[llvm-commits] [llvm] r43910 - in /llvm/trunk: include/llvm/CodeGen/Passes.h lib/CodeGen/BreakCriticalMachineEdges.cpp

Owen Anderson resistor at mac.com
Thu Nov 8 19:06:45 PST 2007


I don't really feel strongly about it one way or another.  I'm just  
used to having passes, so I made it a pass.  I do plan to modify it to  
update dominator info.

--Owen

On Nov 8, 2007, at 5:56 PM, Devang Patel wrote:

> If BreakCriticalMachineEdges is going to require analysis info (e.g.
> dom info) then it may be a good idea to use BCME as a pass, if
> possible. Otherwise, BCME user won't know that about required analysis
> and it may not let pass manager know about those requirements.
>
> We have seen such failures in opt when BreakCriticalEdges() is used as
> a funciton. When it works, it works fine.  However when it breaks, it
> breaks silently.
>
> On Nov 8, 2007, at 3:39 PM, Evan Cheng wrote:
>
>> I know. But why not make it a MachineFunction method and just call  
>> it?
>> I don't see it needing to be an independent pass.
>>
>> Evan
>>
>> On Nov 8, 2007, at 2:28 PM, Owen Anderson wrote:
>>
>>> StrongPHIElimination will require it in the near future.
>>>
>>> --Owen
>>>
>>> On Nov 8, 2007, at 4:24 PM, Evan Cheng wrote:
>>>> Does it need to be a pass? Seems to me it should just be a
>>>> function, no?
>>>>
>>>> Evan
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
> -
> Devang
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2555 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20071108/001ebb9e/attachment.bin>


More information about the llvm-commits mailing list