linux-kernel - Re: [PATCH] rust: fix building firmware abstraction on 32bit arm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <D94KNIHTJOWU.1EHA7217LSC4S@proton.me>
Date: Sat, 12 Apr 2025 10:01:22 +0000
From: Benno Lossin <benno.lossin@...ton.me>
To: Miguel Ojeda <miguel.ojeda.sandonis@...il.com>
Cc: Christian Schrefl <chrisi.schrefl@...il.com>, Luis Chamberlain <mcgrof@...nel.org>, Russ Weight <russ.weight@...ux.dev>, Danilo Krummrich <dakr@...nel.org>, Miguel Ojeda <ojeda@...nel.org>, Alex Gaynor <alex.gaynor@...il.com>, Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>, Björn Roy Baron <bjorn3_gh@...tonmail.com>, Andreas Hindborg <a.hindborg@...nel.org>, Alice Ryhl <aliceryhl@...gle.com>, Trevor Gross <tmgross@...ch.edu>, linux-kernel@...r.kernel.org, rust-for-linux@...r.kernel.org, stable@...r.kernel.org
Subject: Re: [PATCH] rust: fix building firmware abstraction on 32bit arm

On Fri Apr 11, 2025 at 4:15 PM CEST, Miguel Ojeda wrote:
> On Fri, Apr 11, 2025 at 2:46 PM Benno Lossin <benno.lossin@...ton.me> wrote:
>>
>> Ah I overlooked this, you should be using `kernel::ffi` (or
>> `crate::ffi`) instead of `core`. (for `c_char` it doesn't matter, but we
>> shouldn't be using `core::ffi`, since we have our own mappings).
>
> In 6.6, C `char` changed to unsigned, but `core::ffi::c_char` is
> signed (in x86_64 at least).
>
> We should just never use `core::ffi` (except in `rust/ffi.rs`, of
> course) -- I think we should just add the C types to the prelude
> (which we discussed in the past) so that it is easy to avoid the
> mistake (something like the patch attached as the end result, but
> tested and across a kernel cycle or two) and mention it in the Coding
> Guidelines. Thoughts?

Yeah sounds like a good idea.

> I tried to use Clippy's `disallowed-types` too:
>
>     disallowed-types = [
>         { path = "core::ffi::c_void", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_char", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_schar", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_uchar", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_short", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_ushort", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_int", reason = "the `kernel::ffi` types
> should be used instead" },
>         { path = "core::ffi::c_uint", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_long", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_ulong", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_longlong", reason = "the `kernel::ffi`
> types should be used instead" },
>         { path = "core::ffi::c_ulonglong", reason = "the `kernel::ffi`
> types should be used instead" },
>     ]
>
> But it goes across aliases.

We could make the types in `ffi` be transparent newtypes. But not sure
if that could interfere with kCFI or other stuff.

---
Cheers,
Benno