diff --git a/llvm/test/CodeGen/X86/extractelement-load.ll b/llvm/test/CodeGen/X86/extractelement-load.ll
index 78ad6d95963b50..5e1ff78a31c109 100644
--- a/llvm/test/CodeGen/X86/extractelement-load.ll
+++ b/llvm/test/CodeGen/X86/extractelement-load.ll
@@ -338,6 +338,41 @@ define i32 @multi_use_load_scalarization(<4 x i32>* %p) nounwind {
   ret i32 %r
 }
 
+define i32 @multi_use_volatile_load_scalarization(<4 x i32>* %p) nounwind {
+; X32-SSE2-LABEL: multi_use_volatile_load_scalarization:
+; X32-SSE2:       # %bb.0:
+; X32-SSE2-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X32-SSE2-NEXT:    movl (%ecx), %eax
+; X32-SSE2-NEXT:    movdqu (%ecx), %xmm0
+; X32-SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; X32-SSE2-NEXT:    psubd %xmm1, %xmm0
+; X32-SSE2-NEXT:    movdqa %xmm0, (%ecx)
+; X32-SSE2-NEXT:    retl
+;
+; X64-SSSE3-LABEL: multi_use_volatile_load_scalarization:
+; X64-SSSE3:       # %bb.0:
+; X64-SSSE3-NEXT:    movl (%rdi), %eax
+; X64-SSSE3-NEXT:    movdqu (%rdi), %xmm0
+; X64-SSSE3-NEXT:    pcmpeqd %xmm1, %xmm1
+; X64-SSSE3-NEXT:    psubd %xmm1, %xmm0
+; X64-SSSE3-NEXT:    movdqa %xmm0, (%rdi)
+; X64-SSSE3-NEXT:    retq
+;
+; X64-AVX-LABEL: multi_use_volatile_load_scalarization:
+; X64-AVX:       # %bb.0:
+; X64-AVX-NEXT:    movl (%rdi), %eax
+; X64-AVX-NEXT:    vmovdqu (%rdi), %xmm0
+; X64-AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; X64-AVX-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
+; X64-AVX-NEXT:    vmovdqa %xmm0, (%rdi)
+; X64-AVX-NEXT:    retq
+  %v = load volatile <4 x i32>, <4 x i32>* %p, align 1
+  %v1 = add <4 x i32> %v, <i32 1, i32 1, i32 1, i32 1>
+  store <4 x i32> %v1, <4 x i32>* %p
+  %r = extractelement <4 x i32> %v, i64 0
+  ret i32 %r
+}
+
 ; This test is reduced from a C source example that showed a miscompile:
 ; https://github.com/llvm/llvm-project/issues/53695
 ; The scalarized loads from 'zero' in the AVX asm must occur before