+ &pmn_hwcfg->event_enable)) {
+ dev_err(dev, "DT:Couldnot read event-en property!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "evtype-reg",
+ &pmn_hwcfg->evtype_reg0_off)) {
+ dev_err(dev, "DT:Couldnot read evtype-reg!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "module-id",
+ &pmn_hwcfg->module_id)) {
+ dev_err(dev, "DT:Couldnot read module-id property!\n");
+ return -EINVAL;
+ }
+
+ if (of_property_read_u32(node, "cfgen-map",
+ &pmn_hwcfg->bank_cfgen)) {
+ dev_err(dev, "DT:Couldnot read cfgen-map property!\n");
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int init_hisi_mn_data(struct device *dev,
+ struct hisi_pmu *pmn_pmu,
+ struct hisi_djtag_client *client)
+{
+ struct hisi_mn_data *mn_hwmod_data;
+ int ret;
+
+ mn_hwmod_data = kzalloc(sizeof(struct hisi_mn_data),
+ GFP_KERNEL);
+ if (!mn_hwmod_data)
+ return -ENOMEM;
+
+ /* Set the djtag Identifier */
+ mn_hwmod_data->client = client;
+
+ pmn_pmu->hw_events.events = devm_kcalloc(dev,
+ pmn_pmu->num_counters,
+ sizeof(*pmn_pmu->hw_events.events),
+ GFP_KERNEL);
+ if (!pmn_pmu->hw_events.events) {
+ ret = -ENOMEM;
+ goto fail;
+ }
+
+ raw_spin_lock_init(&pmn_pmu->hw_events.pmu_lock);
+
+ pmn_pmu->hwmod_data = mn_hwmod_data;
+
+ ret = init_hisi_mn_hwcfg(dev, mn_hwmod_data);
+ if (ret)
+ goto fail;
+
+ return 0;
+
+fail:
+ hisi_free_mn_data(pmn_pmu);
+ return ret;
+}
+
+static struct attribute *hisi_mn_format_attr[] = {
+ HISI_PMU_FORMAT_ATTR(event, "config:0-11"),
+ NULL,
+};
+
+static struct attribute_group hisi_mn_format_group = {
+ .name = "format",
+ .attrs = hisi_mn_format_attr,
+};
+
+static struct attribute *hisi_mn_events_attr[] = {
+ HISI_PMU_EVENT_ATTR_STR(eo_barrier_req,
+ "event=0x0"),
+ HISI_PMU_EVENT_ATTR_STR(ec_barrier_req,
+ "event=0x01"),
+ HISI_PMU_EVENT_ATTR_STR(dvm_op_req,
+ "event=0x02"),
+ HISI_PMU_EVENT_ATTR_STR(dvm_sync_req,
+ "event=0x03"),
+ HISI_PMU_EVENT_ATTR_STR(read_req,
+ "event=0x04"),
+ HISI_PMU_EVENT_ATTR_STR(write_req,
+ "event=0x05"),
+ NULL,
+};
+
+static struct attribute_group hisi_mn_events_group = {
+ .name = "events",
+ .attrs = hisi_mn_events_attr,
+};
+
+static struct attribute *hisi_mn_attrs[] = {
+ NULL,
+};
+
+struct attribute_group hisi_mn_attr_group = {
+ .attrs = hisi_mn_attrs,
+};
+
+static DEVICE_ATTR(cpumask, 0444, hisi_cpumask_sysfs_show, NULL);
+
+static struct attribute *hisi_mn_cpumask_attrs[] = {
+ &dev_attr_cpumask.attr,
+ NULL,
+};
+
+static const struct attribute_group hisi_mn_cpumask_attr_group = {
+ .attrs = hisi_mn_cpumask_attrs,
+};
+
+static const struct attribute_group *hisi_mn_pmu_attr_groups[] = {
+ &hisi_mn_attr_group,
+ &hisi_mn_format_group,
+ &hisi_mn_events_group,
+ &hisi_mn_cpumask_attr_group,
+ NULL,
+};
+
+static struct hisi_uncore_ops hisi_uncore_mn_ops = {
+ .set_evtype = hisi_set_mn_evtype,
+ .set_event_period = hisi_pmu_set_event_period,
+ .get_event_idx = hisi_mn_get_event_idx,
+ .clear_event_idx = hisi_clear_mn_event_idx,
+ .event_update = hisi_mn_event_update,
+ .enable_counter = hisi_enable_mn_counter,
+ .disable_counter = hisi_disable_mn_counter,
+ .write_counter = hisi_write_mn_counter,
+};
+
+static int hisi_mn_pmu_init(struct device *dev,
+ struct hisi_pmu *pmn_pmu)
+{
+ int ret;
+ /* Read common PMU properties */
+ ret = hisi_uncore_common_fwprop_read(dev, pmn_pmu);
+ if (ret)
+ return ret;
+
+ pmn_pmu->name = kasprintf(GFP_KERNEL, "hisi_mn%d",
+ pmn_pmu->scl_id);
+ pmn_pmu->ops = &hisi_uncore_mn_ops;
+
+ pmn_pmu->dev = dev;
+ /* Pick one core to use for cpumask attributes */
+ cpumask_set_cpu(smp_processor_id(), &pmn_pmu->cpu);
+
+ return 0;
+}
+
+static int hisi_pmu_mn_dev_probe(struct hisi_djtag_client *client)
+{
+ struct hisi_pmu *pmn_pmu = NULL;
+ struct device *dev = &client->dev;
+ int ret;
+
+ pmn_pmu = hisi_pmu_alloc(dev);
+ if (IS_ERR(pmn_pmu))
+ return PTR_ERR(pmn_pmu);
+
+ ret = hisi_mn_pmu_init(dev, pmn_pmu);
+ if (ret)
+ return ret;
+
+ ret = init_hisi_mn_data(dev, pmn_pmu, client);
+ if (ret)
+ goto fail_init;
+
+ /* Register with perf PMU */
+ pmn_pmu->pmu = (struct pmu) {
+ .name = pmn_pmu->name,
+ .task_ctx_nr = perf_invalid_context,
+ .event_init = hisi_uncore_pmu_event_init,
+ .add = hisi_uncore_pmu_add,
+ .del = hisi_uncore_pmu_del,
+ .start = hisi_uncore_pmu_start,
+ .stop = hisi_uncore_pmu_stop,
+ .read = hisi_uncore_pmu_read,
+ .attr_groups = hisi_mn_pmu_attr_groups,
+ };
+
+ ret = hisi_uncore_pmu_setup(pmn_pmu, pmn_pmu->name);
+ if (ret) {
+ dev_err(pmn_pmu->dev, "hisi_uncore_pmu_init FAILED!!\n");
+ goto fail;
+ }
+
+ /* Set the drv data to mn_pmu */
+ dev_set_drvdata(dev, pmn_pmu);
+
+ return 0;
+
+fail:
+ hisi_free_mn_data(pmn_pmu);
+
+fail_init:
+ if (pmn_pmu)
+ devm_kfree(dev, pmn_pmu);
+ dev_err(pmn_pmu->dev, "%s failed\n", __func__);
+
+ return ret;
+}
+
+static int hisi_pmu_mn_dev_remove(struct hisi_djtag_client *client)
+{
+ struct hisi_pmu *pmn_pmu = NULL;
+ struct device *dev = &client->dev;
+
+ pmn_pmu = dev_get_drvdata(dev);
+
+ perf_pmu_unregister(&pmn_pmu->pmu);
+ hisi_free_mn_data(pmn_pmu);
+
+ return 0;
+}
+
+static const struct of_device_id mn_of_match[] = {
+ { .compatible = "hisilicon,hisi-pmu-mn-v1", },
+ {},
+};
+MODULE_DEVICE_TABLE(of, mn_of_match);
+
+static struct hisi_djtag_driver hisi_pmu_mn_driver = {
+ .driver = {
+ .name = "hisi-pmu-mn",
+ .of_match_table = mn_of_match,
+ },
+ .probe = hisi_pmu_mn_dev_probe,
+ .remove = hisi_pmu_mn_dev_remove,
+};
+
+static int __init hisi_pmu_mn_init(void)
+{
+ int rc;
+
+ rc = hisi_djtag_register_driver(THIS_MODULE, &hisi_pmu_mn_driver);
+ if (rc < 0) {
+ pr_err("hisi pmu mn init failed, rc=%d\n", rc);
+ return rc;
+ }
+
+ return 0;
+}
+module_init(hisi_pmu_mn_init);
+
+static void __exit hisi_pmu_mn_exit(void)
+{
+ hisi_djtag_unregister_driver(&hisi_pmu_mn_driver);
+}
+module_exit(hisi_pmu_mn_exit);
+
+MODULE_DESCRIPTION("HiSilicon SoC HIP0x MN PMU driver");
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Shaokun Zhang");
diff --git a/drivers/perf/hisilicon/hisi_uncore_mn.h b/drivers/perf/hisilicon/hisi_uncore_mn.h
new file mode 100644
index 0000000..4cac2f3
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_mn.h
@@ -0,0 +1,68 @@
+/*
+ * HiSilicon SoC MN Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Shaokun Zhang <zhangshaokun@hisilicon.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __HISI_UNCORE_MN_H__
+#define __HISI_UNCORE_MN_H__
+
+#include "hisi_uncore_pmu.h"
+
+/*
+ * ARMv8 HiSilicon MN RAW event types.
