POST MORTEM: llvm-test changes

After the removal of llvm/test/Programs and insertion of llvm-test cvs
module, the nightly tester showed the following:

x86 Linux debug:
  * hard to say. I think it failed because the check in occurred just
     as it was starting up. Could someone verify for me? No results
     were produced.

x86 Linux optimized:
  * everything ran okay
  * loc graph now shows 550k lines (up 350k!) - not sure what that's
     all about. I'll look at the loc counting code. It should have gone
     DOWN.

x86 FreeBSD:
  * hasn't run with changes yet

PPC Darwin:
  * hasn't run with changes yet

SPARC V9 Solaris:
  * hasn't run with changes yet.

We'll have to wait a few more hours for the last 3 platforms. It looks
like things mostly build okay. If you update and try testing, please let
me know how it goes. I'd like to hear about successes too as I'm pretty
worried that its just flat out broken on non-Linux platforms.

Reid

Configure gives me these errors on FreeBSD. I'm proceding with the build anyway to see what happens. Don't worry :slight_smile: It'll only take about 2.5 hours instead of 30+ now.

install: ../test/Programs/Makefile does not exist
config.status: executing test/Programs/Makefile.programs commands
install: ../test/Programs/Makefile.programs does not exist
config.status: executing test/Programs/Makefile.tests commands
install: ../test/Programs/Makefile.tests does not exist
config.status: executing test/Programs/TEST.aa.Makefile commands
install: ../test/Programs/TEST.aa.Makefile does not exist
config.status: executing test/Programs/TEST.dsgraph.report commands
install: ../test/Programs/TEST.dsgraph.report does not exist
config.status: executing test/Programs/TEST.aa.report commands
install: ../test/Programs/TEST.aa.report does not exist
config.status: executing test/Programs/TEST.example.Makefile commands
install: ../test/Programs/TEST.example.Makefile does not exist
config.status: executing test/Programs/TEST.nightly.Makefile commands
install: ../test/Programs/TEST.nightly.Makefile does not exist
config.status: executing test/Programs/TEST.buildrepo.Makefile commands
install: ../test/Programs/TEST.buildrepo.Makefile does not exist
config.status: executing test/Programs/TEST.jit.Makefile commands
install: ../test/Programs/TEST.jit.Makefile does not exist
config.status: executing test/Programs/TEST.nightly.report commands
install: ../test/Programs/TEST.nightly.report does not exist
config.status: executing test/Programs/TEST.dsgraph.Makefile commands
install: ../test/Programs/TEST.dsgraph.Makefile does not exist
config.status: executing test/Programs/TEST.jit.report commands
install: ../test/Programs/TEST.jit.report does not exist
config.status: executing test/Programs/TEST.typesafe.Makefile commands
install: ../test/Programs/TEST.typesafe.Makefile does not exist
config.status: executing test/Programs/TEST.dsgraph.gnuplot commands
install: ../test/Programs/TEST.dsgraph.gnuplot does not exist
config.status: executing test/Programs/TEST.vtl.Makefile commands
install: ../test/Programs/TEST.vtl.Makefile does not exist
cd: can't cd to ../../../../test/Programs/External
config.status: executing test/Programs/External/Makefile commands
install: ../test/Programs/External/Makefile does not exist
cd: can't cd to ../../../../../test/Programs/External/SPEC
config.status: executing test/Programs/External/SPEC/Makefile commands
install: ../test/Programs/External/SPEC/Makefile does not exist
cd: can't cd to ../../../../../test/Programs/External/SPEC
config.status: executing test/Programs/External/SPEC/Makefile.spec commands
install: ../test/Programs/External/SPEC/Makefile.spec does not exist
cd: can't cd to ../../../../../test/Programs/External/SPEC
config.status: executing test/Programs/External/SPEC/Makefile.spec2000 commands
install: ../test/Programs/External/SPEC/Makefile.spec2000 does not exist
cd: can't cd to ../../../../../test/Programs/External/SPEC
config.status: executing test/Programs/External/SPEC/Makefile.spec95 commands
install: ../test/Programs/External/SPEC/Makefile.spec95 does not exist
config.status: executing test/Programs/MultiSource/Makefile commands
install: ../test/Programs/MultiSource/Makefile does not exist
config.status: executing test/Programs/MultiSource/Makefile.multisrc commands
install: ../test/Programs/MultiSource/Makefile.multisrc does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/analyzer
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/analyzer/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/analyzer/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/distray
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/distray/Makefile commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/distray/Makefile does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/distray
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/distray/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/distray/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow/Makefile commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow/Makefile does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/fourinarow/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/mason
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/mason/Makefile commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/mason/Makefile does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/mason
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/mason/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/mason/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/neural
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/neural/Makefile commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/neural/Makefile does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/neural
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/neural/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/neural/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2/Makefile commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2/Makefile does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2/test.in commands
install: ../test/Programs/MultiSource/Benchmarks/FreeBench/pcompress2/test.in does not exist
cd: can't cd to ../../../../../../../test/Programs/MultiSource/Benchmarks/FreeBench/pifft
config.status: executing test/Programs/MultiSource/Benchmarks/FreeBench/pifft/Makefile commands

