Allow ARM if-converter to be run after post allocation scheduling.

- This fixed a number of bugs in if-converter, tail merging, and post-allocation
  scheduler. If-converter now runs branch folding / tail merging first to
  maximize if-conversion opportunities.
- Also changed the t2IT instruction slightly. It now defines the ITSTATE
  register which is read by instructions in the IT block.
- Added Thumb2 specific hazard recognizer to ensure the scheduler doesn't
  change the instruction ordering in the IT block (since IT mask has been
  finalized). It also ensures no other instructions can be scheduled between
  instructions in the IT block.

This is not yet enabled.

llvm-svn: 106344

1 files changed, 10 insertions, 5 deletions

diff --git a/llvm/test/CodeGen/ARM/ifcvt2.ll b/llvm/test/CodeGen/ARM/ifcvt2.ll
index d9cac8022b2..7b9d0cf32cf 100644
--- a/llvm/test/CodeGen/ARM/ifcvt2.ll
+++ b/llvm/test/CodeGen/ARM/ifcvt2.ll
@@ -1,10 +1,8 @@
-; RUN: llc < %s -march=arm > %t
-; RUN: grep bxlt %t | count 1
-; RUN: grep bxgt %t | count 1
-; RUN: not grep bxge %t
-; RUN: not grep bxle %t
+; RUN: llc < %s -march=arm | FileCheck %s
 
 define i32 @t1(i32 %a, i32 %b, i32 %c, i32 %d) {
+; CHECK: t1:
+; CHECK: bxlt lr
 	%tmp2 = icmp sgt i32 %c, 10
 	%tmp5 = icmp slt i32 %d, 4
 	%tmp8 = or i1 %tmp5, %tmp2
@@ -21,6 +19,13 @@ UnifiedReturnBlock:
 }
 
 define i32 @t2(i32 %a, i32 %b, i32 %c, i32 %d) {
+; CHECK: t2:
+; CHECK: bxgt lr
+; CHECK: cmp
+; CHECK: addge
+; CHECK: subge
+; CHECK-NOT: bxge lr
+; CHECK: bx lr
 	%tmp2 = icmp sgt i32 %c, 10
 	%tmp5 = icmp slt i32 %d, 4
 	%tmp8 = and i1 %tmp5, %tmp2