* [PATCH 1/6] MIPS: Octeon: Enable per-CPU IRQs on all CPUs.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-07 2:35 ` [PATCH 2/6] MIPS: Add accessor macros for 64-bit performance counter registers David Daney
` (5 subsequent siblings)
6 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel; +Cc: David Daney, Thomas Gleixner
We cannot use on_each_cpu() from low-level irq code, as it ends up
being run with interrupts disabled (a no-no). Instead use some direct
IPI message bits to enable and disable the MIPS CPU interrupts.
Also, enable the irq on all CPUs for MIPS CPU interrupts.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
---
arch/mips/cavium-octeon/octeon-irq.c | 30 +++++++++++++++++++++++++++---
arch/mips/cavium-octeon/smp.c | 10 ++++++++++
2 files changed, 37 insertions(+), 3 deletions(-)
diff --git a/arch/mips/cavium-octeon/octeon-irq.c b/arch/mips/cavium-octeon/octeon-irq.c
index ce7500c..023cf04 100644
--- a/arch/mips/cavium-octeon/octeon-irq.c
+++ b/arch/mips/cavium-octeon/octeon-irq.c
@@ -56,7 +56,7 @@ static void octeon_irq_core_eoi(unsigned int irq)
set_c0_status(0x100 << bit);
}
-static void octeon_irq_core_enable(unsigned int irq)
+static void octeon_irq_core_enable_local(unsigned int irq)
{
unsigned long flags;
unsigned int bit = irq - OCTEON_IRQ_SW0;
@@ -83,16 +83,40 @@ static void octeon_irq_core_disable_local(unsigned int irq)
local_irq_restore(flags);
}
+extern void octeon_send_ipi_single(int cpu, unsigned int action);
+
static void octeon_irq_core_disable(unsigned int irq)
{
#ifdef CONFIG_SMP
- on_each_cpu((void (*)(void *)) octeon_irq_core_disable_local,
- (void *) (long) irq, 1);
+ unsigned int bit = irq - OCTEON_IRQ_SW0;
+ int cpu;
+ for_each_online_cpu(cpu) {
+ if (cpu == smp_processor_id())
+ octeon_irq_core_disable_local(irq);
+ else
+ octeon_send_ipi_single(cpu, 0x100 << bit);
+ }
#else
octeon_irq_core_disable_local(irq);
#endif
}
+static void octeon_irq_core_enable(unsigned int irq)
+{
+#ifdef CONFIG_SMP
+ unsigned int bit = irq - OCTEON_IRQ_SW0;
+ int cpu;
+ for_each_online_cpu(cpu) {
+ if (cpu == smp_processor_id())
+ octeon_irq_core_enable_local(irq);
+ else
+ octeon_send_ipi_single(cpu, 0x10000 << bit);
+ }
+#else
+ octeon_irq_core_enable_local(irq);
+#endif
+}
+
static struct irq_chip octeon_irq_chip_core = {
.name = "Core",
.enable = octeon_irq_core_enable,
diff --git a/arch/mips/cavium-octeon/smp.c b/arch/mips/cavium-octeon/smp.c
index 391cefe..92d819b 100644
--- a/arch/mips/cavium-octeon/smp.c
+++ b/arch/mips/cavium-octeon/smp.c
@@ -48,6 +48,16 @@ static irqreturn_t mailbox_interrupt(int irq, void *dev_id)
/* Check if we've been told to flush the icache */
if (action & SMP_ICACHE_FLUSH)
asm volatile ("synci 0($0)\n");
+ if (action & 0xff00) {
+ /* Disable MIPS CPU irq*/
+ unsigned int mask = action & 0xff00;
+ clear_c0_status(mask);
+ }
+ if (action & 0xff0000) {
+ /* Enable MIPS CPU irq*/
+ unsigned int mask = (action >> 8) & 0xff00;
+ set_c0_status(mask);
+ }
return IRQ_HANDLED;
}
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* [PATCH 2/6] MIPS: Add accessor macros for 64-bit performance counter registers.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
2011-01-07 2:35 ` [PATCH 1/6] MIPS: Octeon: Enable per-CPU IRQs on all CPUs David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-07 2:35 ` [PATCH 3/6] MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c David Daney
` (4 subsequent siblings)
6 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel; +Cc: David Daney
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
---
arch/mips/include/asm/mipsregs.h | 8 ++++++++
1 files changed, 8 insertions(+), 0 deletions(-)
diff --git a/arch/mips/include/asm/mipsregs.h b/arch/mips/include/asm/mipsregs.h
index d5c9eaa..86c7ec1 100644
--- a/arch/mips/include/asm/mipsregs.h
+++ b/arch/mips/include/asm/mipsregs.h
@@ -1019,18 +1019,26 @@ do { \
#define write_c0_perfctrl0(val) __write_32bit_c0_register($25, 0, val)
#define read_c0_perfcntr0() __read_32bit_c0_register($25, 1)
#define write_c0_perfcntr0(val) __write_32bit_c0_register($25, 1, val)
+#define read_c0_perfcntr0_64() __read_64bit_c0_register($25, 1)
+#define write_c0_perfcntr0_64(val) __write_64bit_c0_register($25, 1, val)
#define read_c0_perfctrl1() __read_32bit_c0_register($25, 2)
#define write_c0_perfctrl1(val) __write_32bit_c0_register($25, 2, val)
#define read_c0_perfcntr1() __read_32bit_c0_register($25, 3)
#define write_c0_perfcntr1(val) __write_32bit_c0_register($25, 3, val)
+#define read_c0_perfcntr1_64() __read_64bit_c0_register($25, 3)
+#define write_c0_perfcntr1_64(val) __write_64bit_c0_register($25, 3, val)
#define read_c0_perfctrl2() __read_32bit_c0_register($25, 4)
#define write_c0_perfctrl2(val) __write_32bit_c0_register($25, 4, val)
#define read_c0_perfcntr2() __read_32bit_c0_register($25, 5)
#define write_c0_perfcntr2(val) __write_32bit_c0_register($25, 5, val)
+#define read_c0_perfcntr2_64() __read_64bit_c0_register($25, 5)
+#define write_c0_perfcntr2_64(val) __write_64bit_c0_register($25, 5, val)
#define read_c0_perfctrl3() __read_32bit_c0_register($25, 6)
#define write_c0_perfctrl3(val) __write_32bit_c0_register($25, 6, val)
#define read_c0_perfcntr3() __read_32bit_c0_register($25, 7)
#define write_c0_perfcntr3(val) __write_32bit_c0_register($25, 7, val)
+#define read_c0_perfcntr3_64() __read_64bit_c0_register($25, 7)
+#define write_c0_perfcntr3_64(val) __write_64bit_c0_register($25, 7, val)
/* RM9000 PerfCount performance counter register */
#define read_c0_perfcount() __read_64bit_c0_register($25, 0)
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* [PATCH 3/6] MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
2011-01-07 2:35 ` [PATCH 1/6] MIPS: Octeon: Enable per-CPU IRQs on all CPUs David Daney
2011-01-07 2:35 ` [PATCH 2/6] MIPS: Add accessor macros for 64-bit performance counter registers David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-07 2:35 ` [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files David Daney
` (3 subsequent siblings)
6 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel
Cc: David Daney, Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo, Deng-Cheng Zhu
Get rid of a bunch of useless inline declarations, and join a bunch of
improperly split lines.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
---
arch/mips/kernel/perf_event.c | 26 +++++-------
arch/mips/kernel/perf_event_mipsxx.c | 68 +++++++++++++---------------------
2 files changed, 37 insertions(+), 57 deletions(-)
diff --git a/arch/mips/kernel/perf_event.c b/arch/mips/kernel/perf_event.c
index a824485..931d957 100644
--- a/arch/mips/kernel/perf_event.c
+++ b/arch/mips/kernel/perf_event.c
@@ -118,10 +118,9 @@ struct mips_pmu {
static const struct mips_pmu *mipspmu;
-static int
-mipspmu_event_set_period(struct perf_event *event,
- struct hw_perf_event *hwc,
- int idx)
+static int mipspmu_event_set_period(struct perf_event *event,
+ struct hw_perf_event *hwc,
+ int idx)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
s64 left = local64_read(&hwc->period_left);
@@ -162,8 +161,8 @@ mipspmu_event_set_period(struct perf_event *event,
}
static void mipspmu_event_update(struct perf_event *event,
- struct hw_perf_event *hwc,
- int idx)
+ struct hw_perf_event *hwc,
+ int idx)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
unsigned long flags;
@@ -422,8 +421,7 @@ static struct pmu pmu = {
.read = mipspmu_read,
};
-static inline unsigned int
-mipspmu_perf_event_encode(const struct mips_perf_event *pev)
+static unsigned int mipspmu_perf_event_encode(const struct mips_perf_event *pev)
{
/*
* Top 8 bits for range, next 16 bits for cntr_mask, lowest 8 bits for
@@ -439,8 +437,7 @@ mipspmu_perf_event_encode(const struct mips_perf_event *pev)
#endif
}
-static const struct mips_perf_event *
-mipspmu_map_general_event(int idx)
+static const struct mips_perf_event *mipspmu_map_general_event(int idx)
{
const struct mips_perf_event *pev;
@@ -451,8 +448,7 @@ mipspmu_map_general_event(int idx)
return pev;
}
-static const struct mips_perf_event *
-mipspmu_map_cache_event(u64 config)
+static const struct mips_perf_event *mipspmu_map_cache_event(u64 config)
{
unsigned int cache_type, cache_op, cache_result;
const struct mips_perf_event *pev;
@@ -515,9 +511,9 @@ static int validate_group(struct perf_event *event)
}
/* This is needed by specific irq handlers in perf_event_*.c */
-static void
-handle_associated_event(struct cpu_hw_events *cpuc,
- int idx, struct perf_sample_data *data, struct pt_regs *regs)
+static void handle_associated_event(struct cpu_hw_events *cpuc,
+ int idx, struct perf_sample_data *data,
+ struct pt_regs *regs)
{
struct perf_event *event = cpuc->events[idx];
struct hw_perf_event *hwc = &event->hw;
diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index c9406d6..a72e7247 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -49,37 +49,32 @@ static int cpu_has_mipsmt_pertccounters;
#endif
/* Copied from op_model_mipsxx.c */
-static inline unsigned int vpe_shift(void)
+static unsigned int vpe_shift(void)
{
if (num_possible_cpus() > 1)
return 1;
return 0;
}
-#else /* !CONFIG_MIPS_MT_SMP */
-#define vpe_id() 0
-
-static inline unsigned int vpe_shift(void)
-{
- return 0;
-}
-#endif /* CONFIG_MIPS_MT_SMP */
-static inline unsigned int
-counters_total_to_per_cpu(unsigned int counters)
+static unsigned int counters_total_to_per_cpu(unsigned int counters)
{
return counters >> vpe_shift();
}
-static inline unsigned int
-counters_per_cpu_to_total(unsigned int counters)
+static unsigned int counters_per_cpu_to_total(unsigned int counters)
{
return counters << vpe_shift();
}
+#else /* !CONFIG_MIPS_MT_SMP */
+#define vpe_id() 0
+
+#endif /* CONFIG_MIPS_MT_SMP */
+
#define __define_perf_accessors(r, n, np) \
\
-static inline unsigned int r_c0_ ## r ## n(void) \
+static unsigned int r_c0_ ## r ## n(void) \
{ \
unsigned int cpu = vpe_id(); \
\
@@ -94,7 +89,7 @@ static inline unsigned int r_c0_ ## r ## n(void) \
return 0; \
} \
\
-static inline void w_c0_ ## r ## n(unsigned int value) \
+static void w_c0_ ## r ## n(unsigned int value) \
{ \
unsigned int cpu = vpe_id(); \
\
@@ -121,7 +116,7 @@ __define_perf_accessors(perfctrl, 1, 3)
__define_perf_accessors(perfctrl, 2, 0)
__define_perf_accessors(perfctrl, 3, 1)
-static inline int __n_counters(void)
+static int __n_counters(void)
{
if (!(read_c0_config1() & M_CONFIG1_PC))
return 0;
@@ -135,7 +130,7 @@ static inline int __n_counters(void)
return 4;
}
-static inline int n_counters(void)
+static int n_counters(void)
{
int counters;
@@ -175,8 +170,7 @@ static void reset_counters(void *arg)
}
}
-static inline u64
-mipsxx_pmu_read_counter(unsigned int idx)
+static u64 mipsxx_pmu_read_counter(unsigned int idx)
{
switch (idx) {
case 0:
@@ -193,8 +187,7 @@ mipsxx_pmu_read_counter(unsigned int idx)
}
}
-static inline void
-mipsxx_pmu_write_counter(unsigned int idx, u64 val)
+static void mipsxx_pmu_write_counter(unsigned int idx, u64 val)
{
switch (idx) {
case 0:
@@ -212,8 +205,7 @@ mipsxx_pmu_write_counter(unsigned int idx, u64 val)
}
}
-static inline unsigned int
-mipsxx_pmu_read_control(unsigned int idx)
+static unsigned int mipsxx_pmu_read_control(unsigned int idx)
{
switch (idx) {
case 0:
@@ -230,8 +222,7 @@ mipsxx_pmu_read_control(unsigned int idx)
}
}
-static inline void
-mipsxx_pmu_write_control(unsigned int idx, unsigned int val)
+static void mipsxx_pmu_write_control(unsigned int idx, unsigned int val)
{
switch (idx) {
case 0:
@@ -483,9 +474,8 @@ static const struct mips_perf_event mipsxx74Kcore_cache_map
};
#ifdef CONFIG_MIPS_MT_SMP
-static void
-check_and_calc_range(struct perf_event *event,
- const struct mips_perf_event *pev)
+static void check_and_calc_range(struct perf_event *event,
+ const struct mips_perf_event *pev)
{
struct hw_perf_event *hwc = &event->hw;
@@ -508,9 +498,8 @@ check_and_calc_range(struct perf_event *event,
hwc->config_base |= M_TC_EN_ALL;
}
#else
-static void
-check_and_calc_range(struct perf_event *event,
- const struct mips_perf_event *pev)
+static void check_and_calc_range(struct perf_event *event,
+ const struct mips_perf_event *pev)
{
}
#endif
@@ -705,8 +694,7 @@ static int mipsxx_pmu_handle_shared_irq(void)
return handled;
}
-static irqreturn_t
-mipsxx_pmu_handle_irq(int irq, void *dev)
+static irqreturn_t mipsxx_pmu_handle_irq(int irq, void *dev)
{
return mipsxx_pmu_handle_shared_irq();
}
@@ -738,9 +726,8 @@ static void mipsxx_pmu_stop(void)
#endif
}
-static int
-mipsxx_pmu_alloc_counter(struct cpu_hw_events *cpuc,
- struct hw_perf_event *hwc)
+static int mipsxx_pmu_alloc_counter(struct cpu_hw_events *cpuc,
+ struct hw_perf_event *hwc)
{
int i;
@@ -769,8 +756,7 @@ mipsxx_pmu_alloc_counter(struct cpu_hw_events *cpuc,
return -EAGAIN;
}
-static void
-mipsxx_pmu_enable_event(struct hw_perf_event *evt, int idx)
+static void mipsxx_pmu_enable_event(struct hw_perf_event *evt, int idx)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
unsigned long flags;
@@ -788,8 +774,7 @@ mipsxx_pmu_enable_event(struct hw_perf_event *evt, int idx)
local_irq_restore(flags);
}
-static void
-mipsxx_pmu_disable_event(int idx)
+static void mipsxx_pmu_disable_event(int idx)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
unsigned long flags;
@@ -864,8 +849,7 @@ mipsxx_pmu_disable_event(int idx)
* then 128 needs to be added to 15 as the input for the event config,
* i.e., 143 (0x8F) to be used.
