Skip to content
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

AMDGPU: Use default shouldRewriteCopySrc #125535

Merged
merged 1 commit into from
Feb 7, 2025
Merged

AMDGPU: Use default shouldRewriteCopySrc #125535

merged 1 commit into from
Feb 7, 2025

Conversation

arsenm
Copy link
Contributor

@arsenm arsenm commented Feb 3, 2025

This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.

@arsenm arsenm mentioned this pull request Feb 3, 2025
Merged
@arsenm arsenm marked this pull request as ready for review February 3, 2025 17:15
@arsenm arsenm mentioned this pull request Feb 3, 2025
Merged
@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Feb 3, 2025 •
edited
Loading

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

@llvmbot
Copy link
Member

llvmbot commented Feb 3, 2025 •
edited
Loading

@llvm/pr-subscribers-backend-systemz
@llvm/pr-subscribers-llvm-globalisel
@llvm/pr-subscribers-backend-x86
@llvm/pr-subscribers-llvm-regalloc

@llvm/pr-subscribers-backend-powerpc

Author: Matt Arsenault (arsenm)

Changes

This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.

Patch is 76.32 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/125535.diff

9 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp (-24)
  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.h (-5)
  • (modified) llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll (+22-22)
  • (modified) llvm/test/CodeGen/AMDGPU/ctpop64.ll (+18-18)
  • (modified) llvm/test/CodeGen/AMDGPU/idot2.ll (+91-91)
  • (modified) llvm/test/CodeGen/AMDGPU/load-global-i32.ll (+42-43)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-fold-reg-sequence-subreg.mir (+4-4)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-regseq-removal.mir (+2-2)
  • (modified) llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll (+239-237)
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
index 6fc57dec6a8264..71c720ed09b5fb 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
@@ -3516,30 +3516,6 @@ bool SIRegisterInfo::opCanUseInlineConstant(unsigned OpType) const {
          OpType <= AMDGPU::OPERAND_SRC_LAST;
 }
 
-bool SIRegisterInfo::shouldRewriteCopySrc(
-  const TargetRegisterClass *DefRC,
-  unsigned DefSubReg,
-  const TargetRegisterClass *SrcRC,
-  unsigned SrcSubReg) const {
-  // We want to prefer the smallest register class possible, so we don't want to
-  // stop and rewrite on anything that looks like a subregister
-  // extract. Operations mostly don't care about the super register class, so we
-  // only want to stop on the most basic of copies between the same register
-  // class.
-  //
-  // e.g. if we have something like
-  // %0 = ...
-  // %1 = ...
-  // %2 = REG_SEQUENCE %0, sub0, %1, sub1, %2, sub2
-  // %3 = COPY %2, sub0
-  //
-  // We want to look through the COPY to find:
-  //  => %3 = COPY %0
-
-  // Plain copy.
-  return getCommonSubClass(DefRC, SrcRC) != nullptr;
-}
-
 bool SIRegisterInfo::opCanUseLiteralConstant(unsigned OpType) const {
   // TODO: 64-bit operands have extending behavior from 32-bit literal.
   return OpType >= AMDGPU::OPERAND_REG_IMM_FIRST &&
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
index 8e481e3ac23043..a434efb70d0525 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
@@ -275,11 +275,6 @@ class SIRegisterInfo final : public AMDGPUGenRegisterInfo {
                            const TargetRegisterClass *SubRC,
                            unsigned SubIdx) const;
 
-  bool shouldRewriteCopySrc(const TargetRegisterClass *DefRC,
-                            unsigned DefSubReg,
-                            const TargetRegisterClass *SrcRC,
-                            unsigned SrcSubReg) const override;
-
   /// \returns True if operands defined with this operand type can accept
   /// a literal constant (i.e. any 32-bit immediate).
   bool opCanUseLiteralConstant(unsigned OpType) const;
diff --git a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
index c6c0b9cf8f027f..cc2f775ff22bc5 100644
--- a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
+++ b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
@@ -163,33 +163,33 @@ define amdgpu_kernel void @test_copy_v4i8_x3(ptr addrspace(1) %out0, ptr addrspa
 define amdgpu_kernel void @test_copy_v4i8_x4(ptr addrspace(1) %out0, ptr addrspace(1) %out1, ptr addrspace(1) %out2, ptr addrspace(1) %out3, ptr addrspace(1) %in) nounwind {
 ; SI-LABEL: test_copy_v4i8_x4:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x11
-; SI-NEXT:    s_mov_b32 s3, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
-; SI-NEXT:    s_mov_b32 s11, s3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x11
+; SI-NEXT:    s_mov_b32 s11, 0xf000
+; SI-NEXT:    s_mov_b32 s2, 0
+; SI-NEXT:    s_mov_b32 s3, s11
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_load_dwordx8 s[4:11], s[4:5], 0x9
-; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    s_mov_b32 s14, s2
-; SI-NEXT:    s_mov_b32 s15, s3
-; SI-NEXT:    s_mov_b32 s18, s2
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
+; SI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x9
+; SI-NEXT:    s_mov_b32 s10, -1
+; SI-NEXT:    s_mov_b32 s14, s10
+; SI-NEXT:    s_mov_b32 s15, s11
+; SI-NEXT:    s_mov_b32 s18, s10
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_mov_b32 s0, s4
-; SI-NEXT:    s_mov_b32 s1, s5
-; SI-NEXT:    s_mov_b32 s19, s3
-; SI-NEXT:    s_mov_b32 s22, s2
-; SI-NEXT:    s_mov_b32 s23, s3
-; SI-NEXT:    s_mov_b32 s12, s6
-; SI-NEXT:    s_mov_b32 s13, s7
-; SI-NEXT:    s_mov_b32 s16, s8
-; SI-NEXT:    s_mov_b32 s17, s9
-; SI-NEXT:    s_mov_b32 s20, s10
-; SI-NEXT:    s_mov_b32 s21, s11
+; SI-NEXT:    s_mov_b32 s8, s0
+; SI-NEXT:    s_mov_b32 s9, s1
+; SI-NEXT:    s_mov_b32 s19, s11
+; SI-NEXT:    s_mov_b32 s22, s10
+; SI-NEXT:    s_mov_b32 s23, s11
+; SI-NEXT:    s_mov_b32 s12, s2
+; SI-NEXT:    s_mov_b32 s13, s3
+; SI-NEXT:    s_mov_b32 s16, s4
+; SI-NEXT:    s_mov_b32 s17, s5
+; SI-NEXT:    s_mov_b32 s20, s6
+; SI-NEXT:    s_mov_b32 s21, s7
 ; SI-NEXT:    s_waitcnt vmcnt(0)
-; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; SI-NEXT:    buffer_store_dword v0, off, s[8:11], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[12:15], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[16:19], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[20:23], 0
diff --git a/llvm/test/CodeGen/AMDGPU/ctpop64.ll b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
index 3504546801c93b..2258f6a7b5483a 100644
--- a/llvm/test/CodeGen/AMDGPU/ctpop64.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
@@ -334,58 +334,58 @@ define amdgpu_kernel void @v_ctpop_v4i64(ptr addrspace(1) noalias %out, ptr addr
 define amdgpu_kernel void @ctpop_i64_in_br(ptr addrspace(1) %out, ptr addrspace(1) %in, i64 %ctpop_arg, i32 %cond) {
 ; SI-LABEL: ctpop_i64_in_br:
 ; SI:       ; %bb.0: ; %entry
-; SI-NEXT:    s_load_dword s8, s[4:5], 0xf
+; SI-NEXT:    s_load_dword s6, s[4:5], 0xf
 ; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
-; SI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0xd
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_cmp_lg_u32 s8, 0
+; SI-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; SI-NEXT:  ; %bb.1: ; %else
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x2
+; SI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x2
 ; SI-NEXT:    s_mov_b64 s[2:3], 0
 ; SI-NEXT:    s_andn2_b64 vcc, exec, s[2:3]
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_mov_b64 vcc, vcc
 ; SI-NEXT:    s_cbranch_vccnz .LBB7_3
 ; SI-NEXT:  .LBB7_2: ; %if
-; SI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; SI-NEXT:    s_mov_b32 s5, 0
+; SI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; SI-NEXT:    s_mov_b32 s7, 0
 ; SI-NEXT:  .LBB7_3: ; %endif
-; SI-NEXT:    v_mov_b32_e32 v0, s4
+; SI-NEXT:    v_mov_b32_e32 v0, s6
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    v_mov_b32_e32 v1, s5
+; SI-NEXT:    v_mov_b32_e32 v1, s7
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ; SI-NEXT:  .LBB7_4:
-; SI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; SI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; SI-NEXT:    s_branch .LBB7_2
 ;
 ; VI-LABEL: ctpop_i64_in_br:
 ; VI:       ; %bb.0: ; %entry
-; VI-NEXT:    s_load_dword s8, s[4:5], 0x3c
+; VI-NEXT:    s_load_dword s6, s[4:5], 0x3c
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
-; VI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x34
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_cmp_lg_u32 s8, 0
+; VI-NEXT:    s_cmp_lg_u32 s6, 0
 ; VI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; VI-NEXT:  ; %bb.1: ; %else
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x8
+; VI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x8
 ; VI-NEXT:    s_cbranch_execnz .LBB7_3
 ; VI-NEXT:  .LBB7_2: ; %if
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; VI-NEXT:    s_mov_b32 s5, 0
+; VI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; VI-NEXT:    s_mov_b32 s7, 0
 ; VI-NEXT:  .LBB7_3: ; %endif
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v0, s4
+; VI-NEXT:    v_mov_b32_e32 v0, s6
 ; VI-NEXT:    s_mov_b32 s3, 0xf000
 ; VI-NEXT:    s_mov_b32 s2, -1
-; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_mov_b32_e32 v1, s7
 ; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ; VI-NEXT:  .LBB7_4:
-; VI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; VI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; VI-NEXT:    s_branch .LBB7_2
 entry:
   %tmp0 = icmp eq i32 %cond, 0
diff --git a/llvm/test/CodeGen/AMDGPU/idot2.ll b/llvm/test/CodeGen/AMDGPU/idot2.ll
index cd85c301e16d5b..b443e654350c5e 100644
--- a/llvm/test/CodeGen/AMDGPU/idot2.ll
+++ b/llvm/test/CodeGen/AMDGPU/idot2.ll
@@ -151,20 +151,20 @@ entry:
 define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MulMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -174,8 +174,8 @@ define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v0, v0, v2
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v3, v1, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s0, v0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MulMul:
@@ -1698,20 +1698,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -1719,10 +1719,10 @@ define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_add1:
@@ -1851,20 +1851,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -1872,10 +1872,10 @@ define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s4
+; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v1, v3, v1, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_add1:
@@ -2004,20 +2004,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -2025,10 +2025,10 @@ define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul1:
@@ -2163,20 +2163,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -2184,10 +2184,10 @@ define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, v4
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v3, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_mul1:
@@ -2322,31 +2322,31 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul2:
@@ -2479,20 +2479,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b...
[truncated]

@llvmbot
Copy link
Member

llvmbot commented Feb 3, 2025

@llvm/pr-subscribers-backend-amdgpu

Author: Matt Arsenault (arsenm)

Changes

This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.

Patch is 76.32 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/125535.diff

9 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp (-24)
  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.h (-5)
  • (modified) llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll (+22-22)
  • (modified) llvm/test/CodeGen/AMDGPU/ctpop64.ll (+18-18)
  • (modified) llvm/test/CodeGen/AMDGPU/idot2.ll (+91-91)
  • (modified) llvm/test/CodeGen/AMDGPU/load-global-i32.ll (+42-43)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-fold-reg-sequence-subreg.mir (+4-4)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-regseq-removal.mir (+2-2)
  • (modified) llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll (+239-237)
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
index 6fc57dec6a8264..71c720ed09b5fb 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
@@ -3516,30 +3516,6 @@ bool SIRegisterInfo::opCanUseInlineConstant(unsigned OpType) const {
          OpType <= AMDGPU::OPERAND_SRC_LAST;
 }
 
-bool SIRegisterInfo::shouldRewriteCopySrc(
-  const TargetRegisterClass *DefRC,
-  unsigned DefSubReg,
-  const TargetRegisterClass *SrcRC,
-  unsigned SrcSubReg) const {
-  // We want to prefer the smallest register class possible, so we don't want to
-  // stop and rewrite on anything that looks like a subregister
-  // extract. Operations mostly don't care about the super register class, so we
-  // only want to stop on the most basic of copies between the same register
-  // class.
-  //
-  // e.g. if we have something like
-  // %0 = ...
-  // %1 = ...
-  // %2 = REG_SEQUENCE %0, sub0, %1, sub1, %2, sub2
-  // %3 = COPY %2, sub0
-  //
-  // We want to look through the COPY to find:
-  //  => %3 = COPY %0
-
-  // Plain copy.
-  return getCommonSubClass(DefRC, SrcRC) != nullptr;
-}
-
 bool SIRegisterInfo::opCanUseLiteralConstant(unsigned OpType) const {
   // TODO: 64-bit operands have extending behavior from 32-bit literal.
   return OpType >= AMDGPU::OPERAND_REG_IMM_FIRST &&
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
index 8e481e3ac23043..a434efb70d0525 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
@@ -275,11 +275,6 @@ class SIRegisterInfo final : public AMDGPUGenRegisterInfo {
                            const TargetRegisterClass *SubRC,
                            unsigned SubIdx) const;
 
-  bool shouldRewriteCopySrc(const TargetRegisterClass *DefRC,
-                            unsigned DefSubReg,
-                            const TargetRegisterClass *SrcRC,
-                            unsigned SrcSubReg) const override;
-
   /// \returns True if operands defined with this operand type can accept
   /// a literal constant (i.e. any 32-bit immediate).
   bool opCanUseLiteralConstant(unsigned OpType) const;
diff --git a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
index c6c0b9cf8f027f..cc2f775ff22bc5 100644
--- a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
+++ b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
@@ -163,33 +163,33 @@ define amdgpu_kernel void @test_copy_v4i8_x3(ptr addrspace(1) %out0, ptr addrspa
 define amdgpu_kernel void @test_copy_v4i8_x4(ptr addrspace(1) %out0, ptr addrspace(1) %out1, ptr addrspace(1) %out2, ptr addrspace(1) %out3, ptr addrspace(1) %in) nounwind {
 ; SI-LABEL: test_copy_v4i8_x4:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x11
-; SI-NEXT:    s_mov_b32 s3, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
-; SI-NEXT:    s_mov_b32 s11, s3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x11
+; SI-NEXT:    s_mov_b32 s11, 0xf000
+; SI-NEXT:    s_mov_b32 s2, 0
+; SI-NEXT:    s_mov_b32 s3, s11
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_load_dwordx8 s[4:11], s[4:5], 0x9
-; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    s_mov_b32 s14, s2
-; SI-NEXT:    s_mov_b32 s15, s3
-; SI-NEXT:    s_mov_b32 s18, s2
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
+; SI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x9
+; SI-NEXT:    s_mov_b32 s10, -1
+; SI-NEXT:    s_mov_b32 s14, s10
+; SI-NEXT:    s_mov_b32 s15, s11
+; SI-NEXT:    s_mov_b32 s18, s10
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_mov_b32 s0, s4
-; SI-NEXT:    s_mov_b32 s1, s5
-; SI-NEXT:    s_mov_b32 s19, s3
-; SI-NEXT:    s_mov_b32 s22, s2
-; SI-NEXT:    s_mov_b32 s23, s3
-; SI-NEXT:    s_mov_b32 s12, s6
-; SI-NEXT:    s_mov_b32 s13, s7
-; SI-NEXT:    s_mov_b32 s16, s8
-; SI-NEXT:    s_mov_b32 s17, s9
-; SI-NEXT:    s_mov_b32 s20, s10
-; SI-NEXT:    s_mov_b32 s21, s11
+; SI-NEXT:    s_mov_b32 s8, s0
+; SI-NEXT:    s_mov_b32 s9, s1
+; SI-NEXT:    s_mov_b32 s19, s11
+; SI-NEXT:    s_mov_b32 s22, s10
+; SI-NEXT:    s_mov_b32 s23, s11
+; SI-NEXT:    s_mov_b32 s12, s2
+; SI-NEXT:    s_mov_b32 s13, s3
+; SI-NEXT:    s_mov_b32 s16, s4
+; SI-NEXT:    s_mov_b32 s17, s5
+; SI-NEXT:    s_mov_b32 s20, s6
+; SI-NEXT:    s_mov_b32 s21, s7
 ; SI-NEXT:    s_waitcnt vmcnt(0)
-; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; SI-NEXT:    buffer_store_dword v0, off, s[8:11], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[12:15], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[16:19], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[20:23], 0
diff --git a/llvm/test/CodeGen/AMDGPU/ctpop64.ll b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
index 3504546801c93b..2258f6a7b5483a 100644
--- a/llvm/test/CodeGen/AMDGPU/ctpop64.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
@@ -334,58 +334,58 @@ define amdgpu_kernel void @v_ctpop_v4i64(ptr addrspace(1) noalias %out, ptr addr
 define amdgpu_kernel void @ctpop_i64_in_br(ptr addrspace(1) %out, ptr addrspace(1) %in, i64 %ctpop_arg, i32 %cond) {
 ; SI-LABEL: ctpop_i64_in_br:
 ; SI:       ; %bb.0: ; %entry
-; SI-NEXT:    s_load_dword s8, s[4:5], 0xf
+; SI-NEXT:    s_load_dword s6, s[4:5], 0xf
 ; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
-; SI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0xd
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_cmp_lg_u32 s8, 0
+; SI-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; SI-NEXT:  ; %bb.1: ; %else
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x2
+; SI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x2
 ; SI-NEXT:    s_mov_b64 s[2:3], 0
 ; SI-NEXT:    s_andn2_b64 vcc, exec, s[2:3]
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_mov_b64 vcc, vcc
 ; SI-NEXT:    s_cbranch_vccnz .LBB7_3
 ; SI-NEXT:  .LBB7_2: ; %if
-; SI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; SI-NEXT:    s_mov_b32 s5, 0
+; SI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; SI-NEXT:    s_mov_b32 s7, 0
 ; SI-NEXT:  .LBB7_3: ; %endif
-; SI-NEXT:    v_mov_b32_e32 v0, s4
+; SI-NEXT:    v_mov_b32_e32 v0, s6
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    v_mov_b32_e32 v1, s5
+; SI-NEXT:    v_mov_b32_e32 v1, s7
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ; SI-NEXT:  .LBB7_4:
-; SI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; SI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; SI-NEXT:    s_branch .LBB7_2
 ;
 ; VI-LABEL: ctpop_i64_in_br:
 ; VI:       ; %bb.0: ; %entry
-; VI-NEXT:    s_load_dword s8, s[4:5], 0x3c
+; VI-NEXT:    s_load_dword s6, s[4:5], 0x3c
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
-; VI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x34
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_cmp_lg_u32 s8, 0
+; VI-NEXT:    s_cmp_lg_u32 s6, 0
 ; VI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; VI-NEXT:  ; %bb.1: ; %else
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x8
+; VI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x8
 ; VI-NEXT:    s_cbranch_execnz .LBB7_3
 ; VI-NEXT:  .LBB7_2: ; %if
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; VI-NEXT:    s_mov_b32 s5, 0
+; VI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; VI-NEXT:    s_mov_b32 s7, 0
 ; VI-NEXT:  .LBB7_3: ; %endif
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v0, s4
+; VI-NEXT:    v_mov_b32_e32 v0, s6
 ; VI-NEXT:    s_mov_b32 s3, 0xf000
 ; VI-NEXT:    s_mov_b32 s2, -1
-; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_mov_b32_e32 v1, s7
 ; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ; VI-NEXT:  .LBB7_4:
-; VI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; VI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; VI-NEXT:    s_branch .LBB7_2
 entry:
   %tmp0 = icmp eq i32 %cond, 0
diff --git a/llvm/test/CodeGen/AMDGPU/idot2.ll b/llvm/test/CodeGen/AMDGPU/idot2.ll
index cd85c301e16d5b..b443e654350c5e 100644
--- a/llvm/test/CodeGen/AMDGPU/idot2.ll
+++ b/llvm/test/CodeGen/AMDGPU/idot2.ll
@@ -151,20 +151,20 @@ entry:
 define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MulMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -174,8 +174,8 @@ define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v0, v0, v2
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v3, v1, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s0, v0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MulMul:
@@ -1698,20 +1698,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -1719,10 +1719,10 @@ define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_add1:
@@ -1851,20 +1851,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -1872,10 +1872,10 @@ define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s4
+; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v1, v3, v1, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_add1:
@@ -2004,20 +2004,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -2025,10 +2025,10 @@ define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul1:
@@ -2163,20 +2163,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -2184,10 +2184,10 @@ define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, v4
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v3, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_mul1:
@@ -2322,31 +2322,31 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul2:
@@ -2479,20 +2479,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b...
[truncated]

@llvmbot
Copy link
Member

llvmbot commented Feb 3, 2025

@llvm/pr-subscribers-backend-aarch64

Author: Matt Arsenault (arsenm)

Changes

This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.

Patch is 76.32 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/125535.diff

9 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp (-24)
  • (modified) llvm/lib/Target/AMDGPU/SIRegisterInfo.h (-5)
  • (modified) llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll (+22-22)
  • (modified) llvm/test/CodeGen/AMDGPU/ctpop64.ll (+18-18)
  • (modified) llvm/test/CodeGen/AMDGPU/idot2.ll (+91-91)
  • (modified) llvm/test/CodeGen/AMDGPU/load-global-i32.ll (+42-43)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-fold-reg-sequence-subreg.mir (+4-4)
  • (modified) llvm/test/CodeGen/AMDGPU/peephole-opt-regseq-removal.mir (+2-2)
  • (modified) llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll (+239-237)
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
index 6fc57dec6a8264..71c720ed09b5fb 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.cpp
@@ -3516,30 +3516,6 @@ bool SIRegisterInfo::opCanUseInlineConstant(unsigned OpType) const {
          OpType <= AMDGPU::OPERAND_SRC_LAST;
 }
 
-bool SIRegisterInfo::shouldRewriteCopySrc(
-  const TargetRegisterClass *DefRC,
-  unsigned DefSubReg,
-  const TargetRegisterClass *SrcRC,
-  unsigned SrcSubReg) const {
-  // We want to prefer the smallest register class possible, so we don't want to
-  // stop and rewrite on anything that looks like a subregister
-  // extract. Operations mostly don't care about the super register class, so we
-  // only want to stop on the most basic of copies between the same register
-  // class.
-  //
-  // e.g. if we have something like
-  // %0 = ...
-  // %1 = ...
-  // %2 = REG_SEQUENCE %0, sub0, %1, sub1, %2, sub2
-  // %3 = COPY %2, sub0
-  //
-  // We want to look through the COPY to find:
-  //  => %3 = COPY %0
-
-  // Plain copy.
-  return getCommonSubClass(DefRC, SrcRC) != nullptr;
-}
-
 bool SIRegisterInfo::opCanUseLiteralConstant(unsigned OpType) const {
   // TODO: 64-bit operands have extending behavior from 32-bit literal.
   return OpType >= AMDGPU::OPERAND_REG_IMM_FIRST &&
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
index 8e481e3ac23043..a434efb70d0525 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.h
@@ -275,11 +275,6 @@ class SIRegisterInfo final : public AMDGPUGenRegisterInfo {
                            const TargetRegisterClass *SubRC,
                            unsigned SubIdx) const;
 
-  bool shouldRewriteCopySrc(const TargetRegisterClass *DefRC,
-                            unsigned DefSubReg,
-                            const TargetRegisterClass *SrcRC,
-                            unsigned SrcSubReg) const override;
-
   /// \returns True if operands defined with this operand type can accept
   /// a literal constant (i.e. any 32-bit immediate).
   bool opCanUseLiteralConstant(unsigned OpType) const;
diff --git a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
index c6c0b9cf8f027f..cc2f775ff22bc5 100644
--- a/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
+++ b/llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll
@@ -163,33 +163,33 @@ define amdgpu_kernel void @test_copy_v4i8_x3(ptr addrspace(1) %out0, ptr addrspa
 define amdgpu_kernel void @test_copy_v4i8_x4(ptr addrspace(1) %out0, ptr addrspace(1) %out1, ptr addrspace(1) %out2, ptr addrspace(1) %out3, ptr addrspace(1) %in) nounwind {
 ; SI-LABEL: test_copy_v4i8_x4:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x11
-; SI-NEXT:    s_mov_b32 s3, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
-; SI-NEXT:    s_mov_b32 s11, s3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x11
+; SI-NEXT:    s_mov_b32 s11, 0xf000
+; SI-NEXT:    s_mov_b32 s2, 0
+; SI-NEXT:    s_mov_b32 s3, s11
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_load_dwordx8 s[4:11], s[4:5], 0x9
-; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    s_mov_b32 s14, s2
-; SI-NEXT:    s_mov_b32 s15, s3
-; SI-NEXT:    s_mov_b32 s18, s2
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
+; SI-NEXT:    s_load_dwordx8 s[0:7], s[4:5], 0x9
+; SI-NEXT:    s_mov_b32 s10, -1
+; SI-NEXT:    s_mov_b32 s14, s10
+; SI-NEXT:    s_mov_b32 s15, s11
+; SI-NEXT:    s_mov_b32 s18, s10
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_mov_b32 s0, s4
-; SI-NEXT:    s_mov_b32 s1, s5
-; SI-NEXT:    s_mov_b32 s19, s3
-; SI-NEXT:    s_mov_b32 s22, s2
-; SI-NEXT:    s_mov_b32 s23, s3
-; SI-NEXT:    s_mov_b32 s12, s6
-; SI-NEXT:    s_mov_b32 s13, s7
-; SI-NEXT:    s_mov_b32 s16, s8
-; SI-NEXT:    s_mov_b32 s17, s9
-; SI-NEXT:    s_mov_b32 s20, s10
-; SI-NEXT:    s_mov_b32 s21, s11
+; SI-NEXT:    s_mov_b32 s8, s0
+; SI-NEXT:    s_mov_b32 s9, s1
+; SI-NEXT:    s_mov_b32 s19, s11
+; SI-NEXT:    s_mov_b32 s22, s10
+; SI-NEXT:    s_mov_b32 s23, s11
+; SI-NEXT:    s_mov_b32 s12, s2
+; SI-NEXT:    s_mov_b32 s13, s3
+; SI-NEXT:    s_mov_b32 s16, s4
+; SI-NEXT:    s_mov_b32 s17, s5
+; SI-NEXT:    s_mov_b32 s20, s6
+; SI-NEXT:    s_mov_b32 s21, s7
 ; SI-NEXT:    s_waitcnt vmcnt(0)
-; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; SI-NEXT:    buffer_store_dword v0, off, s[8:11], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[12:15], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[16:19], 0
 ; SI-NEXT:    buffer_store_dword v0, off, s[20:23], 0
diff --git a/llvm/test/CodeGen/AMDGPU/ctpop64.ll b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
index 3504546801c93b..2258f6a7b5483a 100644
--- a/llvm/test/CodeGen/AMDGPU/ctpop64.ll
+++ b/llvm/test/CodeGen/AMDGPU/ctpop64.ll
@@ -334,58 +334,58 @@ define amdgpu_kernel void @v_ctpop_v4i64(ptr addrspace(1) noalias %out, ptr addr
 define amdgpu_kernel void @ctpop_i64_in_br(ptr addrspace(1) %out, ptr addrspace(1) %in, i64 %ctpop_arg, i32 %cond) {
 ; SI-LABEL: ctpop_i64_in_br:
 ; SI:       ; %bb.0: ; %entry
-; SI-NEXT:    s_load_dword s8, s[4:5], 0xf
+; SI-NEXT:    s_load_dword s6, s[4:5], 0xf
 ; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
-; SI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0xd
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    s_cmp_lg_u32 s8, 0
+; SI-NEXT:    s_cmp_lg_u32 s6, 0
 ; SI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; SI-NEXT:  ; %bb.1: ; %else
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x2
+; SI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x2
 ; SI-NEXT:    s_mov_b64 s[2:3], 0
 ; SI-NEXT:    s_andn2_b64 vcc, exec, s[2:3]
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_mov_b64 vcc, vcc
 ; SI-NEXT:    s_cbranch_vccnz .LBB7_3
 ; SI-NEXT:  .LBB7_2: ; %if
-; SI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; SI-NEXT:    s_mov_b32 s5, 0
+; SI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; SI-NEXT:    s_mov_b32 s7, 0
 ; SI-NEXT:  .LBB7_3: ; %endif
-; SI-NEXT:    v_mov_b32_e32 v0, s4
+; SI-NEXT:    v_mov_b32_e32 v0, s6
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    v_mov_b32_e32 v1, s5
+; SI-NEXT:    v_mov_b32_e32 v1, s7
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ; SI-NEXT:  .LBB7_4:
-; SI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; SI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; SI-NEXT:    s_branch .LBB7_2
 ;
 ; VI-LABEL: ctpop_i64_in_br:
 ; VI:       ; %bb.0: ; %entry
-; VI-NEXT:    s_load_dword s8, s[4:5], 0x3c
+; VI-NEXT:    s_load_dword s6, s[4:5], 0x3c
