[llvm-branch-commits] [llvm] release/20.x: [X86][SSE] Don't emit SSE2 load instructions in SSE1-only mode (#134547) (PR #135191)

[via llvm-branch-commits]

https://github.com/llvmbot updated https://github.com/llvm/llvm-project/pull/135191

>From 91647ae0dffed1dedcd9ef9214e1ea849b391bbe Mon Sep 17 00:00:00 2001
From: Stefan Schmidt <thrimbor at users.noreply.github.com>
Date: Wed, 9 Apr 2025 11:19:26 +0200
Subject: [PATCH] [X86][SSE] Don't emit SSE2 load instructions in SSE1-only
 mode (#134547)

This fixes a regression I traced back to
https://github.com/llvm/llvm-project/commit/8b43c1be23119c1024bed0a8ce392bc73727e2e2
/ https://github.com/llvm/llvm-project/pull/79000

The regression caused an SSE2 instruction, `movsd`, to be emitted as a
replacement for an SSE instruction, `movaps` despite the target
potentially not supporting this instruction, such as when building with
clang using `-march=pentium3`.

Fixes #134607

(cherry picked from commit 08e080ee98832c2aec6f379b04f486bea18730cc)
---
 .../Target/X86/X86FixupVectorConstants.cpp    | 11 ++++++----
 llvm/test/CodeGen/X86/pr134607.ll             | 20 +++++++++++++++++++
 2 files changed, 27 insertions(+), 4 deletions(-)
 create mode 100644 llvm/test/CodeGen/X86/pr134607.ll

diff --git a/llvm/lib/Target/X86/X86FixupVectorConstants.cpp b/llvm/lib/Target/X86/X86FixupVectorConstants.cpp
index 453898e132ca4..9dc392d6e9626 100644
--- a/llvm/lib/Target/X86/X86FixupVectorConstants.cpp
+++ b/llvm/lib/Target/X86/X86FixupVectorConstants.cpp
@@ -333,6 +333,7 @@ bool X86FixupVectorConstantsPass::processInstruction(MachineFunction &MF,
                                                      MachineInstr &MI) {
   unsigned Opc = MI.getOpcode();
   MachineConstantPool *CP = MI.getParent()->getParent()->getConstantPool();
+  bool HasSSE2 = ST->hasSSE2();
   bool HasSSE41 = ST->hasSSE41();
   bool HasAVX2 = ST->hasAVX2();
   bool HasDQI = ST->hasDQI();
@@ -394,11 +395,13 @@ bool X86FixupVectorConstantsPass::processInstruction(MachineFunction &MF,
   case X86::MOVAPDrm:
   case X86::MOVAPSrm:
   case X86::MOVUPDrm:
-  case X86::MOVUPSrm:
+  case X86::MOVUPSrm: {
     // TODO: SSE3 MOVDDUP Handling
-    return FixupConstant({{X86::MOVSSrm, 1, 32, rebuildZeroUpperCst},
-                          {X86::MOVSDrm, 1, 64, rebuildZeroUpperCst}},
-                         128, 1);
+    FixupEntry Fixups[] = {
+        {X86::MOVSSrm, 1, 32, rebuildZeroUpperCst},
+        {HasSSE2 ? X86::MOVSDrm : 0, 1, 64, rebuildZeroUpperCst}};
+    return FixupConstant(Fixups, 128, 1);
+  }
   case X86::VMOVAPDrm:
   case X86::VMOVAPSrm:
   case X86::VMOVUPDrm:
diff --git a/llvm/test/CodeGen/X86/pr134607.ll b/llvm/test/CodeGen/X86/pr134607.ll
new file mode 100644
index 0000000000000..5e824c22e5a22
--- /dev/null
+++ b/llvm/test/CodeGen/X86/pr134607.ll
@@ -0,0 +1,20 @@
+; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=+sse -O3 | FileCheck %s --check-prefixes=X86
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=-sse2,+sse -O3 | FileCheck %s --check-prefixes=X64-SSE1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2,+sse -O3 | FileCheck %s --check-prefixes=X64-SSE2
+
+define void @store_v2f32_constant(ptr %v) {
+; X86-LABEL: store_v2f32_constant:
+; X86:       # %bb.0:
+; X86-NEXT:    movl 4(%esp), %eax
+; X86-NEXT:    movaps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
+
+; X64-SSE1-LABEL: store_v2f32_constant:
+; X64-SSE1:       # %bb.0:
+; X64-SSE1-NEXT:    movaps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+
+; X64-SSE2-LABEL: store_v2f32_constant:
+; X64-SSE2:       # %bb.0:
+; X64-SSE2-NEXT:    movsd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+  store <2 x float> <float 2.560000e+02, float 5.120000e+02>, ptr %v, align 4
+  ret void
+}