Re: [GIT PULL] MM updates for 6.14-rc1

[Linus Torvalds <torvalds@linux-foundation.org>]

On Sun, 26 Jan 2025 at 09:22, Uros Bizjak <ubizjak@gmail.com> wrote:
>
> bindgen ChangeLog does not mention __typeof_unqual__, so I assume that
> the support depends on RUSTC_LLVM_VERSION.

That seems to be a reasonable assumption. I guess some crazy setup
*could* install a bindgen that uses a different llvm version than
rustc itself, but that sounds pretty damn broken.

Let's add Miguel to the participants. Miguel, see

   https://lore.kernel.org/all/CAHk-=whddBhfi5DUi370W3pYs+z3r2E7KYuHjwR=a1eRig5Gxg@mail.gmail.com/

for my "this doesn't work" report, and then Uros' suggested fix in

   https://lore.kernel.org/all/CAFULd4a-2F_zKMeR0Yjo2WhRLmyoOQ1VdR2qdV34BrM-b_cQCQ@mail.gmail.com/

> Does the attached incremental patch work for you?

That does seem to work, but I'd admittedly be happier if we could just
find some way to dynamically disable/enable __typeof_unqual__ based on
which compiler is used, rather than make it a config option. So that
bindgen would not see it, but a recent enough C compiler would.

We already use '__has_attribute()' for some of these things. There's a
'__has_extension()' thing that comes from clang but that gcc also
supports.

But I can't find the list of extensions that that model supports, and
I guess typeof_unqual isn't on that list if I find it.

In the absence of something _clean_ like that, can we just do it in
<linux/compiler-xyz.h> instead? IOW, not use the CONFIG_xyz tests for
this at all, but do something like

  /*
   * bindgen uses LLVM even if our C compiler is gcc, so we cannot
   * rely on CONFIG_CC_HAS_TYPEOF_UNQUAL
   */
  #if __clang_major__ >= 19
  # define CC_HAS_TYPEOF_UNQUAL 1
  #endif

in <linux/compiler-clang.h> and the gcc version check for gcc (I don't
know when __typeof_unqual__ made it)?

This does show that our whole "CC_HAS_XYZ" is kind of broken for the
Rust integration in general.

                 Linus