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
-; VI-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x34
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x34
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_cmp_lg_u32 s8, 0
+; VI-NEXT:    s_cmp_lg_u32 s6, 0
 ; VI-NEXT:    s_cbranch_scc0 .LBB7_4
 ; VI-NEXT:  ; %bb.1: ; %else
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[2:3], 0x8
+; VI-NEXT:    s_load_dwordx2 s[6:7], s[2:3], 0x8
 ; VI-NEXT:    s_cbranch_execnz .LBB7_3
 ; VI-NEXT:  .LBB7_2: ; %if
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_bcnt1_i32_b64 s4, s[6:7]
-; VI-NEXT:    s_mov_b32 s5, 0
+; VI-NEXT:    s_bcnt1_i32_b64 s6, s[4:5]
+; VI-NEXT:    s_mov_b32 s7, 0
 ; VI-NEXT:  .LBB7_3: ; %endif
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v0, s4
+; VI-NEXT:    v_mov_b32_e32 v0, s6
 ; VI-NEXT:    s_mov_b32 s3, 0xf000
 ; VI-NEXT:    s_mov_b32 s2, -1
-; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_mov_b32_e32 v1, s7
 ; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ; VI-NEXT:  .LBB7_4:
-; VI-NEXT:    ; implicit-def: $sgpr4_sgpr5
+; VI-NEXT:    ; implicit-def: $sgpr6_sgpr7
 ; VI-NEXT:    s_branch .LBB7_2
 entry:
   %tmp0 = icmp eq i32 %cond, 0
diff --git a/llvm/test/CodeGen/AMDGPU/idot2.ll b/llvm/test/CodeGen/AMDGPU/idot2.ll
index cd85c301e16d5b..b443e654350c5e 100644
--- a/llvm/test/CodeGen/AMDGPU/idot2.ll
+++ b/llvm/test/CodeGen/AMDGPU/idot2.ll
@@ -151,20 +151,20 @@ entry:
 define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MulMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -174,8 +174,8 @@ define amdgpu_kernel void @udot2_MulMul(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v0, v0, v2
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v3, v1, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s0, v0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MulMul:
@@ -1698,20 +1698,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -1719,10 +1719,10 @@ define amdgpu_kernel void @udot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_add1:
@@ -1851,20 +1851,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_add1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -1872,10 +1872,10 @@ define amdgpu_kernel void @idot2_MultipleUses_add1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s4
+; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v1, v3, v1, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_add1:
@@ -2004,20 +2004,20 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
@@ -2025,10 +2025,10 @@ define amdgpu_kernel void @udot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v0, v2, s0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul1:
@@ -2163,20 +2163,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul1:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_bfe_i32 v1, v2, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v2, 16, v2
@@ -2184,10 +2184,10 @@ define amdgpu_kernel void @idot2_MultipleUses_mul1(ptr addrspace(1) %src1,
 ; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 16
 ; GFX7-NEXT:    v_ashrrev_i32_e32 v0, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_i32_i24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v2, v4
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v3, v1, v0
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot2_MultipleUses_mul1:
@@ -2322,31 +2322,31 @@ entry:
 define amdgpu_kernel void @udot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: udot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s6, 0
-; GFX7-NEXT:    s_mov_b32 s7, s3
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s10, 0
+; GFX7-NEXT:    s_mov_b32 s11, s7
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[8:9]
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[0:1]
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, 0
-; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_mov_b64 s[4:5], s[10:11]
-; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
-; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_mov_b64 s[8:9], s[2:3]
+; GFX7-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
+; GFX7-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX7-NEXT:    s_mov_b32 s6, -1
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 16, v2
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s4
+; GFX7-NEXT:    v_mad_u32_u24 v4, v3, v1, s0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v1, v3, v1, v4
 ; GFX7-NEXT:    v_mad_u32_u24 v0, v0, v2, v1
-; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_MultipleUses_mul2:
@@ -2479,20 +2479,20 @@ entry:
 define amdgpu_kernel void @idot2_MultipleUses_mul2(ptr addrspace(1) %src1,
 ; GFX7-LABEL: idot2_MultipleUses_mul2:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[4:5], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
-; GFX7-NEXT:    s_mov_b...
[truncated]

