From: "tip-bot2 for Ravi Bangoria" <tip-bot2@linutronix.de>
To: linux-tip-commits@vger.kernel.org
Cc: Peter Zijlstra <peterz@infradead.org>,
Ravi Bangoria <ravi.bangoria@amd.com>,
x86@kernel.org, linux-kernel@vger.kernel.org
Subject: [tip: perf/core] perf/core: Rework forwarding of {task|cpu}-clock events
Date: Wed, 10 May 2023 13:33:42 -0000 [thread overview]
Message-ID: <168372562225.404.13530707071831243955.tip-bot2@tip-bot2> (raw)
In-Reply-To: <20230504110003.2548-2-ravi.bangoria@amd.com>
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 0d6d062ca27ec7ef547712d34dcfcfb952bcef53
Gitweb: https://git.kernel.org/tip/0d6d062ca27ec7ef547712d34dcfcfb952bcef53
Author: Ravi Bangoria <ravi.bangoria@amd.com>
AuthorDate: Thu, 04 May 2023 16:30:00 +05:30
Committer: Peter Zijlstra <peterz@infradead.org>
CommitterDate: Mon, 08 May 2023 10:58:30 +02:00
perf/core: Rework forwarding of {task|cpu}-clock events
Currently, PERF_TYPE_SOFTWARE is treated specially since task-clock and
cpu-clock events are interfaced through it but internally gets forwarded
to their own pmus.
Rework this by overwriting event->attr.type in perf_swevent_init() which
will cause perf_init_event() to retry with updated type and event will
automatically get forwarded to right pmu. With the change, SW pmu no
longer needs to be treated specially and can be included in 'pmu_idr'
list.
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230504110003.2548-2-ravi.bangoria@amd.com
---
include/linux/perf_event.h | 10 +++++-
kernel/events/core.c | 77 +++++++++++++++++++------------------
2 files changed, 51 insertions(+), 36 deletions(-)
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index d5628a7..bf4f346 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -295,6 +295,8 @@ struct perf_event_pmu_context;
struct perf_output_handle;
+#define PMU_NULL_DEV ((void *)(~0UL))
+
/**
* struct pmu - generic performance monitoring unit
*/
@@ -827,6 +829,14 @@ struct perf_event {
void *security;
#endif
struct list_head sb_list;
+
+ /*
+ * Certain events gets forwarded to another pmu internally by over-
+ * writing kernel copy of event->attr.type without user being aware
+ * of it. event->orig_type contains original 'type' requested by
+ * user.
+ */
+ __u32 orig_type;
#endif /* CONFIG_PERF_EVENTS */
};
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 68baa81..c01bbe9 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -6647,7 +6647,7 @@ static void perf_sigtrap(struct perf_event *event)
return;
send_sig_perf((void __user *)event->pending_addr,
- event->attr.type, event->attr.sig_data);
+ event->orig_type, event->attr.sig_data);
}
/*
@@ -9951,6 +9951,9 @@ static void sw_perf_event_destroy(struct perf_event *event)
swevent_hlist_put();
}
+static struct pmu perf_cpu_clock; /* fwd declaration */
+static struct pmu perf_task_clock;
+
static int perf_swevent_init(struct perf_event *event)
{
u64 event_id = event->attr.config;
@@ -9966,7 +9969,10 @@ static int perf_swevent_init(struct perf_event *event)
switch (event_id) {
case PERF_COUNT_SW_CPU_CLOCK:
+ event->attr.type = perf_cpu_clock.type;
+ return -ENOENT;
case PERF_COUNT_SW_TASK_CLOCK:
+ event->attr.type = perf_task_clock.type;
return -ENOENT;
default:
@@ -11086,7 +11092,7 @@ static void cpu_clock_event_read(struct perf_event *event)
static int cpu_clock_event_init(struct perf_event *event)
{
- if (event->attr.type != PERF_TYPE_SOFTWARE)
+ if (event->attr.type != perf_cpu_clock.type)
return -ENOENT;
if (event->attr.config != PERF_COUNT_SW_CPU_CLOCK)
@@ -11107,6 +11113,7 @@ static struct pmu perf_cpu_clock = {
.task_ctx_nr = perf_sw_context,
.capabilities = PERF_PMU_CAP_NO_NMI,
+ .dev = PMU_NULL_DEV,
.event_init = cpu_clock_event_init,
.add = cpu_clock_event_add,
@@ -11167,7 +11174,7 @@ static void task_clock_event_read(struct perf_event *event)
static int task_clock_event_init(struct perf_event *event)
{
- if (event->attr.type != PERF_TYPE_SOFTWARE)
+ if (event->attr.type != perf_task_clock.type)
return -ENOENT;
if (event->attr.config != PERF_COUNT_SW_TASK_CLOCK)
@@ -11188,6 +11195,7 @@ static struct pmu perf_task_clock = {
.task_ctx_nr = perf_sw_context,
.capabilities = PERF_PMU_CAP_NO_NMI,
+ .dev = PMU_NULL_DEV,
.event_init = task_clock_event_init,
.add = task_clock_event_add,
@@ -11415,31 +11423,31 @@ int perf_pmu_register(struct pmu *pmu, const char *name, int type)
goto unlock;
pmu->type = -1;
- if (!name)
- goto skip_type;
+ if (WARN_ONCE(!name, "Can not register anonymous pmu.\n")) {
+ ret = -EINVAL;
+ goto free_pdc;
+ }
+
pmu->name = name;
- if (type != PERF_TYPE_SOFTWARE) {
- if (type >= 0)
- max = type;
+ if (type >= 0)
+ max = type;
- ret = idr_alloc(&pmu_idr, pmu, max, 0, GFP_KERNEL);
- if (ret < 0)
- goto free_pdc;
+ ret = idr_alloc(&pmu_idr, pmu, max, 0, GFP_KERNEL);
+ if (ret < 0)
+ goto free_pdc;
- WARN_ON(type >= 0 && ret != type);
+ WARN_ON(type >= 0 && ret != type);
- type = ret;
- }
+ type = ret;
pmu->type = type;
- if (pmu_bus_running) {
+ if (pmu_bus_running && !pmu->dev) {
ret = pmu_dev_alloc(pmu);
if (ret)
goto free_idr;
}
-skip_type:
ret = -ENOMEM;
pmu->cpu_pmu_context = alloc_percpu(struct perf_cpu_pmu_context);
if (!pmu->cpu_pmu_context)
@@ -11481,16 +11489,7 @@ skip_type:
if (!pmu->event_idx)
pmu->event_idx = perf_event_idx_default;
- /*
- * Ensure the TYPE_SOFTWARE PMUs are at the head of the list,
- * since these cannot be in the IDR. This way the linear search
- * is fast, provided a valid software event is provided.
- */
- if (type == PERF_TYPE_SOFTWARE || !name)
- list_add_rcu(&pmu->entry, &pmus);
- else
- list_add_tail_rcu(&pmu->entry, &pmus);
-
+ list_add_rcu(&pmu->entry, &pmus);
atomic_set(&pmu->exclusive_cnt, 0);
ret = 0;
unlock:
@@ -11499,12 +11498,13 @@ unlock:
return ret;
free_dev:
- device_del(pmu->dev);
- put_device(pmu->dev);
+ if (pmu->dev && pmu->dev != PMU_NULL_DEV) {
+ device_del(pmu->dev);
+ put_device(pmu->dev);
+ }
free_idr:
- if (pmu->type != PERF_TYPE_SOFTWARE)
- idr_remove(&pmu_idr, pmu->type);
+ idr_remove(&pmu_idr, pmu->type);
free_pdc:
free_percpu(pmu->pmu_disable_count);
@@ -11525,9 +11525,8 @@ void perf_pmu_unregister(struct pmu *pmu)
synchronize_rcu();
free_percpu(pmu->pmu_disable_count);
- if (pmu->type != PERF_TYPE_SOFTWARE)
- idr_remove(&pmu_idr, pmu->type);
- if (pmu_bus_running) {
+ idr_remove(&pmu_idr, pmu->type);
+ if (pmu_bus_running && pmu->dev && pmu->dev != PMU_NULL_DEV) {
if (pmu->nr_addr_filters)
device_remove_file(pmu->dev, &dev_attr_nr_addr_filters);
device_del(pmu->dev);
@@ -11601,6 +11600,12 @@ static struct pmu *perf_init_event(struct perf_event *event)
idx = srcu_read_lock(&pmus_srcu);
+ /*
+ * Save original type before calling pmu->event_init() since certain
+ * pmus overwrites event->attr.type to forward event to another pmu.
