[LLVMbugs] [Bug 13916] SEGV in SROA in llvm::Instruction::eraseFromParent

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Wed Sep 26 00:43:54 PDT 2012


Chandler Carruth <chandlerc at gmail.com> changed:

           What    |Removed                     |Added
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Chandler Carruth <chandlerc at gmail.com> 2012-09-26 02:43:54 CDT ---
(In reply to comment #6)
> The current defacto is that two pointers are allowed to be equal regardless 
> whether they come from one SSA value, or two different. 
> I agree that we either need to fix the spec or the implementation, but I don't
> have a strong preference.

Yea, I'm not even going to really deal with this. The pass is actually OK with
it currently, and treats memcpy and memmove the same. ;] This is an unrelated
but amusingly tickled bug due to other stuff and the way we handle the
intrinsics at a basic level.

I've fixed this I think in r164669. Let me know if there are specific issues w/
memcpy that show back up, but I suspect that patch gets the real problem out of
the way.

Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the llvm-bugs mailing list