Skip to contentSkip to conversation

perf stat: force error in fallback on :k events

PATCH Applied Ian Rogers 4 participants 4 messages Apr 14, 2020 – Apr 22, 2020 perf events

lore
Share & tools
lore

Ian Rogers

author reviewer
#0 lore raw

PATCH v2 1/1 perf stat: force error in fallback on :k events

From: Stephane Eranian <[email protected]>

When it is not possible for a non-privilege perf command to monitor at the kernel level (:k), the fallback code forces a :u. That works if the event was previously monitoring both levels. But if the event was already constrained to kernel only, then it does not make sense to restrict it to user only.
Given the code works by exclusion, a kernel only event would have:
attr->exclude_user = 1
The fallback code would add:
attr->exclude_kernel = 1;

In the end the end would not monitor in either the user level or kernel level. In other words, it would count nothing.

An event programmed to monitor kernel only cannot be switched to user only without seriously warning the user.

This patch forces an error in this case to make it clear the request cannot really be satisfied.

Behavior with paranoid 1:

$ sudo bash -c "echo 1 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1

Performance counter stats for 'sleep 1':

         1,520,413      cycles:k

       1.002361664 seconds time elapsed

       0.002480000 seconds user
       0.000000000 seconds sys

Old behavior with paranoid 2:

$ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1
 Performance counter stats for 'sleep 1':
                 0      cycles:ku

       1.002358127 seconds time elapsed

       0.002384000 seconds user
       0.000000000 seconds sys

New behavior with paranoid 2:

$ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1
Error:
You may not have permission to collect stats.

Consider tweaking /proc/sys/kernel/perf_event_paranoid, which controls use of the performance events system by unprivileged users (without CAP_PERFMON or CAP_SYS_ADMIN).

The current value is 2:

-1: Allow use of (almost) all events by all users
Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK

Quoted 1 line

= 0: Disallow ftrace function tracepoint by users without CAP_PERFMON or CAP_SYS_ADMIN

Disallow raw tracepoint access by users without CAP_SYS_PERFMON or CAP_SYS_ADMIN

Quoted 2 lines

= 1: Disallow CPU event access by users without CAP_PERFMON or CAP_SYS_ADMIN = 2: Disallow kernel profiling by users without CAP_PERFMON or CAP_SYS_ADMIN

To make this setting permanent, edit /etc/sysctl.conf too, e.g.:

kernel.perf_event_paranoid = -1

v2 of this patch addresses the review feedback from [email protected].

---
 tools/perf/util/evsel.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/perf/util/evsel.c +4 -0
index d23db6755f51..b4e8bcb5ab05 100644
@@ -2446,6 +2446,10 @@ bool perf_evsel__fallback(struct evsel *evsel, int err,
 		char *new_name;
 		const char *sep = ":";
 
+		/* If event has exclude user then don't exclude kernel. */
+		if (evsel->core.attr.exclude_user)
+			return false;
+
 		/* Is there already the separator in the name. */
 		if (strchr(name, '/') ||
 		    strchr(name, ':'))
Signature · 1 line
2.26.0.110.g2183baf09c-goog
Acked-by on v2 1/1
#1 lore raw

On Tue, Apr 14, 2020 at 09:15:50AM -0700, Ian Rogers wrote:

SNIP

Quoted 36 lines

New behavior with paranoid 2:

$ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1
Error:
You may not have permission to collect stats.

Consider tweaking /proc/sys/kernel/perf_event_paranoid, which controls use of the performance events system by unprivileged users (without CAP_PERFMON or CAP_SYS_ADMIN).

The current value is 2:

-1: Allow use of (almost) all events by all users
Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK

Quoted 1 line

= 0: Disallow ftrace function tracepoint by users without CAP_PERFMON or CAP_SYS_ADMIN

Disallow raw tracepoint access by users without CAP_SYS_PERFMON or CAP_SYS_ADMIN

Quoted 2 lines

= 1: Disallow CPU event access by users without CAP_PERFMON or CAP_SYS_ADMIN = 2: Disallow kernel profiling by users without CAP_PERFMON or CAP_SYS_ADMIN

To make this setting permanent, edit /etc/sysctl.conf too, e.g.:

kernel.perf_event_paranoid = -1

v2 of this patch addresses the review feedback from [email protected].

