[X86] Add a test case for idempotent atomic operations with speculative load hardening. Fix an additional issue found by the test.

This test covers the fix from r360475 as well.

llvm-svn: 360511

2 files changed, 24 insertions, 1 deletions

diff --git a/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp b/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
index 02f07d88afc..7b043378819 100644
--- a/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
+++ b/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
@@ -1719,9 +1719,11 @@ void X86SpeculativeLoadHardeningPass::tracePredStateThroughBlocksAndHarden(
 
         // If we have at least one (non-frame-index, non-RIP) register operand,
         // and neither operand is load-dependent, we need to check the load.
+        // Also handle explicit references to RSP as used by idempotent atomic
+        // or with 0.
         unsigned BaseReg = 0, IndexReg = 0;
         if (!BaseMO.isFI() && BaseMO.getReg() != X86::RIP &&
-            BaseMO.getReg() != X86::NoRegister)
+            BaseMO.getReg() != X86::RSP && BaseMO.getReg() != X86::NoRegister)
           BaseReg = BaseMO.getReg();
         if (IndexMO.getReg() != X86::NoRegister)
           IndexReg = IndexMO.getReg();
diff --git a/llvm/test/CodeGen/X86/speculative-load-hardening.ll b/llvm/test/CodeGen/X86/speculative-load-hardening.ll
index 54cde2c124e..5599b88a791 100644
--- a/llvm/test/CodeGen/X86/speculative-load-hardening.ll
+++ b/llvm/test/CodeGen/X86/speculative-load-hardening.ll
@@ -1142,3 +1142,24 @@ entry:
   call void @sink(i32 %e7)
   ret void
 }
+
+; Make sure we don't crash on idempotent atomic operations which have a
+; hardcoded reference to RSP+offset.
+define void @idempotent_atomic(i32* %x) speculative_load_hardening {
+; X64-LABEL: idempotent_atomic:
+; X64:       # %bb.0:
+; X64-NEXT:    movq %rsp, %rax
+; X64-NEXT:    movq $-1, %rcx
+; X64-NEXT:    sarq $63, %rax
+; X64-NEXT:    lock orl $0, (%rsp)
+; X64-NEXT:    shlq $47, %rax
+; X64-NEXT:    orq %rax, %rsp
+; X64-NEXT:    retq
+;
+; X64-LFENCE-LABEL: idempotent_atomic:
+; X64-LFENCE:       # %bb.0:
+; X64-LFENCE-NEXT:    lock orl $0, (%rsp)
+; X64-LFENCE-NEXT:    retq
+  %tmp = atomicrmw or i32* %x, i32 0 seq_cst
+  ret void
+}