[llvm] r197414 - Allow MachineCSE to coalesce trivial subregister copies the same way

Hal Finkel hfinkel at anl.gov
Mon Dec 16 13:42:52 PST 2013


----- Original Message -----
> From: "Rafael EspĂ­ndola" <rafael.espindola at gmail.com>
> To: "Andrew Trick" <atrick at apple.com>
> Cc: "llvm-commits" <llvm-commits at cs.uiuc.edu>
> Sent: Monday, December 16, 2013 3:29:42 PM
> Subject: Re: [llvm] r197414 - Allow MachineCSE to coalesce trivial subregister	copies the same way
> 
> Hi Andrew,
> 
> This broke the bootstrap on ppc64, so I reverted it.  A reduced
> testcase was added in r197426.

For the record, it was this assert:

clang: lib/CodeGen/TwoAddressInstructionPass.cpp:1398: void {anonymous}::TwoAddressInstructionPass::processTiedPairs(llvm::MachineInstr*, {anonymous}::TwoAddressInstructionPass::TiedPairList&, unsigned int&): Assertion `TRI->getMatchingSuperRegClass(MRI->getRegClass(RegB), MRI->getRegClass(RegA), SubRegB) && "tied subregister must be a truncation"' failed.

Andy, please let me know if there is something in the backend itself that needs to be fixed.

 -Hal

> 
> Cheers,
> Rafael
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory




More information about the llvm-commits mailing list