Skip to content

AMDGPU: Fix not folding splat immediate into VGPR MFMA src2 #150628

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Jul 26, 2025
Merged

AMDGPU: Fix not folding splat immediate into VGPR MFMA src2 #150628

merged 1 commit into from
Jul 26, 2025

Conversation

@arsenm
Copy link
Contributor

@arsenm arsenm commented Jul 25, 2025

No description provided.

@arsenm arsenm mentioned this pull request Jul 25, 2025
Merged
@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Jul 25, 2025 •
edited
Loading

This stack of pull requests is managed by Graphite. Learn more about stacking.

@arsenm arsenm mentioned this pull request Jul 25, 2025
Merged
@arsenm arsenm marked this pull request as ready for review July 25, 2025 14:53
@llvmbot
Copy link
Member

llvmbot commented Jul 25, 2025

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

Patch is 20.86 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/150628.diff

3 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/SIFoldOperands.cpp (+4)
  • (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll (+48-114)
  • (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.ll (+18-71)
diff --git a/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp b/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
index e5d1eaad2b8f4..b77da4d612dd4 100644
--- a/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
+++ b/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
@@ -1062,9 +1062,13 @@ bool SIFoldOperandsImpl::tryFoldRegSeqSplat(
     switch (OpTy) {
     case AMDGPU::OPERAND_REG_INLINE_AC_INT32:
     case AMDGPU::OPERAND_REG_INLINE_AC_FP32:
+    case AMDGPU::OPERAND_REG_INLINE_C_INT32:
+    case AMDGPU::OPERAND_REG_INLINE_C_FP32:
       OpRC = TRI->getSubRegisterClass(OpRC, AMDGPU::sub0);
       break;
     case AMDGPU::OPERAND_REG_INLINE_AC_FP64:
+    case AMDGPU::OPERAND_REG_INLINE_C_FP64:
+    case AMDGPU::OPERAND_REG_INLINE_C_INT64:
       OpRC = TRI->getSubRegisterClass(OpRC, AMDGPU::sub0_sub1);
       break;
     default:
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
index adbc2df4b3474..7d5bc3f4c0c39 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
@@ -1083,58 +1083,36 @@ define amdgpu_kernel void @test_mfma_f64_16x16x4f64_splat_imm_1(ptr addrspace(1)
 ; GFX90A-VGPR:       ; %bb.0: ; %bb
 ; GFX90A-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX90A-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v0, 0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v1, 0x3ff00000
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX90A-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[2:3], s[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], v[6:7], v[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[12:13], s[6:7], s[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[6:7], v[4:5], v[4:5] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[4:5], v[2:3], v[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[2:3], v[0:1], v[0:1] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], s[2:3], s[2:3] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[6:7], s[6:7] op_sel:[0,1]
 ; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], 1.0
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX90A-VGPR-NEXT:    s_nop 7
 ; GFX90A-VGPR-NEXT:    s_nop 7
-; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v0, v[6:9], s[0:1] offset:16
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v0, v[2:5], s[0:1]
+; GFX90A-VGPR-NEXT:    s_nop 0
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX90A-VGPR-NEXT:    s_endpgm
 ;
 ; GFX942-VGPR-LABEL: test_mfma_f64_16x16x4f64_splat_imm_1:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 0x3ff00000
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], v[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[12:13], s[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[6:7], v[4:5]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[4:5], v[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[2:3], v[0:1]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], s[2:3]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[6:7]
 ; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], 1.0
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 7
-; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v0, v[6:9], s[0:1] offset:16
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v0, v[2:5], s[0:1]
+; GFX942-VGPR-NEXT:    s_nop 0
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX942-VGPR-NEXT:    s_endpgm
 bb:
   %mai.1 = tail call <4 x double> @llvm.amdgcn.mfma.f64.16x16x4f64(double %a, double %b, <4 x double> splat (double 1.0), i32 0, i32 0, i32 0)
@@ -1184,58 +1162,36 @@ define amdgpu_kernel void @test_mfma_f64_16x16x4f64_splat_imm_neg1(ptr addrspace
 ; GFX90A-VGPR:       ; %bb.0: ; %bb
 ; GFX90A-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX90A-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v0, 0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v1, 0xbff00000
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX90A-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[2:3], s[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], v[6:7], v[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[12:13], s[6:7], s[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[6:7], v[4:5], v[4:5] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[4:5], v[2:3], v[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[2:3], v[0:1], v[0:1] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], s[2:3], s[2:3] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[6:7], s[6:7] op_sel:[0,1]
 ; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], -1.0
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX90A-VGPR-NEXT:    s_nop 7
 ; GFX90A-VGPR-NEXT:    s_nop 7
-; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v0, v[6:9], s[0:1] offset:16
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v0, v[2:5], s[0:1]
+; GFX90A-VGPR-NEXT:    s_nop 0
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX90A-VGPR-NEXT:    s_endpgm
 ;
 ; GFX942-VGPR-LABEL: test_mfma_f64_16x16x4f64_splat_imm_neg1:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 0xbff00000
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], v[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[12:13], s[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[6:7], v[4:5]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[4:5], v[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[2:3], v[0:1]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], s[2:3]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[6:7]
 ; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], -1.0
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 7
-; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v0, v[6:9], s[0:1] offset:16
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v0, v[2:5], s[0:1]
+; GFX942-VGPR-NEXT:    s_nop 0
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX942-VGPR-NEXT:    s_endpgm
 bb:
   %mai.1 = tail call <4 x double> @llvm.amdgcn.mfma.f64.16x16x4f64(double %a, double %b, <4 x double> splat (double -1.0), i32 0, i32 0, i32 0)
@@ -1285,58 +1241,36 @@ define amdgpu_kernel void @test_mfma_f64_16x16x4f64_splat_imm_int_64(ptr addrspa
 ; GFX90A-VGPR:       ; %bb.0: ; %bb
 ; GFX90A-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX90A-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v1, 0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v0, 64
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX90A-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[2:3], s[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], v[6:7], v[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[12:13], s[6:7], s[6:7] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[6:7], v[4:5], v[4:5] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[4:5], v[2:3], v[2:3] op_sel:[0,1]
-; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[2:3], v[0:1], v[0:1] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[8:9], s[2:3], s[2:3] op_sel:[0,1]
+; GFX90A-VGPR-NEXT:    v_pk_mov_b32 v[10:11], s[6:7], s[6:7] op_sel:[0,1]
 ; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], 64
+; GFX90A-VGPR-NEXT:    v_mfma_f64_16x16x4f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 blgp:3
+; GFX90A-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX90A-VGPR-NEXT:    s_nop 7
 ; GFX90A-VGPR-NEXT:    s_nop 7
-; GFX90A-VGPR-NEXT:    s_nop 1
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v1, v[6:9], s[0:1] offset:16
-; GFX90A-VGPR-NEXT:    global_store_dwordx4 v1, v[2:5], s[0:1]
+; GFX90A-VGPR-NEXT:    s_nop 0
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX90A-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX90A-VGPR-NEXT:    s_endpgm
 ;
 ; GFX942-VGPR-LABEL: test_mfma_f64_16x16x4f64_splat_imm_int_64:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 64
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v1
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], v[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[12:13], s[6:7]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[6:7], v[4:5]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[4:5], v[2:3]
-; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[2:3], v[0:1]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[8:9], s[2:3]
+; GFX942-VGPR-NEXT:    v_mov_b64_e32 v[10:11], s[6:7]
 ; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9]
-; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[2:9], v[10:11], v[12:13], v[2:9] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], 64
+; GFX942-VGPR-NEXT:    v_mfma_f64_16x16x4_f64 v[0:7], v[8:9], v[10:11], v[0:7] cbsz:1 abid:2 neg:[1,1,0]
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, 0
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 7
-; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v1, v[6:9], s[0:1] offset:16
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v1, v[2:5], s[0:1]
+; GFX942-VGPR-NEXT:    s_nop 0
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[4:7], s[0:1] offset:16
+; GFX942-VGPR-NEXT:    global_store_dwordx4 v8, v[0:3], s[0:1]
 ; GFX942-VGPR-NEXT:    s_endpgm
 bb:
   %mai.1 = tail call <4 x double> @llvm.amdgcn.mfma.f64.16x16x4f64(double %a, double %b, <4 x double> splat (double bitcast (i64 64 to double)), i32 0, i32 0, i32 0)
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.ll
index d7257a49b00bb..b4a6451908a6f 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.ll
@@ -3238,27 +3238,11 @@ define amdgpu_kernel void @test_mfma_i32_16x16x4i8_splatimm_src2_64(ptr addrspac
 ;
 ; GFX942-VGPR-LABEL: test_mfma_i32_16x16x4i8_splatimm_src2_64:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v17, 1
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 64
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v9, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v10, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v11, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v12, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v13, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v14, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v15, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v18, 2
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 1
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 2
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v16, 0
-; GFX942-VGPR-NEXT:    v_mfma_i32_16x16x4_4b_i8 v[0:15], v17, v18, v[0:15] cbsz:1 abid:2 blgp:3
+; GFX942-VGPR-NEXT:    v_mfma_i32_16x16x4_4b_i8 v[0:15], v0, v1, 64 cbsz:1 abid:2 blgp:3
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 1
@@ -3463,13 +3447,10 @@ define amdgpu_kernel void @test_mfma_i32_4x4x4i8_splat_imm_src2_1(ptr addrspace(
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 1
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, 2
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 2
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, 0
 ; GFX942-VGPR-NEXT:    s_nop 0
-; GFX942-VGPR-NEXT:    v_mfma_i32_4x4x4_16b_i8 v[0:3], v0, v5, v[0:3] cbsz:1 abid:2 blgp:3
+; GFX942-VGPR-NEXT:    v_mfma_i32_4x4x4_16b_i8 v[0:3], v0, v1, 1 cbsz:1 abid:2 blgp:3
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX942-VGPR-NEXT:    s_nop 3
 ; GFX942-VGPR-NEXT:    global_store_dwordx4 v4, v[0:3], s[0:1]
@@ -4483,13 +4464,10 @@ define amdgpu_kernel void @test_mfma_f32_4x4x1f32_imm_splat(ptr addrspace(1) %ar
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 1.0
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, 2.0
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 2.0
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, 0
 ; GFX942-VGPR-NEXT:    s_nop 0
-; GFX942-VGPR-NEXT:    v_mfma_f32_4x4x1_16b_f32 v[0:3], v0, v5, v[0:3]
+; GFX942-VGPR-NEXT:    v_mfma_f32_4x4x1_16b_f32 v[0:3], v0, v1, 1.0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX942-VGPR-NEXT:    s_nop 2
 ; GFX942-VGPR-NEXT:    global_store_dwordx4 v4, v[0:3], s[0:1]
@@ -4627,25 +4605,10 @@ define amdgpu_kernel void @test_mfma_f32_16x16x1f32_imm_splat(ptr addrspace(1) %
 ; GFX942-VGPR-LABEL: test_mfma_f32_16x16x1f32_imm_splat:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 1.0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v9, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v10, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v11, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v12, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v13, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v14, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v15, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v17, 2.0
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, 2.0
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v16, 0
-; GFX942-VGPR-NEXT:    v_mfma_f32_16x16x1_4b_f32 v[0:15], v0, v17, v[0:15]
+; GFX942-VGPR-NEXT:    v_mfma_f32_16x16x1_4b_f32 v[0:15], v0, v1, 1.0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 0
@@ -4797,36 +4760,20 @@ define amdgpu_kernel void @test_mfma_f32_32x32x8f16_imm_splat(ptr addrspace(1) %
 ;
 ; GFX942-VGPR-LABEL: test_mfma_f32_32x32x8f16_imm_splat:
 ; GFX942-VGPR:       ; %bb.0: ; %bb
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v16, 0x3c003c00
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v17, v16
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 1.0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v18, 0x40004000
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v0, 0x3c003c00
 ; GFX942-VGPR-NEXT:    v_mov_b32_e32 v1, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v4, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v5, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v6, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v7, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v8, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v9, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v10, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v11, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v12, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v13, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v14, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v15, v0
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v19, v18
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v2, 0x40004000
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v3, v2
 ; GFX942-VGPR-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
-; GFX942-VGPR-NEXT:    v_mov_b32_e32 v20, 0
-; GFX942-VGPR-NEXT:    v_mfma_f32_32x32x8_f16 v[0:15], v[16:17], v[18:19], v[0:15]
+; GFX942-VGPR-NEXT:    v_mov_b32_e32 v16, 0
+; GFX942-VGPR-NEXT:    v_mfma_f32_32x32x8_f16 v[0:15], v[0:1], v[2:3], 1.0
 ; GFX942-VGPR-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX942-VGPR-NEXT:    s_nop 7
 ; GFX942-VGPR-NEXT:    s_nop 1
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v20, v[12:15], s[0:1] offset:48
-; GFX942-VGPR-NEXT:    global_store_dwordx4 v20, v[8...
[truncated]

