[LLVMbugs] [Bug 10072] Necessary 'indirectbr' instruction is dropped by optimizer [regression]

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Fri Jun 3 18:58:46 PDT 2011


Yuri <yuri at tsoft.com> changed:

           What    |Removed                     |Added
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |

--- Comment #2 from Yuri <yuri at tsoft.com> 2011-06-03 20:58:46 CDT ---
I don't think you are right.

My original testcase has the last label xblkx.end which doesn't do what all
other labels do.

Labels xblkx.begin to xblkx.begin9 (8 of them) all set 'yes' variable to 1
(from 0). But label xblkx.end is also in 'indirectbr' but doesn't do that.

Therefore 'indirectbr' or its equivalent should survive optimization. And it
doesn't. In fact, my really original testcase (that I simplified into this one)
produces wrong result with optimization and correct result without it, which
should never ever happen.

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