From: "tip-bot2 for Peter Zijlstra" <tip-bot2@linutronix.de>
To: linux-tip-commits@vger.kernel.org
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>,
Ingo Molnar <mingo@kernel.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
x86 <x86@kernel.org>, LKML <linux-kernel@vger.kernel.org>
Subject: [tip: core/static_call] x86/perf, static_call: Optimize x86_pmu methods
Date: Tue, 01 Sep 2020 11:48:02 -0000 [thread overview]
Message-ID: <159896088203.20229.18201547304769937601.tip-bot2@tip-bot2> (raw)
In-Reply-To: <20200818135805.338001015@infradead.org>
The following commit has been merged into the core/static_call branch of tip:
Commit-ID: 7c9903c9bf716d89b34f96cc2ed64e28dabf570b
Gitweb: https://git.kernel.org/tip/7c9903c9bf716d89b34f96cc2ed64e28dabf570b
Author: Peter Zijlstra <peterz@infradead.org>
AuthorDate: Tue, 18 Aug 2020 15:57:53 +02:00
Committer: Ingo Molnar <mingo@kernel.org>
CommitterDate: Tue, 01 Sep 2020 09:58:06 +02:00
x86/perf, static_call: Optimize x86_pmu methods
Replace many of the indirect calls with static_call().
The average PMI time, as measured by perf_sample_event_took()*:
PRE: 3283.03 [ns]
POST: 3145.12 [ns]
Which is a ~138 [ns] win per PMI, or a ~4.2% decrease.
[*] on an IVB-EP, using: 'perf record -a -e cycles -- make O=defconfig-build/ -j80'
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/r/20200818135805.338001015@infradead.org
---
arch/x86/events/core.c | 134 ++++++++++++++++++++++++++++------------
1 file changed, 94 insertions(+), 40 deletions(-)
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 1cbf57d..360c395 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -28,6 +28,7 @@
#include <linux/bitops.h>
#include <linux/device.h>
#include <linux/nospec.h>
+#include <linux/static_call.h>
#include <asm/apic.h>
#include <asm/stacktrace.h>
@@ -52,6 +53,34 @@ DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
DEFINE_STATIC_KEY_FALSE(rdpmc_never_available_key);
DEFINE_STATIC_KEY_FALSE(rdpmc_always_available_key);
+/*
+ * This here uses DEFINE_STATIC_CALL_NULL() to get a static_call defined
+ * from just a typename, as opposed to an actual function.
+ */
+DEFINE_STATIC_CALL_NULL(x86_pmu_handle_irq, *x86_pmu.handle_irq);
+DEFINE_STATIC_CALL_NULL(x86_pmu_disable_all, *x86_pmu.disable_all);
+DEFINE_STATIC_CALL_NULL(x86_pmu_enable_all, *x86_pmu.enable_all);
+DEFINE_STATIC_CALL_NULL(x86_pmu_enable, *x86_pmu.enable);
+DEFINE_STATIC_CALL_NULL(x86_pmu_disable, *x86_pmu.disable);
+
+DEFINE_STATIC_CALL_NULL(x86_pmu_add, *x86_pmu.add);
+DEFINE_STATIC_CALL_NULL(x86_pmu_del, *x86_pmu.del);
+DEFINE_STATIC_CALL_NULL(x86_pmu_read, *x86_pmu.read);
+
+DEFINE_STATIC_CALL_NULL(x86_pmu_schedule_events, *x86_pmu.schedule_events);
+DEFINE_STATIC_CALL_NULL(x86_pmu_get_event_constraints, *x86_pmu.get_event_constraints);
+DEFINE_STATIC_CALL_NULL(x86_pmu_put_event_constraints, *x86_pmu.put_event_constraints);
+
+DEFINE_STATIC_CALL_NULL(x86_pmu_start_scheduling, *x86_pmu.start_scheduling);
+DEFINE_STATIC_CALL_NULL(x86_pmu_commit_scheduling, *x86_pmu.commit_scheduling);
+DEFINE_STATIC_CALL_NULL(x86_pmu_stop_scheduling, *x86_pmu.stop_scheduling);
+
+DEFINE_STATIC_CALL_NULL(x86_pmu_sched_task, *x86_pmu.sched_task);
+DEFINE_STATIC_CALL_NULL(x86_pmu_swap_task_ctx, *x86_pmu.swap_task_ctx);
+
+DEFINE_STATIC_CALL_NULL(x86_pmu_drain_pebs, *x86_pmu.drain_pebs);
+DEFINE_STATIC_CALL_NULL(x86_pmu_pebs_aliases, *x86_pmu.pebs_aliases);
+
u64 __read_mostly hw_cache_event_ids
[PERF_COUNT_HW_CACHE_MAX]
[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -660,7 +689,7 @@ static void x86_pmu_disable(struct pmu *pmu)
cpuc->enabled = 0;
barrier();
- x86_pmu.disable_all();
+ static_call(x86_pmu_disable_all)();
}
void x86_pmu_enable_all(int added)
@@ -907,8 +936,7 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
if (cpuc->txn_flags & PERF_PMU_TXN_ADD)
n0 -= cpuc->n_txn;
- if (x86_pmu.start_scheduling)
- x86_pmu.start_scheduling(cpuc);
+ static_call_cond(x86_pmu_start_scheduling)(cpuc);
for (i = 0, wmin = X86_PMC_IDX_MAX, wmax = 0; i < n; i++) {
c = cpuc->event_constraint[i];
@@ -925,7 +953,7 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
* change due to external factors (sibling state, allow_tfa).
*/
if (!c || (c->flags & PERF_X86_EVENT_DYNAMIC)) {
- c = x86_pmu.get_event_constraints(cpuc, i, cpuc->event_list[i]);
+ c = static_call(x86_pmu_get_event_constraints)(cpuc, i, cpuc->event_list[i]);
cpuc->event_constraint[i] = c;
}
@@ -1008,8 +1036,7 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
if (!unsched && assign) {
for (i = 0; i < n; i++) {
e = cpuc->event_list[i];
- if (x86_pmu.commit_scheduling)
- x86_pmu.commit_scheduling(cpuc, i, assign[i]);
+ static_call_cond(x86_pmu_commit_scheduling)(cpuc, i, assign[i]);
}
} else {
for (i = n0; i < n; i++) {
@@ -1018,15 +1045,13 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
/*
* release events that failed scheduling
*/
- if (x86_pmu.put_event_constraints)
- x86_pmu.put_event_constraints(cpuc, e);
+ static_call_cond(x86_pmu_put_event_constraints)(cpuc, e);
cpuc->event_constraint[i] = NULL;
}
}
- if (x86_pmu.stop_scheduling)
- x86_pmu.stop_scheduling(cpuc);
+ static_call_cond(x86_pmu_stop_scheduling)(cpuc);
return unsched ? -EINVAL : 0;
}
@@ -1226,7 +1251,7 @@ static void x86_pmu_enable(struct pmu *pmu)
cpuc->enabled = 1;
barrier();
- x86_pmu.enable_all(added);
+ static_call(x86_pmu_enable_all)(added);
}
static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
@@ -1347,7 +1372,7 @@ static int x86_pmu_add(struct perf_event *event, int flags)
if (cpuc->txn_flags & PERF_PMU_TXN_ADD)
goto done_collect;
- ret = x86_pmu.schedule_events(cpuc, n, assign);
+ ret = static_call(x86_pmu_schedule_events)(cpuc, n, assign);
if (ret)
goto out;
/*
@@ -1365,13 +1390,11 @@ done_collect:
cpuc->n_added += n - n0;
cpuc->n_txn += n - n0;
- if (x86_pmu.add) {
- /*
- * This is before x86_pmu_enable() will call x86_pmu_start(),
- * so we enable LBRs before an event needs them etc..
- */
- x86_pmu.add(event);
- }
+ /*
+ * This is before x86_pmu_enable() will call x86_pmu_start(),
+ * so we enable LBRs before an event needs them etc..
+ */
+ static_call_cond(x86_pmu_add)(event);
ret = 0;
out:
@@ -1399,7 +1422,7 @@ static void x86_pmu_start(struct perf_event *event, int flags)
cpuc->events[idx] = event;
__set_bit(idx, cpuc->active_mask);
__set_bit(idx, cpuc->running);
- x86_pmu.enable(event);
+ static_call(x86_pmu_enable)(event);
perf_event_update_userpage(event);
}
@@ -1469,7 +1492,7 @@ void x86_pmu_stop(struct perf_event *event, int flags)
struct hw_perf_event *hwc = &event->hw;
if (test_bit(hwc->idx, cpuc->active_mask)) {
- x86_pmu.disable(event);
+ static_call(x86_pmu_disable)(event);
__clear_bit(hwc->idx, cpuc->active_mask);
cpuc->events[hwc->idx] = NULL;
WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
@@ -1519,8 +1542,7 @@ static void x86_pmu_del(struct perf_event *event, int flags)
if (i >= cpuc->n_events - cpuc->n_added)
--cpuc->n_added;
- if (x86_pmu.put_event_constraints)
- x86_pmu.put_event_constraints(cpuc, event);
+ static_call_cond(x86_pmu_put_event_constraints)(cpuc, event);
/* Delete the array entry. */
while (++i < cpuc->n_events) {
@@ -1533,13 +1555,12 @@ static void x86_pmu_del(struct perf_event *event, int flags)
perf_event_update_userpage(event);
do_del:
- if (x86_pmu.del) {
- /*
- * This is after x86_pmu_stop(); so we disable LBRs after any
- * event can need them etc..
- */
- x86_pmu.del(event);
- }
+
+ /*
+ * This is after x86_pmu_stop(); so we disable LBRs after any
+ * event can need them etc..
+ */
+ static_call_cond(x86_pmu_del)(event);
}
int x86_pmu_handle_irq(struct pt_regs *regs)
@@ -1617,7 +1638,7 @@ perf_event_nmi_handler(unsigned int cmd, struct pt_regs *regs)
return NMI_DONE;
start_clock = sched_clock();
- ret = x86_pmu.handle_irq(regs);
+ ret = static_call(x86_pmu_handle_irq)(regs);
finish_clock = sched_clock();
perf_sample_event_took(finish_clock - start_clock);
@@ -1830,6 +1851,38 @@ ssize_t x86_event_sysfs_show(char *page, u64 config, u64 event)
static struct attribute_group x86_pmu_attr_group;
static struct attribute_group x86_pmu_caps_group;
+static void x86_pmu_static_call_update(void)
+{
+ static_call_update(x86_pmu_handle_irq, x86_pmu.handle_irq);
+ static_call_update(x86_pmu_disable_all, x86_pmu.disable_all);
+ static_call_update(x86_pmu_enable_all, x86_pmu.enable_all);
+ static_call_update(x86_pmu_enable, x86_pmu.enable);
+ static_call_update(x86_pmu_disable, x86_pmu.disable);
+
+ static_call_update(x86_pmu_add, x86_pmu.add);
+ static_call_update(x86_pmu_del, x86_pmu.del);
+ static_call_update(x86_pmu_read, x86_pmu.read);
+
+ static_call_update(x86_pmu_schedule_events, x86_pmu.schedule_events);
+ static_call_update(x86_pmu_get_event_constraints, x86_pmu.get_event_constraints);
+ static_call_update(x86_pmu_put_event_constraints, x86_pmu.put_event_constraints);
+
+ static_call_update(x86_pmu_start_scheduling, x86_pmu.start_scheduling);
+ static_call_update(x86_pmu_commit_scheduling, x86_pmu.commit_scheduling);
+ static_call_update(x86_pmu_stop_scheduling, x86_pmu.stop_scheduling);
+
+ static_call_update(x86_pmu_sched_task, x86_pmu.sched_task);
+ static_call_update(x86_pmu_swap_task_ctx, x86_pmu.swap_task_ctx);
+
+ static_call_update(x86_pmu_drain_pebs, x86_pmu.drain_pebs);
+ static_call_update(x86_pmu_pebs_aliases, x86_pmu.pebs_aliases);
+}
+
+static void _x86_pmu_read(struct perf_event *event)
+{
+ x86_perf_event_update(event);
+}
+
static int __init init_hw_perf_events(void)
{
struct x86_pmu_quirk *quirk;
@@ -1898,6 +1951,11 @@ static int __init init_hw_perf_events(void)
pr_info("... fixed-purpose events: %d\n", x86_pmu.num_counters_fixed);
pr_info("... event mask: %016Lx\n", x86_pmu.intel_ctrl);
+ if (!x86_pmu.read)
+ x86_pmu.read = _x86_pmu_read;
+
+ x86_pmu_static_call_update();
+
/*
* Install callbacks. Core will call them for each online
* cpu.
