* bpftool uses wrong order of tracefs search
@ 2025-09-11 17:27 Alexei Starovoitov
2025-09-11 21:03 ` Quentin Monnet
0 siblings, 1 reply; 5+ messages in thread
From: Alexei Starovoitov @ 2025-09-11 17:27 UTC (permalink / raw)
To: Quentin Monnet, bpf, Daniel Borkmann, Andrii Nakryiko,
Martin KaFai Lau, Steven Rostedt
Hi Quentin,
since last merge window bpftool triggers a warn:
$ bpftool prog tracelog
[ 72.942082] NOTICE: Automounting of tracing to debugfs is
deprecated and will be removed in 2030
I suspect it happens because get_tracefs_pipe()
accesses debug/tracing first which causes automnount
and triggers this warning.
Pls take a look.
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: bpftool uses wrong order of tracefs search
2025-09-11 17:27 bpftool uses wrong order of tracefs search Alexei Starovoitov
@ 2025-09-11 21:03 ` Quentin Monnet
2025-09-11 21:32 ` Alexei Starovoitov
0 siblings, 1 reply; 5+ messages in thread
From: Quentin Monnet @ 2025-09-11 21:03 UTC (permalink / raw)
To: Alexei Starovoitov, bpf, Daniel Borkmann, Andrii Nakryiko,
Martin KaFai Lau, Steven Rostedt
2025-09-11 10:27 UTC-0700 ~ Alexei Starovoitov
<alexei.starovoitov@gmail.com>
> Hi Quentin,
>
> since last merge window bpftool triggers a warn:
> $ bpftool prog tracelog
> [ 72.942082] NOTICE: Automounting of tracing to debugfs is
> deprecated and will be removed in 2030
>
>
> I suspect it happens because get_tracefs_pipe()
> accesses debug/tracing first which causes automnount
> and triggers this warning.
>
> Pls take a look.
Hi Alexei, sure, I'll look into it.
I haven't tested yet, but the check on /sys/kernel/debug/tracing is
likely to be the cause indeed. If that's the case, would that be OK to
move it to the end of the "known_mnts" list in the function? Or do you
prefer to avoid the warning completely, even at the cost of missing the
tracefs on older systems?
Quentin
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: bpftool uses wrong order of tracefs search
2025-09-11 21:03 ` Quentin Monnet
@ 2025-09-11 21:32 ` Alexei Starovoitov
2025-09-11 21:39 ` Steven Rostedt
0 siblings, 1 reply; 5+ messages in thread
From: Alexei Starovoitov @ 2025-09-11 21:32 UTC (permalink / raw)
To: Quentin Monnet
Cc: bpf, Daniel Borkmann, Andrii Nakryiko, Martin KaFai Lau,
Steven Rostedt
On Thu, Sep 11, 2025 at 2:03 PM Quentin Monnet <qmo@kernel.org> wrote:
>
> 2025-09-11 10:27 UTC-0700 ~ Alexei Starovoitov
> <alexei.starovoitov@gmail.com>
> > Hi Quentin,
> >
> > since last merge window bpftool triggers a warn:
> > $ bpftool prog tracelog
> > [ 72.942082] NOTICE: Automounting of tracing to debugfs is
> > deprecated and will be removed in 2030
> >
> >
> > I suspect it happens because get_tracefs_pipe()
> > accesses debug/tracing first which causes automnount
> > and triggers this warning.
> >
> > Pls take a look.
>
>
> Hi Alexei, sure, I'll look into it.
>
> I haven't tested yet, but the check on /sys/kernel/debug/tracing is
> likely to be the cause indeed. If that's the case, would that be OK to
> move it to the end of the "known_mnts" list in the function? Or do you
> prefer to avoid the warning completely, even at the cost of missing the
> tracefs on older systems?
Older kernels won't have this warn.
Also I don't know what these are:
"/tracing",
"/trace",
I think the fix would be to do:
"/sys/kernel/tracing",
"/sys/kernel/debug/tracing",
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: bpftool uses wrong order of tracefs search
2025-09-11 21:32 ` Alexei Starovoitov
@ 2025-09-11 21:39 ` Steven Rostedt
2025-09-11 21:42 ` Alexei Starovoitov
0 siblings, 1 reply; 5+ messages in thread
From: Steven Rostedt @ 2025-09-11 21:39 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Quentin Monnet, bpf, Daniel Borkmann, Andrii Nakryiko,
Martin KaFai Lau
On Thu, 11 Sep 2025 14:32:50 -0700
Alexei Starovoitov <alexei.starovoitov@gmail.com> wrote:
> Older kernels won't have this warn.
> Also I don't know what these are:
> "/tracing",
> "/trace",
It was common for people to hardcode the tracing and debug directories to
short names. I use to do it, and I believe Arnaldo did it as well.
> I think the fix would be to do:
> "/sys/kernel/tracing",
> "/sys/kernel/debug/tracing",
Note, the libtracefs uses /proc/mounts to find the directory. Feel free to
copy it and tweak it for your needs:
https://git.kernel.org/pub/scm/libs/libtrace/libtracefs.git/tree/src/tracefs-utils.c#n89
It has some legacy code to allow finding the debugfs directory first, which
shouldn't be needed anymore.
-- Steve
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: bpftool uses wrong order of tracefs search
2025-09-11 21:39 ` Steven Rostedt
@ 2025-09-11 21:42 ` Alexei Starovoitov
0 siblings, 0 replies; 5+ messages in thread
From: Alexei Starovoitov @ 2025-09-11 21:42 UTC (permalink / raw)
To: Steven Rostedt
Cc: Quentin Monnet, bpf, Daniel Borkmann, Andrii Nakryiko,
Martin KaFai Lau
On Thu, Sep 11, 2025 at 2:38 PM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Thu, 11 Sep 2025 14:32:50 -0700
> Alexei Starovoitov <alexei.starovoitov@gmail.com> wrote:
>
> > Older kernels won't have this warn.
> > Also I don't know what these are:
> > "/tracing",
> > "/trace",
>
> It was common for people to hardcode the tracing and debug directories to
> short names. I use to do it, and I believe Arnaldo did it as well.
>
> > I think the fix would be to do:
> > "/sys/kernel/tracing",
> > "/sys/kernel/debug/tracing",
>
> Note, the libtracefs uses /proc/mounts to find the directory. Feel free to
> copy it and tweak it for your needs:
>
> https://git.kernel.org/pub/scm/libs/libtrace/libtracefs.git/tree/src/tracefs-utils.c#n89
get_tracefs_pipe() searches /proc on 2nd step,
and I would keep it in that order, since /proc parsing isn't fast.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2025-09-11 21:42 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-09-11 17:27 bpftool uses wrong order of tracefs search Alexei Starovoitov
2025-09-11 21:03 ` Quentin Monnet
2025-09-11 21:32 ` Alexei Starovoitov
2025-09-11 21:39 ` Steven Rostedt
2025-09-11 21:42 ` Alexei Starovoitov
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.