linux-kernel - Re: [PATCH v2 1/4] perf trace: Allocate syscall stats only if summary is on

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAP-5=fXdF41qVwUnZ1XjPM9PuiDa8owYwK=3eRznPEnMKrVkdw@mail.gmail.com>
Date: Tue, 4 Feb 2025 11:30:39 -0800
From: Ian Rogers <irogers@...gle.com>
To: Namhyung Kim <namhyung@...nel.org>
Cc: Arnaldo Carvalho de Melo <acme@...nel.org>, Kan Liang <kan.liang@...ux.intel.com>, 
	Jiri Olsa <jolsa@...nel.org>, Adrian Hunter <adrian.hunter@...el.com>, 
	Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...nel.org>, 
	LKML <linux-kernel@...r.kernel.org>, linux-perf-users@...r.kernel.org, 
	Howard Chu <howardchu95@...il.com>
Subject: Re: [PATCH v2 1/4] perf trace: Allocate syscall stats only if summary
 is on

On Tue, Feb 4, 2025 at 11:21 AM Namhyung Kim <namhyung@...nel.org> wrote:
>
> On Tue, Feb 04, 2025 at 07:59:01AM -0800, Ian Rogers wrote:
> > On Mon, Feb 3, 2025 at 6:59 PM Namhyung Kim <namhyung@...nel.org> wrote:
> > >
> > > On Sat, Feb 01, 2025 at 10:57:00PM -0800, Ian Rogers wrote:
> > > > On Wed, Jan 29, 2025 at 7:05 PM Namhyung Kim <namhyung@...nel.org> wrote:
> > > > >
> > > > > The syscall stats are used only when summary is requested.  Let's avoid
> > > > > unnecessary operations.  Pass 'trace' pointer to check summary and give
> > > > > output file together.
> > > >
> > > > I don't think this last sentence makes sense.
> > >
> > > Thanks for your review.  I'd say:  Pass 'trace' pointer instead of doing
> > > 'summary' option and 'output' file pointer separately.
> >
> > This still doesn't make sense. There is lazier initialization:
> > ```
> > -               ttrace->syscall_stats = intlist__new(NULL);
> > +               if (trace->summary)
> > +                       ttrace->syscall_stats = intlist__new(NULL);
> > ```
> > and there are functions that take a FILE* but now we're going to use
> > the one in trace instead:
>
> Yep, those FILE* (fp) was from trace->output.
>
>
> > ```
> > @@ -1568,7 +1569,7 @@ static struct thread_trace *thread__trace(struct
> > thread *thread, FILE *fp)
> >
> >         return ttrace;
> >  fail:
> > -       color_fprintf(fp, PERF_COLOR_RED,
> > +       color_fprintf(trace->output, PERF_COLOR_RED,
> >                       "WARNING: not enough memory, dropping samples!\n");
> >         return NULL;
> > ```
> > So why does the one passed to trace still exist? Unfortunately names
> > like "fp" and "output" are not intention revealing.
>
> I think "fp" is a conventional name for file pointers (probably from
> K&R?).
>
> >
> > Anyway, from the commit message and the code I don't understand what
> > this change is trying to do.
>
> I don't know where you didn't get it.  Apparently my English is not good
> enough.  So this commit does two things.
>
> 1. check trace->summary before allocating syscall_stats
> 2. change signature of thread__trace from (thread, fp) to (thread,
>    trace) so that it can use trace->output (fp) and trace->summary.
>
> I thought the change #2 is trivial enough to be in the same commit.  But
> I can split that if you want.

Ah, the diff was showing the function signature still taking fp which
was confusing me. Looking above I see:
```
-static struct thread_trace *thread__trace(struct thread *thread, FILE *fp)
+static struct thread_trace *thread__trace(struct thread *thread,
struct trace *trace)
```
Keeping the context in trace and not splitting the patch both make
sense to me. It'd be nice to capture your more verbose description in
the commit message.

Thanks,
Ian