[llvm] [AMDGPU] Mark WAIT_ASYNCMARK as zero-size instruction (PR #195107)

[via llvm-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-amdgpu

Author: Zhewen Yu (Yu-Zhewen)

<details>
<summary>Changes</summary>

`WAIT_ASYNCMARK` emits no bytes but was inheriting `Size = 4` from `SOPP_Pseudo`.

Without the fix, #<!-- -->194362 causes: `Size mismatch for: WAIT_ASYNCMARK 1 Expected exact size: 4 Actual size: 0`

---
Full diff: https://github.com/llvm/llvm-project/pull/195107.diff


1 Files Affected:

- (modified) llvm/lib/Target/AMDGPU/SOPInstructions.td (+1) 


``````````diff
diff --git a/llvm/lib/Target/AMDGPU/SOPInstructions.td b/llvm/lib/Target/AMDGPU/SOPInstructions.td
index eb7ad8bcb04ce..e0bdcf61440f3 100644
--- a/llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1736,6 +1736,7 @@ def ASYNCMARK : SPseudoInstSI<(outs), (ins),
 def WAIT_ASYNCMARK : SOPP_Pseudo <"", (ins s16imm:$simm16), "$simm16",
     [(int_amdgcn_wait_asyncmark timm:$simm16)]> {
     let maybeAtomic = 0;
+    let Size = 0;
     let isMeta = 1;
 }
 }

``````````

</details>


https://github.com/llvm/llvm-project/pull/195107