*/
-static const struct mips_perf_event *
-mipsxx_pmu_map_raw_event(u64 config)
+static const struct mips_perf_event *mipsxx_pmu_map_raw_event(u64 config)
{
unsigned int raw_id = config & 0xff;
unsigned int base_id = raw_id & 0x7f;
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
` (2 preceding siblings ...)
2011-01-07 2:35 ` [PATCH 3/6] MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-20 10:01 ` Deng-Cheng Zhu
2011-01-07 2:35 ` [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters David Daney
` (2 subsequent siblings)
6 siblings, 1 reply; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel
Cc: David Daney, Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo, Dezhong Diao, Gabor Juhos, Grant Likely,
Deng-Cheng Zhu
The contents of arch/mips/kernel/perf_event.c and
arch/mips/kernel/perf_event_mipsxx.c were divided in a seemingly ad
hoc manner, with the first including the second.
I moved all the hardware counter support code to perf_event_mipsxx.c
and removed the gating #ifdefs to the Kconfig and Makefile.
Now perf_event.c contains only the callchain support, everything else
is in perf_event_mipsxx.c
There are no code changes, only moving of functions from one file to
the other, or removing empty unneeded functions.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: David Daney <ddaney@caviumnetworks.com>
Cc: Dezhong Diao <dediao@cisco.com>
Cc: Gabor Juhos <juhosg@openwrt.org>
Cc: Grant Likely <grant.likely@secretlab.ca>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
---
arch/mips/Kconfig | 2 +-
arch/mips/kernel/Makefile | 5 +-
arch/mips/kernel/perf_event.c | 517 +---------------------------------
arch/mips/kernel/perf_event_mipsxx.c | 532 +++++++++++++++++++++++++++++++++-
4 files changed, 534 insertions(+), 522 deletions(-)
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index acfa1fe..88aaec2 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -2016,7 +2016,7 @@ config NODES_SHIFT
config HW_PERF_EVENTS
bool "Enable hardware performance counter support for perf events"
- depends on PERF_EVENTS && !MIPS_MT_SMTC && OPROFILE=n && CPU_MIPS32
+ depends on PERF_EVENTS && !MIPS_MT_SMTC && OPROFILE=n && (CPU_MIPS32 || CPU_MIPS64 || CPU_R10000 || CPU_SB1)
default y
help
Enable hardware performance counter support for perf events. If
diff --git a/arch/mips/kernel/Makefile b/arch/mips/kernel/Makefile
index cedee2b..753b421 100644
--- a/arch/mips/kernel/Makefile
+++ b/arch/mips/kernel/Makefile
@@ -11,6 +11,8 @@ obj-y += cpu-probe.o branch.o entry.o genex.o irq.o process.o \
ifdef CONFIG_FUNCTION_TRACER
CFLAGS_REMOVE_ftrace.o = -pg
CFLAGS_REMOVE_early_printk.o = -pg
+CFLAGS_REMOVE_perf_event.o = -pg
+CFLAGS_REMOVE_perf_event_mipsxx.o = -pg
endif
obj-$(CONFIG_CEVT_BCM1480) += cevt-bcm1480.o
@@ -105,7 +107,8 @@ obj-$(CONFIG_HAVE_STD_PC_SERIAL_PORT) += 8250-platform.o
obj-$(CONFIG_MIPS_CPUFREQ) += cpufreq/
-obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o
+obj-$(CONFIG_PERF_EVENTS) += perf_event.o
+obj-$(CONFIG_HW_PERF_EVENTS) += perf_event_mipsxx.o
obj-$(CONFIG_JUMP_LABEL) += jump_label.o
diff --git a/arch/mips/kernel/perf_event.c b/arch/mips/kernel/perf_event.c
index 931d957..c1cf9c6 100644
--- a/arch/mips/kernel/perf_event.c
+++ b/arch/mips/kernel/perf_event.c
@@ -14,531 +14,16 @@
* published by the Free Software Foundation.
*/
-#include <linux/cpumask.h>
-#include <linux/interrupt.h>
-#include <linux/smp.h>
-#include <linux/kernel.h>
#include <linux/perf_event.h>
-#include <linux/uaccess.h>
-#include <asm/irq.h>
-#include <asm/irq_regs.h>
#include <asm/stacktrace.h>
-#include <asm/time.h> /* For perf_irq */
-
-/* These are for 32bit counters. For 64bit ones, define them accordingly. */
-#define MAX_PERIOD ((1ULL << 32) - 1)
-#define VALID_COUNT 0x7fffffff
-#define TOTAL_BITS 32
-#define HIGHEST_BIT 31
-
-#define MIPS_MAX_HWEVENTS 4
-
-struct cpu_hw_events {
- /* Array of events on this cpu. */
- struct perf_event *events[MIPS_MAX_HWEVENTS];
-
- /*
- * Set the bit (indexed by the counter number) when the counter
- * is used for an event.
- */
- unsigned long used_mask[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
-
- /*
- * The borrowed MSB for the performance counter. A MIPS performance
- * counter uses its bit 31 (for 32bit counters) or bit 63 (for 64bit
- * counters) as a factor of determining whether a counter overflow
- * should be signaled. So here we use a separate MSB for each
- * counter to make things easy.
- */
- unsigned long msbs[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
-
- /*
- * Software copy of the control register for each performance counter.
- * MIPS CPUs vary in performance counters. They use this differently,
- * and even may not use it.
- */
- unsigned int saved_ctrl[MIPS_MAX_HWEVENTS];
-};
-DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
- .saved_ctrl = {0},
-};
-
-/* The description of MIPS performance events. */
-struct mips_perf_event {
- unsigned int event_id;
- /*
- * MIPS performance counters are indexed starting from 0.
- * CNTR_EVEN indicates the indexes of the counters to be used are
- * even numbers.
- */
- unsigned int cntr_mask;
- #define CNTR_EVEN 0x55555555
- #define CNTR_ODD 0xaaaaaaaa
-#ifdef CONFIG_MIPS_MT_SMP
- enum {
- T = 0,
- V = 1,
- P = 2,
- } range;
-#else
- #define T
- #define V
- #define P
-#endif
-};
-
-static struct mips_perf_event raw_event;
-static DEFINE_MUTEX(raw_event_mutex);
-
-#define UNSUPPORTED_PERF_EVENT_ID 0xffffffff
-#define C(x) PERF_COUNT_HW_CACHE_##x
-
-struct mips_pmu {
- const char *name;
- int irq;
- irqreturn_t (*handle_irq)(int irq, void *dev);
- int (*handle_shared_irq)(void);
- void (*start)(void);
- void (*stop)(void);
- int (*alloc_counter)(struct cpu_hw_events *cpuc,
- struct hw_perf_event *hwc);
- u64 (*read_counter)(unsigned int idx);
- void (*write_counter)(unsigned int idx, u64 val);
- void (*enable_event)(struct hw_perf_event *evt, int idx);
- void (*disable_event)(int idx);
- const struct mips_perf_event *(*map_raw_event)(u64 config);
- const struct mips_perf_event (*general_event_map)[PERF_COUNT_HW_MAX];
- const struct mips_perf_event (*cache_event_map)
- [PERF_COUNT_HW_CACHE_MAX]
- [PERF_COUNT_HW_CACHE_OP_MAX]
- [PERF_COUNT_HW_CACHE_RESULT_MAX];
- unsigned int num_counters;
-};
-
-static const struct mips_pmu *mipspmu;
-
-static int mipspmu_event_set_period(struct perf_event *event,
- struct hw_perf_event *hwc,
- int idx)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- s64 left = local64_read(&hwc->period_left);
- s64 period = hwc->sample_period;
- int ret = 0;
- u64 uleft;
- unsigned long flags;
-
- if (unlikely(left <= -period)) {
- left = period;
- local64_set(&hwc->period_left, left);
- hwc->last_period = period;
- ret = 1;
- }
-
- if (unlikely(left <= 0)) {
- left += period;
- local64_set(&hwc->period_left, left);
- hwc->last_period = period;
- ret = 1;
- }
-
- if (left > (s64)MAX_PERIOD)
- left = MAX_PERIOD;
-
- local64_set(&hwc->prev_count, (u64)-left);
-
- local_irq_save(flags);
- uleft = (u64)(-left) & MAX_PERIOD;
- uleft > VALID_COUNT ?
- set_bit(idx, cpuc->msbs) : clear_bit(idx, cpuc->msbs);
- mipspmu->write_counter(idx, (u64)(-left) & VALID_COUNT);
- local_irq_restore(flags);
-
- perf_event_update_userpage(event);
-
- return ret;
-}
-
-static void mipspmu_event_update(struct perf_event *event,
- struct hw_perf_event *hwc,
- int idx)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- unsigned long flags;
- int shift = 64 - TOTAL_BITS;
- s64 prev_raw_count, new_raw_count;
- u64 delta;
-
-again:
- prev_raw_count = local64_read(&hwc->prev_count);
- local_irq_save(flags);
- /* Make the counter value be a "real" one. */
- new_raw_count = mipspmu->read_counter(idx);
- if (new_raw_count & (test_bit(idx, cpuc->msbs) << HIGHEST_BIT)) {
- new_raw_count &= VALID_COUNT;
- clear_bit(idx, cpuc->msbs);
- } else
- new_raw_count |= (test_bit(idx, cpuc->msbs) << HIGHEST_BIT);
- local_irq_restore(flags);
-
- if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
- new_raw_count) != prev_raw_count)
- goto again;
-
- delta = (new_raw_count << shift) - (prev_raw_count << shift);
- delta >>= shift;
-
- local64_add(delta, &event->count);
- local64_sub(delta, &hwc->period_left);
-
- return;
-}
-
-static void mipspmu_start(struct perf_event *event, int flags)
-{
- struct hw_perf_event *hwc = &event->hw;
-
- if (!mipspmu)
- return;
-
- if (flags & PERF_EF_RELOAD)
- WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
-
- hwc->state = 0;
-
- /* Set the period for the event. */
- mipspmu_event_set_period(event, hwc, hwc->idx);
-
- /* Enable the event. */
- mipspmu->enable_event(hwc, hwc->idx);
-}
-
-static void mipspmu_stop(struct perf_event *event, int flags)
-{
- struct hw_perf_event *hwc = &event->hw;
-
- if (!mipspmu)
- return;
-
- if (!(hwc->state & PERF_HES_STOPPED)) {
- /* We are working on a local event. */
- mipspmu->disable_event(hwc->idx);
- barrier();
- mipspmu_event_update(event, hwc, hwc->idx);
- hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
- }
-}
-
-static int mipspmu_add(struct perf_event *event, int flags)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- struct hw_perf_event *hwc = &event->hw;
- int idx;
- int err = 0;
-
- perf_pmu_disable(event->pmu);
-
- /* To look for a free counter for this event. */
- idx = mipspmu->alloc_counter(cpuc, hwc);
- if (idx < 0) {
- err = idx;
- goto out;
- }
-
- /*
- * If there is an event in the counter we are going to use then
- * make sure it is disabled.
- */
- event->hw.idx = idx;
- mipspmu->disable_event(idx);
- cpuc->events[idx] = event;
-
- hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
- if (flags & PERF_EF_START)
- mipspmu_start(event, PERF_EF_RELOAD);
-
- /* Propagate our changes to the userspace mapping. */
- perf_event_update_userpage(event);
-
-out:
- perf_pmu_enable(event->pmu);
- return err;
-}
-
-static void mipspmu_del(struct perf_event *event, int flags)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- struct hw_perf_event *hwc = &event->hw;
- int idx = hwc->idx;
-
- WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
-
- mipspmu_stop(event, PERF_EF_UPDATE);
- cpuc->events[idx] = NULL;
- clear_bit(idx, cpuc->used_mask);
-
- perf_event_update_userpage(event);
-}
-
-static void mipspmu_read(struct perf_event *event)
-{
- struct hw_perf_event *hwc = &event->hw;
-
- /* Don't read disabled counters! */
- if (hwc->idx < 0)
- return;
-
- mipspmu_event_update(event, hwc, hwc->idx);
-}
-
-static void mipspmu_enable(struct pmu *pmu)
-{
- if (mipspmu)
- mipspmu->start();
-}
-
-static void mipspmu_disable(struct pmu *pmu)
-{
- if (mipspmu)
- mipspmu->stop();
-}
-
-static atomic_t active_events = ATOMIC_INIT(0);
-static DEFINE_MUTEX(pmu_reserve_mutex);
-static int (*save_perf_irq)(void);
-
-static int mipspmu_get_irq(void)
-{
- int err;
-
- if (mipspmu->irq >= 0) {
- /* Request my own irq handler. */
- err = request_irq(mipspmu->irq, mipspmu->handle_irq,
- IRQF_DISABLED | IRQF_NOBALANCING,
- "mips_perf_pmu", NULL);
- if (err) {
- pr_warning("Unable to request IRQ%d for MIPS "
- "performance counters!\n", mipspmu->irq);
- }
- } else if (cp0_perfcount_irq < 0) {
- /*
- * We are sharing the irq number with the timer interrupt.
- */
- save_perf_irq = perf_irq;
- perf_irq = mipspmu->handle_shared_irq;
- err = 0;
- } else {
- pr_warning("The platform hasn't properly defined its "
- "interrupt controller.\n");
- err = -ENOENT;
- }
-
- return err;
-}
-
-static void mipspmu_free_irq(void)
-{
- if (mipspmu->irq >= 0)
- free_irq(mipspmu->irq, NULL);
- else if (cp0_perfcount_irq < 0)
- perf_irq = save_perf_irq;
-}
-
-/*
- * mipsxx/rm9000/loongson2 have different performance counters, they have
- * specific low-level init routines.
- */
-static void reset_counters(void *arg);
-static int __hw_perf_event_init(struct perf_event *event);
-
-static void hw_perf_event_destroy(struct perf_event *event)
-{
- if (atomic_dec_and_mutex_lock(&active_events,
- &pmu_reserve_mutex)) {
- /*
- * We must not call the destroy function with interrupts
- * disabled.
