[PATCH] perf tools: Remove evsel__handle_error_quirks()

Namhyung Kim posted 1 patch 8 months, 1 week ago
tools/perf/util/evsel.c | 22 ----------------------
1 file changed, 22 deletions(-)
[PATCH] perf tools: Remove evsel__handle_error_quirks()
Posted by Namhyung Kim 8 months, 1 week ago
The evsel__handle_error_quirks() is to fixup invalid event attributes on
some architecture based on the error code.  Currently it's only used for
AMD to disable precise_ip not to use IBS which has more restrictions.

But the commit c33aea446bf555ab changed call evsel__precise_ip_fallback
for any errors so there's no difference with the above function.  To
make matter worse, it caused a problem with branch stack on Zen3.

The IBS doesn't support branch stack so it should use a regular core
PMU event.  The default event is set precise_max and it starts with 3.
And evsel__precise_ip_fallback() tries with it and reduces the level one
by one.  At last it tries with 0 but it also failed on Zen3 since the
branch stack is not supported for the cycles event.

At this point, evsel__precise_ip_fallback() restores the original
precise_ip value (3) in the hope that it can succeed with other modifier
(like exclude_kernel).  Then evsel__handle_error_quirks() see it has
precise_ip != 0 and make it retry with 0.  This created an infinite
loop.

Before:

  $ perf record -b -vv |& grep removing
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  removing precise_ip on AMD
  ...

After:

  $ perf record -b true
  Error:
  Failure to open event 'cycles:P' on PMU 'cpu' which will be removed.
  Invalid event (cycles:P) in per-thread mode, enable system wide with '-a'.
  Error:
  Failure to open any events for recording.

Fixes: c33aea446bf555ab ("perf tools: Fix precise_ip fallback logic")
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/evsel.c | 22 ----------------------
 1 file changed, 22 deletions(-)

diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 1974395492d7da5e..3c030da2e477c707 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -2566,25 +2566,6 @@ static bool evsel__detect_missing_features(struct evsel *evsel, struct perf_cpu
 	return false;
 }
 
-static bool evsel__handle_error_quirks(struct evsel *evsel, int error)
-{
-	/*
-	 * AMD core PMU tries to forward events with precise_ip to IBS PMU
-	 * implicitly.  But IBS PMU has more restrictions so it can fail with
-	 * supported event attributes.  Let's forward it back to the core PMU
-	 * by clearing precise_ip only if it's from precise_max (:P).
-	 */
-	if ((error == -EINVAL || error == -ENOENT) && x86__is_amd_cpu() &&
-	    evsel->core.attr.precise_ip && evsel->precise_max) {
-		evsel->core.attr.precise_ip = 0;
-		pr_debug2_peo("removing precise_ip on AMD\n");
-		display_attr(&evsel->core.attr);
-		return true;
-	}
-
-	return false;
-}
-
 static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
 		struct perf_thread_map *threads,
 		int start_cpu_map_idx, int end_cpu_map_idx)
@@ -2730,9 +2711,6 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
 	if (evsel__precise_ip_fallback(evsel))
 		goto retry_open;
 
-	if (evsel__handle_error_quirks(evsel, err))
-		goto retry_open;
-
 out_close:
 	if (err)
 		threads->err_thread = thread;
-- 
2.49.0.504.g3bcea36a83-goog
Re: [PATCH] perf tools: Remove evsel__handle_error_quirks()
Posted by Chun-Tse Shao 8 months, 1 week ago
On Wed, Apr 9, 2025 at 6:02 PM Namhyung Kim <namhyung@kernel.org> wrote:
>
> The evsel__handle_error_quirks() is to fixup invalid event attributes on
> some architecture based on the error code.  Currently it's only used for
> AMD to disable precise_ip not to use IBS which has more restrictions.
>
> But the commit c33aea446bf555ab changed call evsel__precise_ip_fallback
> for any errors so there's no difference with the above function.  To
> make matter worse, it caused a problem with branch stack on Zen3.
>
> The IBS doesn't support branch stack so it should use a regular core
> PMU event.  The default event is set precise_max and it starts with 3.
> And evsel__precise_ip_fallback() tries with it and reduces the level one
> by one.  At last it tries with 0 but it also failed on Zen3 since the
> branch stack is not supported for the cycles event.
>
> At this point, evsel__precise_ip_fallback() restores the original
> precise_ip value (3) in the hope that it can succeed with other modifier
> (like exclude_kernel).  Then evsel__handle_error_quirks() see it has
> precise_ip != 0 and make it retry with 0.  This created an infinite
> loop.
>
> Before:
>
>   $ perf record -b -vv |& grep removing
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   removing precise_ip on AMD
>   ...
>
> After:
>
>   $ perf record -b true
>   Error:
>   Failure to open event 'cycles:P' on PMU 'cpu' which will be removed.
>   Invalid event (cycles:P) in per-thread mode, enable system wide with '-a'.
>   Error:
>   Failure to open any events for recording.
>
> Fixes: c33aea446bf555ab ("perf tools: Fix precise_ip fallback logic")
> Cc: Ravi Bangoria <ravi.bangoria@amd.com>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Chun-Tse Shao <ctshao@google.com>
> ---
>  tools/perf/util/evsel.c | 22 ----------------------
>  1 file changed, 22 deletions(-)
>
> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> index 1974395492d7da5e..3c030da2e477c707 100644
> --- a/tools/perf/util/evsel.c
> +++ b/tools/perf/util/evsel.c
> @@ -2566,25 +2566,6 @@ static bool evsel__detect_missing_features(struct evsel *evsel, struct perf_cpu
>         return false;
>  }
>
> -static bool evsel__handle_error_quirks(struct evsel *evsel, int error)
> -{
> -       /*
> -        * AMD core PMU tries to forward events with precise_ip to IBS PMU
> -        * implicitly.  But IBS PMU has more restrictions so it can fail with
> -        * supported event attributes.  Let's forward it back to the core PMU
> -        * by clearing precise_ip only if it's from precise_max (:P).
> -        */
> -       if ((error == -EINVAL || error == -ENOENT) && x86__is_amd_cpu() &&
> -           evsel->core.attr.precise_ip && evsel->precise_max) {
> -               evsel->core.attr.precise_ip = 0;
> -               pr_debug2_peo("removing precise_ip on AMD\n");
> -               display_attr(&evsel->core.attr);
> -               return true;
> -       }
> -
> -       return false;
> -}
> -
>  static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
>                 struct perf_thread_map *threads,
>                 int start_cpu_map_idx, int end_cpu_map_idx)
> @@ -2730,9 +2711,6 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
>         if (evsel__precise_ip_fallback(evsel))
>                 goto retry_open;
>
> -       if (evsel__handle_error_quirks(evsel, err))
> -               goto retry_open;
> -
>  out_close:
>         if (err)
>                 threads->err_thread = thread;
> --
> 2.49.0.504.g3bcea36a83-goog
>
Re: [PATCH] perf tools: Remove evsel__handle_error_quirks()
Posted by Namhyung Kim 8 months, 1 week ago
Hello,

