diff options
| author | Dan Gohman <gohman@apple.com> | 2009-04-27 20:16:15 +0000 |
|---|---|---|
| committer | Dan Gohman <gohman@apple.com> | 2009-04-27 20:16:15 +0000 |
| commit | 76466373794f34303146931969c1f1905ad526fe (patch) | |
| tree | 2171320ee377c70e311d3d55bbaec5eca611cb25 /llvm/test/Transforms/LoopStrengthReduce | |
| parent | 61cac2b2954cf20760b609fc3efd1c976183cf9b (diff) | |
| download | bcm5719-llvm-76466373794f34303146931969c1f1905ad526fe.tar.gz bcm5719-llvm-76466373794f34303146931969c1f1905ad526fe.zip | |
Teach getZeroExtendExpr and getSignExtendExpr to use trip-count
information to simplify [sz]ext({a,+,b}) to {zext(a),+,[zs]ext(b)},
as appropriate.
These functions and the trip count code each call into the other, so
this requires careful handling to avoid infinite recursion. During
the initial trip count computation, conservative SCEVs are used,
which are subsequently discarded once the trip count is actually
known.
Among other benefits, this change lets LSR automatically eliminate
some unnecessary zext-inreg and sext-inreg operation where the
operand is an induction variable.
llvm-svn: 70241
Diffstat (limited to 'llvm/test/Transforms/LoopStrengthReduce')
| -rw-r--r-- | llvm/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll | 10 |
1 files changed, 6 insertions, 4 deletions
diff --git a/llvm/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll b/llvm/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll index 06e231212ed..cb2f3aa5166 100644 --- a/llvm/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll +++ b/llvm/test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-1.ll @@ -1,4 +1,4 @@ -; RUN: llvm-as < %s | llc --x86-asm-syntax=att | grep {cmpl \$8} +; RUN: llvm-as < %s | llc --x86-asm-syntax=att | grep {cmpq \$8} 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" target triple = "x86_64-apple-darwin9" @@ -6,7 +6,7 @@ target triple = "x86_64-apple-darwin9" ; happens after the relevant use, so the comparison stride can be ; easily changed. -define void @foo() { +define void @foo() nounwind { entry: br label %loop @@ -14,9 +14,11 @@ loop: %indvar = phi i32 [ 0, %entry ], [ %i.2.0.us1534, %loop ] ; <i32> [#uses=1] %i.2.0.us1534 = add i32 %indvar, 1 ; <i32> [#uses=3] %tmp628.us1540 = shl i32 %i.2.0.us1534, 1 ; <i32> [#uses=1] - %tmp645646647.us1547 = sext i32 %tmp628.us1540 to i64 ; <i64> [#uses=0] + %tmp645646647.us1547 = sext i32 %tmp628.us1540 to i64 ; <i64> [#uses=1] + store i64 %tmp645646647.us1547, i64* null %tmp611.us1535 = icmp eq i32 %i.2.0.us1534, 4 ; <i1> [#uses=2] - %tmp623.us1538 = select i1 %tmp611.us1535, i32 6, i32 0 ; <i32> [#uses=0] + %tmp623.us1538 = select i1 %tmp611.us1535, i32 6, i32 0 ; <i32> [#uses=1] + store i32 %tmp623.us1538, i32* null br i1 %tmp611.us1535, label %exit, label %loop exit: |
