llvm test results for FreeBSD platform

In attached file.

Vladimir

test2004_06_20.log (86.4 KB)

Thanks Vladimir. That's great! Glad you got it working.

BTW, the failures you're seeing have been experienced by Chris and I as
well. Chris is diligently working on making the LLVM processing more
consistent so he track down the problem. A week ago or so, these tests
passed at 100%.

Reid.

Is it ok sending this results for FreeBSD5.1 at daily/weekly based to this
mail list?

Now results.
Big improvement in llvm tests results from last test result sended.

New regressions:
  Regression.Assembler.ConstantExprFold : FAIL , expected PASS
  Regression.CodeGen.Generic.2004-04-09-SameValueCoalescing: FAIL ,
expected PASS
  Regression.Transforms.PRE.basictest : FAIL , expected PASS
  Regression.Transforms.TailCallElim.tail_call_with_branch: FAIL ,
expected PASS

Full log attached.

test2004_06_21.log (80.5 KB)

Vladimir,

We had a problem with our nightly tester (a change I introduced to
hopefully solve another problem) that caused most of the failures in
your first test. This run looks much more like what we're used to
seeing. I believe the two tests below are known problems.

Reid.

Is it ok sending this results for FreeBSD5.1 at daily/weekly based to this
mail list?

A better list for it would be the llvmbugs list for now.

I beginning to think that we need a new test results mailing list. We
have 3 instances of the nightly tester going now (x86/linux, sparc, ppc)
and may have more in the future. The nightly tester spits out
notification of changes (new test failures, etc) that get sent to whoever
"owns" the tester. It would be good of having a way of keeping track of
these reports.

I don't really like the idea of adding yet another mailing list,
especially because few people will probably subscribe to it, but I also
don't want to spam any of the other lists too badly.

What do others think?

BTW, if you are interested in running the nightly tester on your freebsd
box, let me know and I can help you get it set up.

Now results.
Big improvement in llvm tests results from last test result sended.

Great! That's what we want to see. :slight_smile:

New regressions:
  Regression.Assembler.ConstantExprFold : FAIL , expected PASS
  Regression.CodeGen.Generic.2004-04-09-SameValueCoalescing: FAIL ,
expected PASS
  Regression.Transforms.PRE.basictest : FAIL , expected PASS
  Regression.Transforms.TailCallElim.tail_call_with_branch: FAIL ,
expected PASS

These should all be xfails right now, so this is ok.

     920 tests total
     856 ( 93%) tests as expected
       6 ( 1%) tests unexpected FAIL
      58 ( 6%) tests unexpected PASS

The 58 XPASSes are strange. It looks like qmtest isn't expecting you to
have a working code generator, but you do. qmtest probably needs to be
updated somehow to reflect this...

-Chris

--- TESTS WITH UNEXPECTED
OUTCOMES -------------------------------------------

  Feature.mc.basictest : XPASS

  Feature.mc.calltest : XPASS

  Feature.mc.casttest : XPASS

  Feature.mc.cfgstructures : XPASS

  Feature.mc.constexpr : XPASS

  Feature.mc.constexprbad : XPASS

  Feature.mc.constpointer : XPASS

  Feature.mc.forwardreftest : XPASS

  Feature.mc.globalvars : XPASS

  Feature.mc.indirectcall : XPASS

  Feature.mc.indirectcall2 : XPASS

  Feature.mc.properties : XPASS

  Feature.mc.prototype : XPASS

  Feature.mc.recursivetype : XPASS

  Feature.mc.simplecalltest : XPASS

  Feature.mc.small : XPASS

  Feature.mc.smallest : XPASS

  Feature.mc.testalloca : XPASS

  Feature.mc.testconstants : XPASS

  Feature.mc.testlogical : XPASS

  Feature.mc.testmemory : XPASS

  Feature.mc.testswitch : XPASS

  Feature.mc.testvarargs : XPASS

  Feature.mc.varargs : XPASS

  Feature.mc.weirdnames : XPASS

  Regression.Assembler.ConstantExprFold : FAIL , expected PASS
    Script:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trConstantExprFold.llx/testsc
ript.ConstantExprFold.llx
    Output:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trConstantExprFold.llx/testsc
ript.ConstantExprFold.llx.out