- */
- on_each_cpu(reset_counters,
- (void *)(long)mipspmu->num_counters, 1);
- mipspmu_free_irq();
- mutex_unlock(&pmu_reserve_mutex);
- }
-}
-
-static int mipspmu_event_init(struct perf_event *event)
-{
- int err = 0;
-
- switch (event->attr.type) {
- case PERF_TYPE_RAW:
- case PERF_TYPE_HARDWARE:
- case PERF_TYPE_HW_CACHE:
- break;
-
- default:
- return -ENOENT;
- }
-
- if (!mipspmu || event->cpu >= nr_cpumask_bits ||
- (event->cpu >= 0 && !cpu_online(event->cpu)))
- return -ENODEV;
-
- if (!atomic_inc_not_zero(&active_events)) {
- if (atomic_read(&active_events) > MIPS_MAX_HWEVENTS) {
- atomic_dec(&active_events);
- return -ENOSPC;
- }
-
- mutex_lock(&pmu_reserve_mutex);
- if (atomic_read(&active_events) == 0)
- err = mipspmu_get_irq();
-
- if (!err)
- atomic_inc(&active_events);
- mutex_unlock(&pmu_reserve_mutex);
- }
-
- if (err)
- return err;
-
- err = __hw_perf_event_init(event);
- if (err)
- hw_perf_event_destroy(event);
-
- return err;
-}
-
-static struct pmu pmu = {
- .pmu_enable = mipspmu_enable,
- .pmu_disable = mipspmu_disable,
- .event_init = mipspmu_event_init,
- .add = mipspmu_add,
- .del = mipspmu_del,
- .start = mipspmu_start,
- .stop = mipspmu_stop,
- .read = mipspmu_read,
-};
-
-static unsigned int mipspmu_perf_event_encode(const struct mips_perf_event *pev)
-{
-/*
- * Top 8 bits for range, next 16 bits for cntr_mask, lowest 8 bits for
- * event_id.
- */
-#ifdef CONFIG_MIPS_MT_SMP
- return ((unsigned int)pev->range << 24) |
- (pev->cntr_mask & 0xffff00) |
- (pev->event_id & 0xff);
-#else
- return (pev->cntr_mask & 0xffff00) |
- (pev->event_id & 0xff);
-#endif
-}
-
-static const struct mips_perf_event *mipspmu_map_general_event(int idx)
-{
- const struct mips_perf_event *pev;
-
- pev = ((*mipspmu->general_event_map)[idx].event_id ==
- UNSUPPORTED_PERF_EVENT_ID ? ERR_PTR(-EOPNOTSUPP) :
- &(*mipspmu->general_event_map)[idx]);
-
- return pev;
-}
-
-static const struct mips_perf_event *mipspmu_map_cache_event(u64 config)
-{
- unsigned int cache_type, cache_op, cache_result;
- const struct mips_perf_event *pev;
-
- cache_type = (config >> 0) & 0xff;
- if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
- return ERR_PTR(-EINVAL);
-
- cache_op = (config >> 8) & 0xff;
- if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
- return ERR_PTR(-EINVAL);
-
- cache_result = (config >> 16) & 0xff;
- if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
- return ERR_PTR(-EINVAL);
-
- pev = &((*mipspmu->cache_event_map)
- [cache_type]
- [cache_op]
- [cache_result]);
-
- if (pev->event_id == UNSUPPORTED_PERF_EVENT_ID)
- return ERR_PTR(-EOPNOTSUPP);
-
- return pev;
-
-}
-
-static int validate_event(struct cpu_hw_events *cpuc,
- struct perf_event *event)
-{
- struct hw_perf_event fake_hwc = event->hw;
-
- /* Allow mixed event group. So return 1 to pass validation. */
- if (event->pmu != &pmu || event->state <= PERF_EVENT_STATE_OFF)
- return 1;
-
- return mipspmu->alloc_counter(cpuc, &fake_hwc) >= 0;
-}
-
-static int validate_group(struct perf_event *event)
-{
- struct perf_event *sibling, *leader = event->group_leader;
- struct cpu_hw_events fake_cpuc;
-
- memset(&fake_cpuc, 0, sizeof(fake_cpuc));
-
- if (!validate_event(&fake_cpuc, leader))
- return -ENOSPC;
-
- list_for_each_entry(sibling, &leader->sibling_list, group_entry) {
- if (!validate_event(&fake_cpuc, sibling))
- return -ENOSPC;
- }
-
- if (!validate_event(&fake_cpuc, event))
- return -ENOSPC;
-
- return 0;
-}
-
-/* This is needed by specific irq handlers in perf_event_*.c */
-static void handle_associated_event(struct cpu_hw_events *cpuc,
- int idx, struct perf_sample_data *data,
- struct pt_regs *regs)
-{
- struct perf_event *event = cpuc->events[idx];
- struct hw_perf_event *hwc = &event->hw;
-
- mipspmu_event_update(event, hwc, idx);
- data->period = event->hw.last_period;
- if (!mipspmu_event_set_period(event, hwc, idx))
- return;
-
- if (perf_event_overflow(event, 0, data, regs))
- mipspmu->disable_event(idx);
-}
-
-#include "perf_event_mipsxx.c"
/* Callchain handling code. */
/*
* Leave userspace callchain empty for now. When we find a way to trace
- * the user stack callchains, we add here.
+ * the user stack callchains, we will add it here.
*/
-void perf_callchain_user(struct perf_callchain_entry *entry,
- struct pt_regs *regs)
-{
-}
static void save_raw_perf_callchain(struct perf_callchain_entry *entry,
unsigned long reg29)
diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index a72e7247..498dd85 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -1,5 +1,531 @@
-#if defined(CONFIG_CPU_MIPS32) || defined(CONFIG_CPU_MIPS64) || \
- defined(CONFIG_CPU_R10000) || defined(CONFIG_CPU_SB1)
+/*
+ * Linux performance counter support for MIPS.
+ *
+ * Copyright (C) 2010 MIPS Technologies, Inc.
+ * Author: Deng-Cheng Zhu
+ *
+ * This code is based on the implementation for ARM, which is in turn
+ * based on the sparc64 perf event code and the x86 code. Performance
+ * counter access is based on the MIPS Oprofile code. And the callchain
+ * support references the code of MIPS stacktrace.c.
+ *
+ * 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/cpumask.h>
+#include <linux/interrupt.h>
+#include <linux/smp.h>
+#include <linux/kernel.h>
+#include <linux/perf_event.h>
+#include <linux/uaccess.h>
+
+#include <asm/irq.h>
+#include <asm/irq_regs.h>
+#include <asm/stacktrace.h>
+#include <asm/time.h> /* For perf_irq */
+
+/* These are for 32bit counters. For 64bit ones, define them accordingly. */
+#define MAX_PERIOD ((1ULL << 32) - 1)
+#define VALID_COUNT 0x7fffffff
+#define TOTAL_BITS 32
+#define HIGHEST_BIT 31
+
+#define MIPS_MAX_HWEVENTS 4
+
+struct cpu_hw_events {
+ /* Array of events on this cpu. */
+ struct perf_event *events[MIPS_MAX_HWEVENTS];
+
+ /*
+ * Set the bit (indexed by the counter number) when the counter
+ * is used for an event.
+ */
+ unsigned long used_mask[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
+
+ /*
+ * The borrowed MSB for the performance counter. A MIPS performance
+ * counter uses its bit 31 (for 32bit counters) or bit 63 (for 64bit
+ * counters) as a factor of determining whether a counter overflow
+ * should be signaled. So here we use a separate MSB for each
+ * counter to make things easy.
+ */
+ unsigned long msbs[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
+
+ /*
+ * Software copy of the control register for each performance counter.
+ * MIPS CPUs vary in performance counters. They use this differently,
+ * and even may not use it.
+ */
+ unsigned int saved_ctrl[MIPS_MAX_HWEVENTS];
+};
+DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
+ .saved_ctrl = {0},
+};
+
+/* The description of MIPS performance events. */
+struct mips_perf_event {
+ unsigned int event_id;
+ /*
+ * MIPS performance counters are indexed starting from 0.
+ * CNTR_EVEN indicates the indexes of the counters to be used are
+ * even numbers.
+ */
+ unsigned int cntr_mask;
+ #define CNTR_EVEN 0x55555555
+ #define CNTR_ODD 0xaaaaaaaa
+#ifdef CONFIG_MIPS_MT_SMP
+ enum {
+ T = 0,
+ V = 1,
+ P = 2,
+ } range;
+#else
+ #define T
+ #define V
+ #define P
+#endif
+};
+
+static struct mips_perf_event raw_event;
+static DEFINE_MUTEX(raw_event_mutex);
+
+#define UNSUPPORTED_PERF_EVENT_ID 0xffffffff
+#define C(x) PERF_COUNT_HW_CACHE_##x
+
+struct mips_pmu {
+ const char *name;
+ int irq;
+ irqreturn_t (*handle_irq)(int irq, void *dev);
+ int (*handle_shared_irq)(void);
+ void (*start)(void);
+ void (*stop)(void);
+ int (*alloc_counter)(struct cpu_hw_events *cpuc,
+ struct hw_perf_event *hwc);
+ u64 (*read_counter)(unsigned int idx);
+ void (*write_counter)(unsigned int idx, u64 val);
+ void (*enable_event)(struct hw_perf_event *evt, int idx);
+ void (*disable_event)(int idx);
+ const struct mips_perf_event *(*map_raw_event)(u64 config);
+ const struct mips_perf_event (*general_event_map)[PERF_COUNT_HW_MAX];
+ const struct mips_perf_event (*cache_event_map)
+ [PERF_COUNT_HW_CACHE_MAX]
+ [PERF_COUNT_HW_CACHE_OP_MAX]
+ [PERF_COUNT_HW_CACHE_RESULT_MAX];
+ unsigned int num_counters;
+};
+
+static const struct mips_pmu *mipspmu;
+
+static int mipspmu_event_set_period(struct perf_event *event,
+ struct hw_perf_event *hwc,
+ int idx)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ s64 left = local64_read(&hwc->period_left);
+ s64 period = hwc->sample_period;
+ int ret = 0;
+ u64 uleft;
+ unsigned long flags;
+
+ if (unlikely(left <= -period)) {
+ left = period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+
+ if (unlikely(left <= 0)) {
+ left += period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+
+ if (left > (s64)MAX_PERIOD)
+ left = MAX_PERIOD;
+
+ local64_set(&hwc->prev_count, (u64)-left);
+
+ local_irq_save(flags);
+ uleft = (u64)(-left) & MAX_PERIOD;
+ uleft > VALID_COUNT ?
+ set_bit(idx, cpuc->msbs) : clear_bit(idx, cpuc->msbs);
+ mipspmu->write_counter(idx, (u64)(-left) & VALID_COUNT);
+ local_irq_restore(flags);
+
+ perf_event_update_userpage(event);
+
+ return ret;
+}
+
+static void mipspmu_event_update(struct perf_event *event,
+ struct hw_perf_event *hwc,
+ int idx)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ unsigned long flags;
+ int shift = 64 - TOTAL_BITS;
+ s64 prev_raw_count, new_raw_count;
+ u64 delta;
+
+again:
+ prev_raw_count = local64_read(&hwc->prev_count);
+ local_irq_save(flags);
+ /* Make the counter value be a "real" one. */
+ new_raw_count = mipspmu->read_counter(idx);
+ if (new_raw_count & (test_bit(idx, cpuc->msbs) << HIGHEST_BIT)) {
+ new_raw_count &= VALID_COUNT;
+ clear_bit(idx, cpuc->msbs);
+ } else
+ new_raw_count |= (test_bit(idx, cpuc->msbs) << HIGHEST_BIT);
+ local_irq_restore(flags);
+
+ if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+ new_raw_count) != prev_raw_count)
+ goto again;
+
+ delta = (new_raw_count << shift) - (prev_raw_count << shift);
+ delta >>= shift;
+
+ local64_add(delta, &event->count);
+ local64_sub(delta, &hwc->period_left);
+
+ return;
+}
+
+static void mipspmu_start(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ if (!mipspmu)
+ return;
+
+ if (flags & PERF_EF_RELOAD)
+ WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+
+ hwc->state = 0;
+
+ /* Set the period for the event. */
+ mipspmu_event_set_period(event, hwc, hwc->idx);
+
+ /* Enable the event. */
+ mipspmu->enable_event(hwc, hwc->idx);
+}
+
+static void mipspmu_stop(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ if (!mipspmu)
+ return;
+
+ if (!(hwc->state & PERF_HES_STOPPED)) {
+ /* We are working on a local event. */
+ mipspmu->disable_event(hwc->idx);
+ barrier();
+ mipspmu_event_update(event, hwc, hwc->idx);
+ hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
+ }
+}
+
+static int mipspmu_add(struct perf_event *event, int flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ struct hw_perf_event *hwc = &event->hw;
+ int idx;
+ int err = 0;
+
+ perf_pmu_disable(event->pmu);
+
+ /* To look for a free counter for this event. */
+ idx = mipspmu->alloc_counter(cpuc, hwc);
+ if (idx < 0) {
+ err = idx;
+ goto out;
+ }
+
+ /*
+ * If there is an event in the counter we are going to use then
+ * make sure it is disabled.
+ */
+ event->hw.idx = idx;
+ mipspmu->disable_event(idx);
+ cpuc->events[idx] = event;
+
+ hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
+ if (flags & PERF_EF_START)
+ mipspmu_start(event, PERF_EF_RELOAD);
+
+ /* Propagate our changes to the userspace mapping. */
+ perf_event_update_userpage(event);
+
+out:
+ perf_pmu_enable(event->pmu);
+ return err;
+}
+
+static void mipspmu_del(struct perf_event *event, int flags)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ struct hw_perf_event *hwc = &event->hw;
+ int idx = hwc->idx;
+
+ WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
+
+ mipspmu_stop(event, PERF_EF_UPDATE);
+ cpuc->events[idx] = NULL;
+ clear_bit(idx, cpuc->used_mask);
+
+ perf_event_update_userpage(event);
+}
+
+static void mipspmu_read(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+
+ /* Don't read disabled counters! */
+ if (hwc->idx < 0)
+ return;
+
+ mipspmu_event_update(event, hwc, hwc->idx);
+}
+
+static void mipspmu_enable(struct pmu *pmu)
+{
+ if (mipspmu)
+ mipspmu->start();
+}
+
+static void mipspmu_disable(struct pmu *pmu)
+{
+ if (mipspmu)
+ mipspmu->stop();
+}
+
+static atomic_t active_events = ATOMIC_INIT(0);
+static DEFINE_MUTEX(pmu_reserve_mutex);
+static int (*save_perf_irq)(void);
+
+static int mipspmu_get_irq(void)
+{
+ int err;
+
+ if (mipspmu->irq >= 0) {
+ /* Request my own irq handler. */
+ err = request_irq(mipspmu->irq, mipspmu->handle_irq,
+ IRQF_DISABLED | IRQF_NOBALANCING,
+ "mips_perf_pmu", NULL);
+ if (err) {
+ pr_warning("Unable to request IRQ%d for MIPS "
+ "performance counters!\n", mipspmu->irq);
+ }
+ } else if (cp0_perfcount_irq < 0) {
+ /*
+ * We are sharing the irq number with the timer interrupt.