x86 FreeBSD:
  * hasn't run with changes yet

I manually start script. I use this options (+ -verbose now for testing)

/home/wanderer/pkg/build/llvm/src/llvm/utils/NightlyTest.pl -parallel -enable-linscan -noexternals -noremove :pserver:anon@llvm-cvs.cs.uiuc.edu:2401/var/cvs/llvm /home/wanderer/pkg/build/llvm/night/build /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD

Log:
INITIALIZED
CVS Root = :pserver:anon@llvm-cvs.cs.uiuc.edu:2401/var/cvs/llvm-test
BuildDir = /home/wanderer/pkg/build/llvm/night/build
WebDir = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD
Prefix = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-05
CVSLog = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-05-CVS-Log.txt
BuildLog = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-05-Build-Log.txt
Changing To: CVS checkout directory (/home/wanderer/pkg/build/llvm/night/build)
CHECKOUT STAGE
Syntax error: "(" unexpected (expecting ")")
Changing To: CVS Checkout directory (/home/wanderer/pkg/build/llvm/night/build)
Changing To: llvm source directory (llvm)
Cannot change directory to: llvm source directory (llvm) at /home/wanderer/pkg/build/llvm/src/llvm/utils/NightlyTest.pl line 133.

Vladimir

You really need to do a cvs update, especially on the configure script.

Reid.

Jeff,

Actually, that was my fault. I forgot to remove the non-existent
directories from the configure.ac file. That's done and committed
now, so the advice is still the same: update configure script :slight_smile:

Reid.

That's weird. What is your default shell that Perl invokes with the
"system" command. This works fine with bash-2.05b.

Reid.

That's weird. What is your default shell that Perl invokes with the
"system" command. This works fine with bash-2.05b.

In reality :slight_smile: i use this script in crontab (and run it manually with added verbose flag):

---8X--------------
#!/bin/sh -

Actually, I'm talking about the shell that Perl invokes. When you run
the NightlyTest.pl script, it uses Perl. The system() function in Perl
invokes the "standard" shell to execute the command provided as the
argument to system(). In NightlyTest.pl, I have changed the program to
check out both the regular llvm CVS module as well as the llvm-test
module. This command is what is giving your error. The command is:

(time -p ($NICE cvs $CVSOPT -d $CVSRootDir co -APR llvm; cd llvm/projects ; \
     $NICE cvs $CVSOPT -d $CVSRootDir co -APR llvm-test ) ) > $CVSLog 2>&1

When your shell interprets this, it is complaining about the second open
parenthesis. Bash 2.05b on Linux handles this fine. I was asking what
your "default" system shell is on FreeBSD. Probably /bin/sh, right?
Perhaps you can:

   SHELL=/usr/bin/bash ; export SHELL

in your script below just before it runs NightlyTest.pl? I'm not sure if
that will work or not.

Reid.

Bash 2.05b on Linux handles this fine. I was asking what
your "default" system shell is on FreeBSD. Probably /bin/sh, right?
Perhaps you can:

  SHELL=/usr/bin/bash ; export SHELL

in your script below just before it runs NightlyTest.pl? I'm not sure if
that will work or not.

I have:
SHELL=/bin/tcsh

sh, csh, tcsh not support nested ()

OK, root install bash by my request, but adding SHELL=/usr/local/bin/bash
export SHELL

in start script doesn't help :frowning:

Vladimir

OK, no problems building. I got some minor issues with the simple
regression tests. First:

  Regression.Bytecode.slow.ll : UNTESTED, expected PASS
    Could not load test.

    qmtest.exception:
      qm.test.database.NoSuchTestError: There is no test with the test ID "".

    qmtest.traceback:
      
        File "/lib/qm/qm/test/execution_engine.py", line 181, in _RunTests
         File "/usr/home/llvm/obj/test/QMTest/llvmdb.py", line 91, in GetTest
          raise qm.test.database.NoSuchTestError(self)
      
Next:

  Feature.cc.packed : FAIL , expected PASS
    Failed to convert /usr/home/llvm/obj/test/tmp/feature-cc-packed.ll to C code.

    qmtest.target:
      local

And finally:

  Feature.cc.weirdnames /usr/home/llvm/obj/../test/Regression/C++Frontend/2003-08-28-ForwardType.cpp:3: warning: inline function `void* operator new(unsigned int, void*)' used but n
ever defined
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-02-18-64bitConstant.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-02-18-64bitConstant.c:3: warning: return type of 'main' is not `int'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-04-08-LocalArray.c: In function `test':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-04-08-LocalArray.c:11: warning: `return' with a value, in function returning void
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c: In function `memset_impl':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c:13: warning: cast from pointer to integer of different size
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c:17: warning: cast to pointer from integer of different size
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests3.c:159: warning: conflicting types for built-in function 'strcpy'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests3.c:163: warning: conflicting types for built-in function 'strcat'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-08-02-UnionTest.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-08-02-UnionTest.c:14: warning: return type of 'main' is not `int'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:28: warning: parameter names (without types) in function declaration
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:52: warning: integer constant is too large for "long" type
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:49: warning: return type of 'main' is not `int'
Assertion failed: (0 && "Couldn't expand constructor in this context!"), function llvm_expand_constant_expr, file ../../src/gcc/llvm-expand.c, line 5070.
/usr/home/llvm/obj/../test/Regression/CFrontend/2003-01-30-UnionInit.c:8: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
/usr/home/llvm/obj/../test/Regression/CFrontend/2003-02-12-NonlocalGoto.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2003-02-12-NonlocalGoto.c:7: error: nested functions are not supported!
WARNING: Function declared to have type 'int' but it actually has type 'sbyte'!
Assertion failed: (0 && "Couldn't expand constructor in this context!"), function llvm_expand_constant_expr, file ../../src/gcc/llvm-expand.c, line 5070.
/usr/home/llvm/obj/../test/Regression/CFrontend/2003-08-30-AggregateInitializer.c:13: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
Assertion failed: ((FieldSizeTree != NULL_TREE) && "Struct/Union member of unknown length!"), function llvm_expand_constructor_elements, file ../../src/gcc/llvm-expand.c, line 3791.
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-01-01-UnknownInitSize.c:11: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
UNKNOWN INTEGRAL TYPE SIZE: 40
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-07-BitfieldCrash.c:7: internal compiler error: in llvm_type_get_integer, at llvm-types.c:571
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-16-AsmRegisterCrash.c: In function `foo':

/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-16-AsmRegisterCrash.c:2: warning: llvm doesn't support asm specifier on 'X'
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-05-21-IncompleteEnum.c:1: warning: "enum foo" declared inside parameter list
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-05-21-IncompleteEnum.c:1: warning: its scope is only this definition or declaration, which is probably not what you want
WARNING: Function declared to have type '%enum.foo3.*' but it actually has type '%enum.foo4.*'!
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-06-08-OpaqueStructArg.c:2: warning: parameter has incomplete type
ntend.2003-08-21-EmptyClass : PASS

Yeah, it passes in the end but there are multiple "internal compiler
errors" in the messages above.

Now starting the llvm-test module...

Okay, I'll have to fix NightlyTest.pl not to use shell script syntax
that isn't universal. Look for a commit soon.

Reid.

> Jeff,
>
> Actually, that was my fault. I forgot to remove the non-existent
> directories from the configure.ac file. That's done and committed
> now, so the advice is still the same: update configure script :slight_smile:
>
> Reid.

