Fix a trip-count overflow issue in LoopUnroll.

Currently LoopUnroll generates a prologue loop before the main loop
body to execute first N%UnrollFactor iterations. Also, this loop is
used if trip-count can overflow - it's determined by a runtime check.

However, we've been mistakenly optimizing this loop to a linear code for
UnrollFactor = 2, not taking into account that it also serves as a safe
version of the loop if its trip-count overflows.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@222451 91177308-0d34-0410-b5e6-96231b3b80d8

test/Transforms/LoopUnroll/tripcount-overflow.ll

@@ -0,0 +1,30 @@
+; RUN: opt < %s -S -unroll-runtime -unroll-count=2 -loop-unroll | FileCheck %s
+target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
+
+; When prologue is fully unrolled, the branch on its end is unconditional.
+; Unrolling it is illegal if we can't prove that trip-count+1 doesn't overflow,
+; like in this example, where it comes from an argument.
+;
+; This test is based on an example from here:
+; http://stackoverflow.com/questions/23838661/why-is-clang-optimizing-this-code-out
+;
+; CHECK: while.body.prol:
+; CHECK: br i1
+; CHECK: entry.split:
+
+; Function Attrs: nounwind readnone ssp uwtable
+define i32 @foo(i32 %N) #0 {
+entry:
+  br label %while.body
+
+while.body:                                       ; preds = %while.body, %entry
+  %i = phi i32 [ 0, %entry ], [ %inc, %while.body ]
+  %cmp = icmp eq i32 %i, %N
+  %inc = add i32 %i, 1
+  br i1 %cmp, label %while.end, label %while.body
+
+while.end:                                        ; preds = %while.body
+  ret i32 %i
+}
+
+attributes #0 = { nounwind readnone ssp uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }