[llvm-commits] [llvm] r88806 - /llvm/trunk/lib/CodeGen/MachineFunction.cpp

[Owen Anderson]

On Nov 14, 2009, at 2:27 PM, Jim Grosbach wrote:
> If we're sharing tables, when we're manipulating the destinations of  
> one jump table branch, we're also potentially changing the destination  
> blocks for other branches which use the same table. That'll cause the  
> CFG to be incorrect for those branches, and things go very badly from  
> there.

FWIW, this was a major problem with implementing StrongPHIElimination.  It assumes that all critical edges can be split, and being unable to hack on jump tables severely complicated that.

--Owen
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2620 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20091115/46209112/attachment.bin>