@arsenm arsenm force-pushed the users/arsenm/amdgpu/use-default-shouldRewriteCopySrc branch from e5479af to e7b88d2 Compare February 4, 2025 06:03
@arsenm arsenm force-pushed the users/arsenm/peephole-opt/fix-looking-for-def-of-current-instruction branch from a897dcf to 84db705 Compare February 4, 2025 06:04
This was referenced Feb 4, 2025
Merged
Open
Open
@arsenm arsenm force-pushed the users/arsenm/peephole-opt/fix-looking-for-def-of-current-instruction branch from 84db705 to 0417e7f Compare February 5, 2025 16:25
Base automatically changed from users/arsenm/peephole-opt/fix-looking-for-def-of-current-instruction to main February 5, 2025 16:29
@arsenm
AMDGPU: Use default shouldRewriteCopySrc
bbb7726 
This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.
@arsenm arsenm force-pushed the users/arsenm/amdgpu/use-default-shouldRewriteCopySrc branch from e7b88d2 to bbb7726 Compare February 5, 2025 16:33
@arsenm Graphite App
Copy link
Contributor Author

arsenm commented Feb 7, 2025 •
edited
Loading

Merge activity

  • Feb 7, 12:29 AM EST: A user started a stack merge that includes this pull request via Graphite.
  • Feb 7, 12:31 AM EST: A user merged this pull request with Graphite.

@arsenm arsenm merged commit d21fc58 into main Feb 7, 2025
8 checks passed
@arsenm arsenm deleted the users/arsenm/amdgpu/use-default-shouldRewriteCopySrc branch February 7, 2025 05:31
Icohedron pushed a commit to Icohedron/llvm-project that referenced this pull request Feb 11, 2025
@arsenm @Icohedron
AMDGPU: Use default shouldRewriteCopySrc (llvm#125535)
a896000 
This was ultimately working around bugs in subregister handling
in peephole-opt. In the common case, it would give up on folding
anything into a subregister extract copy.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@qcolombet qcolombet qcolombet approved these changes

@jayfoad jayfoad Awaiting requested review from jayfoad

@kparzysz kparzysz Awaiting requested review from kparzysz

@RKSimon RKSimon Awaiting requested review from RKSimon

@sdesmalen-arm sdesmalen-arm Awaiting requested review from sdesmalen-arm

@wangpc-pp wangpc-pp Awaiting requested review from wangpc-pp

Assignees
No one assigned
Labels
backend:AArch64 backend:AMDGPU backend:ARM backend:PowerPC backend:SystemZ backend:X86 llvm:codegen llvm:globalisel llvm:regalloc
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@arsenm @llvmbot @qcolombet