  Regression.C++Frontend.2003-08-28-ForwardType : XPASS

  Regression.C++Frontend.2003-09-30-ForIncrementExprBug2: XPASS

  Regression.CFrontend.2003-01-30-UnionInit : FAIL , expected PASS
    Compiling C code failed

  Regression.CFrontend.2003-06-29-MultipleFunctionDefinition: XPASS

  Regression.CFrontend.2003-08-30-AggregateInitializer: FAIL , expected
PASS
    Compiling C code failed

  Regression.CFrontend.2003-08-30-LargeIntegerBitfieldMember: XPASS

  Regression.CodeGen.Generic.2004-04-09-SameValueCoalescing: FAIL ,
expected PASS
    Script:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/tr2004-04-09-SameValueCoalesc
ing.llx/testscript.2004-04-09-SameValueCoalescing.llx
    Output:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/tr2004-04-09-SameValueCoalesc
ing.llx/testscript.2004-04-09-SameValueCoalescing.llx.out

  Regression.LLC.2002-04-14-UnexpectedUnsignedType: XPASS

  Regression.LLC.2002-04-16-StackFrameSizeAlignment: XPASS

  Regression.LLC.2003-05-27-phifcmpd : XPASS

  Regression.LLC.2003-05-27-useboolinotherbb : XPASS

  Regression.LLC.2003-05-27-usefsubasbool : XPASS

  Regression.LLC.2003-05-28-ManyArgs : XPASS

  Regression.LLC.2003-05-30-BadFoldGEP : XPASS

  Regression.LLC.2003-05-30-BadPreselectPhi : XPASS

  Regression.LLC.2003-07-06-BadIntCmp : XPASS

  Regression.LLC.2003-07-07-BadLongConst : XPASS

  Regression.LLC.2003-07-08-BadCastToBool : XPASS

  Regression.LLC.2003-07-29-BadConstSbyte : XPASS

  Regression.LLC.BurgBadRegAlloc : XPASS

  Regression.LLC.badCallArgLRLLVM : XPASS

  Regression.LLC.badFoldGEP : XPASS

  Regression.LLC.badarg6 : XPASS

  Regression.LLC.badlive : XPASS

  Regression.LLC.constindices : XPASS

  Regression.LLC.fwdtwice : XPASS

  Regression.LLC.negintconst : XPASS

  Regression.LLC.sched : XPASS

  Regression.LLC.select : XPASS

  Regression.LLC.spillccr : XPASS

  Regression.Transforms.CorrelatedExprs.2002-10-03-PHIPropogate: XPASS

  Regression.Transforms.CorrelatedExprs.whet : XPASS

  Regression.Transforms.IndVarsSimplify.2003-09-23-NotAtTop: XPASS

  Regression.Transforms.Inline.alloca_test : XPASS

  Regression.Transforms.LICM.2003-08-04-TrappingInstHoist: XPASS

  Regression.Transforms.LICM.2003-08-04-TrappingInstOkHoist: XPASS

  Regression.Transforms.PRE.basictest : FAIL , expected PASS
    Script:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trbasictest.ll/testscript.bas
ictest.ll
    Output:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trbasictest.ll/testscript.bas
ictest.ll.out

  Regression.Transforms.TailCallElim.tail_call_with_branch: FAIL ,
expected PASS
    Script:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trtail_call_with_branch.ll/te
stscript.tail_call_with_branch.ll
    Output:
/usr/home/wanderer/pkg/build/llvm/obj/test/tmp/trtail_call_with_branch.ll/te
stscript.tail_call_with_branch.ll.out

-Chris

> Is it ok sending this results for FreeBSD5.1 at daily/weekly based to

this

> mail list?

A better list for it would be the llvmbugs list for now.

ok

BTW, if you are interested in running the nightly tester on your freebsd
box, let me know and I can help you get it set up.

What i must install or have installed ?
I doesn't have root access at this machine, but admin ready to help.

Vladimir

I think its a great idea. I'd subscribe.

Reid

Just off the top of my head ...

GCC 3.3.1 or 3.4 (but not 3.3.2 or 3.3.3)
Perl 5.8
Python 2.X
QMTest 2.0.3
GNUPlot (latest)
libgd
libfreetype2
libjpeg
libpng
bison 1.35 (not 1.875)
flex

Where I've mentioned the version, the version matters. For others, you
can probably just use the latest release.

Reid.