diff options
author | Evan Cheng <evan.cheng@apple.com> | 2009-05-28 00:35:15 +0000 |
---|---|---|
committer | Evan Cheng <evan.cheng@apple.com> | 2009-05-28 00:35:15 +0000 |
commit | 8b944d39b356135676459152385f05c496951f6c (patch) | |
tree | 6a1c962c410e68929b12e2753749314181570188 /test/CodeGen/X86/narrow_op-2.ll | |
parent | c2695eba5700c785c2ae144eede7e1a932e3f5f3 (diff) | |
download | external_llvm-8b944d39b356135676459152385f05c496951f6c.zip external_llvm-8b944d39b356135676459152385f05c496951f6c.tar.gz external_llvm-8b944d39b356135676459152385f05c496951f6c.tar.bz2 |
Added optimization that narrow load / op / store and the 'op' is a bit twiddling instruction and its second operand is an immediate. If bits that are touched by 'op' can be done with a narrower instruction, reduce the width of the load and store as well. This happens a lot with bitfield manipulation code.
e.g.
orl $65536, 8(%rax)
=>
orb $1, 10(%rax)
Since narrowing is not always a win, e.g. i32 -> i16 is a loss on x86, dag combiner consults with the target before performing the optimization.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72507 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/narrow_op-2.ll')
-rw-r--r-- | test/CodeGen/X86/narrow_op-2.ll | 23 |
1 files changed, 23 insertions, 0 deletions
diff --git a/test/CodeGen/X86/narrow_op-2.ll b/test/CodeGen/X86/narrow_op-2.ll new file mode 100644 index 0000000..b441794 --- /dev/null +++ b/test/CodeGen/X86/narrow_op-2.ll @@ -0,0 +1,23 @@ +; RUN: llvm-as < %s | llc -march=x86-64 | grep andb | count 2 +; RUN: llvm-as < %s | llc -march=x86-64 | grep andb | grep 254 +; RUN: llvm-as < %s | llc -march=x86-64 | grep andb | grep 253 + + %struct.bf = type { i64, i16, i16, i32 } +@bfi = external global %struct.bf* + +define void @t1() nounwind ssp { +entry: + %0 = load %struct.bf** @bfi, align 8 + %1 = getelementptr %struct.bf* %0, i64 0, i32 1 + %2 = bitcast i16* %1 to i32* + %3 = load i32* %2, align 1 + %4 = and i32 %3, -65537 + store i32 %4, i32* %2, align 1 + %5 = load %struct.bf** @bfi, align 8 + %6 = getelementptr %struct.bf* %5, i64 0, i32 1 + %7 = bitcast i16* %6 to i32* + %8 = load i32* %7, align 1 + %9 = and i32 %8, -131073 + store i32 %9, i32* %7, align 1 + ret void +} |