Add LoopSimplifyCFG pass

Loop transformations can sometimes fail because the loop, while in
valid rotated LCSSA form, is not in a canonical CFG form. This is
an extremely simple pass that just merges obviously redundant
blocks, which can be used to fix some known failure cases. In the
future, it may be enhanced with more cases (and have code shared with
SimplifyCFG).

This allows us to run LoopSimplifyCFG -> LoopRotate -> LoopUnroll,
so that SimplifyCFG cleans up the loop before Rotate tries to run.

Not currently used in the pass manager, since this pass doesn't do
anything unless you can hook it up in an LPM with other loop passes.
It'll be added once Chandler cleans up things to allow this.

Tested in a custom pipeline out of tree to confirm it works in
practice (in addition to the included trivial test).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@259256 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 34 insertions, 0 deletions

diff --git a/test/Transforms/LoopSimplifyCFG/merge-header.ll b/test/Transforms/LoopSimplifyCFG/merge-header.ll
new file mode 100644
index 00000000000..2e032ef22eb
--- /dev/null
+++ b/test/Transforms/LoopSimplifyCFG/merge-header.ll
@@ -0,0 +1,34 @@
+; RUN: opt -S -loop-simplifycfg < %s | FileCheck %s
+
+; CHECK-LABEL: foo
+; CHECK:      entry:
+; CHECK-NEXT:   br label %[[LOOP:[a-z]+]]
+; CHECK:      [[LOOP]]:
+; CHECK-NEXT:   phi
+; CHECK-NOT:    br label
+; CHECK:        br i1
+define i32 @foo(i32* %P, i64* %Q) {
+entry:
+  br label %outer
+
+outer:                                            ; preds = %outer.latch2, %entry
+  %y.2 = phi i32 [ 0, %entry ], [ %y.inc2, %outer.latch2 ]
+  br label %inner
+
+inner:                                            ; preds = %outer
+  store i32 0, i32* %P
+  store i32 1, i32* %P
+  store i32 2, i32* %P
+  %y.inc2 = add nsw i32 %y.2, 1
+  %exitcond.outer = icmp eq i32 %y.inc2, 3
+  store i32 %y.2, i32* %P
+  br i1 %exitcond.outer, label %exit, label %outer.latch2
+
+outer.latch2:                                     ; preds = %inner
+  %t = sext i32 %y.inc2 to i64
+  store i64 %t, i64* %Q
+  br label %outer
+
+exit:                                             ; preds = %inner
+  ret i32 0
+}