@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Jul 25, 2025 •
edited
Loading

Merge activity

@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Jul 25, 2025

I suspect this regressed in 3601849 and just didn't have tests, so perhaps this should go to the release branch

@arsenm arsenm force-pushed the users/arsenm/amdgpu/add-tests-mfma-src2-imm branch from ad766da to 4ee7be5 Compare July 25, 2025 23:48
@arsenm arsenm force-pushed the users/arsenm/amdgpu/fix-not-folding-splat-imm-vgpr-mfma branch from 3c67b08 to 8ccb289 Compare July 25, 2025 23:48
@arsenm arsenm force-pushed the users/arsenm/amdgpu/add-tests-mfma-src2-imm branch from 4ee7be5 to 2f95bc3 Compare July 26, 2025 00:47
@arsenm arsenm force-pushed the users/arsenm/amdgpu/fix-not-folding-splat-imm-vgpr-mfma branch from 8ccb289 to 4a00166 Compare July 26, 2025 00:48
@arsenm arsenm force-pushed the users/arsenm/amdgpu/add-tests-mfma-src2-imm branch 7 times, most recently from 54837d8 to 4d8c839 Compare July 26, 2025 02:25
Base automatically changed from users/arsenm/amdgpu/add-tests-mfma-src2-imm to main July 26, 2025 02:58
@arsenm arsenm force-pushed the users/arsenm/amdgpu/fix-not-folding-splat-imm-vgpr-mfma branch 4 times, most recently from 859c60f to 5c71bc4 Compare July 26, 2025 03:59
@arsenm arsenm force-pushed the users/arsenm/amdgpu/fix-not-folding-splat-imm-vgpr-mfma branch from 5c71bc4 to 4cbec19 Compare July 26, 2025 04:25
@arsenm arsenm merged commit 5f3eea7 into main Jul 26, 2025
9 checks passed
@arsenm arsenm deleted the users/arsenm/amdgpu/fix-not-folding-splat-imm-vgpr-mfma branch July 26, 2025 04:54
@llvm-ci
Copy link
Collaborator

llvm-ci commented Jul 26, 2025

LLVM Buildbot has detected a new failure on builder fuchsia-x86_64-linux running on fuchsia-debian-64-us-central1-a-1 while building llvm at step 4 "annotate".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/11/builds/20420