+ */
+enum armv8_hisi_mn_event_types {
+ HISI_HWEVENT_MN_EO_BARR_REQ = 0x0,
+ HISI_HWEVENT_MN_EC_BARR_REQ = 0x01,
+ HISI_HWEVENT_MN_DVM_OP_REQ = 0x02,
+ HISI_HWEVENT_MN_DVM_SYNC_REQ = 0x03,
+ HISI_HWEVENT_MN_READ_REQ = 0x04,
+ HISI_HWEVENT_MN_WRITE_REQ = 0x05,
+ HISI_HWEVENT_MN_COPYBK_REQ = 0x06,
+ HISI_HWEVENT_MN_OTHER_REQ = 0x07,
+ HISI_HWEVENT_MN_EVENT_MAX = 0x08,
+};
+
+/*
+ * ARMv8 HiSilicon Hardware counter Index.
+ */
+enum armv8_hisi_mn_counters {
+ HISI_IDX_MN_COUNTER0 = 0x0,
+ HISI_IDX_MN_COUNTER_MAX = 0x4,
+};
+
+#define HISI_MAX_CFG_MN_CNTR 0x04
+
+struct hisi_mn_hwcfg {
+ u32 evtype_reg0_off;
+ u32 counter_reg0_off;
+ u32 event_ctrl_reg_off;
+ u32 event_enable;
+ u32 module_id;
+ u32 bank_cfgen;
+};
+
+struct hisi_mn_data {
+ struct hisi_djtag_client *client;
+ DECLARE_BITMAP(hisi_mn_event_used_mask,
+ HISI_MAX_CFG_MN_CNTR);
+ struct hisi_mn_hwcfg mn_hwcfg;
+};
+
+#endif /* __HISI_UNCORE_MN_H__ */
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 10/11] perf: hisi: Support for Hisilicon DDRC PMU.
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (8 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 09/11] perf: hisi: Miscellanous node(MN) event counting in perf Anurup M
@ 2016-11-02 15:42 ` Anurup M
2016-11-02 15:42 ` [PATCH v1 11/11] dts: arm64: hip06: Add Hisilicon SoC PMU support Anurup M
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1. Add support for counting Hisilicon DDRC
statistics events in perf.
2. Support a total of 13 statistics events.
3. Events listed in /sys/devices/<pmu_name>/
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
---
drivers/perf/hisilicon/Makefile | 2 +-
drivers/perf/hisilicon/hisi_uncore_ddrc.c | 444 ++++++++++++++++++++++++++++++
drivers/perf/hisilicon/hisi_uncore_ddrc.h | 73 +++++
3 files changed, 518 insertions(+), 1 deletion(-)
create mode 100644 drivers/perf/hisilicon/hisi_uncore_ddrc.c
create mode 100644 drivers/perf/hisilicon/hisi_uncore_ddrc.h
diff --git a/drivers/perf/hisilicon/Makefile b/drivers/perf/hisilicon/Makefile
index 8975104..8e9df2e 100644
--- a/drivers/perf/hisilicon/Makefile
+++ b/drivers/perf/hisilicon/Makefile
@@ -1 +1 @@
-obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c.o hisi_uncore_mn.o
+obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o hisi_uncore_l3c.o hisi_uncore_mn.o hisi_uncore_ddrc.o
diff --git a/drivers/perf/hisilicon/hisi_uncore_ddrc.c b/drivers/perf/hisilicon/hisi_uncore_ddrc.c
new file mode 100644
index 0000000..b89a72e
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_ddrc.c
@@ -0,0 +1,444 @@
+/*
+ * HiSilicon SoC DDRC Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#include <linux/bitmap.h>
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/perf_event.h>
+#include "hisi_uncore_ddrc.h"
+
+static inline int hisi_ddrc_counter_valid(int idx, struct hisi_pmu *ddrc_pmu)
+{
+ return (idx >= 0 && idx < ddrc_pmu->num_counters);
+}
+
+static u32 hisi_ddrc_read32_relaxed(void __iomem *regs_base, u32 off)
+{
+ void __iomem *reg_addr = regs_base + off;
+
+ return readl_relaxed(reg_addr);
+}
+
+static void hisi_ddrc_write32(void __iomem *regs_base, u32 off, u32 val)
+{
+ void __iomem *reg_addr = regs_base + off;
+
+ writel(val, reg_addr);
+}
+
+static u32 hisi_read_ddrc_counter(struct hisi_ddrc_data *ddrc_hwmod_data,
+ unsigned long event_code, int idx)
+{
+ u32 value;
+ u32 reg_off;
+
+ reg_off = HISI_DDRC_FLUX_WR_REG_OFF + (event_code * 4);
+
+ value = hisi_ddrc_read32_relaxed(ddrc_hwmod_data->regs_base,
+ reg_off);
+ return value;
+}
+
