[PATCH] perf/annotate: Use architecture-agnostic register limit

[PATCH] perf/annotate: Use architecture-agnostic register limit

[Suchit Karunakaran]

Remove the arch-specific guard around TYPE_STATE_MAX_REGS and define it
as 32 for all architectures. The architecture that perf is built on may
not match the architecture that produced the perf.data file, so relying
on __powerpc__ or similar is fragile. Using 32 as a fixed upper bound is
safe since it is greater than the previous maximum of 16.
Add a comment to clarify that TYPE_STATE_MAX_REGS is an arch-independent
maximum rather than a build-time choice.

Suggested-by: Ian Rogers <irogers@google.com>
Signed-off-by: Suchit Karunakaran <suchitkarunakaran@gmail.com>
---
 tools/perf/util/annotate-data.h | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/tools/perf/util/annotate-data.h b/tools/perf/util/annotate-data.h
index 541fee1a5f0a..1f76885facb0 100644
--- a/tools/perf/util/annotate-data.h
+++ b/tools/perf/util/annotate-data.h
@@ -189,12 +189,15 @@ struct type_state_stack {
 	u8 kind;
 };
 
-/* FIXME: This should be arch-dependent */
-#ifdef __powerpc__
+/*
+ * Maximum number of registers tracked in type_state.
+ *
+ * This limit must cover all supported architectures, since perf
+ * may analyze perf.data files generated on systems with a different
+ * register set. Use 32 as a safe upper bound instead of relying on
+ * build-arch specific values.
+ */
 #define TYPE_STATE_MAX_REGS  32
-#else
-#define TYPE_STATE_MAX_REGS  16
-#endif
 
 /*
  * State table to maintain type info in each register and stack location.
-- 
2.51.0

Re: [PATCH] perf/annotate: Use architecture-agnostic register limit

[Ian Rogers]

On Tue, Sep 23, 2025 at 10:43 AM Suchit Karunakaran
<suchitkarunakaran@gmail.com> wrote:
>
> Remove the arch-specific guard around TYPE_STATE_MAX_REGS and define it
> as 32 for all architectures. The architecture that perf is built on may
> not match the architecture that produced the perf.data file, so relying
> on __powerpc__ or similar is fragile. Using 32 as a fixed upper bound is
> safe since it is greater than the previous maximum of 16.
> Add a comment to clarify that TYPE_STATE_MAX_REGS is an arch-independent
> maximum rather than a build-time choice.
>
> Suggested-by: Ian Rogers <irogers@google.com>
> Signed-off-by: Suchit Karunakaran <suchitkarunakaran@gmail.com>

Reviewed-by: Ian Rogers <irogers@google.com>

Thanks,
Ian

> ---
>  tools/perf/util/annotate-data.h | 13 ++++++++-----
>  1 file changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/tools/perf/util/annotate-data.h b/tools/perf/util/annotate-data.h
> index 541fee1a5f0a..1f76885facb0 100644
> --- a/tools/perf/util/annotate-data.h
> +++ b/tools/perf/util/annotate-data.h
> @@ -189,12 +189,15 @@ struct type_state_stack {
>         u8 kind;
>  };
>
> -/* FIXME: This should be arch-dependent */
> -#ifdef __powerpc__
> +/*
> + * Maximum number of registers tracked in type_state.
> + *
> + * This limit must cover all supported architectures, since perf
> + * may analyze perf.data files generated on systems with a different
> + * register set. Use 32 as a safe upper bound instead of relying on
> + * build-arch specific values.
> + */
>  #define TYPE_STATE_MAX_REGS  32
> -#else
> -#define TYPE_STATE_MAX_REGS  16
> -#endif
>
>  /*
>   * State table to maintain type info in each register and stack location.
> --
> 2.51.0
>

Re: [PATCH] perf/annotate: Use architecture-agnostic register limit

[Arnaldo Carvalho de Melo]

On Tue, Sep 23, 2025 at 12:11:27PM -0700, Ian Rogers wrote:
> On Tue, Sep 23, 2025 at 10:43 AM Suchit Karunakaran
> <suchitkarunakaran@gmail.com> wrote:
> >
> > Remove the arch-specific guard around TYPE_STATE_MAX_REGS and define it
> > as 32 for all architectures. The architecture that perf is built on may
> > not match the architecture that produced the perf.data file, so relying
> > on __powerpc__ or similar is fragile. Using 32 as a fixed upper bound is
> > safe since it is greater than the previous maximum of 16.
> > Add a comment to clarify that TYPE_STATE_MAX_REGS is an arch-independent
> > maximum rather than a build-time choice.
> >
> > Suggested-by: Ian Rogers <irogers@google.com>
> > Signed-off-by: Suchit Karunakaran <suchitkarunakaran@gmail.com>
> 
> Reviewed-by: Ian Rogers <irogers@google.com>

Thanks, applied to perf-tools-next,

- Arnaldo