public inbox for stable@vger.kernel.org
 help / color / mirror / Atom feed
From: Miguel Ojeda <ojeda@kernel.org>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Sasha Levin <sashal@kernel.org>,
	stable@vger.kernel.org
Cc: Danilo Krummrich <dakr@kernel.org>,
	Alice Ryhl <aliceryhl@google.com>, Alyssa Ross <hi@alyssa.is>,
	NoisyCoil <noisycoil@disroot.org>,
	patches@lists.linux.dev, Miguel Ojeda <ojeda@kernel.org>
Subject: [PATCH 6.12.y 57/60] rust: fix size_t in bindgen prototypes of C builtins
Date: Fri,  7 Mar 2025 23:50:04 +0100	[thread overview]
Message-ID: <20250307225008.779961-58-ojeda@kernel.org> (raw)
In-Reply-To: <20250307225008.779961-1-ojeda@kernel.org>

From: Gary Guo <gary@garyguo.net>

commit 75c1fd41a671a0843b89d1526411a837a7163fa2 upstream.

Without `-fno-builtin`, for functions like memcpy/memmove (and many
others), bindgen seems to be using the clang-provided prototype. This
prototype is ABI-wise compatible, but the issue is that it does not have
the same information as the source code w.r.t. typedefs.

For example, bindgen generates the following:

    extern "C" {
        pub fn strlen(s: *const core::ffi::c_char) -> core::ffi::c_ulong;
    }

note that the return type is `c_ulong` (i.e. unsigned long), despite the
size_t-is-usize behavior (this is default, and we have not opted out
from it using --no-size_t-is-usize).

Similarly, memchr's size argument should be of type `__kernel_size_t`,
but bindgen generates `c_ulong` directly.

We want to ensure any `size_t` is translated to Rust `usize` so that we
can avoid having them be different type on 32-bit and 64-bit
architectures, and hence would require a lot of excessive type casts
when calling FFI functions.

I found that this bindgen behavior (which probably is caused by
libclang) can be disabled by `-fno-builtin`. Using the flag for compiled
code can result in less optimisation because compiler cannot assume
about their properties anymore, but this should not affect bindgen.

[ Trevor asked: "I wonder how reliable this behavior is. Maybe bindgen
  could do a better job controlling this, is there an open issue?".

  Gary replied: ..."apparently this is indeed the suggested approach in
  https://github.com/rust-lang/rust-bindgen/issues/1770". - Miguel ]

Signed-off-by: Gary Guo <gary@garyguo.net>
Link: https://lore.kernel.org/r/20240913213041.395655-2-gary@garyguo.net
[ Formatted comment. - Miguel ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
---
 rust/Makefile | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/rust/Makefile b/rust/Makefile
index 5e7612f69cea..a6e80caa42c1 100644
--- a/rust/Makefile
+++ b/rust/Makefile
@@ -264,7 +264,11 @@ else
 bindgen_c_flags_lto = $(bindgen_c_flags)
 endif
 
-bindgen_c_flags_final = $(bindgen_c_flags_lto) -D__BINDGEN__
+# `-fno-builtin` is passed to avoid `bindgen` from using `clang` builtin
+# prototypes for functions like `memcpy` -- if this flag is not passed,
+# `bindgen`-generated prototypes use `c_ulong` or `c_uint` depending on
+# architecture instead of generating `usize`.
+bindgen_c_flags_final = $(bindgen_c_flags_lto) -fno-builtin -D__BINDGEN__
 
 # Each `bindgen` release may upgrade the list of Rust target versions. By
 # default, the highest stable release in their list is used. Thus we need to set
-- 
2.48.1


  parent reply	other threads:[~2025-03-07 22:53 UTC|newest]

Thread overview: 72+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-07 22:49 [PATCH 6.12.y 00/60] `alloc`, `#[expect]` and "Custom FFI" Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 01/60] rust: workqueue: remove unneeded ``#[allow(clippy::new_ret_no_self)]` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 02/60] rust: sort global Rust flags Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 03/60] rust: types: avoid repetition in `{As,From}Bytes` impls Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 04/60] rust: enable `clippy::undocumented_unsafe_blocks` lint Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 05/60] rust: enable `clippy::unnecessary_safety_comment` lint Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 06/60] rust: enable `clippy::unnecessary_safety_doc` lint Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 07/60] rust: enable `clippy::ignored_unit_patterns` lint Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 08/60] rust: enable `rustdoc::unescaped_backticks` lint Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 09/60] rust: init: remove unneeded `#[allow(clippy::disallowed_names)]` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 10/60] rust: sync: remove unneeded `#[allow(clippy::non_send_fields_in_send_ty)]` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 11/60] rust: introduce `.clippy.toml` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 12/60] rust: replace `clippy::dbg_macro` with `disallowed_macros` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 13/60] rust: provide proper code documentation titles Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 14/60] rust: enable Clippy's `check-private-items` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 15/60] Documentation: rust: add coding guidelines on lints Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 16/60] rust: start using the `#[expect(...)]` attribute Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 17/60] Documentation: rust: discuss `#[expect(...)]` in the guidelines Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 18/60] rust: error: make conversion functions public Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 19/60] rust: error: optimize error type to use nonzero Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 20/60] rust: alloc: add `Allocator` trait Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 21/60] rust: alloc: separate `aligned_size` from `krealloc_aligned` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 22/60] rust: alloc: rename `KernelAllocator` to `Kmalloc` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 23/60] rust: alloc: implement `ReallocFunc` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 24/60] rust: alloc: make `allocator` module public Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 25/60] rust: alloc: implement `Allocator` for `Kmalloc` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 26/60] rust: alloc: add module `allocator_test` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 27/60] rust: alloc: implement `Vmalloc` allocator Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 28/60] rust: alloc: implement `KVmalloc` allocator Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 29/60] rust: alloc: add __GFP_NOWARN to `Flags` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 30/60] rust: alloc: implement kernel `Box` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 31/60] rust: treewide: switch to our kernel `Box` type Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 32/60] rust: alloc: remove extension of std's `Box` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 33/60] rust: alloc: add `Box` to prelude Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 34/60] rust: alloc: introduce `ArrayLayout` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 35/60] rust: alloc: implement kernel `Vec` type Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 36/60] rust: alloc: implement `IntoIterator` for `Vec` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 37/60] rust: alloc: implement `collect` for `IntoIter` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 38/60] rust: treewide: switch to the kernel `Vec` type Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 39/60] rust: alloc: remove `VecExt` extension Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 40/60] rust: alloc: add `Vec` to prelude Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 41/60] rust: error: use `core::alloc::LayoutError` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 42/60] rust: error: check for config `test` in `Error::name` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 43/60] rust: alloc: implement `contains` for `Flags` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 44/60] rust: alloc: implement `Cmalloc` in module allocator_test Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 45/60] rust: str: test: replace `alloc::format` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 46/60] rust: alloc: update module comment of alloc.rs Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 47/60] kbuild: rust: remove the `alloc` crate and `GlobalAlloc` Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 48/60] MAINTAINERS: add entry for the Rust `alloc` module Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 49/60] drm/panic: avoid reimplementing Iterator::find Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 50/60] drm/panic: remove unnecessary borrow in alignment_pattern Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 51/60] drm/panic: prefer eliding lifetimes Miguel Ojeda
2025-03-07 22:49 ` [PATCH 6.12.y 52/60] drm/panic: remove redundant field when assigning value Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 53/60] drm/panic: correctly indent continuation of line in list item Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 54/60] drm/panic: allow verbose boolean for clarity Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 55/60] drm/panic: allow verbose version check Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 56/60] rust: kbuild: expand rusttest target for macros Miguel Ojeda
2025-03-07 22:50 ` Miguel Ojeda [this message]
2025-03-07 22:50 ` [PATCH 6.12.y 58/60] rust: map `__kernel_size_t` and friends also to usize/isize Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 59/60] rust: use custom FFI integer types Miguel Ojeda
2025-03-07 22:50 ` [PATCH 6.12.y 60/60] rust: alloc: Fix `ArrayLayout` allocations Miguel Ojeda
2025-03-09  9:47 ` [PATCH 6.12.y 00/60] `alloc`, `#[expect]` and "Custom FFI" Greg Kroah-Hartman
2025-03-09 12:41 ` Ilya K
2025-03-09 14:20   ` Miguel Ojeda
2025-03-09 16:27     ` Greg Kroah-Hartman
2025-03-09 20:42       ` [PATCH 6.12.y 0/2] The two missing ones Miguel Ojeda
2025-03-09 20:42         ` [PATCH 6.12.y 1/2] rust: finish using custom FFI integer types Miguel Ojeda
2025-03-13  9:01           ` Sasha Levin
2025-03-09 20:42         ` [PATCH 6.12.y 2/2] rust: map `long` to `isize` and `char` to `u8` Miguel Ojeda
2025-03-13  9:01           ` Sasha Levin
2025-03-13 10:59             ` Miguel Ojeda
2025-03-09 21:02         ` [PATCH 6.12.y 0/2] The two missing ones Greg Kroah-Hartman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250307225008.779961-58-ojeda@kernel.org \
    --to=ojeda@kernel.org \
    --cc=aliceryhl@google.com \
    --cc=dakr@kernel.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=hi@alyssa.is \
    --cc=noisycoil@disroot.org \
    --cc=patches@lists.linux.dev \
    --cc=sashal@kernel.org \
    --cc=stable@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox