Re: [PATCH v3 00/35] Compiler-Based Capability- and Locking-Analysis

On Fri, 14 Nov 2025 at 05:38, Nathan Chancellor <[email protected]> wrote:
> On Thu, Nov 13, 2025 at 03:30:08PM +0100, Marco Elver wrote:
> > On Fri, 19 Sept 2025 at 11:10, Marco Elver <[email protected]> wrote:
> > [..]
> > > I went with "context guard" to refer to the objects themselves, as that
> > > doesn't look too odd. It does match the concept of "guard" in
> > > <linux/cleanup.h>.
> > >
> > > See second attempt below.
> > [..]
> >
> > I finally got around baking this into a renamed series, that now calls
> > it "Context Analysis" - here's a preview:
> > https://git.kernel.org/pub/scm/linux/kernel/git/melver/linux.git/log/?h=ctx-analysis/dev
> >
> > As for when we should give this v4 another try: I'm 50/50 on sending
> > this now vs. waiting for final Clang 22 to be released (~March 2026).
> >
> > Preferences?
>
> For the record, I can continue to upload clang snapshots for testing and
> validating this plus the sooner this hits a tree that goes into -next,
> the sooner the ClangBuiltLinux infrastructure can start testing it. I
> assume there will not need to be many compiler side fixes but if

I hope so ... Famous last words. ;-)

> __counted_by has shown us anything, it is that getting this stuff
> deployed and into the hands of people who want to use it is the only
> real way to find corner cases to address. No strong objection from me if
> you want to wait for clang-22 to actually be released though for more
> access.

Thanks, Nathan - having ClangBuiltLinux infra help test would be very helpful.
Unless I hear otherwise, I can send v4 next week for review - in case
of a v5 I will wait until ~March (as that coincides with Clang 22
release, and for lack of time on my end between Jan and March).
Could also skip the subsystem-enablement patches for now; only the
patches until the MAINTAINERS patch are the bare minimum, the rest can
be taken later by individual maintainers.

Thanks,
-- Marco