diff options
| author | Cameron Zwarich <zwarich@apple.com> | 2011-03-16 00:13:44 +0000 |
|---|---|---|
| committer | Cameron Zwarich <zwarich@apple.com> | 2011-03-16 00:13:44 +0000 |
| commit | 85b0f468cf8390fca3ec356cd498ce0039dbad4f (patch) | |
| tree | 798870eaa63fe0b259845b06b3bad494451a73ab /test | |
| parent | 0a0f6f342b9acf8b90618cd3f4e61584370837a3 (diff) | |
| download | external_llvm-85b0f468cf8390fca3ec356cd498ce0039dbad4f.zip external_llvm-85b0f468cf8390fca3ec356cd498ce0039dbad4f.tar.gz external_llvm-85b0f468cf8390fca3ec356cd498ce0039dbad4f.tar.bz2 | |
Only convert allocas to scalars if it is profitable. The profitability metric I
chose is having a non-memcpy/memset use and being larger than any native integer
type. Originally I chose having an access of a size smaller than the total size
of the alloca, but this caused some minor issues on the spirit benchmark where
SRoA runs again after some inlining.
This fixes <rdar://problem/8613163>.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127718 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test')
| -rw-r--r-- | test/Transforms/ScalarRepl/only-memcpy-uses.ll | 27 |
1 files changed, 27 insertions, 0 deletions
diff --git a/test/Transforms/ScalarRepl/only-memcpy-uses.ll b/test/Transforms/ScalarRepl/only-memcpy-uses.ll new file mode 100644 index 0000000..cfb88bd --- /dev/null +++ b/test/Transforms/ScalarRepl/only-memcpy-uses.ll @@ -0,0 +1,27 @@ +; RUN: opt < %s -scalarrepl -S | FileCheck %s +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-apple-darwin10.0.0" + +%struct.S = type { [12 x i32] } + +; CHECK: @bar4 +define void @bar4(%struct.S* byval %s) nounwind ssp { +entry: +; CHECK: alloca +; CHECK-NOT: load +; CHECK: memcpy + %t = alloca %struct.S, align 4 + %agg.tmp = alloca %struct.S, align 4 + %tmp = bitcast %struct.S* %t to i8* + %tmp1 = bitcast %struct.S* %s to i8* + call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* %tmp1, i64 48, i32 4, i1 false) + %tmp2 = bitcast %struct.S* %agg.tmp to i8* + %tmp3 = bitcast %struct.S* %t to i8* + call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp2, i8* %tmp3, i64 48, i32 4, i1 false) + %call = call i32 (...)* @bazz(%struct.S* byval %agg.tmp) + ret void +} + +declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind + +declare i32 @bazz(...) |