Re: Fix build ID parsing logic in stable trees

[Jiri Olsa <olsajiri@gmail.com>]

On Tue, Nov 19, 2024 at 12:58:21PM +0100, Greg KH wrote:

SNIP

> > > > >
> > > > > ok, so the fix the issue in 6.11 with upstream backports we'd need both:
> > > > >
> > > > >   1) de3ec364c3c3 lib/buildid: add single folio-based file reader abstraction
> > > > >   2) 60c845b4896b lib/buildid: take into account e_phoff when fetching program headers
> > > > >
> > > > > 2) is needed because 1) seems to omit ehdr->e_phoff addition (patch below)
> > > > > which is added back in 2)
> > > > >
> > > > > IMO 6.11 is close to upstream and by taking above upstream fixes it will be
> > > > > easier to backport other possible fixes in the future, for other trees I'd
> > > > > take the original one line fix I posted
> > > > 
> > > > I still maintain that very minimal is the way to go instead of risking
> > > > bringing new potential regressions by partially backporting folio
> > > > rework patchset.
> > > > 
> > > > Jiri, there is no point in risking this, best to fix this quickly and
> > > > minimally. If we ever need to backport further fixes, *then* we can
> > > > think about folio-based implementation backport.
> > > 
> > > ok, make sense, the original plan works for me as well
> > > 
> > > jirka
> > 
> > Greg, could you please queue up Jiri's one line fixes for 5.15, 6.1,
> > 6.6, and 6.11?
> 
> Ok, will do, but hopefully you all will help out if there's any problems
> with the change going forward...

no worries, will help with that

thanks,
jirka