[LLVMbugs] [Bug 9234] New: CodeGen/X86 regress tests error on MinGW

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=9234

           Summary: CodeGen/X86 regress tests error on MinGW
           Product: new-bugs
           Version: trunk
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dmalyshev at accesssoftek.com
                CC: llvmbugs at cs.uiuc.edu


Is anybody aware of the regress tests failures on MinGW?
The following tests fail:

    LLVM :: CodeGen/X86/add.ll
    LLVM :: CodeGen/X86/apm.ll
    LLVM :: CodeGen/X86/avoid-lea-scale2.ll
    LLVM :: CodeGen/X86/break-sse-dep.ll
    LLVM :: CodeGen/X86/byval.ll
    LLVM :: CodeGen/X86/byval2.ll
    LLVM :: CodeGen/X86/byval3.ll
    LLVM :: CodeGen/X86/byval4.ll
    LLVM :: CodeGen/X86/byval5.ll
    LLVM :: CodeGen/X86/coalescer-commute2.ll
    LLVM :: CodeGen/X86/codegen-prepare-extload.ll
    LLVM :: CodeGen/X86/constant-pool-remat-0.ll
    LLVM :: CodeGen/X86/constant-pool-sharing.ll
    LLVM :: CodeGen/X86/convert-2-addr-3-addr-inc64.ll
    LLVM :: CodeGen/X86/ctpop-combine.ll
    LLVM :: CodeGen/X86/fast-isel-cmp-branch.ll
    LLVM :: CodeGen/X86/fast-isel-gep.ll
    LLVM :: CodeGen/X86/fold-mul-lohi.ll
    LLVM :: CodeGen/X86/gather-addresses.ll
    LLVM :: CodeGen/X86/h-register-store.ll
    LLVM :: CodeGen/X86/h-registers-0.ll
    LLVM :: CodeGen/X86/i128-ret.ll
    LLVM :: CodeGen/X86/i64-mem-copy.ll
    LLVM :: CodeGen/X86/lea-3.ll
    LLVM :: CodeGen/X86/lea.ll
    LLVM :: CodeGen/X86/lsr-overflow.ll
    LLVM :: CodeGen/X86/lsr-reuse-trunc.ll
    LLVM :: CodeGen/X86/memcmp.ll
    LLVM :: CodeGen/X86/mmx-copy-gprs.ll
    LLVM :: CodeGen/X86/movgs.ll
    LLVM :: CodeGen/X86/optimize-max-3.ll
    LLVM :: CodeGen/X86/peep-vector-extract-concat.ll
    LLVM :: CodeGen/X86/pmulld.ll
    LLVM :: CodeGen/X86/pr9127.ll
    LLVM :: CodeGen/X86/red-zone.ll
    LLVM :: CodeGen/X86/remat-mov-0.ll
    LLVM :: CodeGen/X86/scalar-min-max-fill-operand.ll
    LLVM :: CodeGen/X86/sse-align-0.ll
    LLVM :: CodeGen/X86/sse-align-3.ll
    LLVM :: CodeGen/X86/sse-align-7.ll
    LLVM :: CodeGen/X86/sse-commute.ll
    LLVM :: CodeGen/X86/sse_reload_fold.ll
    LLVM :: CodeGen/X86/stdarg.ll
    LLVM :: CodeGen/X86/stride-nine-with-base-reg.ll
    LLVM :: CodeGen/X86/stride-reuse.ll
    LLVM :: CodeGen/X86/tailcallbyval64.ll
    LLVM :: CodeGen/X86/test-shrink.ll
    LLVM :: CodeGen/X86/use-add-flags.ll
    LLVM :: CodeGen/X86/v2f32.ll
    LLVM :: CodeGen/X86/vec_cast.ll
    LLVM :: CodeGen/X86/vec_set-8.ll
    LLVM :: CodeGen/X86/vec_shuffle-17.ll
    LLVM :: CodeGen/X86/vec_shuffle-37.ll
    LLVM :: CodeGen/X86/widen_load-0.ll
    LLVM :: CodeGen/X86/x86-64-malloc.ll
    LLVM :: CodeGen/X86/xor.ll

The reason of the failures is the same for all, the output from llc
-march=x86-64 is different then expected, because different registers get used.
In some of them tests the generated commands are different as well.

All these tests pass just file for -march=x86.

Is the problem with the code generation for MinGW or with the tests?

-- 
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.