diff options
| author | Bill Wendling <isanbard@gmail.com> | 2013-12-01 03:03:42 +0000 |
|---|---|---|
| committer | Bill Wendling <isanbard@gmail.com> | 2013-12-01 03:03:42 +0000 |
| commit | 3f297541c5440c4758b34214fdbbf9ae5414c0f1 (patch) | |
| tree | 4a305ba6efdb16b9515bb31d7a965e3e1381b3c5 /test/Transforms | |
| parent | c4795e39fa0de4e7d9cbf4c30a31e1d26e108ca0 (diff) | |
| download | external_llvm-3f297541c5440c4758b34214fdbbf9ae5414c0f1.zip external_llvm-3f297541c5440c4758b34214fdbbf9ae5414c0f1.tar.gz external_llvm-3f297541c5440c4758b34214fdbbf9ae5414c0f1.tar.bz2 | |
Merging r195787:
------------------------------------------------------------------------
r195787 | arnolds | 2013-11-26 14:11:23 -0800 (Tue, 26 Nov 2013) | 8 lines
LoopVectorizer: Truncate i64 trip counts of i32 phis if necessary
In signed arithmetic we could end up with an i64 trip count for an i32 phi.
Because it is signed arithmetic we know that this is only defined if the i32
does not wrap. It is therefore safe to truncate the i64 trip count to a i32
value.
Fixes PR18049.
------------------------------------------------------------------------
git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/release_34@195991 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/Transforms')
| -rw-r--r-- | test/Transforms/LoopVectorize/X86/tripcount.ll | 39 |
1 files changed, 39 insertions, 0 deletions
diff --git a/test/Transforms/LoopVectorize/X86/tripcount.ll b/test/Transforms/LoopVectorize/X86/tripcount.ll new file mode 100644 index 0000000..6b38bac --- /dev/null +++ b/test/Transforms/LoopVectorize/X86/tripcount.ll @@ -0,0 +1,39 @@ +; RUN: opt -S -loop-vectorize -force-vector-width=2 -force-vector-unroll=1 -mcpu=prescott < %s | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128" +target triple = "i386-unknown-freebsd11.0" + +@big = external global [0 x i32] + +; PR18049 +; We need to truncate the exit count to i32. This is legal because the +; arithmetic is signed (%inc is nsw). + +; CHECK-LABEL: tripcount +; CHECK: trunc i64 %count to i32 + +define void @tripcount(i64 %count) { +entry: + %cmp6 = icmp sgt i64 %count, 0 + br i1 %cmp6, label %for.body.preheader, label %for.end + +for.body.preheader: + br label %for.body + +for.body: + %i.07 = phi i32 [ %inc, %for.body ], [ 0, %for.body.preheader ] + %arrayidx = getelementptr inbounds [0 x i32]* @big, i32 0, i32 %i.07 + %0 = load i32* %arrayidx, align 4 + %neg = xor i32 %0, -1 + store i32 %neg, i32* %arrayidx, align 4 + %inc = add nsw i32 %i.07, 1 + %conv = sext i32 %inc to i64 + %cmp = icmp slt i64 %conv, %count + br i1 %cmp, label %for.body, label %for.end.loopexit + +for.end.loopexit: + br label %for.end + +for.end: + ret void +} |