[AMDGPU] Introduce AMDGPU::SGPR_SPILL asm comment flag (#67091)

yashssh · web-flow · commit 7ac532efc8ae · 2023-09-29T11:15:01.000+05:30
Use this flag to give more context to implicit def comments in assembly. Reviewed on phabricator: https://reviews.llvm.org/D153754
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp b/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
@@ -250,6 +250,21 @@ void AMDGPUAsmPrinter::emitFunctionBodyEnd() {
   Streamer.popSection();
 }
 
+void AMDGPUAsmPrinter::emitImplicitDef(const MachineInstr *MI) const {
+  Register RegNo = MI->getOperand(0).getReg();
+
+  SmallString<128> Str;
+  raw_svector_ostream OS(Str);
+  OS << "implicit-def: "
+     << printReg(RegNo, MF->getSubtarget().getRegisterInfo());
+
+  if (MI->getAsmPrinterFlags() & AMDGPU::SGPR_SPILL)
+    OS << " : SGPR spill to VGPR lane";
+
+  OutStreamer->AddComment(OS.str());
+  OutStreamer->addBlankLine();
+}
+
 void AMDGPUAsmPrinter::emitFunctionEntryLabel() {
   if (TM.getTargetTriple().getOS() == Triple::AMDHSA) {
     AsmPrinter::emitFunctionEntryLabel();
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.h b/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.h
@@ -116,6 +116,8 @@ class AMDGPUAsmPrinter final : public AsmPrinter {
 
   void emitFunctionBodyEnd() override;
 
+  void emitImplicitDef(const MachineInstr *MI) const override;
+
   void emitFunctionEntryLabel() override;
 
   void emitBasicBlockStart(const MachineBasicBlock &MBB) override;
diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.h b/llvm/lib/Target/AMDGPU/SIInstrInfo.h
@@ -1397,6 +1397,13 @@ namespace AMDGPU {
 
 } // end namespace AMDGPU
 
+namespace AMDGPU {
+enum AsmComments {
+  // For sgpr to vgpr spill instructions
+  SGPR_SPILL = MachineInstr::TAsmComments
+};
+} // namespace AMDGPU
+
 namespace SI {
 namespace KernelInputOffsets {
 
diff --git a/llvm/lib/Target/AMDGPU/SILowerSGPRSpills.cpp b/llvm/lib/Target/AMDGPU/SILowerSGPRSpills.cpp
@@ -281,6 +281,8 @@ void SILowerSGPRSpills::extendWWMVirtRegLiveness(MachineFunction &MF,
       auto MIB = BuildMI(*SaveBlock, *InsertBefore, InsertBefore->getDebugLoc(),
                          TII->get(AMDGPU::IMPLICIT_DEF), Reg);
       MFI->setFlag(Reg, AMDGPU::VirtRegFlag::WWM_REG);
+      // Set SGPR_SPILL asm printer flag
+      MIB->setAsmPrinterFlag(AMDGPU::SGPR_SPILL);
       if (LIS) {
         LIS->InsertMachineInstrInMaps(*MIB);
       }
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/image-waterfall-loop-O0.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/image-waterfall-loop-O0.ll
@@ -15,7 +15,7 @@ define <4 x float> @waterfall_loop(<8 x i32> %vgpr_srd) {
 ; CHECK-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:80 ; 4-byte Folded Spill
 ; CHECK-NEXT:    buffer_store_dword v2, off, s[0:3], s32 offset:84 ; 4-byte Folded Spill
 ; CHECK-NEXT:    s_mov_b32 exec_lo, s4
-; CHECK-NEXT:    ; implicit-def: $vgpr8
+; CHECK-NEXT:    ; implicit-def: $vgpr8 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_mov_b32_e32 v8, v0
 ; CHECK-NEXT:    s_or_saveexec_b32 s21, -1
 ; CHECK-NEXT:    buffer_load_dword v0, off, s[0:3], s32 ; 4-byte Folded Reload
diff --git a/llvm/test/CodeGen/AMDGPU/cf-loop-on-constant.ll b/llvm/test/CodeGen/AMDGPU/cf-loop-on-constant.ll
@@ -36,7 +36,7 @@ define amdgpu_kernel void @test_loop(ptr addrspace(3) %ptr, i32 %n) nounwind {
 ; GCN_DBG-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN_DBG-NEXT:    s_add_u32 s12, s12, s11
 ; GCN_DBG-NEXT:    s_addc_u32 s13, s13, 0
-; GCN_DBG-NEXT:    ; implicit-def: $vgpr0
+; GCN_DBG-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN_DBG-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN_DBG-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN_DBG-NEXT:    v_writelane_b32 v0, s0, 0
@@ -144,7 +144,7 @@ define amdgpu_kernel void @loop_const_true(ptr addrspace(3) %ptr, i32 %n) nounwi
 ; GCN_DBG-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN_DBG-NEXT:    s_add_u32 s12, s12, s11
 ; GCN_DBG-NEXT:    s_addc_u32 s13, s13, 0
-; GCN_DBG-NEXT:    ; implicit-def: $vgpr0
+; GCN_DBG-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN_DBG-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN_DBG-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN_DBG-NEXT:    v_writelane_b32 v0, s0, 0
@@ -232,7 +232,7 @@ define amdgpu_kernel void @loop_const_false(ptr addrspace(3) %ptr, i32 %n) nounw
 ; GCN_DBG-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN_DBG-NEXT:    s_add_u32 s12, s12, s11
 ; GCN_DBG-NEXT:    s_addc_u32 s13, s13, 0
-; GCN_DBG-NEXT:    ; implicit-def: $vgpr0
+; GCN_DBG-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN_DBG-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN_DBG-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN_DBG-NEXT:    v_writelane_b32 v0, s0, 0
@@ -321,7 +321,7 @@ define amdgpu_kernel void @loop_const_undef(ptr addrspace(3) %ptr, i32 %n) nounw
 ; GCN_DBG-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN_DBG-NEXT:    s_add_u32 s12, s12, s11
 ; GCN_DBG-NEXT:    s_addc_u32 s13, s13, 0
-; GCN_DBG-NEXT:    ; implicit-def: $vgpr0
+; GCN_DBG-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN_DBG-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN_DBG-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN_DBG-NEXT:    v_writelane_b32 v0, s0, 0
@@ -422,7 +422,7 @@ define amdgpu_kernel void @loop_arg_0(ptr addrspace(3) %ptr, i32 %n) nounwind {
 ; GCN_DBG-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN_DBG-NEXT:    s_add_u32 s12, s12, s11
 ; GCN_DBG-NEXT:    s_addc_u32 s13, s13, 0
-; GCN_DBG-NEXT:    ; implicit-def: $vgpr0
+; GCN_DBG-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN_DBG-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN_DBG-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN_DBG-NEXT:    v_writelane_b32 v0, s0, 0
diff --git a/llvm/test/CodeGen/AMDGPU/collapse-endcf.ll b/llvm/test/CodeGen/AMDGPU/collapse-endcf.ll
@@ -48,7 +48,7 @@ define amdgpu_kernel void @simple_nested_if(ptr addrspace(1) nocapture %arg) {
 ; GCN-O0-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN-O0-NEXT:    s_add_u32 s12, s12, s11
 ; GCN-O0-NEXT:    s_addc_u32 s13, s13, 0
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[8:9], -1
 ; GCN-O0-NEXT:    buffer_load_dword v0, off, s[12:15], 0 offset:4 ; 4-byte Folded Reload
@@ -221,7 +221,7 @@ define amdgpu_kernel void @uncollapsable_nested_if(ptr addrspace(1) nocapture %a
 ; GCN-O0-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN-O0-NEXT:    s_add_u32 s12, s12, s11
 ; GCN-O0-NEXT:    s_addc_u32 s13, s13, 0
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[8:9], -1
 ; GCN-O0-NEXT:    buffer_load_dword v0, off, s[12:15], 0 offset:4 ; 4-byte Folded Reload
@@ -430,7 +430,7 @@ define amdgpu_kernel void @nested_if_if_else(ptr addrspace(1) nocapture %arg) {
 ; GCN-O0-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN-O0-NEXT:    s_add_u32 s12, s12, s11
 ; GCN-O0-NEXT:    s_addc_u32 s13, s13, 0
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[6:7], -1
 ; GCN-O0-NEXT:    buffer_load_dword v0, off, s[12:15], 0 offset:4 ; 4-byte Folded Reload
@@ -676,7 +676,7 @@ define amdgpu_kernel void @nested_if_else_if(ptr addrspace(1) nocapture %arg) {
 ; GCN-O0-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN-O0-NEXT:    s_add_u32 s12, s12, s11
 ; GCN-O0-NEXT:    s_addc_u32 s13, s13, 0
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[8:9], -1
 ; GCN-O0-NEXT:    buffer_load_dword v0, off, s[12:15], 0 offset:4 ; 4-byte Folded Reload
@@ -931,7 +931,7 @@ define amdgpu_kernel void @s_endpgm_unsafe_barrier(ptr addrspace(1) nocapture %a
 ; GCN-O0-NEXT:    s_mov_b32 s15, 0xe8f000
 ; GCN-O0-NEXT:    s_add_u32 s12, s12, s11
 ; GCN-O0-NEXT:    s_addc_u32 s13, s13, 0
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[6:7], -1
 ; GCN-O0-NEXT:    buffer_load_dword v0, off, s[12:15], 0 offset:4 ; 4-byte Folded Reload
@@ -1080,7 +1080,7 @@ define void @scc_liveness(i32 %arg) local_unnamed_addr #0 {
 ; GCN-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:72 ; 4-byte Folded Spill
 ; GCN-O0-NEXT:    buffer_store_dword v4, off, s[0:3], s32 offset:76 ; 4-byte Folded Spill
 ; GCN-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; GCN-O0-NEXT:    ; implicit-def: $vgpr1
+; GCN-O0-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-O0-NEXT:    v_mov_b32_e32 v1, v0
 ; GCN-O0-NEXT:    s_or_saveexec_b64 s[14:15], -1
 ; GCN-O0-NEXT:    s_waitcnt expcnt(1)
diff --git a/llvm/test/CodeGen/AMDGPU/flat-scratch-init.ll b/llvm/test/CodeGen/AMDGPU/flat-scratch-init.ll
@@ -117,7 +117,7 @@ define amdgpu_kernel void @test(ptr addrspace(1) %out, i32 %in) {
 ; FLAT_SCR_OPT-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s2
 ; FLAT_SCR_OPT-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s3
 ; FLAT_SCR_OPT-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x0
-; FLAT_SCR_OPT-NEXT:    ; implicit-def: $vgpr0
+; FLAT_SCR_OPT-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; FLAT_SCR_OPT-NEXT:    s_waitcnt lgkmcnt(0)
 ; FLAT_SCR_OPT-NEXT:    v_writelane_b32 v0, s2, 0
 ; FLAT_SCR_OPT-NEXT:    v_writelane_b32 v0, s3, 1
@@ -240,7 +240,7 @@ define amdgpu_kernel void @test(ptr addrspace(1) %out, i32 %in) {
 ; FLAT_SCR_ARCH-LABEL: test:
 ; FLAT_SCR_ARCH:       ; %bb.0:
 ; FLAT_SCR_ARCH-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x0
-; FLAT_SCR_ARCH-NEXT:    ; implicit-def: $vgpr0
+; FLAT_SCR_ARCH-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; FLAT_SCR_ARCH-NEXT:    s_waitcnt lgkmcnt(0)
 ; FLAT_SCR_ARCH-NEXT:    v_writelane_b32 v0, s2, 0
 ; FLAT_SCR_ARCH-NEXT:    v_writelane_b32 v0, s3, 1
diff --git a/llvm/test/CodeGen/AMDGPU/kernel-vgpr-spill-mubuf-with-voffset.ll b/llvm/test/CodeGen/AMDGPU/kernel-vgpr-spill-mubuf-with-voffset.ll
@@ -13,7 +13,7 @@ define amdgpu_kernel void @test_kernel(i32 %val) #0 {
 ; CHECK-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; CHECK-NEXT:    s_add_u32 s0, s0, s17
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
-; CHECK-NEXT:    ; implicit-def: $vgpr3
+; CHECK-NEXT:    ; implicit-def: $vgpr3 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_writelane_b32 v3, s16, 0
 ; CHECK-NEXT:    s_or_saveexec_b64 s[34:35], -1
 ; CHECK-NEXT:    s_add_i32 s12, s33, 0x100200
diff --git a/llvm/test/CodeGen/AMDGPU/mubuf-legalize-operands-non-ptr-intrinsics.ll b/llvm/test/CodeGen/AMDGPU/mubuf-legalize-operands-non-ptr-intrinsics.ll
@@ -144,7 +144,7 @@ define float @mubuf_vgpr(<4 x i32> %i, i32 %c) #0 {
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:28 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:32 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr5
+; W64-O0-NEXT:    ; implicit-def: $vgpr5 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v4, off, s[0:3], s32 offset:20 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    v_mov_b32_e32 v5, v3
 ; W64-O0-NEXT:    v_mov_b32_e32 v6, v2
@@ -497,7 +497,7 @@ define void @mubuf_vgpr_adjacent_in_block(<4 x i32> %i, <4 x i32> %j, i32 %c, pt
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:76 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:80 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr13
+; W64-O0-NEXT:    ; implicit-def: $vgpr13 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v11, off, s[0:3], s32 offset:56 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v9, off, s[0:3], s32 offset:52 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v8, off, s[0:3], s32 offset:64 ; 4-byte Folded Spill
@@ -1019,7 +1019,7 @@ define void @mubuf_vgpr_outside_entry(<4 x i32> %i, <4 x i32> %j, i32 %c, ptr ad
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:76 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:80 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr8
+; W64-O0-NEXT:    ; implicit-def: $vgpr8 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v31, off, s[0:3], s32 offset:56 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v11, off, s[0:3], s32 offset:52 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    v_mov_b32_e32 v8, v6
diff --git a/llvm/test/CodeGen/AMDGPU/mubuf-legalize-operands.ll b/llvm/test/CodeGen/AMDGPU/mubuf-legalize-operands.ll
@@ -143,7 +143,7 @@ define float @mubuf_vgpr(ptr addrspace(8) %i, i32 %c) #0 {
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:32 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:36 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr5
+; W64-O0-NEXT:    ; implicit-def: $vgpr5 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v4, off, s[0:3], s32 offset:24 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v3, off, s[0:3], s32 offset:20 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    v_mov_b32_e32 v6, v2
@@ -511,7 +511,7 @@ define void @mubuf_vgpr_adjacent_in_block(ptr addrspace(8) %i, ptr addrspace(8)
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:72 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:76 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr13
+; W64-O0-NEXT:    ; implicit-def: $vgpr13 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v11, off, s[0:3], s32 offset:56 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v9, off, s[0:3], s32 offset:52 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v8, off, s[0:3], s32 offset:60 ; 4-byte Folded Spill
@@ -1058,7 +1058,7 @@ define void @mubuf_vgpr_outside_entry(ptr addrspace(8) %i, ptr addrspace(8) %j,
 ; W64-O0-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:100 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v1, off, s[0:3], s32 offset:104 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    s_mov_b64 exec, s[4:5]
-; W64-O0-NEXT:    ; implicit-def: $vgpr8
+; W64-O0-NEXT:    ; implicit-def: $vgpr8 : SGPR spill to VGPR lane
 ; W64-O0-NEXT:    buffer_store_dword v31, off, s[0:3], s32 offset:64 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v11, off, s[0:3], s32 offset:48 ; 4-byte Folded Spill
 ; W64-O0-NEXT:    buffer_store_dword v6, off, s[0:3], s32 offset:60 ; 4-byte Folded Spill
diff --git a/llvm/test/CodeGen/AMDGPU/need-fp-from-vgpr-spills.ll b/llvm/test/CodeGen/AMDGPU/need-fp-from-vgpr-spills.ll
@@ -71,7 +71,7 @@ define amdgpu_kernel void @kernel_call() {
 ; CHECK-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; CHECK-NEXT:    s_add_u32 s0, s0, s17
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
-; CHECK-NEXT:    ; implicit-def: $vgpr3
+; CHECK-NEXT:    ; implicit-def: $vgpr3 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_writelane_b32 v3, s16, 0
 ; CHECK-NEXT:    s_or_saveexec_b64 s[24:25], -1
 ; CHECK-NEXT:    buffer_store_dword v3, off, s[0:3], 0 offset:4 ; 4-byte Folded Spill
@@ -139,7 +139,7 @@ define amdgpu_kernel void @kernel_tailcall() {
 ; CHECK-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; CHECK-NEXT:    s_add_u32 s0, s0, s17
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
-; CHECK-NEXT:    ; implicit-def: $vgpr3
+; CHECK-NEXT:    ; implicit-def: $vgpr3 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_writelane_b32 v3, s16, 0
 ; CHECK-NEXT:    s_or_saveexec_b64 s[24:25], -1
 ; CHECK-NEXT:    buffer_store_dword v3, off, s[0:3], 0 offset:4 ; 4-byte Folded Spill
@@ -260,7 +260,7 @@ define protected amdgpu_kernel void @kernel() {
 ; CHECK-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; CHECK-NEXT:    s_add_u32 s0, s0, s17
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
-; CHECK-NEXT:    ; implicit-def: $vgpr3
+; CHECK-NEXT:    ; implicit-def: $vgpr3 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_writelane_b32 v3, s16, 0
 ; CHECK-NEXT:    s_or_saveexec_b64 s[24:25], -1
 ; CHECK-NEXT:    buffer_store_dword v3, off, s[0:3], 0 offset:4 ; 4-byte Folded Spill
diff --git a/llvm/test/CodeGen/AMDGPU/partial-sgpr-to-vgpr-spills.ll b/llvm/test/CodeGen/AMDGPU/partial-sgpr-to-vgpr-spills.ll
@@ -17,9 +17,9 @@ define amdgpu_kernel void @spill_sgprs_to_multiple_vgprs(ptr addrspace(1) %out,
 ; GCN-NEXT:    s_mov_b32 s95, 0xe8f000
 ; GCN-NEXT:    s_add_u32 s92, s92, s11
 ; GCN-NEXT:    s_addc_u32 s93, s93, 0
-; GCN-NEXT:    ; implicit-def: $vgpr0
-; GCN-NEXT:    ; implicit-def: $vgpr1
-; GCN-NEXT:    ; implicit-def: $vgpr2
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr2 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_load_dword s0, s[4:5], 0xb
