Thank you for the introduction Matthew.

I’ve proposed following clang builtins for libc++ C++26 P3008 implementation:
https://github.com/llvm/llvm-project/pull/187139


  *   __atomic_fetch_fminimum
  *   __atomic_fetch_fmaximum
  *   __atomic_fetch_fminimum_num
  *   __atomic_fetch_fmaximum_num

These builtin names are analogous to __atomic_fetch_min/max .

I’d like to avoid introducing any incompatibilities between gcc and clang here.
Would the proposed builtins cause any?

Thank you
Gonzalo

From: Matthew Malcomson <[email protected]>
Date: Monday, 30. March 2026 at 15:44
To: [email protected] <[email protected]>, Gonzalo Brito <[email protected]>
Cc: Soumya AR <[email protected]>, Jakub Jelinek <[email protected]>, Jonathan 
Wakely <[email protected]>, [email protected] 
<[email protected]>, Richard Biener <[email protected]>, Joseph Myers 
<[email protected]>
Subject: LLVM FP atomic min/max variants alignment
Hello,

Gonzalo is working on adding the new FP atomic min/max variants to
libc++.  He'd like to ensure that whatever builtins he adds will not
conflict with anything we decide to do in the future.

I'll let him explain the details and list the points he'd like to align
on.  I'm sending this email to make introductions and to ease finding
the addresses to Cc.

I believe the only thing to be concerned about are the names (because
IIUC clang doesn't specify the FP environment semantics) but am Cc'ing
in Joseph Myers for visibility in case I'm wrong about that.

Cheers,
Matthew

Reply via email to