[Mono-bugs] [Bug 503077] New: The "branch" optimization doesn't remove the opcodes compare+branch atomically (at least sometimes).

[bugzilla_noreply at novell.com]

http://bugzilla.novell.com/show_bug.cgi?id=503077


           Summary: The "branch" optimization doesn't remove the opcodes
                    compare+branch atomically (at least sometimes).
    Classification: Mono
           Product: Mono: Runtime
           Version: 2.4.x
          Platform: All
        OS/Version: Linux
            Status: NEW
          Severity: Minor
          Priority: P5 - None
         Component: JIT
        AssignedTo: lupus at novell.com
        ReportedBy: cedric.vincent at st.com
         QAContact: mono-bugs at lists.ximian.com
          Found By: ---


User-Agent:       Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.0.10)
Gecko/2009042315 Firefox/3.0.10

Because the SH4 CPU uses a very reduced instruction set, the Mono/SH4
back-end has to converts the following code sequence:

     compare
     branch_if_greater

to:

     compare_greater
     branch_if_true

However the "branch" optimization does not remove the two opcodes
atomically (at least sometimes), so the SH4 lowering-pass is actually
confused.

Reproducible: Always

Steps to Reproduce:
Here comes a sample to reproduce this behavior with both Mono/x86 and
Mono/SH4 (attachement). You will notice it happens when JIT compiling
the method "System.IO.Path:GetDirectoryName":

    AFTER METHOD-TO-IR 18: [IN:  BB16(0), OUT:  BB17(0) BB19(0) ]
     iconst R101 <- [702263270]
     iconst R102 <- [47]
     iconst R103 <- [92]
     icompare R102 R103
     int_bne_un

    ...

    HANDLE-GLOBAL-VREGS BLOCK 18:
     iconst R101 <- [702263270]
     iconst R102 <- [47]
     iconst R103 <- [92]
     icompare_imm R102
     nop

The virtual register R102 does not seem to be used elsewhere. Maybe
the SH4 port of Mono should replace the opcode "icompare_imm" with a
"nop" for this case.

-- 
Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.