[PATCH] D72222: [Driver][CodeGen] Add -fpatchable-function-entry=N[,0]

[Fangrui Song via Phabricator via cfe-commits]

MaskRay added a comment.

In D72222#1847167 <https://reviews.llvm.org/D72222#1847167>, @mrutland wrote:

> In D72222#1846796 <https://reviews.llvm.org/D72222#1846796>, @mrutland wrote:
>
> > In D72222#1839207 <https://reviews.llvm.org/D72222#1839207>, @MaskRay wrote:
> >
> > > I shall also mention that we are essentially making decisions for x86 people's endbr32/endbr64. I hope you can engage in x86 maintainers. Clang's current layout is recorded at D73071 <https://reviews.llvm.org/D73071>.
> >
> >
> > That's a good point w.r.t. x86; I will get in touch with the people working on ftrace and live-patching there
>
>
> I spoke with Steven Rostedt (ftrace maintainer), and Josh Poimboeuf (livepatching maintainer) in the OFTC/#linux-rt IRC channel. Today x86 uses `-mfentry`, and they have no reason to move to `-fpatchable-function-entry` so long as the `ENDBR32/ENDBR64` instructions compose nicely with `-mfentry`.
>
> Given that, I don't think x86 kernel folk care either way.
>
> On the GCC side I was under the impression that x86 would go the same way as arm64, per https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92424#c4
>
> There's a GCC ticket for x86: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93492
>
> Thanks,
>  Mark.


I created D73680 <https://reviews.llvm.org/D73680> to place the patch label after BTI.

@hans Is there still time to cherry pick the patch to release/10.x? See above, Linux developers really want the Clang release to have compatible behavior with GCC.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72222/new/

https://reviews.llvm.org/D72222