Re: [PATCH 14/14] scripts/sorttable: ftrace: Do not add weak functions to available_filter_functions
From: Peter Zijlstra
Date: Thu Jan 02 2025 - 15:24:30 EST
On Thu, Jan 02, 2025 at 02:55:01PM -0500, Steven Rostedt wrote:
> On Thu, 2 Jan 2025 20:48:14 +0100
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> > *sigh*.. can we please just either add the 'hole' symbols in symtab, or
> > fix symtab to have entry size?
> >
> > You're just fixing your one problem and leaving everybody else that has
> > extra data inside the dead weak things up a creek :/
> >
> > Eg. if might make sense to also ignore alternative / static_branch /
> > static_call patching for such 'dead' code. Yes, that's not an immediate
> > problem atm, but just fixing __mcount_loc seems very short sighted.
>
> Read my reply to the email that I forgot to add to the cover letter (but
> mention in the last patch). Fixing kallsyms does not remove the place
> holders in the available_filter_functions. This has nothing to do with
> kallsyms. I need to remove the fentry/mcount references in the mcount_loc
> section.
>
> The kallsyms is a completely different issue.
It is not. If kallsyms is fixed, you can use that to tell which
fentry/mcount sites are 'invalid'.
Better yet, other people can then also tell if their things are inside
dead weak code or not.