On Wed, Apr 09, 2025 at 08:26:07PM -0700, Chun-Tse Shao wrote:
> On Wed, Apr 9, 2025 at 6:02 PM Namhyung Kim <namhyung@kernel.org> wrote:
> >
> > The evsel__handle_error_quirks() is to fixup invalid event attributes on
> > some architecture based on the error code.  Currently it's only used for
> > AMD to disable precise_ip not to use IBS which has more restrictions.
> >
> > But the commit c33aea446bf555ab changed call evsel__precise_ip_fallback
> > for any errors so there's no difference with the above function.  To
> > make matter worse, it caused a problem with branch stack on Zen3.
> >
> > The IBS doesn't support branch stack so it should use a regular core
> > PMU event.  The default event is set precise_max and it starts with 3.
> > And evsel__precise_ip_fallback() tries with it and reduces the level one
> > by one.  At last it tries with 0 but it also failed on Zen3 since the
> > branch stack is not supported for the cycles event.
> >
> > At this point, evsel__precise_ip_fallback() restores the original
> > precise_ip value (3) in the hope that it can succeed with other modifier
> > (like exclude_kernel).  Then evsel__handle_error_quirks() see it has
> > precise_ip != 0 and make it retry with 0.  This created an infinite
> > loop.
> >
> > Before:
> >
> >   $ perf record -b -vv |& grep removing
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   removing precise_ip on AMD
> >   ...
> >
> > After:
> >
> >   $ perf record -b true
> >   Error:
> >   Failure to open event 'cycles:P' on PMU 'cpu' which will be removed.
> >   Invalid event (cycles:P) in per-thread mode, enable system wide with '-a'.
> >   Error:
> >   Failure to open any events for recording.
> >
> > Fixes: c33aea446bf555ab ("perf tools: Fix precise_ip fallback logic")
> > Cc: Ravi Bangoria <ravi.bangoria@amd.com>
> > Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> Tested-by: Chun-Tse Shao <ctshao@google.com>

Thanks for the test.

Arnaldo, I think we can have this in the perf-tools tree.

Thanks,
Namhyung


> > ---
> >  tools/perf/util/evsel.c | 22 ----------------------
> >  1 file changed, 22 deletions(-)
> >
> > diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
> > index 1974395492d7da5e..3c030da2e477c707 100644
> > --- a/tools/perf/util/evsel.c
> > +++ b/tools/perf/util/evsel.c
> > @@ -2566,25 +2566,6 @@ static bool evsel__detect_missing_features(struct evsel *evsel, struct perf_cpu
> >         return false;
> >  }
> >
> > -static bool evsel__handle_error_quirks(struct evsel *evsel, int error)
> > -{
> > -       /*
> > -        * AMD core PMU tries to forward events with precise_ip to IBS PMU
> > -        * implicitly.  But IBS PMU has more restrictions so it can fail with
> > -        * supported event attributes.  Let's forward it back to the core PMU
> > -        * by clearing precise_ip only if it's from precise_max (:P).
> > -        */
> > -       if ((error == -EINVAL || error == -ENOENT) && x86__is_amd_cpu() &&
> > -           evsel->core.attr.precise_ip && evsel->precise_max) {
> > -               evsel->core.attr.precise_ip = 0;
> > -               pr_debug2_peo("removing precise_ip on AMD\n");
> > -               display_attr(&evsel->core.attr);
> > -               return true;
> > -       }
> > -
> > -       return false;
> > -}
> > -
> >  static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
> >                 struct perf_thread_map *threads,
> >                 int start_cpu_map_idx, int end_cpu_map_idx)
> > @@ -2730,9 +2711,6 @@ static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus,
> >         if (evsel__precise_ip_fallback(evsel))
> >                 goto retry_open;
> >
> > -       if (evsel__handle_error_quirks(evsel, err))
> > -               goto retry_open;
> > -
> >  out_close:
> >         if (err)
> >                 threads->err_thread = thread;
> > --
> > 2.49.0.504.g3bcea36a83-goog
> >