[PATCH v2] perf report: Skip unsupported new event types

[PATCH v2] perf report: Skip unsupported new event types

[Chun-Tse Shao]

`perf report` currently halts with an error when encountering
unsupported new event types (`event.type >= PERF_RECORD_HEADER_MAX`).
This patch modifies the behavior to skip these samples and continue
processing the remaining events.

Signed-off-by: Chun-Tse Shao <ctshao@google.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Suggested-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/session.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 60fb9997ea0d..b4c7973c34e2 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -1639,8 +1639,17 @@ static s64 perf_session__process_event(struct perf_session *session,
 	if (session->header.needs_swap)
 		event_swap(event, evlist__sample_id_all(evlist));
 
-	if (event->header.type >= PERF_RECORD_HEADER_MAX)
-		return -EINVAL;
+	if (event->header.type >= PERF_RECORD_HEADER_MAX) {
+		if (!PERF_ALIGN(event->header.size, sizeof(u64)))
+			/* perf should not support unaligned event */
+			ui__warning("Skip unaligned event type %u.\n", event->header.type);
+		else
+			/* This perf is outdated and does not support the latest event type. */
+			ui__warning("Unsupported type %u, please considering update perf.\n",
+				    event->header.type);
+		/* Skip unsupported/unaligned event by returning its size. */
+		return event->header.size;
+	}
 
 	events_stats__inc(&evlist->stats, event->header.type);
 
-- 
2.49.0.504.g3bcea36a83-goog

Re: [PATCH v2] perf report: Skip unsupported new event types

[Namhyung Kim]

On Wed, Apr 09, 2025 at 11:39:30AM -0700, Chun-Tse Shao wrote:
> `perf report` currently halts with an error when encountering
> unsupported new event types (`event.type >= PERF_RECORD_HEADER_MAX`).
> This patch modifies the behavior to skip these samples and continue
> processing the remaining events.
> 
> Signed-off-by: Chun-Tse Shao <ctshao@google.com>
> Suggested-by: Arnaldo Carvalho de Melo <acme@kernel.org>
> Suggested-by: Namhyung Kim <namhyung@kernel.org>
> ---
>  tools/perf/util/session.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> index 60fb9997ea0d..b4c7973c34e2 100644
> --- a/tools/perf/util/session.c
> +++ b/tools/perf/util/session.c
> @@ -1639,8 +1639,17 @@ static s64 perf_session__process_event(struct perf_session *session,
>  	if (session->header.needs_swap)
>  		event_swap(event, evlist__sample_id_all(evlist));
>  
> -	if (event->header.type >= PERF_RECORD_HEADER_MAX)
> -		return -EINVAL;
> +	if (event->header.type >= PERF_RECORD_HEADER_MAX) {
> +		if (!PERF_ALIGN(event->header.size, sizeof(u64)))

I think you wanted:

		if (event->header.size % sizeof(u64))


> +			/* perf should not support unaligned event */
> +			ui__warning("Skip unaligned event type %u.\n", event->header.type);

And I think we should stop here.  Going with unaligned data will cause
more troubles I guess.

Thanks,
Namhyung


> +		else
> +			/* This perf is outdated and does not support the latest event type. */
> +			ui__warning("Unsupported type %u, please considering update perf.\n",
> +				    event->header.type);
> +		/* Skip unsupported/unaligned event by returning its size. */
> +		return event->header.size;
> +	}
>  
>  	events_stats__inc(&evlist->stats, event->header.type);
>  
> -- 
> 2.49.0.504.g3bcea36a83-goog
>