OK, no problems building.

Great!

I got some minor issues with the simple
regression tests. First:

  Regression.Bytecode.slow.ll : UNTESTED, expected PASS
    Could not load test.

    qmtest.exception:
      qm.test.database.NoSuchTestError: There is no test with the test ID "".

    qmtest.traceback:
      
        File "/lib/qm/qm/test/execution_engine.py", line 181, in _RunTests
         File "/usr/home/llvm/obj/test/QMTest/llvmdb.py", line 91, in GetTest
          raise qm.test.database.NoSuchTestError(self)
    
Known problem. Test needs to be added to the database. :frowning:

John?

Next:

  Feature.cc.packed : FAIL , expected PASS
    Failed to convert /usr/home/llvm/obj/test/tmp/feature-cc-packed.ll to C code.

    qmtest.target:
      local

Incomplete feature. The Packed lowering pass hasn't been implemented
yet.

And finally:

  Feature.cc.weirdnames /usr/home/llvm/obj/../test/Regression/C++Frontend/2003-08-28-ForwardType.cpp:3: warning: inline function `void* operator new(unsigned int, void*)' used but n
ever defined
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-02-18-64bitConstant.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-02-18-64bitConstant.c:3: warning: return type of 'main' is not `int'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-04-08-LocalArray.c: In function `test':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-04-08-LocalArray.c:11: warning: `return' with a value, in function returning void
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c: In function `memset_impl':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c:13: warning: cast from pointer to integer of different size
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests.c:17: warning: cast to pointer from integer of different size
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests3.c:159: warning: conflicting types for built-in function 'strcpy'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-07-14-MiscTests3.c:163: warning: conflicting types for built-in function 'strcat'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-08-02-UnionTest.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-08-02-UnionTest.c:14: warning: return type of 'main' is not `int'
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:28: warning: parameter names (without types) in function declaration
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:52: warning: integer constant is too large for "long" type
/usr/home/llvm/obj/../test/Regression/CFrontend/2002-11-07-Redefinition.c:49: warning: return type of 'main' is not `int'
Assertion failed: (0 && "Couldn't expand constructor in this context!"), function llvm_expand_constant_expr, file ../../src/gcc/llvm-expand.c, line 5070.
/usr/home/llvm/obj/../test/Regression/CFrontend/2003-01-30-UnionInit.c:8: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.

Could you follow those instructions please? For some reason this is a
bug on your platform.

/usr/home/llvm/obj/../test/Regression/CFrontend/2003-02-12-NonlocalGoto.c: In function `main':

/usr/home/llvm/obj/../test/Regression/CFrontend/2003-02-12-NonlocalGoto.c:7: error: nested functions are not supported!
WARNING: Function declared to have type 'int' but it actually has type 'sbyte'!
Assertion failed: (0 && "Couldn't expand constructor in this context!"), function llvm_expand_constant_expr, file ../../src/gcc/llvm-expand.c, line 5070.
/usr/home/llvm/obj/../test/Regression/CFrontend/2003-08-30-AggregateInitializer.c:13: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
Assertion failed: ((FieldSizeTree != NULL_TREE) && "Struct/Union member of unknown length!"), function llvm_expand_constructor_elements, file ../../src/gcc/llvm-expand.c, line 3791.
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-01-01-UnknownInitSize.c:11: internal compiler error: Abort trap
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
UNKNOWN INTEGRAL TYPE SIZE: 40
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-07-BitfieldCrash.c:7: internal compiler error: in llvm_type_get_integer, at llvm-types.c:571
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.cs.uiuc.edu> for instructions.
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-16-AsmRegisterCrash.c: In function `foo':

/usr/home/llvm/obj/../test/Regression/CFrontend/2004-03-16-AsmRegisterCrash.c:2: warning: llvm doesn't support asm specifier on 'X'
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-05-21-IncompleteEnum.c:1: warning: "enum foo" declared inside parameter list
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-05-21-IncompleteEnum.c:1: warning: its scope is only this definition or declaration, which is probably not what you want
WARNING: Function declared to have type '%enum.foo3.*' but it actually has type '%enum.foo4.*'!
/usr/home/llvm/obj/../test/Regression/CFrontend/2004-06-08-OpaqueStructArg.c:2: warning: parameter has incomplete type
ntend.2003-08-21-EmptyClass : PASS

Yeah, it passes in the end but there are multiple "internal compiler
errors" in the messages above.