+ */
+ save_perf_irq = perf_irq;
+ perf_irq = mipspmu->handle_shared_irq;
+ err = 0;
+ } else {
+ pr_warning("The platform hasn't properly defined its "
+ "interrupt controller.\n");
+ err = -ENOENT;
+ }
+
+ return err;
+}
+
+static void mipspmu_free_irq(void)
+{
+ if (mipspmu->irq >= 0)
+ free_irq(mipspmu->irq, NULL);
+ else if (cp0_perfcount_irq < 0)
+ perf_irq = save_perf_irq;
+}
+
+/*
+ * mipsxx/rm9000/loongson2 have different performance counters, they have
+ * specific low-level init routines.
+ */
+static void reset_counters(void *arg);
+static int __hw_perf_event_init(struct perf_event *event);
+
+static void hw_perf_event_destroy(struct perf_event *event)
+{
+ if (atomic_dec_and_mutex_lock(&active_events,
+ &pmu_reserve_mutex)) {
+ /*
+ * We must not call the destroy function with interrupts
+ * disabled.
+ */
+ on_each_cpu(reset_counters,
+ (void *)(long)mipspmu->num_counters, 1);
+ mipspmu_free_irq();
+ mutex_unlock(&pmu_reserve_mutex);
+ }
+}
+
+static int mipspmu_event_init(struct perf_event *event)
+{
+ int err = 0;
+
+ switch (event->attr.type) {
+ case PERF_TYPE_RAW:
+ case PERF_TYPE_HARDWARE:
+ case PERF_TYPE_HW_CACHE:
+ break;
+
+ default:
+ return -ENOENT;
+ }
+
+ if (!mipspmu || event->cpu >= nr_cpumask_bits ||
+ (event->cpu >= 0 && !cpu_online(event->cpu)))
+ return -ENODEV;
+
+ if (!atomic_inc_not_zero(&active_events)) {
+ if (atomic_read(&active_events) > MIPS_MAX_HWEVENTS) {
+ atomic_dec(&active_events);
+ return -ENOSPC;
+ }
+
+ mutex_lock(&pmu_reserve_mutex);
+ if (atomic_read(&active_events) == 0)
+ err = mipspmu_get_irq();
+
+ if (!err)
+ atomic_inc(&active_events);
+ mutex_unlock(&pmu_reserve_mutex);
+ }
+
+ if (err)
+ return err;
+
+ err = __hw_perf_event_init(event);
+ if (err)
+ hw_perf_event_destroy(event);
+
+ return err;
+}
+
+static struct pmu pmu = {
+ .pmu_enable = mipspmu_enable,
+ .pmu_disable = mipspmu_disable,
+ .event_init = mipspmu_event_init,
+ .add = mipspmu_add,
+ .del = mipspmu_del,
+ .start = mipspmu_start,
+ .stop = mipspmu_stop,
+ .read = mipspmu_read,
+};
+
+static unsigned int mipspmu_perf_event_encode(const struct mips_perf_event *pev)
+{
+/*
+ * Top 8 bits for range, next 16 bits for cntr_mask, lowest 8 bits for
+ * event_id.
+ */
+#ifdef CONFIG_MIPS_MT_SMP
+ return ((unsigned int)pev->range << 24) |
+ (pev->cntr_mask & 0xffff00) |
+ (pev->event_id & 0xff);
+#else
+ return (pev->cntr_mask & 0xffff00) |
+ (pev->event_id & 0xff);
+#endif
+}
+
+static const struct mips_perf_event *mipspmu_map_general_event(int idx)
+{
+ const struct mips_perf_event *pev;
+
+ pev = ((*mipspmu->general_event_map)[idx].event_id ==
+ UNSUPPORTED_PERF_EVENT_ID ? ERR_PTR(-EOPNOTSUPP) :
+ &(*mipspmu->general_event_map)[idx]);
+
+ return pev;
+}
+
+static const struct mips_perf_event *mipspmu_map_cache_event(u64 config)
+{
+ unsigned int cache_type, cache_op, cache_result;
+ const struct mips_perf_event *pev;
+
+ cache_type = (config >> 0) & 0xff;
+ if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
+ return ERR_PTR(-EINVAL);
+
+ cache_op = (config >> 8) & 0xff;
+ if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
+ return ERR_PTR(-EINVAL);
+
+ cache_result = (config >> 16) & 0xff;
+ if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
+ return ERR_PTR(-EINVAL);
+
+ pev = &((*mipspmu->cache_event_map)
+ [cache_type]
+ [cache_op]
+ [cache_result]);
+
+ if (pev->event_id == UNSUPPORTED_PERF_EVENT_ID)
+ return ERR_PTR(-EOPNOTSUPP);
+
+ return pev;
+
+}
+
+static int validate_event(struct cpu_hw_events *cpuc,
+ struct perf_event *event)
+{
+ struct hw_perf_event fake_hwc = event->hw;
+
+ /* Allow mixed event group. So return 1 to pass validation. */
+ if (event->pmu != &pmu || event->state <= PERF_EVENT_STATE_OFF)
+ return 1;
+
+ return mipspmu->alloc_counter(cpuc, &fake_hwc) >= 0;
+}
+
+static int validate_group(struct perf_event *event)
+{
+ struct perf_event *sibling, *leader = event->group_leader;
+ struct cpu_hw_events fake_cpuc;
+
+ memset(&fake_cpuc, 0, sizeof(fake_cpuc));
+
+ if (!validate_event(&fake_cpuc, leader))
+ return -ENOSPC;
+
+ list_for_each_entry(sibling, &leader->sibling_list, group_entry) {
+ if (!validate_event(&fake_cpuc, sibling))
+ return -ENOSPC;
+ }
+
+ if (!validate_event(&fake_cpuc, event))
+ return -ENOSPC;
+
+ return 0;
+}
+
+/* This is needed by specific irq handlers in perf_event_*.c */
+static void handle_associated_event(struct cpu_hw_events *cpuc,
+ int idx, struct perf_sample_data *data,
+ struct pt_regs *regs)
+{
+ struct perf_event *event = cpuc->events[idx];
+ struct hw_perf_event *hwc = &event->hw;
+
+ mipspmu_event_update(event, hwc, idx);
+ data->period = event->hw.last_period;
+ if (!mipspmu_event_set_period(event, hwc, idx))
+ return;
+
+ if (perf_event_overflow(event, 0, data, regs))
+ mipspmu->disable_event(idx);
+}
#define M_CONFIG1_PC (1 << 4)
@@ -1034,5 +1560,3 @@ init_hw_perf_events(void)
return 0;
}
arch_initcall(init_hw_perf_events);
-
-#endif /* defined(CONFIG_CPU_MIPS32)... */
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* Re: [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files.
2011-01-07 2:35 ` [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files David Daney
@ 2011-01-20 10:01 ` Deng-Cheng Zhu
2011-01-20 17:51 ` David Daney
0 siblings, 1 reply; 14+ messages in thread
From: Deng-Cheng Zhu @ 2011-01-20 10:01 UTC (permalink / raw)
To: David Daney
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo, Dezhong Diao, Gabor Juhos,
Grant Likely
2011/1/7 David Daney <ddaney@caviumnetworks.com>:
> @@ -1034,5 +1560,3 @@ init_hw_perf_events(void)
> return 0;
> }
> arch_initcall(init_hw_perf_events);
> -
> -#endif /* defined(CONFIG_CPU_MIPS32)... */
arch_initcall should be now early_initcall.
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files.
2011-01-20 10:01 ` Deng-Cheng Zhu
@ 2011-01-20 17:51 ` David Daney
0 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-20 17:51 UTC (permalink / raw)
To: Deng-Cheng Zhu
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo, Dezhong Diao, Gabor Juhos,
Grant Likely
On 01/20/2011 02:01 AM, Deng-Cheng Zhu wrote:
> 2011/1/7 David Daney<ddaney@caviumnetworks.com>:
>> @@ -1034,5 +1560,3 @@ init_hw_perf_events(void)
>> return 0;
>> }
>> arch_initcall(init_hw_perf_events);
>> -
>> -#endif /* defined(CONFIG_CPU_MIPS32)... */
>
> arch_initcall should be now early_initcall.
>
Yes, with my current patch set, it is as you indicate.
I will try to send a new patch set later today.
David Daney
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
` (3 preceding siblings ...)
2011-01-07 2:35 ` [PATCH 4/6] MIPS: perf: Reorganize contents of perf support files David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-20 10:06 ` Deng-Cheng Zhu
2011-01-07 2:35 ` [PATCH 6/6] MIPS: perf: Add Octeon support for hardware perf David Daney
2011-01-20 9:59 ` [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters Deng-Cheng Zhu
6 siblings, 1 reply; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel
Cc: David Daney, Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo, Deng-Cheng Zhu
The hard coded constants are moved to struct mips_pmu. All counter
register access move to the read_counter and write_counter function
pointers, which are set to either 32-bit or 64-bit access methods at
initialization time.
Many of the function pointers in struct mips_pmu were not needed as
there was only a single implementation, these were removed.
I couldn't figure out what made struct cpu_hw_events.msbs[] at all
useful, so I removed it too.
Some functions and other declarations were reordered to reduce the
need for forward declarations.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
---
arch/mips/kernel/perf_event_mipsxx.c | 838 ++++++++++++++++------------------
1 files changed, 385 insertions(+), 453 deletions(-)
diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index 498dd85..9bdac86 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -2,6 +2,7 @@
* Linux performance counter support for MIPS.
*
* Copyright (C) 2010 MIPS Technologies, Inc.
+ * Copyright (C) 2011 Cavium Networks, Inc.
* Author: Deng-Cheng Zhu
*
* This code is based on the implementation for ARM, which is in turn
@@ -26,12 +27,6 @@
#include <asm/stacktrace.h>
#include <asm/time.h> /* For perf_irq */
-/* These are for 32bit counters. For 64bit ones, define them accordingly. */
-#define MAX_PERIOD ((1ULL << 32) - 1)
-#define VALID_COUNT 0x7fffffff
-#define TOTAL_BITS 32
-#define HIGHEST_BIT 31
-
#define MIPS_MAX_HWEVENTS 4
struct cpu_hw_events {
@@ -45,15 +40,6 @@ struct cpu_hw_events {
unsigned long used_mask[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
/*
- * The borrowed MSB for the performance counter. A MIPS performance
- * counter uses its bit 31 (for 32bit counters) or bit 63 (for 64bit
- * counters) as a factor of determining whether a counter overflow
- * should be signaled. So here we use a separate MSB for each
- * counter to make things easy.
- */
- unsigned long msbs[BITS_TO_LONGS(MIPS_MAX_HWEVENTS)];
-
- /*
* Software copy of the control register for each performance counter.
* MIPS CPUs vary in performance counters. They use this differently,
* and even may not use it.
@@ -75,6 +61,7 @@ struct mips_perf_event {
unsigned int cntr_mask;
#define CNTR_EVEN 0x55555555
#define CNTR_ODD 0xaaaaaaaa
+ #define CNTR_ALL 0xffffffff
#ifdef CONFIG_MIPS_MT_SMP
enum {
T = 0,
@@ -95,18 +82,13 @@ static DEFINE_MUTEX(raw_event_mutex);
#define C(x) PERF_COUNT_HW_CACHE_##x
struct mips_pmu {
+ u64 max_period;
+ u64 valid_count;
+ u64 overflow;
const char *name;
int irq;
- irqreturn_t (*handle_irq)(int irq, void *dev);
- int (*handle_shared_irq)(void);
- void (*start)(void);
- void (*stop)(void);
- int (*alloc_counter)(struct cpu_hw_events *cpuc,
- struct hw_perf_event *hwc);
u64 (*read_counter)(unsigned int idx);
void (*write_counter)(unsigned int idx, u64 val);
- void (*enable_event)(struct hw_perf_event *evt, int idx);
- void (*disable_event)(int idx);
const struct mips_perf_event *(*map_raw_event)(u64 config);
const struct mips_perf_event (*general_event_map)[PERF_COUNT_HW_MAX];
const struct mips_perf_event (*cache_event_map)
@@ -116,43 +98,300 @@ struct mips_pmu {
unsigned int num_counters;
};
-static const struct mips_pmu *mipspmu;
+static struct mips_pmu mipspmu;
+
+#define M_CONFIG1_PC (1 << 4)
+
+#define M_PERFCTL_EXL (1 << 0)
+#define M_PERFCTL_KERNEL (1 << 1)
+#define M_PERFCTL_SUPERVISOR (1 << 2)
+#define M_PERFCTL_USER (1 << 3)
+#define M_PERFCTL_INTERRUPT_ENABLE (1 << 4)
+#define M_PERFCTL_EVENT(event) (((event) & 0x3ff) << 5)
+#define M_PERFCTL_VPEID(vpe) ((vpe) << 16)
+#define M_PERFCTL_MT_EN(filter) ((filter) << 20)
+#define M_TC_EN_ALL M_PERFCTL_MT_EN(0)
+#define M_TC_EN_VPE M_PERFCTL_MT_EN(1)
+#define M_TC_EN_TC M_PERFCTL_MT_EN(2)
+#define M_PERFCTL_TCID(tcid) ((tcid) << 22)
+#define M_PERFCTL_WIDE (1 << 30)
+#define M_PERFCTL_MORE (1 << 31)
+
+#define M_PERFCTL_COUNT_EVENT_WHENEVER (M_PERFCTL_EXL | \
+ M_PERFCTL_KERNEL | \
+ M_PERFCTL_USER | \
+ M_PERFCTL_SUPERVISOR | \
+ M_PERFCTL_INTERRUPT_ENABLE)
+
+#ifdef CONFIG_MIPS_MT_SMP
+#define M_PERFCTL_CONFIG_MASK 0x3fff801f
+#else
+#define M_PERFCTL_CONFIG_MASK 0x1f
+#endif
+#define M_PERFCTL_EVENT_MASK 0xfe0
+
+
+#ifdef CONFIG_MIPS_MT_SMP
+static int cpu_has_mipsmt_pertccounters;
+
+/*
+ * FIXME: For VSMP, vpe_id() is redefined for Perf-events, because
+ * cpu_data[cpuid].vpe_id reports 0 for _both_ CPUs.
+ */
+#if defined(CONFIG_HW_PERF_EVENTS)
+#define vpe_id() (cpu_has_mipsmt_pertccounters ? \
+ 0 : smp_processor_id())
+#else
+#define vpe_id() (cpu_has_mipsmt_pertccounters ? \
+ 0 : cpu_data[smp_processor_id()].vpe_id)
+#endif
+
+/* Copied from op_model_mipsxx.c */
+static unsigned int vpe_shift(void)
+{
+ if (num_possible_cpus() > 1)
+ return 1;
+
+ return 0;
+}
+
+static unsigned int counters_total_to_per_cpu(unsigned int counters)
+{
+ return counters >> vpe_shift();
+}
+
+static unsigned int counters_per_cpu_to_total(unsigned int counters)
+{
+ return counters << vpe_shift();
+}
+
+#else /* !CONFIG_MIPS_MT_SMP */
+#define vpe_id() 0
+
+#endif /* CONFIG_MIPS_MT_SMP */
+
+static void resume_local_counters(void);
+static void pause_local_counters(void);
+static irqreturn_t mipsxx_pmu_handle_irq(int, void *);
+static int mipsxx_pmu_handle_shared_irq(void);
+
+static unsigned int mipsxx_pmu_swizzle_perf_idx(unsigned int idx)
+{
+ if (vpe_id() == 1)
+ idx = (idx + 2) & 3;
+ return idx;
+}
+
+static u64 mipsxx_pmu_read_counter(unsigned int idx)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ return read_c0_perfcntr0();
+ case 1:
+ return read_c0_perfcntr1();
+ case 2:
+ return read_c0_perfcntr2();
+ case 3:
+ return read_c0_perfcntr3();
+ default:
+ WARN_ONCE(1, "Invalid performance counter number (%d)\n", idx);
+ return 0;
+ }
+}
+
+static u64 mipsxx_pmu_read_counter_64(unsigned int idx)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ return read_c0_perfcntr0_64();
+ case 1:
+ return read_c0_perfcntr1_64();
+ case 2:
+ return read_c0_perfcntr2_64();
+ case 3:
+ return read_c0_perfcntr3_64();
+ default:
+ WARN_ONCE(1, "Invalid performance counter number (%d)\n", idx);
+ return 0;
+ }
+}
+
+static void mipsxx_pmu_write_counter(unsigned int idx, u64 val)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ write_c0_perfcntr0(val);
+ return;
+ case 1:
+ write_c0_perfcntr1(val);
+ return;
+ case 2:
+ write_c0_perfcntr2(val);
+ return;
+ case 3:
+ write_c0_perfcntr3(val);
+ return;
+ }
+}
+
+static void mipsxx_pmu_write_counter_64(unsigned int idx, u64 val)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ write_c0_perfcntr0_64(val);
+ return;
+ case 1:
+ write_c0_perfcntr1_64(val);
+ return;
+ case 2:
+ write_c0_perfcntr2_64(val);
+ return;
+ case 3:
+ write_c0_perfcntr3_64(val);
+ return;
+ }
+}
+
+static unsigned int mipsxx_pmu_read_control(unsigned int idx)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ return read_c0_perfctrl0();
+ case 1:
+ return read_c0_perfctrl1();
+ case 2:
+ return read_c0_perfctrl2();
+ case 3:
+ return read_c0_perfctrl3();
+ default:
+ WARN_ONCE(1, "Invalid performance counter number (%d)\n", idx);
+ return 0;
+ }
+}
+
+static void mipsxx_pmu_write_control(unsigned int idx, unsigned int val)
+{
+ idx = mipsxx_pmu_swizzle_perf_idx(idx);
+
+ switch (idx) {
+ case 0:
+ write_c0_perfctrl0(val);
+ return;
+ case 1:
+ write_c0_perfctrl1(val);
+ return;
+ case 2:
+ write_c0_perfctrl2(val);
+ return;
+ case 3:
+ write_c0_perfctrl3(val);
+ return;
+ }
+}
+
+static int mipsxx_pmu_alloc_counter(struct cpu_hw_events *cpuc,
+ struct hw_perf_event *hwc)
+{
+ int i;
+
+ /*
+ * We only need to care the counter mask. The range has been
+ * checked definitely.
+ */
+ unsigned long cntr_mask = (hwc->event_base >> 8) & 0xffff;
+
+ for (i = mipspmu.num_counters - 1; i >= 0; i--) {
+ /*
+ * Note that some MIPS perf events can be counted by both
+ * even and odd counters, wheresas many other are only by
+ * even _or_ odd counters. This introduces an issue that
+ * when the former kind of event takes the counter the
+ * latter kind of event wants to use, then the "counter
+ * allocation" for the latter event will fail. In fact if
+ * they can be dynamically swapped, they both feel happy.
+ * But here we leave this issue alone for now.
+ */
+ if (test_bit(i, &cntr_mask) &&
+ !test_and_set_bit(i, cpuc->used_mask))
+ return i;
+ }
+
+ return -EAGAIN;
+}
+
+static void mipsxx_pmu_enable_event(struct hw_perf_event *evt, int idx)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ unsigned long flags;
+
+ WARN_ON(idx < 0 || idx >= mipspmu.num_counters);
+
+ local_irq_save(flags);
+ cpuc->saved_ctrl[idx] = M_PERFCTL_EVENT(evt->event_base & 0xff) |
+ (evt->config_base & M_PERFCTL_CONFIG_MASK) |
+ /* Make sure interrupt enabled. */
+ M_PERFCTL_INTERRUPT_ENABLE;
+ /*
+ * We do not actually let the counter run. Leave it until start().
+ */
+ local_irq_restore(flags);
+}
+
+static void mipsxx_pmu_disable_event(int idx)
+{
+ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+ unsigned long flags;
+
+ WARN_ON(idx < 0 || idx >= mipspmu.num_counters);
+
+ local_irq_save(flags);
+ cpuc->saved_ctrl[idx] = mipsxx_pmu_read_control(idx) &
+ ~M_PERFCTL_COUNT_EVENT_WHENEVER;
+ mipsxx_pmu_write_control(idx, cpuc->saved_ctrl[idx]);
+ local_irq_restore(flags);
+}
static int mipspmu_event_set_period(struct perf_event *event,
struct hw_perf_event *hwc,
int idx)
{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- s64 left = local64_read(&hwc->period_left);
- s64 period = hwc->sample_period;
+ u64 left = local64_read(&hwc->period_left);
+ u64 period = hwc->sample_period;
int ret = 0;
- u64 uleft;
unsigned long flags;
- if (unlikely(left <= -period)) {
+ if (unlikely((left + period) & (1ULL << 63))) {
left = period;
local64_set(&hwc->period_left, left);
hwc->last_period = period;
ret = 1;
}
- if (unlikely(left <= 0)) {
+
+ if (unlikely((left + period) <= period)) {
left += period;
local64_set(&hwc->period_left, left);
hwc->last_period = period;
ret = 1;
}
- if (left > (s64)MAX_PERIOD)
- left = MAX_PERIOD;
+ if (left > mipspmu.max_period)
+ left = mipspmu.max_period;
- local64_set(&hwc->prev_count, (u64)-left);
+ local64_set(&hwc->prev_count, mipspmu.overflow - left);
local_irq_save(flags);
- uleft = (u64)(-left) & MAX_PERIOD;
- uleft > VALID_COUNT ?
- set_bit(idx, cpuc->msbs) : clear_bit(idx, cpuc->msbs);
- mipspmu->write_counter(idx, (u64)(-left) & VALID_COUNT);
+ mipspmu.write_counter(idx, mipspmu.overflow - left);
local_irq_restore(flags);
perf_event_update_userpage(event);
@@ -164,30 +403,22 @@ static void mipspmu_event_update(struct perf_event *event,
struct hw_perf_event *hwc,
int idx)
{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
unsigned long flags;
- int shift = 64 - TOTAL_BITS;
- s64 prev_raw_count, new_raw_count;
+ u64 prev_raw_count, new_raw_count;
u64 delta;
again:
prev_raw_count = local64_read(&hwc->prev_count);
local_irq_save(flags);
/* Make the counter value be a "real" one. */
- new_raw_count = mipspmu->read_counter(idx);
- if (new_raw_count & (test_bit(idx, cpuc->msbs) << HIGHEST_BIT)) {
- new_raw_count &= VALID_COUNT;
- clear_bit(idx, cpuc->msbs);
- } else
- new_raw_count |= (test_bit(idx, cpuc->msbs) << HIGHEST_BIT);
+ new_raw_count = mipspmu.read_counter(idx);
local_irq_restore(flags);
if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
new_raw_count) != prev_raw_count)
goto again;
- delta = (new_raw_count << shift) - (prev_raw_count << shift);
- delta >>= shift;
+ delta = new_raw_count - prev_raw_count;
local64_add(delta, &event->count);
local64_sub(delta, &hwc->period_left);
@@ -199,9 +430,6 @@ static void mipspmu_start(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
- if (!mipspmu)
- return;
-
if (flags & PERF_EF_RELOAD)
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
@@ -211,19 +439,16 @@ static void mipspmu_start(struct perf_event *event, int flags)
mipspmu_event_set_period(event, hwc, hwc->idx);
/* Enable the event. */
- mipspmu->enable_event(hwc, hwc->idx);
+ mipsxx_pmu_enable_event(hwc, hwc->idx);
}
static void mipspmu_stop(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
- if (!mipspmu)
- return;
-
if (!(hwc->state & PERF_HES_STOPPED)) {
/* We are working on a local event. */
- mipspmu->disable_event(hwc->idx);
+ mipsxx_pmu_disable_event(hwc->idx);
barrier();
mipspmu_event_update(event, hwc, hwc->idx);
hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
@@ -240,7 +465,7 @@ static int mipspmu_add(struct perf_event *event, int flags)
perf_pmu_disable(event->pmu);
/* To look for a free counter for this event. */
- idx = mipspmu->alloc_counter(cpuc, hwc);
+ idx = mipsxx_pmu_alloc_counter(cpuc, hwc);
if (idx < 0) {
err = idx;
goto out;
@@ -251,7 +476,7 @@ static int mipspmu_add(struct perf_event *event, int flags)
* make sure it is disabled.
*/
event->hw.idx = idx;
- mipspmu->disable_event(idx);
+ mipsxx_pmu_disable_event(idx);
cpuc->events[idx] = event;
hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
@@ -272,7 +497,7 @@ static void mipspmu_del(struct perf_event *event, int flags)
struct hw_perf_event *hwc = &event->hw;
int idx = hwc->idx;
- WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
+ WARN_ON(idx < 0 || idx >= mipspmu.num_counters);
mipspmu_stop(event, PERF_EF_UPDATE);
cpuc->events[idx] = NULL;
@@ -294,14 +519,29 @@ static void mipspmu_read(struct perf_event *event)
static void mipspmu_enable(struct pmu *pmu)
{
- if (mipspmu)
- mipspmu->start();
+#ifdef CONFIG_MIPS_MT_SMP
+ write_unlock(&pmuint_rwlock);
+#endif
+ resume_local_counters();
}
+/*
+ * MIPS performance counters can be per-TC. The control registers can
+ * not be directly accessed accross CPUs. Hence if we want to do global
+ * control, we need cross CPU calls. on_each_cpu() can help us, but we
+ * can not make sure this function is called with interrupts enabled. So
+ * here we pause local counters and then grab a rwlock and leave the
+ * counters on other CPUs alone. If any counter interrupt raises while
+ * we own the write lock, simply pause local counters on that CPU and
+ * spin in the handler. Also we know we won't be switched to another
+ * CPU after pausing local counters and before grabbing the lock.
+ */
static void mipspmu_disable(struct pmu *pmu)
{
- if (mipspmu)
- mipspmu->stop();
+ pause_local_counters();
+#ifdef CONFIG_MIPS_MT_SMP
+ write_lock(&pmuint_rwlock);
+#endif
}
static atomic_t active_events = ATOMIC_INIT(0);
@@ -312,21 +552,21 @@ static int mipspmu_get_irq(void)
{
int err;
- if (mipspmu->irq >= 0) {
+ if (mipspmu.irq >= 0) {
/* Request my own irq handler. */
- err = request_irq(mipspmu->irq, mipspmu->handle_irq,
- IRQF_DISABLED | IRQF_NOBALANCING,
+ err = request_irq(mipspmu.irq, mipsxx_pmu_handle_irq,
+ IRQF_PERCPU | IRQF_NOBALANCING,
"mips_perf_pmu", NULL);
if (err) {
pr_warning("Unable to request IRQ%d for MIPS "
- "performance counters!\n", mipspmu->irq);
+ "performance counters!\n", mipspmu.irq);
}
} else if (cp0_perfcount_irq < 0) {
/*
* We are sharing the irq number with the timer interrupt.
*/
save_perf_irq = perf_irq;
- perf_irq = mipspmu->handle_shared_irq;
+ perf_irq = mipsxx_pmu_handle_shared_irq;
err = 0;
} else {
pr_warning("The platform hasn't properly defined its "
@@ -339,8 +579,8 @@ static int mipspmu_get_irq(void)
static void mipspmu_free_irq(void)
{
- if (mipspmu->irq >= 0)
- free_irq(mipspmu->irq, NULL);
+ if (mipspmu.irq >= 0)
+ free_irq(mipspmu.irq, NULL);
else if (cp0_perfcount_irq < 0)
perf_irq = save_perf_irq;
}
@@ -361,7 +601,7 @@ static void hw_perf_event_destroy(struct perf_event *event)
* disabled.
