Skip to content
This repository was archived by the owner on Feb 5, 2019. It is now read-only.

Commit 707f677

Browse files
pcwaltonpftbest
authored andcommitted
Add support for i1 compare operations to X86 FastISel, and ignore llvm.assume()
intrinsics in the target-independent FastISel.
1 parent 4daef48 commit 707f677

File tree

2 files changed

+3
-0
lines changed

2 files changed

+3
-0
lines changed

lib/CodeGen/SelectionDAG/FastISel.cpp

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1102,6 +1102,8 @@ bool FastISel::selectIntrinsicCall(const IntrinsicInst *II) {
11021102
// At -O0 we don't care about the lifetime intrinsics.
11031103
case Intrinsic::lifetime_start:
11041104
case Intrinsic::lifetime_end:
1105+
// At -O0 we don't care about assumptions.
1106+
case Intrinsic::assume:
11051107
// The donothing intrinsic does, well, nothing.
11061108
case Intrinsic::donothing:
11071109
return true;

lib/Target/X86/X86FastISel.cpp

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1330,6 +1330,7 @@ static unsigned X86ChooseCmpOpcode(EVT VT, const X86Subtarget *Subtarget) {
13301330

13311331
switch (VT.getSimpleVT().SimpleTy) {
13321332
default: return 0;
1333+
case MVT::i1: return X86::CMP8rr;
13331334
case MVT::i8: return X86::CMP8rr;
13341335
case MVT::i16: return X86::CMP16rr;
13351336
case MVT::i32: return X86::CMP32rr;

0 commit comments

Comments
 (0)