 ; GCN-NEXT:    ;;#ASMSTART
 ; GCN-NEXT:    ; def s[4:11]
@@ -488,8 +488,8 @@ define amdgpu_kernel void @split_sgpr_spill_2_vgprs(ptr addrspace(1) %out, i32 %
 ; GCN-NEXT:    s_mov_b32 s55, 0xe8f000
 ; GCN-NEXT:    s_add_u32 s52, s52, s11
 ; GCN-NEXT:    s_addc_u32 s53, s53, 0
-; GCN-NEXT:    ; implicit-def: $vgpr0
-; GCN-NEXT:    ; implicit-def: $vgpr1
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_load_dword s0, s[4:5], 0xb
 ; GCN-NEXT:    ;;#ASMSTART
 ; GCN-NEXT:    ; def s[4:19]
@@ -738,8 +738,8 @@ define amdgpu_kernel void @no_vgprs_last_sgpr_spill(ptr addrspace(1) %out, i32 %
 ; GCN-NEXT:    s_mov_b32 s55, 0xe8f000
 ; GCN-NEXT:    s_add_u32 s52, s52, s11
 ; GCN-NEXT:    s_addc_u32 s53, s53, 0
-; GCN-NEXT:    ; implicit-def: $vgpr0
-; GCN-NEXT:    ; implicit-def: $vgpr0
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_load_dword s0, s[4:5], 0xb
 ; GCN-NEXT:    ;;#ASMSTART
 ; GCN-NEXT:    ;;#ASMEND
@@ -989,8 +989,8 @@ define amdgpu_kernel void @no_vgprs_last_sgpr_spill_live_v0(i32 %in) #1 {
 ; GCN-NEXT:    s_mov_b32 s55, 0xe8f000
 ; GCN-NEXT:    s_add_u32 s52, s52, s11
 ; GCN-NEXT:    s_addc_u32 s53, s53, 0
-; GCN-NEXT:    ; implicit-def: $vgpr0
-; GCN-NEXT:    ; implicit-def: $vgpr0
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_load_dword s0, s[4:5], 0x9
 ; GCN-NEXT:    ;;#ASMSTART
 ; GCN-NEXT:    ;;#ASMEND
diff --git a/llvm/test/CodeGen/AMDGPU/scc-clobbered-sgpr-to-vmem-spill.ll b/llvm/test/CodeGen/AMDGPU/scc-clobbered-sgpr-to-vmem-spill.ll
@@ -8,7 +8,7 @@
 define amdgpu_kernel void @kernel0(ptr addrspace(1) %out, i32 %in) #1 {
 ; CHECK-LABEL: kernel0:
 ; CHECK:       ; %bb.0:
-; CHECK-NEXT:    ; implicit-def: $vgpr23
+; CHECK-NEXT:    ; implicit-def: $vgpr23 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    ;;#ASMSTART
 ; CHECK-NEXT:    ;;#ASMEND
 ; CHECK-NEXT:    ;;#ASMSTART
diff --git a/llvm/test/CodeGen/AMDGPU/sgpr-spill-no-vgprs.ll b/llvm/test/CodeGen/AMDGPU/sgpr-spill-no-vgprs.ll
@@ -9,8 +9,8 @@ define amdgpu_kernel void @partial_no_vgprs_last_sgpr_spill(ptr addrspace(1) %ou
 ; GCN:       ; %bb.0:
 ; GCN-NEXT:    s_add_u32 s0, s0, s15
 ; GCN-NEXT:    s_addc_u32 s1, s1, 0
-; GCN-NEXT:    ; implicit-def: $vgpr0
-; GCN-NEXT:    ; implicit-def: $vgpr0
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_load_dword s4, s[8:9], 0x2
 ; GCN-NEXT:    ;;#ASMSTART
 ; GCN-NEXT:    ;;#ASMEND
diff --git a/llvm/test/CodeGen/AMDGPU/sgpr-spill-update-only-slot-indexes.ll b/llvm/test/CodeGen/AMDGPU/sgpr-spill-update-only-slot-indexes.ll
@@ -13,7 +13,7 @@ define amdgpu_kernel void @kernel() {
 ; GCN-NEXT:    s_mov_b32 s37, SCRATCH_RSRC_DWORD1
 ; GCN-NEXT:    s_mov_b32 s38, -1
 ; GCN-NEXT:    s_mov_b32 s39, 0xe00000
-; GCN-NEXT:    ; implicit-def: $vgpr3
+; GCN-NEXT:    ; implicit-def: $vgpr3 : SGPR spill to VGPR lane
 ; GCN-NEXT:    s_add_u32 s36, s36, s11
 ; GCN-NEXT:    v_writelane_b32 v3, s4, 0
 ; GCN-NEXT:    s_movk_i32 s32, 0x400
diff --git a/llvm/test/CodeGen/AMDGPU/sgpr-spills-split-regalloc.ll b/llvm/test/CodeGen/AMDGPU/sgpr-spills-split-regalloc.ll
@@ -715,7 +715,7 @@ define void @spill_sgpr_with_sgpr_uses() #0 {
 ; GCN-NEXT:    buffer_store_dword v251, off, s[0:3], s32 offset:8 ; 4-byte Folded Spill
