Re: [PATCH 1/1] alloc_tag: mark inaccurate allocation counters in /proc/allocinfo output

[Suren Baghdasaryan <surenb@google.com>]

On Thu, Sep 11, 2025 at 2:31 PM Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Thu, 11 Sep 2025 12:00:23 -0400 Usama Arif <usamaarif642@gmail.com> wrote:
>
> > > I think simply adding * to the end of function name or filename is sufficient
> > > as they are already str.
> > >
> >
> > Instead of:
> >
> > 49152*      48* arch/x86/kernel/cpu/mce/core.c:2709 func:mce_device_create
> >
> > Could we do something like:
> >
> > 49152      48 arch/x86/kernel/cpu/mce/core.c:2709 func:mce_device_create(inaccurate)
>
> Can we add another row, saying "the previous row was inaccurate"?  I
> guess that would break parsers also.
>
>
>
> I don't know if this was by design, but the present format does provide
> extensibility.  It is basically
>
>         NNNN NNN name:value name:value
>
> one could arguably append a third name:value and hope that authors of
> existing parsers figured this out.

Actually that sounds like the best idea so far. Currently the format is:

<bytes> <count> <file>:<line> [<module>] func:<function>

We can adopt a rule that after this, the line can contain additional
key:value pairs. In that case for inaccurate lines we can add:

49152      48 arch/x86/kernel/cpu/mce/core.c:2709
func:mce_device_create accurate:no

In the future we can append more key:value pairs if we need them.
Parsers which don't know how to parse a new key can simply ignore
them.

Does that sound good to everyone?

>
>
> Whatev.  I'll drop this version from mm.git.