+ */
+ event->orig_type = event->attr.type;
+
/* Try parent's PMU first: */
if (event->parent && event->parent->pmu) {
pmu = event->parent->pmu;
@@ -13640,8 +13645,8 @@ void __init perf_event_init(void)
perf_event_init_all_cpus();
init_srcu_struct(&pmus_srcu);
perf_pmu_register(&perf_swevent, "software", PERF_TYPE_SOFTWARE);
- perf_pmu_register(&perf_cpu_clock, NULL, -1);
- perf_pmu_register(&perf_task_clock, NULL, -1);
+ perf_pmu_register(&perf_cpu_clock, "cpu_clock", -1);
+ perf_pmu_register(&perf_task_clock, "task_clock", -1);
perf_tp_register();
perf_event_init_cpu(smp_processor_id());
register_reboot_notifier(&perf_reboot_notifier);
@@ -13684,7 +13689,7 @@ static int __init perf_event_sysfs_init(void)
goto unlock;
list_for_each_entry(pmu, &pmus, entry) {
- if (!pmu->name || pmu->type < 0)
+ if (pmu->dev)
continue;
ret = pmu_dev_alloc(pmu);
next prev parent reply other threads:[~2023-05-10 13:34 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-05-04 10:59 [PATCH v4 0/4] perf: Rework event forwarding logic Ravi Bangoria
2023-05-04 11:00 ` [PATCH v4 1/4] perf/core: Rework forwarding of {task|cpu}-clock events Ravi Bangoria
2023-05-10 13:33 ` tip-bot2 for Ravi Bangoria [this message]
2024-02-20 8:41 ` Pengfei Xu
2024-02-23 5:27 ` Ravi Bangoria
2024-02-28 12:49 ` Ravi Bangoria
2024-02-29 3:41 ` Pengfei Xu
2023-05-04 11:00 ` [PATCH v4 2/4] perf/ibs: Fix interface via core pmu events Ravi Bangoria
2023-05-10 13:33 ` [tip: perf/core] " tip-bot2 for Ravi Bangoria
2023-05-04 11:00 ` [PATCH v4 3/4] perf/core: Remove pmu linear searching code Ravi Bangoria
2023-05-10 13:33 ` [tip: perf/core] " tip-bot2 for Ravi Bangoria
2023-05-24 21:41 ` [PATCH v4 3/4] " Nathan Chancellor
2023-05-25 5:16 ` Ravi Bangoria
2023-05-25 7:11 ` Oliver Upton
2023-05-25 14:20 ` Peter Zijlstra
2023-05-25 15:56 ` Oliver Upton
2023-05-26 23:00 ` Ian Rogers
2023-05-27 13:32 ` Marc Zyngier
2023-05-27 17:00 ` Ian Rogers
2023-05-27 17:05 ` Ian Rogers
2023-05-27 18:38 ` Marc Zyngier
2023-05-27 19:50 ` Ian Rogers
2023-05-30 7:45 ` Thomas Richter
2023-05-30 14:00 ` Ian Rogers
2023-05-31 9:09 ` Thomas Richter
[not found] ` <CAP-5=fVw3vSEu7yGFDnZi8c3VUFJu_XEZtY9ZcTsD7ip2imR6Q@mail.gmail.com>
2023-06-01 11:02 ` Thomas Richter
2023-06-01 11:18 ` Peter Zijlstra
2023-06-01 11:20 ` Peter Zijlstra
2023-05-25 15:55 ` Nathan Chancellor
2023-05-04 11:00 ` [PATCH v4 4/4] perf test: Add selftest to test IBS invocation via core pmu events Ravi Bangoria
2023-05-05 9:16 ` Peter Zijlstra
2023-05-15 21:31 ` Arnaldo Carvalho de Melo
2023-05-15 21:33 ` Arnaldo Carvalho de Melo
2023-05-10 13:33 ` [tip: perf/core] " tip-bot2 for Ravi Bangoria
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=168372562225.404.13530707071831243955.tip-bot2@tip-bot2 \
--to=tip-bot2@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-tip-commits@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=ravi.bangoria@amd.com \
--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®