[PATCH] perf test: Record kcore for the CoreSight callchain test

[PATCH] perf test: Record kcore for the CoreSight callchain test

[Leo Yan]

The CoreSight callchain test traces both userspace and kernel execution
and checks the callchain across a syscall.

Add --kcore to perf record to save the running kernel image alongside
the trace data so perf script can use it to decode the kernel trace.

Add a Fixes tag so the change can be backported to stable kernels and
improve the test reliability.

Fixes: ca0e19074bd6 ("perf test: Add Arm CoreSight callchain test")
Signed-off-by: Leo Yan <leo.yan@arm.com>
---
 tools/perf/tests/shell/coresight/callchain.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/perf/tests/shell/coresight/callchain.sh b/tools/perf/tests/shell/coresight/callchain.sh
index 13cca7dc1118..b69187f1a37f 100755
--- a/tools/perf/tests/shell/coresight/callchain.sh
+++ b/tools/perf/tests/shell/coresight/callchain.sh
@@ -43,7 +43,7 @@ record_trace()
 
 	mkfifo "$cf" "$af"
 
-	perf record -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \
+	perf record --kcore -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \
 		perf test --record-ctl fifo:"$cf","$af" -w callchain >/dev/null 2>&1 &&
 
 	# It is safe to use 'i3i' with a three-instruction interval, since the
-- 
2.34.1

Re: [PATCH] perf test: Record kcore for the CoreSight callchain test

[James Clark]

On 28/09/2026 10:06, Leo Yan wrote:
> The CoreSight callchain test traces both userspace and kernel execution
> and checks the callchain across a syscall.
> 
> Add --kcore to perf record to save the running kernel image alongside
> the trace data so perf script can use it to decode the kernel trace.
> 
> Add a Fixes tag so the change can be backported to stable kernels and
> improve the test reliability.
> 
> Fixes: ca0e19074bd6 ("perf test: Add Arm CoreSight callchain test")
> Signed-off-by: Leo Yan <leo.yan@arm.com>
> ---
>   tools/perf/tests/shell/coresight/callchain.sh | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tools/perf/tests/shell/coresight/callchain.sh b/tools/perf/tests/shell/coresight/callchain.sh
> index 13cca7dc1118..b69187f1a37f 100755
> --- a/tools/perf/tests/shell/coresight/callchain.sh
> +++ b/tools/perf/tests/shell/coresight/callchain.sh
> @@ -43,7 +43,7 @@ record_trace()
>   
>   	mkfifo "$cf" "$af"
>   
> -	perf record -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \
> +	perf record --kcore -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \
>   		perf test --record-ctl fifo:"$cf","$af" -w callchain >/dev/null 2>&1 &&
>   
>   	# It is safe to use 'i3i' with a three-instruction interval, since the

Reviewed-by: James Clark <james.clark@linaro.org>

Re: [PATCH] perf test: Record kcore for the CoreSight callchain test

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] perf/test: fail instead of skip when /proc/kcore is inaccessible
--

commit f13d58799b60cb0c738cb7c9e2cb5ed0a5fa9e32
Author: Leo Yan <leo.yan@arm.com>
Subject: perf test: Record kcore for the CoreSight callchain test

The CoreSight callchain test traces both userspace and kernel execution
and checks the callchain across a syscall.

Add --kcore to perf record to save the running kernel image alongside
the trace data so perf script can use it to decode the kernel trace.

> diff --git a/tools/perf/tests/shell/coresight/callchain.sh b/tools/perf/tests/shell/coresight/callchain.sh
> index 13cca7dc11184..b69187f1a37f5 100755
> --- a/tools/perf/tests/shell/coresight/callchain.sh
> +++ b/tools/perf/tests/shell/coresight/callchain.sh
> @@ -43,7 +43,7 @@ record_trace()
>  
>  	mkfifo "$cf" "$af"
>  
> -	perf record -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \
> +	perf record --kcore -o "$data" -e cs_etm// --per-thread -D -1 --control fifo:"$cf","$af" -- \

[Severity: Medium]
Will this unconditional --kcore flag cause the test to fail instead of 
skipping when /proc/kcore is inaccessible?

If /proc/kcore is missing (e.g., due to CONFIG_PROC_KCORE=n) or restricted
by kernel lockdown, perf record --kcore returns an error. This error causes
record_trace() to fail, which propagates an overall test failure (exit
code 1) instead of gracefully skipping (exit code 2). This may cause false
positive regression reports in kernel CI environments.

Should skip_if_system_is_not_ready() be updated to check [ -r /proc/kcore ]
to ensure the test degrades gracefully?

>  		perf test --record-ctl fifo:"$cf","$af" -w callchain >/dev/null 2>&1 &&
>  
>  	# It is safe to use 'i3i' with a three-instruction interval, since the

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260928090644.24346-1-leo.yan@arm.com?part=1