+static u64 hisi_ddrc_event_update(struct perf_event *event,
+ struct hw_perf_event *hwc, int idx)
+{
+ struct hisi_pmu *ddrc_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_ddrc_data *ddrc_hwmod_data;
+ u64 delta, prev_raw_count, new_raw_count = 0;
+
+ if (!hisi_ddrc_counter_valid(idx, ddrc_pmu)) {
+ dev_err(ddrc_pmu->dev,
+ "%s: Unsupported event index:%d!\n", __func__, idx);
+ return 0;
+ }
+
+ ddrc_hwmod_data = ddrc_pmu->hwmod_data;
+
+ /* Check if the DDRC data is initialized for this SCCL */
+ if (!ddrc_hwmod_data->regs_base) {
+ dev_err(ddrc_pmu->dev, "DDRC registers not mapped!\n");
+ return 0;
+ }
+
+ do {
+ prev_raw_count = local64_read(&hwc->prev_count);
+ new_raw_count =
+ hisi_read_ddrc_counter(ddrc_hwmod_data,
+ hwc->config_base, idx);
+ delta = (new_raw_count - prev_raw_count) &
+ HISI_MAX_PERIOD;
+
+ local64_add(delta, &event->count);
+ } while (local64_cmpxchg(
+ &hwc->prev_count, prev_raw_count, new_raw_count) !=
+ prev_raw_count);
+
+ return new_raw_count;
+}
+
+static u32 hisi_write_ddrc_counter(struct hisi_pmu *ddrc_pmu,
+ struct hw_perf_event *hwc, u32 value)
+{
+ struct hisi_ddrc_data *ddrc_hwmod_data = ddrc_pmu->hwmod_data;
+ u32 reg_off;
+ u32 event_code = hwc->config_base;
+
+ if (!(event_code >= HISI_HWEVENT_DDRC_FLUX_WR &&
+ event_code < HISI_HWEVENT_DDRC_MAX_EVENT)) {
+ dev_err(ddrc_pmu->dev, "Unknown DDR evevnt!");
+ return 0;
+ }
+
+ if (!ddrc_hwmod_data->regs_base) {
+ dev_err(ddrc_pmu->dev,
+ "DDR reg address not mapped!\n");
+ return 0;
+ }
+
+ reg_off = HISI_DDRC_FLUX_WR_REG_OFF + (event_code * 4);
+
+ hisi_ddrc_write32(ddrc_hwmod_data->regs_base,
+ reg_off, value);
+
+ return value;
+}
+
+static void hisi_ddrc_set_event_period(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hisi_pmu *ddrc_pmu = to_hisi_pmu(event->pmu);
+ struct hisi_ddrc_data *ddrc_hwmod_data = ddrc_pmu->hwmod_data;
+ u32 event_code = hwc->config_base;
+ u32 reg_off;
+ u32 value;
+
+ reg_off = HISI_DDRC_FLUX_WR_REG_OFF + (event_code * 4);
+ /*
+ * For Hisilicon DDRC PMU we save the current counter value
+ * to prev_count, as we have enabled continuous counting for
+ * DDRC.
+ */
+ value = hisi_ddrc_read32_relaxed(ddrc_hwmod_data->regs_base,
+ reg_off);
+ local64_set(&hwc->prev_count, value);
+}
+
+static void hisi_clear_ddrc_event_idx(struct hisi_pmu *ddrc_pmu, int idx)
+{
+ struct hisi_ddrc_data *ddrc_hwmod_data = ddrc_pmu->hwmod_data;
+ void *bitmap_addr;
+
+ if (!hisi_ddrc_counter_valid(idx, ddrc_pmu)) {
+ dev_err(ddrc_pmu->dev,
+ "%s:Unsupported event index:%d!\n", __func__, idx);
+ return;
+ }
+
+ bitmap_addr = ddrc_hwmod_data->hisi_ddrc_event_used_mask;
+
+ __clear_bit(idx, bitmap_addr);
+}
+
+static int hisi_ddrc_get_event_idx(struct hisi_pmu *ddrc_pmu)
+{
+ struct hisi_ddrc_data *ddrc_hwmod_data = ddrc_pmu->hwmod_data;
+ int event_idx;
+
+ event_idx =
+ find_first_zero_bit(
+ ddrc_hwmod_data->hisi_ddrc_event_used_mask,
+ ddrc_pmu->num_counters);
+
+ if (event_idx == ddrc_pmu->num_counters)
+ return -EAGAIN;
+
+ __set_bit(event_idx,
+ ddrc_hwmod_data->hisi_ddrc_event_used_mask);
+
+ return event_idx;
+}
+
+static void hisi_free_ddrc_data(struct hisi_pmu *ddrc_pmu)
+{
+ kfree(ddrc_pmu->hwmod_data);
+ ddrc_pmu->hwmod_data = NULL;
+}
+
+static void init_hisi_ddr(void __iomem *reg_base)
+{
+ u32 value;
+
+ hisi_ddrc_write32(reg_base, HISI_DDRC_CTRL_PERF_REG_OFF, 0);
+
+ value = hisi_ddrc_read32_relaxed(reg_base, HISI_DDRC_CFG_PERF_REG_OFF);
+ value &= 0x2fffffff;
+ hisi_ddrc_write32(reg_base, HISI_DDRC_CFG_PERF_REG_OFF, value);
+
+ /* Enable Continuous counting */
+ hisi_ddrc_write32(reg_base, HISI_DDRC_CTRL_PERF_REG_OFF, 1);
+}
+
+static int init_hisi_ddrc_dts_data(struct platform_device *pdev,
+ struct hisi_ddrc_data *ddrc_hwmod_data)
+{
+ struct resource *res;
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ /* Continue for zero entries */
+ if (!res) {
+ dev_err(&pdev->dev, "No DDR reg resorces!\n");
+ return -EINVAL;
+ }
+
+ if (!resource_size(res)) {
+ dev_err(&pdev->dev, "Zero DDR reg entry!\n");
+ return -EINVAL;
+ }
+
+ ddrc_hwmod_data->regs_base = devm_ioremap_resource(&pdev->dev, res);
+ if (IS_ERR(ddrc_hwmod_data->regs_base))
+ return PTR_ERR(ddrc_hwmod_data->regs_base);
+
+ init_hisi_ddr(ddrc_hwmod_data->regs_base);
+
+ return 0;
+}
+
+static int init_hisi_ddrc_data(struct platform_device *pdev,
+ struct hisi_pmu *ddrc_pmu)
+{
+ struct device *dev = &pdev->dev;
+ int ret;
+
+ ddrc_pmu->hw_events.events = devm_kcalloc(dev,
+ ddrc_pmu->num_counters,
+ sizeof(*ddrc_pmu->hw_events.events),
+ GFP_KERNEL);
+ if (!ddrc_pmu->hw_events.events) {
+ ret = -ENOMEM;
+ goto fail;
+ }
+
+ raw_spin_lock_init(&ddrc_pmu->hw_events.pmu_lock);
+
+ init_hisi_ddrc_dts_data(pdev, ddrc_pmu->hwmod_data);
+
+ return 0;
+
+fail:
+ hisi_free_ddrc_data(ddrc_pmu);
+ return ret;
+}
+
+static struct attribute *hisi_ddrc_format_attr[] = {
+ HISI_PMU_FORMAT_ATTR(event, "config:0-11"),
+ NULL,
+};
+
+static struct attribute_group hisi_ddrc_format_group = {
+ .name = "format",
+ .attrs = hisi_ddrc_format_attr,
+};
+
+static struct attribute *hisi_ddrc_events_attr[] = {
+ HISI_PMU_EVENT_ATTR_STR(flux_write, "event=0x00"),
+ HISI_PMU_EVENT_ATTR_STR(flux_read, "event=0x01"),
+ HISI_PMU_EVENT_ATTR_STR(flux_write_cmd, "event=0x02"),
+ HISI_PMU_EVENT_ATTR_STR(flux_read_cmd, "event=0x03"),
+ HISI_PMU_EVENT_ATTR_STR(fluxid_write, "event=0x04"),
+ HISI_PMU_EVENT_ATTR_STR(fluxid_read, "event=0x05"),
+ HISI_PMU_EVENT_ATTR_STR(fluxid_write_cmd, "event=0x06"),
+ HISI_PMU_EVENT_ATTR_STR(fluxid_read_cmd, "event=0x07"),
+ HISI_PMU_EVENT_ATTR_STR(write_latency_cnt0,
+ "event=0x08"),
+ HISI_PMU_EVENT_ATTR_STR(read_latency_cnt0,
+ "event=0x09"),
+ HISI_PMU_EVENT_ATTR_STR(write_latency_cnt1,
+ "event=0x0A"),
+ HISI_PMU_EVENT_ATTR_STR(read_latency_cnt1,
+ "event=0x0B"),
+ HISI_PMU_EVENT_ATTR_STR(read_latency_cnt_inher,
+ "event=0x0C"),
+ NULL,
+};
+
+static struct attribute_group hisi_ddrc_events_group = {
+ .name = "events",
+ .attrs = hisi_ddrc_events_attr,
+};
+
+static struct attribute *hisi_ddrc_attrs[] = {
+ NULL,
+};
+
+struct attribute_group hisi_ddrc_attr_group = {
+ .attrs = hisi_ddrc_attrs,
+};
+
+static DEVICE_ATTR(cpumask, 0444, hisi_cpumask_sysfs_show, NULL);
+
+static struct attribute *hisi_ddrc_cpumask_attrs[] = {
+ &dev_attr_cpumask.attr,
+ NULL,
+};
+
+static const struct attribute_group hisi_ddrc_cpumask_attr_group = {
+ .attrs = hisi_ddrc_cpumask_attrs,
+};
+
+static const struct attribute_group *hisi_ddrc_pmu_attr_groups[] = {
+ &hisi_ddrc_attr_group,
+ &hisi_ddrc_format_group,
+ &hisi_ddrc_events_group,
+ &hisi_ddrc_cpumask_attr_group,
+ NULL,
+};
+
+static struct hisi_uncore_ops hisi_uncore_ddrc_ops = {
+ .set_event_period = hisi_ddrc_set_event_period,
+ .get_event_idx = hisi_ddrc_get_event_idx,
+ .clear_event_idx = hisi_clear_ddrc_event_idx,
+ .event_update = hisi_ddrc_event_update,
+ .write_counter = hisi_write_ddrc_counter,
+};
+
+static int hisi_ddrc_pmu_init(struct device *dev,
+ struct hisi_pmu *ddrc_pmu)
+{
+ struct hisi_ddrc_data *ddrc_hwmod_data;
+ struct hisi_ddrc_hwcfg *ddrc_hwcfg;
+ int ret;
