[llvm-commits] [llvm] r133858 - /llvm/trunk/lib/Target/X86/X86ISelLowering.cpp

[Chad Rosier]

On Jun 24, 2011, at 8:44 PM, Eli Friedman wrote:

> On Fri, Jun 24, 2011 at 7:04 PM, Chad Rosier <mcrosier at apple.com> wrote:
>> Author: mcrosier
>> Date: Fri Jun 24 21:04:56 2011
>> New Revision: 133858
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=133858&view=rev
>> Log:
>> Enable tail call optimization in the presence of a byval (x86-32 and x86-64).
>> <rdar://problem/9483883>
>> 
>> Modified:
>>    llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>> 
>> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=133858&r1=133857&r2=133858&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Fri Jun 24 21:04:56 2011
>> @@ -2505,6 +2505,10 @@
>>     if (!FINode)
>>       return false;
>>     FI = FINode->getIndex();
>> +  } else if (Arg.getOpcode() == ISD::FrameIndex && Flags.isByVal()) {
>> +    FrameIndexSDNode *FINode = dyn_cast<FrameIndexSDNode>(Arg);
>> +    FI = FINode->getIndex();
>> +    Bytes = Flags.getByValSize();
>>   } else
>>     return false;
> 
> An interesting side-note: byval+tail calls was already partially
> working, but before this commit it only worked outside the entry
> block. :)

Makes sense given the minimal amount of effort to get this going.

 Chad

> -Eli