Re: [PATCH v1 0/2] RISC-V: enable rust

[Conor Dooley]

On Thu, Jan 25, 2024 at 01:50:05PM +0100, Miguel Ojeda wrote:
> On Thu, Jan 25, 2024 at 1:31 PM Conor Dooley <conor.dooley@xxxxxxxxxxxxx> wrote:

> > > Recently, there has been a thread in our Zulip and a couple people are
> > > experimenting: https://rust-for-linux.zulipchat.com/#narrow/stream/288089-General/topic/Bindgen.20--.20GCC.20backend.20port
> >
> > That link for me goes to a message on 22/01, so later than the email you
> > sent.
> 
> Zulip seems to scroll to the latest message in the topic -- you should
> be able to scroll a bit up, but if that doesn't work, this link should
> go to the first message:
> https://rust-for-linux.zulipchat.com/#narrow/stream/288089-General/topic/Bindgen.20--.20GCC.20backend.20port/near/412609074

Ah, thanks for the direct link :)

> 
> > That said, I gave things another spin today, in a different environment,
> > as a final check before sending and found an issue causing kernel
> > panics. RISC-V (and x86/arm64) supports kcfi (CFI_CLANG) but enabling

I mention x86 and arm64 here, because my grepping didn't see the flag
being set for x86 (in tree) or arm64 (in that series) if CFI_CLANG was
set or any mutual exclusion. Has noone tried CFI_CLANG + RUST there or
just not run into any issues?

> > sanitisers seems to be a nightly only option for rustc. The kernel I
> > built today had CFI_CLANG enabled and that caused panics when the rust
> > samples were loaded.
> >
> > The CFI_CLANG Kconfig entry has a cc-option test for whether the option
> > is supported, but from a quick check I don't see a comparable test to
> > use for rust. Even if a test was added, the current flag is an unstable
> > one, so I am not sure if testing for it is the right call in the first
> > place, given the stabilised flag would be entirely different?
> 
> Yeah, KCFI and other mitigations is WIP -- Cc'ing Ramon and Matthew
> who may be able to tell us the latest status.

Also CC Sami I guess, since he is the one who added the CFI_CLANG bits
to the kernel, and can probably comment on the suitability of adding a
check etc.

> Testing for unstable flags is fine, i.e. we only support a single
> compiler, so we can change the name when we do the upgrade.

Actually, thinking about it for a moment - if only a single compiler
version is supported (the minimum, right?) then you could just add the
-Zsanitizer=kcfi flag whenever CFI_CLANG and RUST are both set.

I'm not sure if that is a better option though. It's a choice between
CFI_CLANG being disabled if the check is not updated when the toolchain
is bumped versus being enabled for C and not for RUST. I think I prefer
the former though, tracking down the cause of the latter I would rather
not wish on a user.

I vote for having the check, even if it can only ever be true at the
moment.

Cheers,
Conor.

Attachment: signature.asc
Description: PGP signature