---
 tools/perf/util/evsel.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/perf/util/evsel.c +0 -0
index d23db6755f51..b4e8bcb5ab05 100644

tools/perf/util/evsel.c bool perf_evsel__fallback(struct evsel *evsel, int err, · line 2452

Quoted 7 lines
tools/perf/util/evsel.c +4 -0
@@ -2446,6 +2446,10 @@ bool perf_evsel__fallback(struct evsel *evsel, int err,
 		char *new_name;
 		const char *sep = ":";
 
+		/* If event has exclude user then don't exclude kernel. */
+		if (evsel->core.attr.exclude_user)
+			return false;
+

nice, it's much simpler

thanks,
jirka

Quoted 6 lines
tools/perf/util/evsel.c +0 -0
 		/* Is there already the separator in the name. */
 		if (strchr(name, '/') ||
 		    strchr(name, ':'))
Signature · 1 line
2.26.0.110.g2183baf09c-goog
#2 lore raw
In reply to Ian Rogers

Em Tue, Apr 14, 2020 at 09:15:50AM -0700, Ian Rogers escreveu:

Quoted 62 lines

From: Stephane Eranian <[email protected]>

When it is not possible for a non-privilege perf command to monitor at the kernel level (:k), the fallback code forces a :u. That works if the event was previously monitoring both levels. But if the event was already constrained to kernel only, then it does not make sense to restrict it to user only.
Given the code works by exclusion, a kernel only event would have:
attr->exclude_user = 1
The fallback code would add:
attr->exclude_kernel = 1;

In the end the end would not monitor in either the user level or kernel level. In other words, it would count nothing.

An event programmed to monitor kernel only cannot be switched to user only without seriously warning the user.

This patch forces an error in this case to make it clear the request cannot really be satisfied.

Behavior with paranoid 1:

$ sudo bash -c "echo 1 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1

Performance counter stats for 'sleep 1':

         1,520,413      cycles:k

       1.002361664 seconds time elapsed

       0.002480000 seconds user
       0.000000000 seconds sys

Old behavior with paranoid 2:

$ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1
 Performance counter stats for 'sleep 1':
                 0      cycles:ku

       1.002358127 seconds time elapsed

       0.002384000 seconds user
       0.000000000 seconds sys

New behavior with paranoid 2:

$ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
$ perf stat -e cycles:k sleep 1
Error:
You may not have permission to collect stats.

Consider tweaking /proc/sys/kernel/perf_event_paranoid, which controls use of the performance events system by unprivileged users (without CAP_PERFMON or CAP_SYS_ADMIN).

The current value is 2:

-1: Allow use of (almost) all events by all users
Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK

Quoted 1 line

= 0: Disallow ftrace function tracepoint by users without CAP_PERFMON or CAP_SYS_ADMIN

Disallow raw tracepoint access by users without CAP_SYS_PERFMON or CAP_SYS_ADMIN

Quoted 8 lines
Quoted 2 lines

= 1: Disallow CPU event access by users without CAP_PERFMON or CAP_SYS_ADMIN = 2: Disallow kernel profiling by users without CAP_PERFMON or CAP_SYS_ADMIN

To make this setting permanent, edit /etc/sysctl.conf too, e.g.:

kernel.perf_event_paranoid = -1

v2 of this patch addresses the review feedback from [email protected].

Hey, thanks _a lot_ for providing the before/after output, this allows me to review the code and then just go mechanically follow your steps to reproduce the results, awesome, keep doing it that way.

And Jiri is right, one-liners are fantastic, this one came close enough :-)

I did some reflowing of the commit text, check if you're curious when this goes upstream, no original content changed, just added some spaces, separator blank lines and moved the examples two spaces to the right.

