[LLVMbugs] [Bug 4435] New: missed optimization, not folding away two equivalent BBs

bugzilla-daemon at cs.uiuc.edu bugzilla-daemon at cs.uiuc.edu
Tue Jun 23 13:21:20 PDT 2009


           Summary: missed optimization, not folding away two equivalent BBs
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu, jyasskin at google.com

The attached hand-written .ll file contains four BBs in an if-statement
diamond. The %cond_true and %cond_false blocks are both very similar; all
instructions are the same except for the parameter of the GEP which is the same
type in either case. The optimization I'd like to see is that one change being
folded away into a phi node and then the blocks getting merged into the %done

This is very similar to an optimization simplify-cfg does, and is also an
example of partial redundancy elimination and instruction sinking. Where should
this optimization belong?

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