Summary:
LLVM's MI level notion of invariant_load is different from LLVM's IR
level notion of invariant_load with respect to dereferenceability. The
IR notion of invariant_load only guarantees that all *non-faulting*
invariant loads result in the same value. The MI notion of invariant
load guarantees that the load can be legally moved to any location
within its containing function. The MI notion of invariant_load is
stronger than the IR notion of invariant_load -- an MI invariant_load is
an IR invariant_load + a guarantee that the location being loaded from
is dereferenceable throughout the function's lifetime.
Reviewers: hfinkel, reames
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D10075
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@238881
91177308-0d34-0410-b5e6-
96231b3b80d8
bool isVolatile = I.isVolatile();
bool isNonTemporal = I.getMetadata(LLVMContext::MD_nontemporal) != nullptr;
- bool isInvariant = I.getMetadata(LLVMContext::MD_invariant_load) != nullptr;
+
+ // The IR notion of invariant_load only guarantees that all *non-faulting*
+ // invariant loads result in the same value. The MI notion of invariant load
+ // guarantees that the load can be legally moved to any location within its
+ // containing function. The MI notion of invariant_load is stronger than the
+ // IR notion of invariant_load -- an MI invariant_load is an IR invariant_load
+ // with a guarantee that the location being loaded from is dereferenceable
+ // throughout the function's lifetime.
+
+ bool isInvariant = I.getMetadata(LLVMContext::MD_invariant_load) != nullptr &&
+ isDereferenceablePointer(SV, *DAG.getTarget().getDataLayout());
unsigned Alignment = I.getAlignment();
AAMDNodes AAInfo;
--- /dev/null
+; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s
+
+declare void @free_v()
+
+define void @f(i32* %x, i32 %c32, i32* %y) {
+; CHECK-LABEL: f
+ entry:
+ %v = load i32, i32* %x, !invariant.load !0
+; CHECK: movl (%rdi), %ebx
+; CHECK: free_v
+; CHECK-NOT: movl (%rdi), %ebx
+ call void @free_v()
+ %c = icmp ne i32 %c32, 0
+ br i1 %c, label %left, label %merge
+
+ left:
+ store i32 %v, i32* %y
+ br label %merge
+
+ merge:
+ ret void
+}
+
+!0 = !{}