+
+ /* Read common PMU properties */
+ ret = hisi_uncore_common_fwprop_read(dev, ddrc_pmu);
+ if (ret)
+ return ret;
+
+ ddrc_hwmod_data = kzalloc(sizeof(struct hisi_ddrc_data),
+ GFP_KERNEL);
+ if (!ddrc_hwmod_data)
+ return -ENOMEM;
+
+ ddrc_hwcfg = &ddrc_hwmod_data->ddrc_hwcfg;
+ if (of_property_read_u32(dev->of_node, "ch-id",
+ &ddrc_hwcfg->channel_id)) {
+ kfree(ddrc_hwmod_data);
+ return -EINVAL;
+ }
+
+ ddrc_pmu->name = kasprintf(GFP_KERNEL, "hisi_ddrc%d_%d",
+ ddrc_pmu->scl_id, ddrc_hwcfg->channel_id);
+
+ ddrc_pmu->ops = &hisi_uncore_ddrc_ops;
+ ddrc_pmu->hwmod_data = ddrc_hwmod_data;
+ ddrc_pmu->dev = dev;
+
+ /* Pick one core to use for cpumask attributes */
+ cpumask_set_cpu(smp_processor_id(), &ddrc_pmu->cpu);
+
+ return 0;
+}
+
+static const struct of_device_id ddrc_of_match[] = {
+ { .compatible = "hisilicon,hisi-pmu-ddrc-v1", },
+ {},
+};
+MODULE_DEVICE_TABLE(of, ddrc_of_match);
+
+static int hisi_pmu_ddrc_dev_probe(struct platform_device *pdev)
+{
+ struct hisi_pmu *ddrc_pmu;
+ struct device *dev = &pdev->dev;
+ const struct of_device_id *of_id;
+ int ret;
+
+ of_id = of_match_device(ddrc_of_match, dev);
+ if (!of_id)
+ return -EINVAL;
+
+ ddrc_pmu = hisi_pmu_alloc(dev);
+ if (IS_ERR(ddrc_pmu))
+ return PTR_ERR(ddrc_pmu);
+
+ ret = hisi_ddrc_pmu_init(dev, ddrc_pmu);
+ if (ret)
+ return ret;
+
+ ret = init_hisi_ddrc_data(pdev, ddrc_pmu);
+ if (ret)
+ goto fail_init;
+
+ /* Register with perf PMU */
+ ddrc_pmu->pmu = (struct pmu) {
+ .name = ddrc_pmu->name,
+ .task_ctx_nr = perf_invalid_context,
+ .event_init = hisi_uncore_pmu_event_init,
+ .add = hisi_uncore_pmu_add,
+ .del = hisi_uncore_pmu_del,
+ .start = hisi_uncore_pmu_start,
+ .stop = hisi_uncore_pmu_stop,
+ .read = hisi_uncore_pmu_read,
+ .attr_groups = hisi_ddrc_pmu_attr_groups,
+ };
+
+ ret = hisi_uncore_pmu_setup(ddrc_pmu, ddrc_pmu->name);
+ if (ret) {
+ dev_err(ddrc_pmu->dev, "hisi_uncore_pmu_init FAILED!!\n");
+ goto fail;
+ }
+
+ platform_set_drvdata(pdev, ddrc_pmu);
+
+ return 0;
+fail:
+ hisi_free_ddrc_data(ddrc_pmu);
+
+fail_init:
+ dev_err(ddrc_pmu->dev, "%s failed\n", __func__);
+
+ return ret;
+}
+
+static int hisi_pmu_ddrc_dev_remove(struct platform_device *pdev)
+{
+ struct hisi_pmu *ddrc_pmu = platform_get_drvdata(pdev);
+
+ perf_pmu_unregister(&ddrc_pmu->pmu);
+ hisi_free_ddrc_data(ddrc_pmu);
+ platform_set_drvdata(pdev, NULL);
+ return 0;
+}
+
+static struct platform_driver hisi_pmu_ddrc_driver = {
+ .driver = {
+ .name = "hisi-pmu-ddrc",
+ .of_match_table = ddrc_of_match,
+ },
+ .probe = hisi_pmu_ddrc_dev_probe,
+ .remove = hisi_pmu_ddrc_dev_remove,
+};
+module_platform_driver(hisi_pmu_ddrc_driver);
+
+MODULE_DESCRIPTION("HiSilicon SoC HIP0x DDRC PMU driver");
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Anurup M");
diff --git a/drivers/perf/hisilicon/hisi_uncore_ddrc.h b/drivers/perf/hisilicon/hisi_uncore_ddrc.h
new file mode 100644
index 0000000..89eab6b
--- /dev/null
+++ b/drivers/perf/hisilicon/hisi_uncore_ddrc.h
@@ -0,0 +1,73 @@
+/*
+ * HiSilicon SoC DDRC Hardware event counters support
+ *
+ * Copyright (C) 2016 Huawei Technologies Limited
+ * Author: Anurup M <anurup.m@huawei.com>
+ *
+ * This code is based on the uncore PMU's like arm-cci and
+ * arm-ccn.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * 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, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __HISI_UNCORE_DDRC_H__
+#define __HISI_UNCORE_DDRC_H__
+
+#include "hisi_uncore_pmu.h"
+
+/*
+ * ARMv8 HiSilicon DDRC event types.
