diff options
author | Dan Gohman <gohman@apple.com> | 2009-06-04 22:49:04 +0000 |
---|---|---|
committer | Dan Gohman <gohman@apple.com> | 2009-06-04 22:49:04 +0000 |
commit | ae3a0be92e33bc716722aa600983fc1535acb122 (patch) | |
tree | 768333097a76cc105813c7c636daf6259e6a0fc7 /test/CodeGen/PowerPC/return-val-i128.ll | |
parent | d18e31ae17390d9c6f6cf93d18badf962452031d (diff) | |
download | llvm-ae3a0be92e33bc716722aa600983fc1535acb122.tar.gz llvm-ae3a0be92e33bc716722aa600983fc1535acb122.tar.bz2 llvm-ae3a0be92e33bc716722aa600983fc1535acb122.tar.xz |
Split the Add, Sub, and Mul instruction opcodes into separate
integer and floating-point opcodes, introducing
FAdd, FSub, and FMul.
For now, the AsmParser, BitcodeReader, and IRBuilder all preserve
backwards compatability, and the Core LLVM APIs preserve backwards
compatibility for IR producers. Most front-ends won't need to change
immediately.
This implements the first step of the plan outlined here:
http://nondot.org/sabre/LLVMNotes/IntegerOverflow.txt
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@72897 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/PowerPC/return-val-i128.ll')
-rw-r--r-- | test/CodeGen/PowerPC/return-val-i128.ll | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/test/CodeGen/PowerPC/return-val-i128.ll b/test/CodeGen/PowerPC/return-val-i128.ll index 6e68ee3c52..27a5004bd1 100644 --- a/test/CodeGen/PowerPC/return-val-i128.ll +++ b/test/CodeGen/PowerPC/return-val-i128.ll @@ -14,7 +14,7 @@ entry: br i1 %toBool, label %bb, label %bb8 bb: ; preds = %entry %tmp4 = load float* %a_addr, align 4 ; <float> [#uses=1] - %tmp5 = sub float -0.000000e+00, %tmp4 ; <float> [#uses=1] + %tmp5 = fsub float -0.000000e+00, %tmp4 ; <float> [#uses=1] %tmp6 = call i128 @__fixunssfDI( float %tmp5 ) nounwind ; <i128> [#uses=1] %tmp7 = sub i128 0, %tmp6 ; <i128> [#uses=1] store i128 %tmp7, i128* %tmp, align 16 |