Here is the relevant piece of the build log for the reference 
Step 4 (annotate) failure: 'python ../llvm-zorg/zorg/buildbot/builders/annotated/fuchsia-linux.py ...' (failure)
...
[499/2522] Building CXX object libc/src/complex/generic/CMakeFiles/libc.src.complex.generic.cprojl.dir/cprojl.cpp.obj
[500/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_has_single_bit_us.dir/stdc_has_single_bit_us.cpp.obj
[501/2522] Building CXX object libc/src/__support/StringUtil/CMakeFiles/libc.src.__support.StringUtil.error_to_string.dir/error_to_string.cpp.obj
[502/2522] Building CXX object libc/src/complex/generic/CMakeFiles/libc.src.complex.generic.cproj.dir/cproj.cpp.obj
[503/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignalingf.dir/issignalingf.cpp.obj
[504/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignaling.dir/issignaling.cpp.obj
[505/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonicalf.dir/iscanonicalf.cpp.obj
[506/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_has_single_bit_ul.dir/stdc_has_single_bit_ul.cpp.obj
[507/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freelist.dir/freelist.cpp.obj
[508/2522] Building CXX object libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj
FAILED: libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj 
/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/./bin/clang++ --target=armv6m-none-eabi -DLIBC_NAMESPACE=__llvm_libc_22_0_0_git -I/var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc -isystem /var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/include/armv6m-unknown-none-eabi --target=armv6m-none-eabi -Wno-atomic-alignment "-Dvfprintf(stream, format, vlist)=vprintf(format, vlist)" "-Dfprintf(stream, format, ...)=printf(format)" "-Dfputs(string, stream)=puts(string)" -D_LIBCPP_PRINT=1 -mthumb -fPIC -fno-semantic-interposition -fvisibility-inlines-hidden -Werror=date-time -Werror=unguarded-availability-new -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -Wimplicit-fallthrough -Wcovered-switch-default -Wno-noexcept-type -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wsuggest-override -Wstring-conversion -Wmisleading-indentation -Wctad-maybe-unsupported -ffunction-sections -fdata-sections -ffile-prefix-map=/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/runtimes/runtimes-armv6m-none-eabi-bins=../../../../llvm-project -ffile-prefix-map=/var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/= -no-canonical-prefixes -Os -DNDEBUG --target=armv6m-none-eabi -DLIBC_QSORT_IMPL=LIBC_QSORT_HEAP_SORT -DLIBC_TYPES_TIME_T_IS_32_BIT -DLIBC_ADD_NULL_CHECKS "-DLIBC_MATH=(LIBC_MATH_SKIP_ACCURATE_PASS | LIBC_MATH_SMALL_TABLES)" -DLIBC_ERRNO_MODE=LIBC_ERRNO_MODE_EXTERNAL -fpie -ffreestanding -DLIBC_FULL_BUILD -nostdlibinc -ffixed-point -fno-builtin -fno-exceptions -fno-lax-vector-conversions -fno-unwind-tables -fno-asynchronous-unwind-tables -fno-rtti -ftrivial-auto-var-init=pattern -fno-omit-frame-pointer -Wall -Wextra -Werror -Wconversion -Wno-sign-conversion -Wdeprecated -Wno-c99-extensions -Wno-gnu-imaginary-constant -Wno-pedantic -Wimplicit-fallthrough -Wwrite-strings -Wextra-semi -Wnewline-eof -Wnonportable-system-include-path -Wstrict-prototypes -Wthread-safety -Wglobal-constructors -DLIBC_COPT_PUBLIC_PACKAGING -MD -MT libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj -MF libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj.d -o libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj -c /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/vsscanf.cpp
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/vsscanf.cpp:14:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/scanf_main.h:14:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/converter.h:15:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/core_structs.h:16:
/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/./lib/clang/22/include/inttypes.h:24:15: fatal error: 'inttypes.h' file not found
   24 | #include_next <inttypes.h>
      |               ^~~~~~~~~~~~
1 error generated.
[509/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_uc.dir/stdc_bit_floor_uc.cpp.obj
[510/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonicall.dir/iscanonicall.cpp.obj
[511/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonical.dir/iscanonical.cpp.obj
[512/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignalingl.dir/issignalingl.cpp.obj
[513/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freetrie.dir/freetrie.cpp.obj
[514/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_us.dir/stdc_bit_ceil_us.cpp.obj
[515/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_us.dir/stdc_bit_floor_us.cpp.obj
[516/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ui.dir/stdc_bit_ceil_ui.cpp.obj
