Displaying 2 results from an estimated 2 matches for "evermadechange".
2011 Jul 29
1
[LLVMdev] alignment checking in isSafeToEliminateVarargsCast
...s a problem if the target is mips-o32, since a function call uses
different registers or different stack locations to pass arguments depending
on the alignment of the type of the argument that is passed.
Breakpoint 2 at
llvm/lib/Transforms/InstCombine/InstructionCombining.cpp:1667
1667 return EverMadeChange;
(gdb) p F.dump()
define i32 @main() nounwind {
entry:
%dw = alloca i64, align 8
%tmpcast = bitcast i64* %dw to %struct.DWordS_struct*
%ls = alloca %struct.LargeS_struct, align 8
store i64 416611827730, i64* %dw, align 8
%i = getelementptr inbounds %struct.LargeS_struct* %ls, i32 0, i32 0...
2018 Jun 29
2
Cleaning up ‘br i1 false’ cases in CodeGenPrepare
...&F) {
SunkAddrs.clear();
if (!DisableBranchOpts) {
+ MadeChange = true;
+ while (MadeChange) {
MadeChange = false;
SmallPtrSet<BasicBlock*, 8> WorkList;
for (BasicBlock &BB : F) {
@@ -352,6 +354,7 @@ bool CodeGenPrepare::runOnFunction(Function &F) {
EverMadeChange |= MadeChange;
}
+ }
if (!DisableGCOpts) {
SmallVector<Instruction *, 2> Statepoints;
Testing the patch, I got a regression with
llvm/test/CodeGen/AMDGPU/nested-loop-conditions.ll. I am unsure if
this requires remastering the test case to adjust with the new results
or if this is a...