[llvm-branch-commits] [llvm] release/18.x: [AMDGPU] Fix GFX12 encoding of s_wait_event export_ready (#89622) (PR #91034)

Tom Stellard via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Wed May 8 20:17:38 PDT 2024


https://github.com/tstellar updated https://github.com/llvm/llvm-project/pull/91034

>From bce9393291a2daa8006d1da629aa2765e00f4e70 Mon Sep 17 00:00:00 2001
From: Jay Foad <jay.foad at amd.com>
Date: Tue, 23 Apr 2024 14:38:45 +0100
Subject: [PATCH] [AMDGPU] Fix GFX12 encoding of s_wait_event export_ready
 (#89622)

As well as flipping the sense of the bit, GFX12 moved it from bit 0 to
bit 1 in the encoded simm16 operand.

(cherry picked from commit e0a763c490d8ef58dca867e0ef834978ccf8e17d)
---
 llvm/lib/Target/AMDGPU/SOPInstructions.td            |  2 +-
 llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.wait.event.ll | 10 +++-------
 2 files changed, 4 insertions(+), 8 deletions(-)

diff --git a/llvm/lib/Target/AMDGPU/SOPInstructions.td b/llvm/lib/Target/AMDGPU/SOPInstructions.td
index ae5ef0541929b..5762efde73f02 100644
--- a/llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1786,7 +1786,7 @@ def : GCNPat<
 let SubtargetPredicate = isNotGFX12Plus in
   def : GCNPat <(int_amdgcn_s_wait_event_export_ready), (S_WAIT_EVENT (i16 0))>;
 let SubtargetPredicate = isGFX12Plus in
-  def : GCNPat <(int_amdgcn_s_wait_event_export_ready), (S_WAIT_EVENT (i16 1))>;
+  def : GCNPat <(int_amdgcn_s_wait_event_export_ready), (S_WAIT_EVENT (i16 2))>;
 
 // The first 10 bits of the mode register are the core FP mode on all
 // subtargets.
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.wait.event.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.wait.event.ll
index 08c77148f6ae1..433fefa434988 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.wait.event.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.wait.event.ll
@@ -5,14 +5,10 @@
 
 ; GCN-LABEL: {{^}}test_wait_event:
 ; GFX11: s_wait_event 0x0
-; GFX12: s_wait_event 0x1
+; GFX12: s_wait_event 0x2
 
-define amdgpu_ps void @test_wait_event() #0 {
+define amdgpu_ps void @test_wait_event() {
 entry:
-  call void @llvm.amdgcn.s.wait.event.export.ready() #0
+  call void @llvm.amdgcn.s.wait.event.export.ready()
   ret void
 }
-
-declare void @llvm.amdgcn.s.wait.event.export.ready() #0
-
-attributes #0 = { nounwind }



More information about the llvm-branch-commits mailing list