[517/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ull.dir/stdc_bit_floor_ull.cpp.obj
[518/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ui.dir/stdc_bit_floor_ui.cpp.obj
[519/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_uc.dir/stdc_bit_ceil_uc.cpp.obj
[520/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ul.dir/stdc_bit_ceil_ul.cpp.obj
[521/2522] Building CXX object libc/src/inttypes/CMakeFiles/libc.src.inttypes.imaxdiv.dir/imaxdiv.cpp.obj
[522/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freelist_heap.dir/freelist_heap.cpp.obj
[523/2522] Building CXX object libc/src/inttypes/CMakeFiles/libc.src.inttypes.imaxabs.dir/imaxabs.cpp.obj
[524/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ull.dir/stdc_bit_ceil_ull.cpp.obj
[525/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ul.dir/stdc_bit_floor_ul.cpp.obj
[526/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ui.dir/stdc_first_leading_zero_ui.cpp.obj
[527/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_trailing_zero_us.dir/stdc_first_trailing_zero_us.cpp.obj
[528/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_uc.dir/stdc_first_leading_zero_uc.cpp.obj
[529/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_us.dir/stdc_first_leading_zero_us.cpp.obj
[530/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ull.dir/stdc_first_leading_zero_ull.cpp.obj
[531/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ul.dir/stdc_first_leading_one_ul.cpp.obj
[532/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_trailing_zero_ui.dir/stdc_first_trailing_zero_ui.cpp.obj
[533/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_us.dir/stdc_first_leading_one_us.cpp.obj
[534/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ui.dir/stdc_first_leading_one_ui.cpp.obj
[535/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ul.dir/stdc_first_leading_zero_ul.cpp.obj
[536/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ull.dir/stdc_first_leading_one_ull.cpp.obj
[537/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_uc.dir/stdc_first_leading_one_uc.cpp.obj
Step 6 (build) failure: build (failure)
...
[499/2522] Building CXX object libc/src/complex/generic/CMakeFiles/libc.src.complex.generic.cprojl.dir/cprojl.cpp.obj
[500/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_has_single_bit_us.dir/stdc_has_single_bit_us.cpp.obj
[501/2522] Building CXX object libc/src/__support/StringUtil/CMakeFiles/libc.src.__support.StringUtil.error_to_string.dir/error_to_string.cpp.obj
[502/2522] Building CXX object libc/src/complex/generic/CMakeFiles/libc.src.complex.generic.cproj.dir/cproj.cpp.obj
[503/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignalingf.dir/issignalingf.cpp.obj
[504/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignaling.dir/issignaling.cpp.obj
[505/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonicalf.dir/iscanonicalf.cpp.obj
[506/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_has_single_bit_ul.dir/stdc_has_single_bit_ul.cpp.obj
[507/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freelist.dir/freelist.cpp.obj
[508/2522] Building CXX object libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj
FAILED: libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj 
/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/./bin/clang++ --target=armv6m-none-eabi -DLIBC_NAMESPACE=__llvm_libc_22_0_0_git -I/var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc -isystem /var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/include/armv6m-unknown-none-eabi --target=armv6m-none-eabi -Wno-atomic-alignment "-Dvfprintf(stream, format, vlist)=vprintf(format, vlist)" "-Dfprintf(stream, format, ...)=printf(format)" "-Dfputs(string, stream)=puts(string)" -D_LIBCPP_PRINT=1 -mthumb -fPIC -fno-semantic-interposition -fvisibility-inlines-hidden -Werror=date-time -Werror=unguarded-availability-new -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -Wimplicit-fallthrough -Wcovered-switch-default -Wno-noexcept-type -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wsuggest-override -Wstring-conversion -Wmisleading-indentation -Wctad-maybe-unsupported -ffunction-sections -fdata-sections -ffile-prefix-map=/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/runtimes/runtimes-armv6m-none-eabi-bins=../../../../llvm-project -ffile-prefix-map=/var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/= -no-canonical-prefixes -Os -DNDEBUG --target=armv6m-none-eabi -DLIBC_QSORT_IMPL=LIBC_QSORT_HEAP_SORT -DLIBC_TYPES_TIME_T_IS_32_BIT -DLIBC_ADD_NULL_CHECKS "-DLIBC_MATH=(LIBC_MATH_SKIP_ACCURATE_PASS | LIBC_MATH_SMALL_TABLES)" -DLIBC_ERRNO_MODE=LIBC_ERRNO_MODE_EXTERNAL -fpie -ffreestanding -DLIBC_FULL_BUILD -nostdlibinc -ffixed-point -fno-builtin -fno-exceptions -fno-lax-vector-conversions -fno-unwind-tables -fno-asynchronous-unwind-tables -fno-rtti -ftrivial-auto-var-init=pattern -fno-omit-frame-pointer -Wall -Wextra -Werror -Wconversion -Wno-sign-conversion -Wdeprecated -Wno-c99-extensions -Wno-gnu-imaginary-constant -Wno-pedantic -Wimplicit-fallthrough -Wwrite-strings -Wextra-semi -Wnewline-eof -Wnonportable-system-include-path -Wstrict-prototypes -Wthread-safety -Wglobal-constructors -DLIBC_COPT_PUBLIC_PACKAGING -MD -MT libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj -MF libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj.d -o libc/src/stdio/CMakeFiles/libc.src.stdio.vsscanf.dir/vsscanf.cpp.obj -c /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/vsscanf.cpp