*/
on_each_cpu(reset_counters,
- (void *)(long)mipspmu->num_counters, 1);
+ (void *)(long)mipspmu.num_counters, 1);
mipspmu_free_irq();
mutex_unlock(&pmu_reserve_mutex);
}
@@ -381,8 +621,8 @@ static int mipspmu_event_init(struct perf_event *event)
return -ENOENT;
}
- if (!mipspmu || event->cpu >= nr_cpumask_bits ||
- (event->cpu >= 0 && !cpu_online(event->cpu)))
+ if (event->cpu >= nr_cpumask_bits ||
+ (event->cpu >= 0 && !cpu_online(event->cpu)))
return -ENODEV;
if (!atomic_inc_not_zero(&active_events)) {
@@ -441,9 +681,9 @@ static const struct mips_perf_event *mipspmu_map_general_event(int idx)
{
const struct mips_perf_event *pev;
- pev = ((*mipspmu->general_event_map)[idx].event_id ==
+ pev = ((*mipspmu.general_event_map)[idx].event_id ==
UNSUPPORTED_PERF_EVENT_ID ? ERR_PTR(-EOPNOTSUPP) :
- &(*mipspmu->general_event_map)[idx]);
+ &(*mipspmu.general_event_map)[idx]);
return pev;
}
@@ -465,7 +705,7 @@ static const struct mips_perf_event *mipspmu_map_cache_event(u64 config)
if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
return ERR_PTR(-EINVAL);
- pev = &((*mipspmu->cache_event_map)
+ pev = &((*mipspmu.cache_event_map)
[cache_type]
[cache_op]
[cache_result]);
@@ -486,7 +726,7 @@ static int validate_event(struct cpu_hw_events *cpuc,
if (event->pmu != &pmu || event->state <= PERF_EVENT_STATE_OFF)
return 1;
- return mipspmu->alloc_counter(cpuc, &fake_hwc) >= 0;
+ return mipsxx_pmu_alloc_counter(cpuc, &fake_hwc) >= 0;
}
static int validate_group(struct perf_event *event)
@@ -524,123 +764,9 @@ static void handle_associated_event(struct cpu_hw_events *cpuc,
return;
if (perf_event_overflow(event, 0, data, regs))
- mipspmu->disable_event(idx);
-}
-
-#define M_CONFIG1_PC (1 << 4)
-
-#define M_PERFCTL_EXL (1UL << 0)
-#define M_PERFCTL_KERNEL (1UL << 1)
-#define M_PERFCTL_SUPERVISOR (1UL << 2)
-#define M_PERFCTL_USER (1UL << 3)
-#define M_PERFCTL_INTERRUPT_ENABLE (1UL << 4)
-#define M_PERFCTL_EVENT(event) (((event) & 0x3ff) << 5)
-#define M_PERFCTL_VPEID(vpe) ((vpe) << 16)
-#define M_PERFCTL_MT_EN(filter) ((filter) << 20)
-#define M_TC_EN_ALL M_PERFCTL_MT_EN(0)
-#define M_TC_EN_VPE M_PERFCTL_MT_EN(1)
-#define M_TC_EN_TC M_PERFCTL_MT_EN(2)
-#define M_PERFCTL_TCID(tcid) ((tcid) << 22)
-#define M_PERFCTL_WIDE (1UL << 30)
-#define M_PERFCTL_MORE (1UL << 31)
-
-#define M_PERFCTL_COUNT_EVENT_WHENEVER (M_PERFCTL_EXL | \
- M_PERFCTL_KERNEL | \
- M_PERFCTL_USER | \
- M_PERFCTL_SUPERVISOR | \
- M_PERFCTL_INTERRUPT_ENABLE)
-
-#ifdef CONFIG_MIPS_MT_SMP
-#define M_PERFCTL_CONFIG_MASK 0x3fff801f
-#else
-#define M_PERFCTL_CONFIG_MASK 0x1f
-#endif
-#define M_PERFCTL_EVENT_MASK 0xfe0
-
-#define M_COUNTER_OVERFLOW (1UL << 31)
-
-#ifdef CONFIG_MIPS_MT_SMP
-static int cpu_has_mipsmt_pertccounters;
-
-/*
- * FIXME: For VSMP, vpe_id() is redefined for Perf-events, because
- * cpu_data[cpuid].vpe_id reports 0 for _both_ CPUs.
- */
-#if defined(CONFIG_HW_PERF_EVENTS)
-#define vpe_id() (cpu_has_mipsmt_pertccounters ? \
- 0 : smp_processor_id())
-#else
-#define vpe_id() (cpu_has_mipsmt_pertccounters ? \
- 0 : cpu_data[smp_processor_id()].vpe_id)
-#endif
-
-/* Copied from op_model_mipsxx.c */
-static unsigned int vpe_shift(void)
-{
- if (num_possible_cpus() > 1)
- return 1;
-
- return 0;
-}
-
-static unsigned int counters_total_to_per_cpu(unsigned int counters)
-{
- return counters >> vpe_shift();
-}
-
-static unsigned int counters_per_cpu_to_total(unsigned int counters)
-{
- return counters << vpe_shift();
+ mipsxx_pmu_disable_event(idx);
}
-#else /* !CONFIG_MIPS_MT_SMP */
-#define vpe_id() 0
-
-#endif /* CONFIG_MIPS_MT_SMP */
-
-#define __define_perf_accessors(r, n, np) \
- \
-static unsigned int r_c0_ ## r ## n(void) \
-{ \
- unsigned int cpu = vpe_id(); \
- \
- switch (cpu) { \
- case 0: \
- return read_c0_ ## r ## n(); \
- case 1: \
- return read_c0_ ## r ## np(); \
- default: \
- BUG(); \
- } \
- return 0; \
-} \
- \
-static void w_c0_ ## r ## n(unsigned int value) \
-{ \
- unsigned int cpu = vpe_id(); \
- \
- switch (cpu) { \
- case 0: \
- write_c0_ ## r ## n(value); \
- return; \
- case 1: \
- write_c0_ ## r ## np(value); \
- return; \
- default: \
- BUG(); \
- } \
- return; \
-} \
-
-__define_perf_accessors(perfcntr, 0, 2)
-__define_perf_accessors(perfcntr, 1, 3)
-__define_perf_accessors(perfcntr, 2, 0)
-__define_perf_accessors(perfcntr, 3, 1)
-
-__define_perf_accessors(perfctrl, 0, 2)
-__define_perf_accessors(perfctrl, 1, 3)
-__define_perf_accessors(perfctrl, 2, 0)
-__define_perf_accessors(perfctrl, 3, 1)
static int __n_counters(void)
{
@@ -682,87 +808,17 @@ static void reset_counters(void *arg)
int counters = (int)(long)arg;
switch (counters) {
case 4:
- w_c0_perfctrl3(0);
- w_c0_perfcntr3(0);
+ mipsxx_pmu_write_control(3, 0);
+ mipspmu.write_counter(3, 0);
case 3:
- w_c0_perfctrl2(0);
- w_c0_perfcntr2(0);
+ mipsxx_pmu_write_control(2, 0);
+ mipspmu.write_counter(2, 0);
case 2:
- w_c0_perfctrl1(0);
- w_c0_perfcntr1(0);
- case 1:
- w_c0_perfctrl0(0);
- w_c0_perfcntr0(0);
- }
-}
-
-static u64 mipsxx_pmu_read_counter(unsigned int idx)
-{
- switch (idx) {
- case 0:
- return r_c0_perfcntr0();
+ mipsxx_pmu_write_control(1, 0);
+ mipspmu.write_counter(1, 0);
case 1:
- return r_c0_perfcntr1();
- case 2:
- return r_c0_perfcntr2();
- case 3:
- return r_c0_perfcntr3();
- default:
- WARN_ONCE(1, "Invalid performance counter number (%d)\n", idx);
- return 0;
- }
-}
-
-static void mipsxx_pmu_write_counter(unsigned int idx, u64 val)
-{
- switch (idx) {
- case 0:
- w_c0_perfcntr0(val);
- return;
- case 1:
- w_c0_perfcntr1(val);
- return;
- case 2:
- w_c0_perfcntr2(val);
- return;
- case 3:
- w_c0_perfcntr3(val);
- return;
- }
-}
-
-static unsigned int mipsxx_pmu_read_control(unsigned int idx)
-{
- switch (idx) {
- case 0:
- return r_c0_perfctrl0();
- case 1:
- return r_c0_perfctrl1();
- case 2:
- return r_c0_perfctrl2();
- case 3:
- return r_c0_perfctrl3();
- default:
- WARN_ONCE(1, "Invalid performance counter number (%d)\n", idx);
- return 0;
- }
-}
-
-static void mipsxx_pmu_write_control(unsigned int idx, unsigned int val)
-{
- switch (idx) {
- case 0:
- w_c0_perfctrl0(val);
- return;
- case 1:
- w_c0_perfctrl1(val);
- return;
- case 2:
- w_c0_perfctrl2(val);
- return;
- case 3:
- w_c0_perfctrl3(val);
- return;
+ mipsxx_pmu_write_control(0, 0);
+ mipspmu.write_counter(0, 0);
}
}
@@ -1047,7 +1103,7 @@ static int __hw_perf_event_init(struct perf_event *event)
} else if (PERF_TYPE_RAW == event->attr.type) {
/* We are working on the global raw event. */
mutex_lock(&raw_event_mutex);
- pev = mipspmu->map_raw_event(event->attr.config);
+ pev = mipspmu.map_raw_event(event->attr.config);
} else {
/* The event type is not (yet) supported. */
return -EOPNOTSUPP;
@@ -1092,7 +1148,7 @@ static int __hw_perf_event_init(struct perf_event *event)
hwc->config = 0;
if (!hwc->sample_period) {
- hwc->sample_period = MAX_PERIOD;
+ hwc->sample_period = mipspmu.max_period;
hwc->last_period = hwc->sample_period;
local64_set(&hwc->period_left, hwc->sample_period);
}
@@ -1105,55 +1161,38 @@ static int __hw_perf_event_init(struct perf_event *event)
}
event->destroy = hw_perf_event_destroy;
-
return err;
}
static void pause_local_counters(void)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- int counters = mipspmu->num_counters;
+ int ctr = mipspmu.num_counters;
unsigned long flags;
local_irq_save(flags);
- switch (counters) {
- case 4:
- cpuc->saved_ctrl[3] = r_c0_perfctrl3();
- w_c0_perfctrl3(cpuc->saved_ctrl[3] &
- ~M_PERFCTL_COUNT_EVENT_WHENEVER);
- case 3:
- cpuc->saved_ctrl[2] = r_c0_perfctrl2();
- w_c0_perfctrl2(cpuc->saved_ctrl[2] &
- ~M_PERFCTL_COUNT_EVENT_WHENEVER);
- case 2:
- cpuc->saved_ctrl[1] = r_c0_perfctrl1();
- w_c0_perfctrl1(cpuc->saved_ctrl[1] &
- ~M_PERFCTL_COUNT_EVENT_WHENEVER);
- case 1:
- cpuc->saved_ctrl[0] = r_c0_perfctrl0();
- w_c0_perfctrl0(cpuc->saved_ctrl[0] &
- ~M_PERFCTL_COUNT_EVENT_WHENEVER);
- }
+ do {
+ ctr--;
+ cpuc->saved_ctrl[ctr] = mipsxx_pmu_read_control(ctr);
+ mipsxx_pmu_write_control(ctr, cpuc->saved_ctrl[ctr] &
+ ~M_PERFCTL_COUNT_EVENT_WHENEVER);
+ } while (ctr > 0);
local_irq_restore(flags);
}
static void resume_local_counters(void)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- int counters = mipspmu->num_counters;
+ int ctr = mipspmu.num_counters;
unsigned long flags;
local_irq_save(flags);
- switch (counters) {
- case 4:
- w_c0_perfctrl3(cpuc->saved_ctrl[3]);
- case 3:
- w_c0_perfctrl2(cpuc->saved_ctrl[2]);
- case 2:
- w_c0_perfctrl1(cpuc->saved_ctrl[1]);
- case 1:
- w_c0_perfctrl0(cpuc->saved_ctrl[0]);
- }
+
+ do {
+ ctr--;
+ mipsxx_pmu_write_control(ctr, cpuc->saved_ctrl[ctr]);
+ } while (ctr > 0);
+
local_irq_restore(flags);
}
@@ -1161,14 +1200,13 @@ static int mipsxx_pmu_handle_shared_irq(void)
{
struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
struct perf_sample_data data;
- unsigned int counters = mipspmu->num_counters;
- unsigned int counter;
+ unsigned int counters = mipspmu.num_counters;
+ u64 counter;
int handled = IRQ_NONE;
struct pt_regs *regs;
if (cpu_has_mips_r2 && !(read_c0_cause() & (1 << 26)))
return handled;
-
/*
* First we pause the local counters, so that when we are locked
* here, the counters are all paused. When it gets locked due to
@@ -1189,13 +1227,9 @@ static int mipsxx_pmu_handle_shared_irq(void)
#define HANDLE_COUNTER(n) \
case n + 1: \
if (test_bit(n, cpuc->used_mask)) { \
- counter = r_c0_perfcntr ## n(); \
- if (counter & M_COUNTER_OVERFLOW) { \
- w_c0_perfcntr ## n(counter & \
- VALID_COUNT); \
- if (test_and_change_bit(n, cpuc->msbs)) \
- handle_associated_event(cpuc, \
- n, &data, regs); \
+ counter = mipspmu.read_counter(n); \
+ if (counter & mipspmu.overflow) { \
+ handle_associated_event(cpuc, n, &data, regs); \
handled = IRQ_HANDLED; \
} \
}
@@ -1225,95 +1259,6 @@ static irqreturn_t mipsxx_pmu_handle_irq(int irq, void *dev)
return mipsxx_pmu_handle_shared_irq();
}
-static void mipsxx_pmu_start(void)
-{
-#ifdef CONFIG_MIPS_MT_SMP
- write_unlock(&pmuint_rwlock);
-#endif
- resume_local_counters();
-}
-
-/*
- * MIPS performance counters can be per-TC. The control registers can
- * not be directly accessed accross CPUs. Hence if we want to do global
- * control, we need cross CPU calls. on_each_cpu() can help us, but we
- * can not make sure this function is called with interrupts enabled. So
- * here we pause local counters and then grab a rwlock and leave the
- * counters on other CPUs alone. If any counter interrupt raises while
- * we own the write lock, simply pause local counters on that CPU and
- * spin in the handler. Also we know we won't be switched to another
- * CPU after pausing local counters and before grabbing the lock.
- */
-static void mipsxx_pmu_stop(void)
-{
- pause_local_counters();
-#ifdef CONFIG_MIPS_MT_SMP
- write_lock(&pmuint_rwlock);
-#endif
-}
-
-static int mipsxx_pmu_alloc_counter(struct cpu_hw_events *cpuc,
- struct hw_perf_event *hwc)
-{
- int i;
-
- /*
- * We only need to care the counter mask. The range has been
- * checked definitely.
- */
- unsigned long cntr_mask = (hwc->event_base >> 8) & 0xffff;
-
- for (i = mipspmu->num_counters - 1; i >= 0; i--) {
- /*
- * Note that some MIPS perf events can be counted by both
- * even and odd counters, wheresas many other are only by
- * even _or_ odd counters. This introduces an issue that
- * when the former kind of event takes the counter the
- * latter kind of event wants to use, then the "counter
- * allocation" for the latter event will fail. In fact if
- * they can be dynamically swapped, they both feel happy.
- * But here we leave this issue alone for now.
- */
- if (test_bit(i, &cntr_mask) &&
- !test_and_set_bit(i, cpuc->used_mask))
- return i;
- }
-
- return -EAGAIN;
-}
-
-static void mipsxx_pmu_enable_event(struct hw_perf_event *evt, int idx)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- unsigned long flags;
-
- WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
-
- local_irq_save(flags);
- cpuc->saved_ctrl[idx] = M_PERFCTL_EVENT(evt->event_base & 0xff) |
- (evt->config_base & M_PERFCTL_CONFIG_MASK) |
- /* Make sure interrupt enabled. */
- M_PERFCTL_INTERRUPT_ENABLE;
- /*
- * We do not actually let the counter run. Leave it until start().
- */
- local_irq_restore(flags);
-}
-
-static void mipsxx_pmu_disable_event(int idx)
-{
- struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
- unsigned long flags;
-
- WARN_ON(idx < 0 || idx >= mipspmu->num_counters);
-
- local_irq_save(flags);
- cpuc->saved_ctrl[idx] = mipsxx_pmu_read_control(idx) &
- ~M_PERFCTL_COUNT_EVENT_WHENEVER;
- mipsxx_pmu_write_control(idx, cpuc->saved_ctrl[idx]);
- local_irq_restore(flags);
-}
-
/* 24K */
#define IS_UNSUPPORTED_24K_EVENT(r, b) \
((b) == 12 || (r) == 151 || (r) == 152 || (b) == 26 || \
@@ -1452,40 +1397,11 @@ static const struct mips_perf_event *mipsxx_pmu_map_raw_event(u64 config)
return &raw_event;
}
-static struct mips_pmu mipsxxcore_pmu = {
- .handle_irq = mipsxx_pmu_handle_irq,
- .handle_shared_irq = mipsxx_pmu_handle_shared_irq,
- .start = mipsxx_pmu_start,
- .stop = mipsxx_pmu_stop,
- .alloc_counter = mipsxx_pmu_alloc_counter,
- .read_counter = mipsxx_pmu_read_counter,
- .write_counter = mipsxx_pmu_write_counter,
- .enable_event = mipsxx_pmu_enable_event,
- .disable_event = mipsxx_pmu_disable_event,
- .map_raw_event = mipsxx_pmu_map_raw_event,
- .general_event_map = &mipsxxcore_event_map,
- .cache_event_map = &mipsxxcore_cache_map,
-};
-
-static struct mips_pmu mipsxx74Kcore_pmu = {
- .handle_irq = mipsxx_pmu_handle_irq,
- .handle_shared_irq = mipsxx_pmu_handle_shared_irq,
- .start = mipsxx_pmu_start,
- .stop = mipsxx_pmu_stop,
- .alloc_counter = mipsxx_pmu_alloc_counter,
- .read_counter = mipsxx_pmu_read_counter,
- .write_counter = mipsxx_pmu_write_counter,
- .enable_event = mipsxx_pmu_enable_event,
- .disable_event = mipsxx_pmu_disable_event,
- .map_raw_event = mipsxx_pmu_map_raw_event,
- .general_event_map = &mipsxx74Kcore_event_map,
- .cache_event_map = &mipsxx74Kcore_cache_map,
-};
-
static int __init
init_hw_perf_events(void)
{
int counters, irq;
+ int counter_bits;
pr_info("Performance counters: ");
@@ -1517,32 +1433,28 @@ init_hw_perf_events(void)
}
#endif
- on_each_cpu(reset_counters, (void *)(long)counters, 1);
+ mipspmu.map_raw_event = mipsxx_pmu_map_raw_event;
switch (current_cpu_type()) {
case CPU_24K:
- mipsxxcore_pmu.name = "mips/24K";
- mipsxxcore_pmu.num_counters = counters;
- mipsxxcore_pmu.irq = irq;
- mipspmu = &mipsxxcore_pmu;
+ mipspmu.name = "mips/24K";
+ mipspmu.general_event_map = &mipsxxcore_event_map;
+ mipspmu.cache_event_map = &mipsxxcore_cache_map;
break;
case CPU_34K:
- mipsxxcore_pmu.name = "mips/34K";
- mipsxxcore_pmu.num_counters = counters;
- mipsxxcore_pmu.irq = irq;
- mipspmu = &mipsxxcore_pmu;
+ mipspmu.name = "mips/34K";
+ mipspmu.general_event_map = &mipsxxcore_event_map;
+ mipspmu.cache_event_map = &mipsxxcore_cache_map;
break;
case CPU_74K:
- mipsxx74Kcore_pmu.name = "mips/74K";
- mipsxx74Kcore_pmu.num_counters = counters;
- mipsxx74Kcore_pmu.irq = irq;
- mipspmu = &mipsxx74Kcore_pmu;
+ mipspmu.name = "mips/74K";
+ mipspmu.general_event_map = &mipsxx74Kcore_event_map;
+ mipspmu.cache_event_map = &mipsxx74Kcore_cache_map;
break;
case CPU_1004K:
- mipsxxcore_pmu.name = "mips/1004K";
- mipsxxcore_pmu.num_counters = counters;
- mipsxxcore_pmu.irq = irq;
- mipspmu = &mipsxxcore_pmu;
+ mipspmu.name = "mips/1004K";
+ mipspmu.general_event_map = &mipsxxcore_event_map;
+ mipspmu.cache_event_map = &mipsxxcore_cache_map;
break;
default:
pr_cont("Either hardware does not support performance "
@@ -1550,10 +1462,30 @@ init_hw_perf_events(void)
return -ENODEV;
}
- if (mipspmu)
- pr_cont("%s PMU enabled, %d counters available to each "
- "CPU, irq %d%s\n", mipspmu->name, counters, irq,
- irq < 0 ? " (share with timer interrupt)" : "");
+ mipspmu.num_counters = counters;
+ mipspmu.irq = irq;
+
+ if (read_c0_perfctrl0() & M_PERFCTL_WIDE) {
+ mipspmu.max_period = (1ULL << 63) - 1;
+ mipspmu.valid_count = (1ULL << 63) - 1;
+ mipspmu.overflow = 1ULL << 63;
+ mipspmu.read_counter = mipsxx_pmu_read_counter_64;
+ mipspmu.write_counter = mipsxx_pmu_write_counter_64;
+ counter_bits = 64;
+ } else {
+ mipspmu.max_period = (1ULL << 32) - 1;
+ mipspmu.valid_count = (1ULL << 31) - 1;
+ mipspmu.overflow = 1ULL << 31;
+ mipspmu.read_counter = mipsxx_pmu_read_counter;
+ mipspmu.write_counter = mipsxx_pmu_write_counter;
+ counter_bits = 32;
+ }
+
+ on_each_cpu(reset_counters, (void *)(long)counters, 1);
+
+ pr_cont("%s PMU enabled, %d %d-bit counters available to each "
+ "CPU, irq %d%s\n", mipspmu.name, counters, counter_bits, irq,
+ irq < 0 ? " (share with timer interrupt)" : "");
perf_pmu_register(&pmu);
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* Re: [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters.
2011-01-07 2:35 ` [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters David Daney
@ 2011-01-20 10:06 ` Deng-Cheng Zhu
2011-01-20 17:48 ` David Daney
0 siblings, 1 reply; 14+ messages in thread
From: Deng-Cheng Zhu @ 2011-01-20 10:06 UTC (permalink / raw)
To: David Daney
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo
2011/1/7 David Daney <ddaney@caviumnetworks.com>:
> @@ -294,14 +519,29 @@ static void mipspmu_read(struct perf_event *event)
>
> static void mipspmu_enable(struct pmu *pmu)
> {
> - if (mipspmu)
> - mipspmu->start();
> +#ifdef CONFIG_MIPS_MT_SMP
> + write_unlock(&pmuint_rwlock);
> +#endif
> + resume_local_counters();
> }
When working with CONFIG_MIPS_MT_SMP, the compiler says 'pmuint_rwlock
undeclared' because of its improper place of definition.
> @@ -1550,10 +1462,30 @@ init_hw_perf_events(void)
> return -ENODEV;
> }
>
> - if (mipspmu)
> - pr_cont("%s PMU enabled, %d counters available to each "
> - "CPU, irq %d%s\n", mipspmu->name, counters, irq,
> - irq < 0 ? " (share with timer interrupt)" : "");
> + mipspmu.num_counters = counters;
> + mipspmu.irq = irq;
> +
> + if (read_c0_perfctrl0() & M_PERFCTL_WIDE) {
> + mipspmu.max_period = (1ULL << 63) - 1;
> + mipspmu.valid_count = (1ULL << 63) - 1;
> + mipspmu.overflow = 1ULL << 63;
> + mipspmu.read_counter = mipsxx_pmu_read_counter_64;
> + mipspmu.write_counter = mipsxx_pmu_write_counter_64;
> + counter_bits = 64;
> + } else {
> + mipspmu.max_period = (1ULL << 32) - 1;
> + mipspmu.valid_count = (1ULL << 31) - 1;
> + mipspmu.overflow = 1ULL << 31;
> + mipspmu.read_counter = mipsxx_pmu_read_counter;
> + mipspmu.write_counter = mipsxx_pmu_write_counter;
> + counter_bits = 32;
> + }
> +
> + on_each_cpu(reset_counters, (void *)(long)counters, 1);
> +
> + pr_cont("%s PMU enabled, %d %d-bit counters available to each "
> + "CPU, irq %d%s\n", mipspmu.name, counters, counter_bits, irq,
> + irq < 0 ? " (share with timer interrupt)" : "");
>
> perf_pmu_register(&pmu);
>
perf_pmu_register(&pmu) should be now changed to perf_pmu_register(&pmu,
"cpu", PERF_TYPE_RAW).
Deng-Cheng
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters.
2011-01-20 10:06 ` Deng-Cheng Zhu
@ 2011-01-20 17:48 ` David Daney
0 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-20 17:48 UTC (permalink / raw)
To: Deng-Cheng Zhu
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo
On 01/20/2011 02:06 AM, Deng-Cheng Zhu wrote:
> 2011/1/7 David Daney<ddaney@caviumnetworks.com>:
>> @@ -294,14 +519,29 @@ static void mipspmu_read(struct perf_event *event)
>>
>> static void mipspmu_enable(struct pmu *pmu)
>> {
>> - if (mipspmu)
>> - mipspmu->start();
>> +#ifdef CONFIG_MIPS_MT_SMP
>> + write_unlock(&pmuint_rwlock);
>> +#endif
>> + resume_local_counters();
>> }
>
> When working with CONFIG_MIPS_MT_SMP, the compiler says 'pmuint_rwlock
> undeclared' because of its improper place of definition.
>
OK, I will try to fix it.
>
>> @@ -1550,10 +1462,30 @@ init_hw_perf_events(void)
>> return -ENODEV;
>> }
>>
>> - if (mipspmu)
>> - pr_cont("%s PMU enabled, %d counters available to each "
>> - "CPU, irq %d%s\n", mipspmu->name, counters, irq,
>> - irq< 0 ? " (share with timer interrupt)" : "");
>> + mipspmu.num_counters = counters;
>> + mipspmu.irq = irq;
>> +
>> + if (read_c0_perfctrl0()& M_PERFCTL_WIDE) {
>> + mipspmu.max_period = (1ULL<< 63) - 1;
>> + mipspmu.valid_count = (1ULL<< 63) - 1;
>> + mipspmu.overflow = 1ULL<< 63;
>> + mipspmu.read_counter = mipsxx_pmu_read_counter_64;
>> + mipspmu.write_counter = mipsxx_pmu_write_counter_64;
>> + counter_bits = 64;
>> + } else {
>> + mipspmu.max_period = (1ULL<< 32) - 1;
>> + mipspmu.valid_count = (1ULL<< 31) - 1;
>> + mipspmu.overflow = 1ULL<< 31;
>> + mipspmu.read_counter = mipsxx_pmu_read_counter;
>> + mipspmu.write_counter = mipsxx_pmu_write_counter;
>> + counter_bits = 32;
>> + }
>> +
>> + on_each_cpu(reset_counters, (void *)(long)counters, 1);
>> +
>> + pr_cont("%s PMU enabled, %d %d-bit counters available to each "
>> + "CPU, irq %d%s\n", mipspmu.name, counters, counter_bits, irq,
>> + irq< 0 ? " (share with timer interrupt)" : "");
>>
>> perf_pmu_register(&pmu);
>>
>
> perf_pmu_register(&pmu) should be now changed to perf_pmu_register(&pmu,
> "cpu", PERF_TYPE_RAW).
Yes, I already have that locally.
David Daney
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 6/6] MIPS: perf: Add Octeon support for hardware perf.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
` (4 preceding siblings ...)
2011-01-07 2:35 ` [PATCH 5/6] MIPS: perf: Add support for 64-bit perf counters David Daney
@ 2011-01-07 2:35 ` David Daney
2011-01-20 9:59 ` [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters Deng-Cheng Zhu
6 siblings, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-07 2:35 UTC (permalink / raw)
To: linux-mips, ralf, linux-kernel
Cc: David Daney, Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo, Deng-Cheng Zhu
Enable hardware counters for Octeon, and add the corresponding event
mappings.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
---
arch/mips/Kconfig | 2 +-
arch/mips/kernel/perf_event_mipsxx.c | 147 ++++++++++++++++++++++++++++++++++
2 files changed, 148 insertions(+), 1 deletions(-)
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index 88aaec2..6694f23 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -2016,7 +2016,7 @@ config NODES_SHIFT
config HW_PERF_EVENTS
bool "Enable hardware performance counter support for perf events"
- depends on PERF_EVENTS && !MIPS_MT_SMTC && OPROFILE=n && (CPU_MIPS32 || CPU_MIPS64 || CPU_R10000 || CPU_SB1)
+ depends on PERF_EVENTS && !MIPS_MT_SMTC && OPROFILE=n && (CPU_MIPS32 || CPU_MIPS64 || CPU_R10000 || CPU_SB1 || CPU_CAVIUM_OCTEON)
default y
help
Enable hardware performance counter support for perf events. If
diff --git a/arch/mips/kernel/perf_event_mipsxx.c b/arch/mips/kernel/perf_event_mipsxx.c
index 9bdac86..527b1b9 100644
--- a/arch/mips/kernel/perf_event_mipsxx.c
+++ b/arch/mips/kernel/perf_event_mipsxx.c
@@ -850,6 +850,16 @@ static const struct mips_perf_event mipsxx74Kcore_event_map
[PERF_COUNT_HW_BUS_CYCLES] = { UNSUPPORTED_PERF_EVENT_ID },
};
+static const struct mips_perf_event octeon_event_map[PERF_COUNT_HW_MAX] = {
+ [PERF_COUNT_HW_CPU_CYCLES] = { 0x01, CNTR_ALL },
+ [PERF_COUNT_HW_INSTRUCTIONS] = { 0x03, CNTR_ALL },
+ [PERF_COUNT_HW_CACHE_REFERENCES] = { 0x2b, CNTR_ALL },
+ [PERF_COUNT_HW_CACHE_MISSES] = { 0x2e, CNTR_ALL },
+ [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = { 0x08, CNTR_ALL },
+ [PERF_COUNT_HW_BRANCH_MISSES] = { 0x09, CNTR_ALL },
+ [PERF_COUNT_HW_BUS_CYCLES] = { 0x25, CNTR_ALL },
+};
+
/* 24K/34K/1004K cores can share the same cache event map. */
static const struct mips_perf_event mipsxxcore_cache_map
[PERF_COUNT_HW_CACHE_MAX]
@@ -1055,6 +1065,102 @@ static const struct mips_perf_event mipsxx74Kcore_cache_map
},
};
+
+static const struct mips_perf_event octeon_cache_map
+ [PERF_COUNT_HW_CACHE_MAX]
+ [PERF_COUNT_HW_CACHE_OP_MAX]
+ [PERF_COUNT_HW_CACHE_RESULT_MAX] = {
+[C(L1D)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x2b, CNTR_ALL },
+ [C(RESULT_MISS)] = { 0x2e, CNTR_ALL },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { 0x30, CNTR_ALL },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+[C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { 0x18, CNTR_ALL },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { 0x19, CNTR_ALL },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+[C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+[C(DTLB)] = {
+ /*
+ * Only general DTLB misses are counted use the same event for
+ * read and write.
+ */
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { 0x35, CNTR_ALL },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { 0x35, CNTR_ALL },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+[C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { 0x37, CNTR_ALL },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+[C(BPU)] = {
+ /* Using the same code for *HW_BRANCH* */
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ [C(RESULT_MISS)] = { UNSUPPORTED_PERF_EVENT_ID },
+ },
+},
+};
+
#ifdef CONFIG_MIPS_MT_SMP
static void check_and_calc_range(struct perf_event *event,
const struct mips_perf_event *pev)
@@ -1397,6 +1503,39 @@ static const struct mips_perf_event *mipsxx_pmu_map_raw_event(u64 config)
return &raw_event;
}
+static const struct mips_perf_event *octeon_pmu_map_raw_event(u64 config)
+{
+ unsigned int raw_id = config & 0xff;
+ unsigned int base_id = raw_id & 0x7f;
+
+
+ raw_event.cntr_mask = CNTR_ALL;
+ raw_event.event_id = base_id;
+
+ if (current_cpu_type() == CPU_CAVIUM_OCTEON2) {
+ if (base_id > 0x42)
+ return ERR_PTR(-EOPNOTSUPP);
+ } else {
+ if (base_id > 0x3a)
+ return ERR_PTR(-EOPNOTSUPP);
+ }
+
+ switch (base_id) {
+ case 0x00:
+ case 0x0f:
+ case 0x1e:
+ case 0x1f:
+ case 0x2f:
+ case 0x34:
+ case 0x3b ... 0x3f:
+ return ERR_PTR(-EOPNOTSUPP);
+ default:
+ break;
+ }
+
+ return &raw_event;
+}
+
static int __init
init_hw_perf_events(void)
{
@@ -1456,6 +1595,14 @@ init_hw_perf_events(void)
mipspmu.general_event_map = &mipsxxcore_event_map;
mipspmu.cache_event_map = &mipsxxcore_cache_map;
break;
+ case CPU_CAVIUM_OCTEON:
+ case CPU_CAVIUM_OCTEON_PLUS:
+ case CPU_CAVIUM_OCTEON2:
+ mipspmu.name = "octeon";
+ mipspmu.general_event_map = &octeon_event_map;
+ mipspmu.cache_event_map = &octeon_cache_map;
+ mipspmu.map_raw_event = octeon_pmu_map_raw_event;
+ break;
default:
pr_cont("Either hardware does not support performance "
"counters, or not yet implemented.\n");
--
1.7.2.3
^ permalink raw reply related [flat|nested] 14+ messages in thread* Re: [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters.
2011-01-07 2:35 [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters David Daney
` (5 preceding siblings ...)
2011-01-07 2:35 ` [PATCH 6/6] MIPS: perf: Add Octeon support for hardware perf David Daney
@ 2011-01-20 9:59 ` Deng-Cheng Zhu
2011-01-20 17:55 ` David Daney
2011-01-22 0:17 ` David Daney
6 siblings, 2 replies; 14+ messages in thread
From: Deng-Cheng Zhu @ 2011-01-20 9:59 UTC (permalink / raw)
To: David Daney
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo
Hi, David
Today I did a quick test against your patch set on my MIPS32 Malta board.
After fixing a small compiling issue (see my comment for patch #5), I
successfully built the kernel based on my previous mainline-sync changes.
And when doing the test, I was using the an previously compiled 'perf'
tool, because the latest perf tool needs arch specific DWARF register
mapping definitions (and currently we have not yet submitted this patch).
And here's the test result:
# When this patch set is built in, the simple 'perf stat' command takes
very long time (182 seconds for the ls command). See following:
-sh-4.0# perf stat -e cycles -e instructions ls /
bin dev home lost+found opt root share tmp usr
boot etc lib mnt proc sbin sys trans var
Performance counter stats for 'ls /':
2825998290 cycles
2148970283 instructions # 0.760 IPC
181.901999444 seconds time elapsed
# When this patch is NOT used, namely, only the mainline-sync changes are
built in, the time looks reasonable:
-sh-4.0# perf stat -e cycles -e instructions ls /
bin dev home lost+found opt root share tmp usr
boot etc lib mnt proc sbin sys trans var
Performance counter stats for 'ls /':
2051461 cycles
1041512 instructions # 0.508 IPC
0.046426513 seconds time elapsed
I noticed that you changed quite a lot of original logics in MIPS
Perf-events, including the deletion of the 'msbs' member in the struct
cpu_hw_events. Honestly speaking, I have not yet taken a careful look into
the patch set to find out how you deal with the MIPS specific 0x80000000
counter overflow (certainly, the value is for MIPS32), instead of
0xffffffff. But maybe this code logic could be related to the test result.
Deng-Cheng
2011/1/7 David Daney <ddaney@caviumnetworks.com>:
> The existing MIPS perf hardware counter support only handles 32-bit
> wide counters. Some CPUs (like Octeon) have the 64-bit wide variety.
> This patch set allows perf to work on Octeon, and I hope not break
> existing systems. I have not tested it on non-Octeon systems, so it
> would be good if someone could test that.
>
> Summary of the patches:
>
> 1) Fix faulty Octeon interrupt controller code.
>
> 2) Add some register definitions.
>
> 3,4) Clean up existing code.
>
> 5) 64-bit perf counter support.
>
> 6) Octeon perf event bindings.
>
> Patch 4/6 is the biggest and has the highest chance of having broken
> something.
>
> This patch set depends on a couple of others that have previously been
> sent to Ralf:
>
> http://patchwork.linux-mips.org/patch/1927/
> http://patchwork.linux-mips.org/patch/1850/
> http://patchwork.linux-mips.org/patch/1851/
> http://patchwork.linux-mips.org/patch/1852/
> http://patchwork.linux-mips.org/patch/1853/
> http://patchwork.linux-mips.org/patch/1854/
>
> David Daney (6):
> MIPS: Octeon: Enable per-CPU IRQs on all CPUs.
> MIPS: Add accessor macros for 64-bit performance counter registers.
> MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c
> MIPS: perf: Reorganize contents of perf support files.
> MIPS: perf: Add support for 64-bit perf counters.
> MIPS: perf: Add Octeon support for hardware perf.
>
> arch/mips/Kconfig | 2 +-
> arch/mips/cavium-octeon/octeon-irq.c | 30 +-
> arch/mips/cavium-octeon/smp.c | 10 +
> arch/mips/include/asm/mipsregs.h | 8 +
> arch/mips/kernel/Makefile | 5 +-
> arch/mips/kernel/perf_event.c | 521 +--------------
> arch/mips/kernel/perf_event_mipsxx.c | 1265 +++++++++++++++++++++++++---------
> 7 files changed, 977 insertions(+), 864 deletions(-)
>
> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
> --
> 1.7.2.3
>
>
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters.
2011-01-20 9:59 ` [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters Deng-Cheng Zhu
@ 2011-01-20 17:55 ` David Daney
2011-01-22 0:17 ` David Daney
1 sibling, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-20 17:55 UTC (permalink / raw)
To: Deng-Cheng Zhu
Cc: linux-mips, ralf, linux-kernel, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Arnaldo Carvalho de Melo
On 01/20/2011 01:59 AM, Deng-Cheng Zhu wrote:
> Hi, David
>
>
> Today I did a quick test against your patch set on my MIPS32 Malta board.
> After fixing a small compiling issue (see my comment for patch #5), I
> successfully built the kernel based on my previous mainline-sync changes.
> And when doing the test, I was using the an previously compiled 'perf'
> tool, because the latest perf tool needs arch specific DWARF register
> mapping definitions (and currently we have not yet submitted this patch).
>
> And here's the test result:
>
> # When this patch set is built in, the simple 'perf stat' command takes
> very long time (182 seconds for the ls command). See following:
>
> -sh-4.0# perf stat -e cycles -e instructions ls /
> bin dev home lost+found opt root share tmp usr
> boot etc lib mnt proc sbin sys trans var
>
> Performance counter stats for 'ls /':
>
> 2825998290 cycles
> 2148970283 instructions # 0.760 IPC
>
> 181.901999444 seconds time elapsed
>
> # When this patch is NOT used, namely, only the mainline-sync changes are
> built in, the time looks reasonable:
>
> -sh-4.0# perf stat -e cycles -e instructions ls /
> bin dev home lost+found opt root share tmp usr
> boot etc lib mnt proc sbin sys trans var
>
> Performance counter stats for 'ls /':
>
> 2051461 cycles
> 1041512 instructions # 0.508 IPC
>
> 0.046426513 seconds time elapsed
>
> I noticed that you changed quite a lot of original logics in MIPS
> Perf-events, including the deletion of the 'msbs' member in the struct
> cpu_hw_events. Honestly speaking, I have not yet taken a careful look into
> the patch set to find out how you deal with the MIPS specific 0x80000000
> counter overflow (certainly, the value is for MIPS32), instead of
> 0xffffffff. But maybe this code logic could be related to the test result.
>
I will try to figure out what is happening. Most of my testing was done
with an explicit '-c' argument to the perf tool.
Thanks for testing it.
If it is OK with you, and once we have tested it, we could use the last
patch set you posted to lmo
(http://patchwork.linux-mips.org/patch/1850/), and add mine on top of it.
David Daney
>
> Deng-Cheng
>
>
> 2011/1/7 David Daney<ddaney@caviumnetworks.com>:
>> The existing MIPS perf hardware counter support only handles 32-bit
>> wide counters. Some CPUs (like Octeon) have the 64-bit wide variety.
>> This patch set allows perf to work on Octeon, and I hope not break
>> existing systems. I have not tested it on non-Octeon systems, so it
>> would be good if someone could test that.
>>
>> Summary of the patches:
>>
>> 1) Fix faulty Octeon interrupt controller code.
>>
>> 2) Add some register definitions.
>>
>> 3,4) Clean up existing code.
>>
>> 5) 64-bit perf counter support.
>>
>> 6) Octeon perf event bindings.
>>
>> Patch 4/6 is the biggest and has the highest chance of having broken
>> something.
>>
>> This patch set depends on a couple of others that have previously been
>> sent to Ralf:
>>
>> http://patchwork.linux-mips.org/patch/1927/
>> http://patchwork.linux-mips.org/patch/1850/
>> http://patchwork.linux-mips.org/patch/1851/
>> http://patchwork.linux-mips.org/patch/1852/
>> http://patchwork.linux-mips.org/patch/1853/
>> http://patchwork.linux-mips.org/patch/1854/
>>
>> David Daney (6):
>> MIPS: Octeon: Enable per-CPU IRQs on all CPUs.
>> MIPS: Add accessor macros for 64-bit performance counter registers.
>> MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c
>> MIPS: perf: Reorganize contents of perf support files.
>> MIPS: perf: Add support for 64-bit perf counters.
>> MIPS: perf: Add Octeon support for hardware perf.
>>
>> arch/mips/Kconfig | 2 +-
>> arch/mips/cavium-octeon/octeon-irq.c | 30 +-
>> arch/mips/cavium-octeon/smp.c | 10 +
>> arch/mips/include/asm/mipsregs.h | 8 +
>> arch/mips/kernel/Makefile | 5 +-
>> arch/mips/kernel/perf_event.c | 521 +--------------
>> arch/mips/kernel/perf_event_mipsxx.c | 1265 +++++++++++++++++++++++++---------
>> 7 files changed, 977 insertions(+), 864 deletions(-)
>>
>> Cc: Peter Zijlstra<a.p.zijlstra@chello.nl>
>> Cc: Paul Mackerras<paulus@samba.org>
>> Cc: Ingo Molnar<mingo@elte.hu>
>> Cc: Arnaldo Carvalho de Melo<acme@redhat.com>
>> Cc: Deng-Cheng Zhu<dengcheng.zhu@gmail.com>
>> --
>> 1.7.2.3
>>
>>
>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters.
2011-01-20 9:59 ` [PATCH 0/6] MIPS: perf: Make perf work for 64-bit/Octeon counters Deng-Cheng Zhu
2011-01-20 17:55 ` David Daney
@ 2011-01-22 0:17 ` David Daney
1 sibling, 0 replies; 14+ messages in thread
From: David Daney @ 2011-01-22 0:17 UTC (permalink / raw)
To: Deng-Cheng Zhu; +Cc: linux-mips, ralf
On 01/20/2011 01:59 AM, Deng-Cheng Zhu wrote:
> Hi, David
>
>
> Today I did a quick test against your patch set on my MIPS32 Malta board.
> After fixing a small compiling issue (see my comment for patch #5), I
> successfully built the kernel based on my previous mainline-sync changes.
> And when doing the test, I was using the an previously compiled 'perf'
> tool, because the latest perf tool needs arch specific DWARF register
> mapping definitions (and currently we have not yet submitted this patch).
>
> And here's the test result:
>
> # When this patch set is built in, the simple 'perf stat' command takes
> very long time (182 seconds for the ls command). See following:
>
> -sh-4.0# perf stat -e cycles -e instructions ls /
> bin dev home lost+found opt root share tmp usr
> boot etc lib mnt proc sbin sys trans var
>
> Performance counter stats for 'ls /':
>
> 2825998290 cycles
> 2148970283 instructions # 0.760 IPC
>
> 181.901999444 seconds time elapsed
>
> # When this patch is NOT used, namely, only the mainline-sync changes are
> built in, the time looks reasonable:
>
> -sh-4.0# perf stat -e cycles -e instructions ls /
> bin dev home lost+found opt root share tmp usr
> boot etc lib mnt proc sbin sys trans var
>
> Performance counter stats for 'ls /':
>
> 2051461 cycles
> 1041512 instructions # 0.508 IPC
>
> 0.046426513 seconds time elapsed
>
> I noticed that you changed quite a lot of original logics in MIPS
> Perf-events, including the deletion of the 'msbs' member in the struct
> cpu_hw_events. Honestly speaking, I have not yet taken a careful look into
> the patch set to find out how you deal with the MIPS specific 0x80000000
> counter overflow (certainly, the value is for MIPS32), instead of
> 0xffffffff. But maybe this code logic could be related to the test result.
>
>
Can you test the new version of my patches I just posted? I think I may
have fixed this issue, but I cannot actually test 32-bit counters.
I think I was initializing the counters to the wrong value in the 32-bit
case. This would have caused an almost unending stream of counter
overflow interrupts, thus slowing things down
David Daney
^ permalink raw reply [flat|nested] 14+ messages in thread