Thanks,

- Arnaldo

Quoted 24 lines
---
 tools/perf/util/evsel.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/perf/util/evsel.c +4 -0
index d23db6755f51..b4e8bcb5ab05 100644
@@ -2446,6 +2446,10 @@ bool perf_evsel__fallback(struct evsel *evsel, int err,
 		char *new_name;
 		const char *sep = ":";
 
+		/* If event has exclude user then don't exclude kernel. */
+		if (evsel->core.attr.exclude_user)
+			return false;
+
 		/* Is there already the separator in the name. */
 		if (strchr(name, '/') ||
 		    strchr(name, ':'))
Signature · 1 line
2.26.0.110.g2183baf09c-goog
Signature · 1 line
- Arnaldo
#3 lore raw

[tip: perf/core] perf stat: Force error in fallback on :k events

In reply to Ian Rogers
The following commit has been merged into the perf/core branch of tip:

The following commit has been merged into the perf/core branch of tip:

Commit-ID:     bec49a9e05db3dbdca696fa07c62c52638fb6371
Gitweb:        https://git.kernel.org/tip/bec49a9e05db3dbdca696fa07c62c52638fb6371
Author:        Stephane Eranian <[email protected]>
AuthorDate:    Tue, 14 Apr 2020 09:15:50 -07:00
Committer:     Arnaldo Carvalho de Melo <[email protected]>
CommitterDate: Sat, 18 Apr 2020 09:05:00 -03:00

perf stat: Force error in fallback on :k events

When it is not possible for a non-privilege perf command to monitor at the kernel level (:k), the fallback code forces a :u. That works if the event was previously monitoring both levels. But if the event was already constrained to kernel only, then it does not make sense to restrict it to user only.

Given the code works by exclusion, a kernel only event would have:

attr->exclude_user = 1

The fallback code would add:

attr->exclude_kernel = 1

In the end the end would not monitor in either the user level or kernel level. In other words, it would count nothing.

An event programmed to monitor kernel only cannot be switched to user only without seriously warning the user.

This patch forces an error in this case to make it clear the request cannot really be satisfied.

Behavior with paranoid 1:

  $ sudo bash -c "echo 1 > /proc/sys/kernel/perf_event_paranoid"
  $ perf stat -e cycles:k sleep 1

Performance counter stats for 'sleep 1':

           1,520,413      cycles:k

         1.002361664 seconds time elapsed

         0.002480000 seconds user
         0.000000000 seconds sys

Old behavior with paranoid 2:

  $ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
  $ perf stat -e cycles:k sleep 1
   Performance counter stats for 'sleep 1':
                   0      cycles:ku

         1.002358127 seconds time elapsed

         0.002384000 seconds user
         0.000000000 seconds sys

New behavior with paranoid 2:

  $ sudo bash -c "echo 2 > /proc/sys/kernel/perf_event_paranoid"
  $ perf stat -e cycles:k sleep 1
  Error:
  You may not have permission to collect stats.

Consider tweaking /proc/sys/kernel/perf_event_paranoid, which controls use of the performance events system by unprivileged users (without CAP_PERFMON or CAP_SYS_ADMIN).

The current value is 2:

-1: Allow use of (almost) all events by all users
Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK

Quoted 1 line

= 0: Disallow ftrace function tracepoint by users without CAP_PERFMON or CAP_SYS_ADMIN

Disallow raw tracepoint access by users without CAP_SYS_PERFMON or CAP_SYS_ADMIN

Quoted 2 lines

= 1: Disallow CPU event access by users without CAP_PERFMON or CAP_SYS_ADMIN = 2: Disallow kernel profiling by users without CAP_PERFMON or CAP_SYS_ADMIN

To make this setting permanent, edit /etc/sysctl.conf too, e.g.:

kernel.perf_event_paranoid = -1

v2 of this patch addresses the review feedback from [email protected].

---
 tools/perf/util/evsel.c | 4 ++++
 1 file changed, 4 insertions(+)
tools/perf/util/evsel.c +4 -0
index 8300e8c..6a571d3 100644
@@ -2427,6 +2427,10 @@ bool perf_evsel__fallback(struct evsel *evsel, int err,
 		char *new_name;
 		const char *sep = ":";
 
+		/* If event has exclude user then don't exclude kernel. */
+		if (evsel->core.attr.exclude_user)
+			return false;
+
 		/* Is there already the separator in the name. */
 		if (strchr(name, '/') ||
 		    strchr(name, ':'))