On Thu, May 28, 2020 at 12:00:10PM -0500, Josh Poimboeuf wrote:
> On Thu, May 28, 2020 at 09:02:47AM -0700, Matt Helsley wrote:
> > On Thu, May 28, 2020 at 09:09:16AM -0500, Josh Poimboeuf wrote:
> > > On Wed, May 27, 2020 at 09:42:32AM -0700, Matt Helsley wrote:
> > > > Currently objtool uses a naming heuristic to find the "base"
> > > > section to apply the relocation(s) to. The standard defines
> > > > the SHF_INFO_LINK flag (SHF => in the section header flags)
> > > > which indicates when the section header's sh_info field can
> > > > be used to find the necessary section.
> > > >
> > > > Warns when the heuristic is used as a fallback and changes
> > > > the name heuristic calculation to handle rela (explicit
> > > > addend) and now rel (implicit addend) relocations.
> > >
> > > Does this fallback case actually happen?
> >
> > Not that I could see. I was thinking about taking it out but
> > I haven't tried this set with clang or other toolchains. So
> > I was wondering if you think holding off before removing it
> > would be wise or if you'd rather just remove it.
>
> I just realized somebody already submitted an almost identical patch:
>
> https://lkml.kernel.org/r/[email protected]
>
> Which I'll be merging soon... so you can just drop this one.
OK, I'll drop this patch.
Cheers,
-Matt