+ */
+enum armv8_hisi_ddrc_event_types {
+ HISI_HWEVENT_DDRC_FLUX_WR = 0x0,
+ HISI_HWEVENT_DDRC_FLUX_RD = 0x01,
+ HISI_HWEVENT_DDRC_FLUX_WCMD = 0x02,
+ HISI_HWEVENT_DDRC_FLUX_RCMD = 0x03,
+ HISI_HWEVENT_DDRC_FLUXID_WD = 0x04,
+ HISI_HWEVENT_DDRC_FLUXID_RD = 0x05,
+ HISI_HWEVENT_DDRC_FLUXID_WCMD = 0x06,
+ HISI_HWEVENT_DDRC_FLUXID_RCMD = 0x07,
+ HISI_HWEVENT_DDRC_WLAT_CNT0 = 0x08,
+ HISI_HWEVENT_DDRC_RLAT_CNT0 = 0x09,
+ HISI_HWEVENT_DDRC_WLAT_CNT1 = 0x0A,
+ HISI_HWEVENT_DDRC_RLAT_CNT1 = 0x0B,
+ HISI_HWEVENT_DDRC_INHERE_RLAT_CNT = 0x0C,
+ HISI_HWEVENT_DDRC_MAX_EVENT,
+};
+
+#define HISI_DDRC_CTRL_PERF_REG_OFF 0x010
+#define HISI_DDRC_CFG_PERF_REG_OFF 0x270
+#define HISI_DDRC_FLUX_WR_REG_OFF 0x380
+#define HISI_DDRC_FLUX_RD_REG_OFF 0x384
+#define HISI_DDRC_FLUX_WCMD_REG_OFF 0x388
+#define HISI_DDRC_FLUX_RCMD_REG_OFF 0x38C
+#define HISI_DDRC_FLUXID_WR_REG_OFF 0x390
+#define HISI_DDRC_FLUXID_RD_REG_OFF 0x394
+#define HISI_DDRC_FLUXID_WCMD_REG_OFF 0x398
+#define HISI_DDRC_FLUXID_RCMD_REG_OFF 0x39C
+#define HISI_DDRC_FLUX_WLATCNT0_REG_OFF 0x3A0
+#define HISI_DDRC_FLUX_RLAT_CNT0_REG_OFF 0x3A4
+#define HISI_DDRC_FLUX_WLATCNT1_REG_OFF 0x3A8
+#define HISI_DDRC_FLUX_RLAT_CNT1_REG_OFF 0x3AC
+
+struct hisi_ddrc_hwcfg {
+ u32 channel_id;
+};
+
+struct hisi_ddrc_data {
+ void __iomem *regs_base;
+ DECLARE_BITMAP(hisi_ddrc_event_used_mask,
+ HISI_HWEVENT_DDRC_MAX_EVENT);
+ struct hisi_ddrc_hwcfg ddrc_hwcfg;
+};
+
+#endif /* __HISI_UNCORE_DDRC_H__ */
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread* [PATCH v1 11/11] dts: arm64: hip06: Add Hisilicon SoC PMU support
2016-11-02 15:42 [PATCH v1 00/11] perf: arm64: Support for Hisilicon SoC Hardware event counters Anurup M
` (9 preceding siblings ...)
2016-11-02 15:42 ` [PATCH v1 10/11] perf: hisi: Support for Hisilicon DDRC PMU Anurup M
@ 2016-11-02 15:42 ` Anurup M
10 siblings, 0 replies; 33+ messages in thread
From: Anurup M @ 2016-11-02 15:42 UTC (permalink / raw)