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/vsscanf.cpp:14:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/scanf_main.h:14:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/converter.h:15:
In file included from /var/lib/buildbot/fuchsia-x86_64-linux/llvm-project/libc/src/stdio/scanf_core/core_structs.h:16:
/var/lib/buildbot/fuchsia-x86_64-linux/build/llvm-build-68pxb90k/./lib/clang/22/include/inttypes.h:24:15: fatal error: 'inttypes.h' file not found
   24 | #include_next <inttypes.h>
      |               ^~~~~~~~~~~~
1 error generated.
[509/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_uc.dir/stdc_bit_floor_uc.cpp.obj
[510/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonicall.dir/iscanonicall.cpp.obj
[511/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.iscanonical.dir/iscanonical.cpp.obj
[512/2522] Building CXX object libc/src/math/generic/CMakeFiles/libc.src.math.generic.issignalingl.dir/issignalingl.cpp.obj
[513/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freetrie.dir/freetrie.cpp.obj
[514/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_us.dir/stdc_bit_ceil_us.cpp.obj
[515/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_us.dir/stdc_bit_floor_us.cpp.obj
[516/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ui.dir/stdc_bit_ceil_ui.cpp.obj
[517/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ull.dir/stdc_bit_floor_ull.cpp.obj
[518/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ui.dir/stdc_bit_floor_ui.cpp.obj
[519/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_uc.dir/stdc_bit_ceil_uc.cpp.obj
[520/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ul.dir/stdc_bit_ceil_ul.cpp.obj
[521/2522] Building CXX object libc/src/inttypes/CMakeFiles/libc.src.inttypes.imaxdiv.dir/imaxdiv.cpp.obj
[522/2522] Building CXX object libc/src/__support/CMakeFiles/libc.src.__support.freelist_heap.dir/freelist_heap.cpp.obj
[523/2522] Building CXX object libc/src/inttypes/CMakeFiles/libc.src.inttypes.imaxabs.dir/imaxabs.cpp.obj
[524/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_ceil_ull.dir/stdc_bit_ceil_ull.cpp.obj
[525/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_bit_floor_ul.dir/stdc_bit_floor_ul.cpp.obj
[526/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ui.dir/stdc_first_leading_zero_ui.cpp.obj
[527/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_trailing_zero_us.dir/stdc_first_trailing_zero_us.cpp.obj
[528/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_uc.dir/stdc_first_leading_zero_uc.cpp.obj
[529/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_us.dir/stdc_first_leading_zero_us.cpp.obj
[530/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ull.dir/stdc_first_leading_zero_ull.cpp.obj
[531/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ul.dir/stdc_first_leading_one_ul.cpp.obj
[532/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_trailing_zero_ui.dir/stdc_first_trailing_zero_ui.cpp.obj
[533/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_us.dir/stdc_first_leading_one_us.cpp.obj
[534/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ui.dir/stdc_first_leading_one_ui.cpp.obj
[535/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_zero_ul.dir/stdc_first_leading_zero_ul.cpp.obj
[536/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_ull.dir/stdc_first_leading_one_ull.cpp.obj
[537/2522] Building CXX object libc/src/stdbit/CMakeFiles/libc.src.stdbit.stdc_first_leading_one_uc.dir/stdc_first_leading_one_uc.cpp.obj

mahesh-attarde pushed a commit to mahesh-attarde/llvm-project that referenced this pull request Jul 28, 2025
@arsenm @mahesh-attarde
AMDGPU: Fix not folding splat immediate into VGPR MFMA src2 (llvm#150628
a36482f 
)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@shiltian shiltian shiltian approved these changes

@jrbyrnes jrbyrnes Awaiting requested review from jrbyrnes

@rampitec rampitec Awaiting requested review from rampitec

@srpande srpande Awaiting requested review from srpande

Assignees

No one assigned

Labels

backend:AMDGPU

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@arsenm @llvmbot @llvm-ci @shiltian