[all-commits] [llvm/llvm-project] b03b58: [NVPTX] Fix scoped atomic when given runtime value...

[Joseph Huber via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b03b58be38f9e89f44ef7785bcc8f9a823309c77
      https://github.com/llvm/llvm-project/commit/b03b58be38f9e89f44ef7785bcc8f9a823309c77
  Author: Joseph Huber <huberjn at outlook.com>
  Date:   2026-03-17 (Tue, 17 Mar 2026)

  Changed paths:
    M llvm/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
    R llvm/test/CodeGen/NVPTX/fence-cluster.ll
    R llvm/test/CodeGen/NVPTX/fence-nocluster.ll
    A llvm/test/CodeGen/NVPTX/fence.ll
    M llvm/test/CodeGen/NVPTX/fence.py

  Log Message:
  -----------
  [NVPTX] Fix scoped atomic when given runtime values (#185883)

Summary:
The `__scoped_` atomic builtins will expand to each supported scope in
the case of a runtime value. There are two problems:
1. Singlethreaded scope doesn't exist -> treat as no-op
2. Cluster is not supported on all tagets -> Widen to device scope

This is exactly how the AMDGPU backend handles it.

---------

Co-authored-by: gonzalobg <65027571+gonzalobg at users.noreply.github.com>



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications