diff options
| author | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2011-10-15 00:57:13 +0000 |
|---|---|---|
| committer | Jakob Stoklund Olesen <stoklund@2pi.dk> | 2011-10-15 00:57:13 +0000 |
| commit | 534849687c8cc0e4b538d64be71670e2a028a1be (patch) | |
| tree | f2d0adf03845531890eddc522b729f83a37addb0 /lib/Target/ARM/ARMInstrThumb.td | |
| parent | ba087df98228d4f1b7be6f93d659bb86ed2f0e59 (diff) | |
| download | llvm-534849687c8cc0e4b538d64be71670e2a028a1be.tar.gz llvm-534849687c8cc0e4b538d64be71670e2a028a1be.tar.bz2 llvm-534849687c8cc0e4b538d64be71670e2a028a1be.tar.xz | |
Mark tADDrSPi as having side effects again.
It really doesn't, but when r141929 removed the hasSideEffects flag from
this instruction, it caused miscompilations. I am guessing that it got
moved across a stack pointer update.
Also clear isRematerializable after checking that this instruction is
in fact never rematerialized in the nightly test suite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@142030 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Target/ARM/ARMInstrThumb.td')
| -rw-r--r-- | lib/Target/ARM/ARMInstrThumb.td | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/lib/Target/ARM/ARMInstrThumb.td b/lib/Target/ARM/ARMInstrThumb.td index 3e81516726..cedb54799d 100644 --- a/lib/Target/ARM/ARMInstrThumb.td +++ b/lib/Target/ARM/ARMInstrThumb.td @@ -310,9 +310,9 @@ def tPICADD : TIt<(outs GPR:$dst), (ins GPR:$lhs, pclabel:$cp), IIC_iALUr, "", } // ADD <Rd>, sp, #<imm8> -// This is rematerializable, which is particularly useful for taking the -// address of locals. -let isReMaterializable = 1, neverHasSideEffects = 1 in +// FIXME: This should not be marked as having side effects, and it should be +// rematerializable. Clearing the side effect bit causes miscompilations, +// probably because the instruction can be moved around. def tADDrSPi : T1pI<(outs tGPR:$dst), (ins GPRsp:$sp, t_imm0_1020s4:$imm), IIC_iALUi, "add", "\t$dst, $sp, $imm", []>, T1Encoding<{1,0,1,0,1,?}> { |