@@ -1934,11 +1992,9 @@ out:
}
early_initcall(init_hw_perf_events);
-static inline void x86_pmu_read(struct perf_event *event)
+static void x86_pmu_read(struct perf_event *event)
{
- if (x86_pmu.read)
- return x86_pmu.read(event);
- x86_perf_event_update(event);
+ static_call(x86_pmu_read)(event);
}
/*
@@ -2015,7 +2071,7 @@ static int x86_pmu_commit_txn(struct pmu *pmu)
if (!x86_pmu_initialized())
return -EAGAIN;
- ret = x86_pmu.schedule_events(cpuc, n, assign);
+ ret = static_call(x86_pmu_schedule_events)(cpuc, n, assign);
if (ret)
return ret;
@@ -2308,15 +2364,13 @@ static const struct attribute_group *x86_pmu_attr_groups[] = {
static void x86_pmu_sched_task(struct perf_event_context *ctx, bool sched_in)
{
- if (x86_pmu.sched_task)
- x86_pmu.sched_task(ctx, sched_in);
+ static_call_cond(x86_pmu_sched_task)(ctx, sched_in);
}
static void x86_pmu_swap_task_ctx(struct perf_event_context *prev,
struct perf_event_context *next)
{
- if (x86_pmu.swap_task_ctx)
- x86_pmu.swap_task_ctx(prev, next);
+ static_call_cond(x86_pmu_swap_task_ctx)(prev, next);
}
void perf_check_microcode(void)
prev parent reply other threads:[~2020-09-01 11:55 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-08-18 13:57 [PATCH v7 00/18] Add static_call Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 01/18] notifier: Fix broken error handling pattern Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 02/18] module: Fix up module_notifier return values Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 03/18] module: Properly propagate MODULE_STATE_COMING failure Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 04/18] jump_label,module: Fix module lifetime for __jump_label_mod_text_reserved Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] jump_label,module: Fix module lifetime for __jump_label_mod_text_reserved() tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 05/18] compiler.h: Make __ADDRESSABLE() symbol truly unique Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Josh Poimboeuf
2020-08-18 13:57 ` [PATCH v7 06/18] static_call: Add basic static call infrastructure Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Josh Poimboeuf
2020-08-18 13:57 ` [PATCH v7 07/18] static_call: Add inline " Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Josh Poimboeuf
2020-08-18 13:57 ` [PATCH v7 08/18] static_call: Avoid kprobes on inline static_call()s Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-09-02 1:35 ` [PATCH v7 08/18] " Masami Hiramatsu
2020-09-02 9:48 ` peterz
2020-09-02 10:16 ` Masami Hiramatsu
2020-09-02 12:01 ` peterz
2020-08-18 13:57 ` [PATCH v7 09/18] x86/static_call: Add out-of-line static call implementation Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Josh Poimboeuf
2020-08-18 13:57 ` [PATCH v7 10/18] x86/static_call: Add inline static call implementation for x86-64 Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Josh Poimboeuf
2020-08-18 13:57 ` [PATCH v7 11/18] static_call: Simple self-test Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] static_call: Add simple self-test for static calls tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 12/18] x86/alternatives: Teach text_poke_bp() to emulate RET Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 13/18] static_call: Add static_call_cond() Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 14/18] static_call: Handle tail-calls Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 15/18] static_call: Add some validation Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 16/18] static_call: Allow early init Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Peter Zijlstra
2020-08-18 13:57 ` [PATCH v7 17/18] tracepoint: Optimize using static_call() Peter Zijlstra
2020-09-01 11:48 ` [tip: core/static_call] " tip-bot2 for Steven Rostedt (VMware)
2020-08-18 13:57 ` [PATCH v7 18/18] x86/perf, static_call: Optimize x86_pmu methods Peter Zijlstra
2020-09-01 11:48 ` tip-bot2 for Peter Zijlstra [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=159896088203.20229.18201547304769937601.tip-bot2@tip-bot2 \
--to=tip-bot2@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-tip-commits@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=peterz@infradead.org \
--cc=torvalds@linux-foundation.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®