 ; GCN-NEXT:    buffer_store_dword v252, off, s[0:3], s32 offset:4 ; 4-byte Folded Spill
 ; GCN-NEXT:    buffer_store_dword v253, off, s[0:3], s32 ; 4-byte Folded Spill
-; GCN-NEXT:    ; implicit-def: $vgpr0
+; GCN-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; GCN-NEXT:    v_mov_b32_e32 v0, 0
 ; GCN-NEXT:    buffer_store_dword v0, off, s[0:3], s32 offset:440
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
diff --git a/llvm/test/CodeGen/AMDGPU/spill-vgpr-to-agpr-update-regscavenger.ll b/llvm/test/CodeGen/AMDGPU/spill-vgpr-to-agpr-update-regscavenger.ll
@@ -13,7 +13,7 @@ define void @test() {
 ; CHECK-NEXT:    s_xor_saveexec_b64 s[4:5], -1
 ; CHECK-NEXT:    buffer_store_dword v0, off, s[0:3], s32 ; 4-byte Folded Spill
 ; CHECK-NEXT:    s_mov_b64 exec, s[4:5]
-; CHECK-NEXT:    ; implicit-def: $vgpr0
+; CHECK-NEXT:    ; implicit-def: $vgpr0 : SGPR spill to VGPR lane
 ; CHECK-NEXT:  .LBB0_1: ; %bb.1
 ; CHECK-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; CHECK-NEXT:    s_cbranch_scc1 .LBB0_3
diff --git a/llvm/test/CodeGen/AMDGPU/swdev380865.ll b/llvm/test/CodeGen/AMDGPU/swdev380865.ll
@@ -16,7 +16,7 @@ define amdgpu_kernel void @_Z6kernelILi4000ELi1EEvPd(ptr addrspace(1) %x.coerce)
 ; CHECK:       ; %bb.0: ; %entry
 ; CHECK-NEXT:    s_mov_b64 s[0:1], 0
 ; CHECK-NEXT:    s_load_dword s2, s[0:1], 0x0
-; CHECK-NEXT:    ; implicit-def: $vgpr2
+; CHECK-NEXT:    ; implicit-def: $vgpr2 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    ; kill: killed $sgpr0_sgpr1
 ; CHECK-NEXT:    s_mov_b32 s7, 0x401c0000
 ; CHECK-NEXT:    s_mov_b32 s5, 0x40280000
diff --git a/llvm/test/CodeGen/AMDGPU/vgpr-spill-placement-issue61083.ll b/llvm/test/CodeGen/AMDGPU/vgpr-spill-placement-issue61083.ll
@@ -13,7 +13,7 @@ define amdgpu_kernel void @__omp_offloading_16_dd2df_main_l9()  {
 ; CHECK:       ; %bb.0: ; %bb
 ; CHECK-NEXT:    s_add_u32 s0, s0, s15
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
-; CHECK-NEXT:    ; implicit-def: $vgpr1
+; CHECK-NEXT:    ; implicit-def: $vgpr1 : SGPR spill to VGPR lane
 ; CHECK-NEXT:    v_mov_b32_e32 v2, v0
 ; CHECK-NEXT:    s_or_saveexec_b64 s[8:9], -1
 ; CHECK-NEXT:    buffer_load_dword v0, off, s[0:3], 0 offset:4 ; 4-byte Folded Reload
diff --git a/llvm/test/CodeGen/AMDGPU/vgpr_constant_to_sgpr.ll b/llvm/test/CodeGen/AMDGPU/vgpr_constant_to_sgpr.ll
diff --git a/llvm/test/CodeGen/AMDGPU/whole-wave-register-copy.ll b/llvm/test/CodeGen/AMDGPU/whole-wave-register-copy.ll
diff --git a/llvm/test/CodeGen/AMDGPU/whole-wave-register-spill.ll b/llvm/test/CodeGen/AMDGPU/whole-wave-register-spill.ll
diff --git a/llvm/test/CodeGen/AMDGPU/wwm-reserved-spill.ll b/llvm/test/CodeGen/AMDGPU/wwm-reserved-spill.ll

Original file line number	Diff line number	Diff line change
`@@ -281,6 +281,8 @@ void SILowerSGPRSpills::extendWWMVirtRegLiveness(MachineFunction &MF,`
`281`	`281`	`auto MIB = BuildMI(SaveBlock, InsertBefore, InsertBefore->getDebugLoc(),`
`282`	`282`	`TII->get(AMDGPU::IMPLICIT_DEF), Reg);`
`283`	`283`	`MFI->setFlag(Reg, AMDGPU::VirtRegFlag::WWM_REG);`
	`284`	`+ // Set SGPR_SPILL asm printer flag`
	`285`	`+ MIB->setAsmPrinterFlag(AMDGPU::SGPR_SPILL);`
`284`	`286`	`if (LIS) {`
`285`	`287`	`LIS->InsertMachineInstrInMaps(*MIB);`
`286`	`288`	`}`