On Thu, 31 Oct 2024 17:13:58 -0700 Suren Baghdasaryan <surenb@xxxxxxxxxx> wrote:
> On Thu, Oct 31, 2024 at 5:00 PM Suren Baghdasaryan <surenb@xxxxxxxxxx> wrote:
> >
> > When a module does not have any allocations, it's allocation tag section
> > is empty and codetag_alloc_module_section() returns NULL. However this
> > condition should never happen because codetag_needs_module_section() will
> > detect an empty section and avoid calling codetag_alloc_module_section().
> > Change codetag_alloc_module_section() to never return NULL, which should
> > prevent static checker warnings. Add a WARN_ON() and a proper error
> > reporting in case codetag_alloc_module_section() returns NULL, to prevent
> > future codetag type implementations from returning NULL from their
> > cttype->desc.alloc_section_mem() operation.
> >
> > Fixes: 61c9e58f3a10 ("alloc_tag: load module tags into separate contiguous memory")
> > Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> > Closes: https://lore.kernel.org/all/50f12fa1-17c1-4940-a6bf-beaf61f6b17a@stanley.mountain/
> > Signed-off-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>
>
> Andrew, I was going to respin v5 of my patchset and include all these
> small fixes in it but it's a bit tricky because I would have to revert
> another unrelated patch [1] from mm-unstable which refactors relevant
> code. So far the fixes are rather small, so I think you should not
> have much trouble folding them into the original patchset when the
> time comes, but if that becomes a problem I can prepare a new version.
>
No probs, thanks, I figured it out. The descriptions of
where-it-fits-and-why are helpful.
Of course we could just pile everything onto mm-unstable HEAD and live
with a messier commit history, but I think things are manageable at
present.
