Re: [RFC PATCH v7 02/10] trace: use the stackmap from the ftrace stack recording path

From: Li Pengfei

Date: Sat Nov 07 2026 - 00:39:47 EST


From: Pengfei Li <lipengfei28@xxxxxxxxxx>

On Tue, 3 Nov 2026 19:34:21 -0500
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> FYI, the tracing subsystem requires starting subjects with capital letters:
>
> trace: Use the stackmap from the ftrace stack recording path

Will fix for all patches, and use the "tracing:" prefix that the rest of
kernel/trace uses.

> Also, did you use AI for this? Most of the comments are useless, and sound
> like they come from AI.

Yes. I used an AI assistant to help draft a substantial portion of the
code, nearly all of the comments and changelogs, and most of the test
programs. The design decisions are mine, and I checked the test results
and went through every version, but I should have trimmed the comments
myself before sending. Many of them were added in response to automated
review findings rather than to explain the code, and they stayed. I
should also have stated the AI use from the first version.

For v8 I am going through every comment in the series, removing the ones
that only restate the code or the review history, and keeping only what
the code cannot say. v8 carries Assisted-by tags, and the cover letter
says what the assistant was used for.

The rest of your comments on this patch will be in v8: a record_stackmap()
helper with a stub, the depth check before loading the map, a named enum,
the tracefs setup moved into a helper, a stub for
stackmap_mark_init_failed() so the #ifdef goes away, and the comments you
pointed out removed.

Thanks,
Pengfei Li