To: anurup.m, linux-arm-kernel, linux-kernel
Cc: will.deacon, mark.rutland, zhangshaokun, xuwei5, john.garry,
gabriele.paoloni, sanil.kumar, shyju.pv, tanxiaojun, shiju.jose,
linuxarm
1. Add nodes for hip06 L3 cache to support uncore events.
2. Add nodes for hip06 MN to support uncore events.
3. Add nodes for hip06 DDRC to support uncore events.
Signed-off-by: Anurup M <anurup.m@huawei.com>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
---
arch/arm64/boot/dts/hisilicon/hip06.dtsi | 116 +++++++++++++++++++++++++++++++
1 file changed, 116 insertions(+)
diff --git a/arch/arm64/boot/dts/hisilicon/hip06.dtsi b/arch/arm64/boot/dts/hisilicon/hip06.dtsi
index cb9e018..9ff3afe 100644
--- a/arch/arm64/boot/dts/hisilicon/hip06.dtsi
+++ b/arch/arm64/boot/dts/hisilicon/hip06.dtsi
@@ -980,6 +980,122 @@
status = "disabled";
};
+ djtag0: djtag@60010000 {
+ compatible = "hisilicon,hip06-cpu-djtag-v1";
+ reg = <0x0 0x60010000 0x0 0x10000>;
+
+ /* L3 cache for socket0 CPU die scl#2 */
+ pmul3c0 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x02>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04 0x04 0x04 0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
+
+ /* Miscellaneous node for socket0
+ * CPU die scl#2
+ */
+ pmumn0 {
+ compatible = "hisilicon,hisi-pmu-mn-v1";
+ scl-id = <0x02>;
+ num-events = <0x09>;
+ num-counters = <0x04>;
+ module-id = <0x0b>;
+ cfgen-map = <0x01>;
+ counter-reg = <0x30>;
+ evctrl-reg = <0x40>;
+ event-en = <0x01>;
+ evtype-reg = <0x48>;
+ };
+ };
+
+ djtag1: djtag@40010000 {
+ compatible = "hisilicon,hip06-cpu-djtag-v1";
+ reg = <0x0 0x40010000 0x0 0x10000>;
+
+ /* L3 cache for socket0 CPU die scl#1 */
+ pmul3c1 {
+ compatible = "hisilicon,hisi-pmu-l3c-v1";
+ scl-id = <0x01>;
+ num-events = <0x16>;
+ num-counters = <0x08>;
+ module-id = <0x04 0x04 0x04 0x04>;
+ num-banks = <0x04>;
+ cfgen-map = <0x02 0x04 0x01 0x08>;
+ counter-reg = <0x170>;
+ evctrl-reg = <0x04>;
+ event-en = <0x1000000>;
+ evtype-reg = <0x140>;
+ };
+
+ /* Miscellaneous node for socket0
+ * CPU die scl#1
+ */
+ pmumn1 {
+ compatible = "hisilicon,hisi-pmu-mn-v1";
+ scl-id = <0x01>;
+ num-events = <0x09>;
+ num-counters = <0x04>;
+ module-id = <0x0b>;
+ cfgen-map = <0x01>;
+ counter-reg = <0x30>;
+ evctrl-reg = <0x40>;
+ event-en = <0x01>;
+ evtype-reg = <0x48>;
+ };
+ };
+
+ /* DDRC for CPU die scl #1 Channel #0 */
+ pmu_sccl0_ddrc0: pmu_ddrc0@40348000 {
+ compatible = "hisilicon,hisi-pmu-ddrc-v1";
+ scl-id = <0x01>;
+ ch-id = <0x0>;
+ num-events = <0x0d>;
+ num-counters = <0x04>;
+ reg = <0x0 0x40348000 0x0 0x10000>; /* TOTEMA DDRC0 */
+ status = "okay";
+ };
+
+ /* DDRC for CPU die scl #1 Channel #1 */
+ pmu_sccl0_ddrc1: pmu_ddrc1@40358000 {
+ compatible = "hisilicon,hisi-pmu-ddrc-v1";
+ scl-id = <0x01>;
+ ch-id = <0x01>;
+ num-events = <0x0d>;
+ num-counters = <0x04>;
+ reg = <0x0 0x40358000 0x0 0x10000>; /* TOTEMA DDRC1 */
+ status = "okay";
+ };
+
+ /* DDRC for CPU die scl #2 Channel #0 */
+ pmu_sccl1_ddrc0: pmu_ddrc0@60348000 {
+ compatible = "hisilicon,hisi-pmu-ddrc-v1";
+ scl-id = <0x02>;
+ ch-id = <0x0>;
+ num-events = <0x0d>;
+ num-counters = <0x04>;
+ reg = <0x0 0x60348000 0x0 0x10000>; /* TOTEMC DDRC0 */
+ status = "okay";
+ };
+
+ /* DDRC for CPU die scl #2 Channel #1 */
+ pmu_sccl1_ddrc1: pmu_ddrc1@60358000 {
+ compatible = "hisilicon,hisi-pmu-ddrc-v1";
+ scl-id = <0x02>;
+ ch-id = <0x01>;
+ num-events = <0x0d>;
+ num-counters = <0x04>;
+ reg = <0x0 0x60358000 0x0 0x10000>; /* TOTEMC DDRC1 */
+ status = "okay";
+ };
+
sas1: sas@a2000000 {
compatible = "hisilicon,hip06-sas-v2";
reg = <0 0xa2000000 0 0x10000>;
--
2.1.4
^ permalink raw reply related [flat|nested] 33+ messages in thread