Re: [PATCH] x86/Kconfig: rust: Patchable function Rust compat

[Matthew Maurer]

On Mon, Dec 11, 2023 at 7:36 AM Miguel Ojeda
<miguel.ojeda.sandonis@xxxxxxxxx> wrote:
>
> On Mon, Dec 11, 2023 at 4:08 PM Matthew Maurer <mmaurer@xxxxxxxxxx> wrote:
> >
> > +       def_bool $(rs-option,-Zpatchable-function-entry=16,16)
>
> We don't have `rs-option` in mainline yet -- missing dependency? We
> will likely eventually need it, but currently we only support a single
> Rust version anyway, so we could add it (and the flag check itself)
> when we upgrade (especially if it is going to be supported soon).
Sorry, I just realized this was only in a local patch. I hadn't sent
it previously because,
as you pointed out, we currently only support one compiler revision.

I was taking this approach because Android's compilers can have patches
backported onto them when needed, so our 1.73.0 could have this flag
and make use
of it.
>
>
> Speaking of which, I can't find the flag in upstream Rust (i.e.
> outside the LLVM submodule), and I couldn't find a PR adding it. Could
> you please add a `Link:` tag to the tracking issue / PR / ... if it is
> submitted / when it is submitted? Or am I very confused?
I haven't uploaded it yet. I'm hoping to send it up later today. I can
wait until it's
uploaded for a v2 of the patch series so I can link to it directly.
>
> Cheers,
> Miguel

If I don't get the PR for `-Zpatchable-function-entry` done in a
timely fashion, I'll send
up an alternate version of this patch that just makes it depend on
!RUST, as this can
currently cause random runtime failures if features which assume
patchable entry are
used with Rust.

Re: Martin's comments (unfortunately they aren't on the same email so
I can't reply inline)
would you like me to file an issue against the R4L repository about
this before sending a v2?
I thought that repository was just for staging/discussion, and this
didn't seem likely to need it.