[llvm] r230141 - Calling memmove on a MachineOperand is totally safe.
benny.kra at gmail.com
Sun Feb 22 04:11:42 PST 2015
> On 22.02.2015, at 01:58, Pete Cooper <peter_cooper at apple.com> wrote:
> Are you sure this is safe? I thought register operands had use lists which memmove isn't going to know to update?
Yes, that's what the other code path in this function does. The path I changed is only used when the MachineInstr containing the operand is not part of a MachineFunction and thus has no use lists.
> Sent from my iPhone
>> On Feb 21, 2015, at 8:22 AM, Benjamin Kramer <benny.kra at googlemail.com> wrote:
>> Author: d0k
>> Date: Sat Feb 21 10:22:48 2015
>> New Revision: 230141
>> URL: http://llvm.org/viewvc/llvm-project?rev=230141&view=rev
>> Calling memmove on a MachineOperand is totally safe.
>> While it's not POD due to the user-defined constructor, it's still a trivially
>> copyable type. No functional change.
>> Modified: llvm/trunk/lib/CodeGen/MachineInstr.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineInstr.cpp?rev=230141&r1=230140&r2=230141&view=diff
>> --- llvm/trunk/lib/CodeGen/MachineInstr.cpp (original)
>> +++ llvm/trunk/lib/CodeGen/MachineInstr.cpp Sat Feb 21 10:22:48 2015
>> @@ -674,14 +674,8 @@ static void moveOperands(MachineOperand
>> if (MRI)
>> return MRI->moveOperands(Dst, Src, NumOps);
>> - // Here it would be convenient to call memmove, so that isn't allowed because
>> - // MachineOperand has a constructor and so isn't a POD type.
>> - if (Dst < Src)
>> - for (unsigned i = 0; i != NumOps; ++i)
>> - new (Dst + i) MachineOperand(Src[i]);
>> - else
>> - for (unsigned i = NumOps; i ; --i)
>> - new (Dst + i - 1) MachineOperand(Src[i - 1]);
>> + // MachineOperand is a trivially copyable type so we can just use memmove.
>> + std::memmove(Dst, Src, NumOps * sizeof(MachineOperand));
>> /// addOperand - Add the specified operand to the instruction. If it is an
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
More information about the llvm-commits