None of those are good. You shouldn't get ICE's from the llvm-gcc
compiler. Not sure what's up with any of that, it doesn't happen on
other platforms. Could you file bugs for these and make sure they are
marked specific to your platform? (FreeBSD?). Thanks. I'm not the person
to diagnose these, Chris will have to do it when he gets back.

Now starting the llvm-test module...

Good luck !

Reid

After fixing nested () problem manual run nighttest finished successfully
with one remarkable logged problem:

INITIALIZED
CVS Root = :pserver:anon@llvm-cvs.cs.uiuc.edu:2401/var/cvs/llvm
BuildDir = /home/wanderer/pkg/build/llvm/night/build
WebDir = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD
Prefix = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-06
CVSLog = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-06-CVS-Log.txt
BuildLog = /home/wanderer/pkg/build/llvm/night/testresults-X86-FreeBSD/2004-09-06-Build-Log.txt
Changing To: CVS checkout directory (/home/wanderer/pkg/build/llvm/night/build)
CHECKOUT STAGE
Changing To: CVS Checkout directory (/home/wanderer/pkg/build/llvm/night/build)
Changing To: llvm source directory (llvm)
UPDATE STAGE
READING TEMPLATE
wc: Argument list too long
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Use of uninitialized value in pattern match (m//) at /home/wanderer/pkg/build/llvm/src/llvm/utils/NightlyTest.pl line 105.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
CONFIGURE STAGE
[log tail skiped]

Vladimir

The wc command that failed is because it uses "getsrcs.sh" and there are
too many sources to fit on your command line. The failure of the wc
command leaves the GetRegex function with a null argument which is what
Perl is complaining about with "uninitialized value in pattern match".
The solution is to fix getsrcs.sh. I'll work something up on that. :slight_smile:

Reid.

Vladimir,

The Lines Of Code problem has been fixed with these two patches:

http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20040906/018078.html
http://mail.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20040906/018079.html

Please update and try again. This should fix the last of the nightly
test problems. Thanks for pointing it out.

Reid.

> Assertion failed: ((FieldSizeTree != NULL_TREE) && "Struct/Union member of unknown length!"), function llvm_expand_constructor_elements, file ../../src/gcc/llvm-expand.c, line 3791.
> /usr/home/llvm/obj/../test/Regression/CFrontend/2004-01-01-UnknownInitSize.c:11: internal compiler error: Abort trap
> Please submit a full bug report,
> with preprocessed source if appropriate.

None of those are good. You shouldn't get ICE's from the llvm-gcc
compiler. Not sure what's up with any of that, it doesn't happen on
other platforms. Could you file bugs for these and make sure they are
marked specific to your platform? (FreeBSD?). Thanks. I'm not the person
to diagnose these, Chris will have to do it when he gets back.

I haven't got around to this yet but I will. The odds are good the
problem is in a BSD system header file so I need to capture the
preprocessed source.

> Now starting the llvm-test module...

Good luck !

Reid

I had the same failures I had before. If they pass everywhere else then
I guess they are FreeBSD specific failures also. Can't blame Pentium II
compatibility problems as I'm now running them on a Celeron. I'll post
a detail description of the failures by this weekend.

I haven't got around to this yet but I will. The odds are good the
problem is in a BSD system header file so I need to capture the
preprocessed source.

I guess not... the file 2004-01-01-UnknownInitSize.c does not include
anything else, so there is no preprocessed source to capture...

After thinking about this some more, I doubt it has anything to do with
FreeBSD--at least not directly. It does indirectly in that I was forced
to build my own front end binaries. I used the tarball that was on the
1.3 download page. I'm beginning to think that source was out of date.
I don't know how else a regression test that was written to make sure a
bug didn't reappear now shows it has reappeared--only for me--and it's
hard to see what the flavor of Unix has to do with it.

I'm going to pull the latest front end source from CVS and start over.

And not only do all the old failures still occur, I now get a new failure:

  Feature.mc.packed : FAIL , expected PASS
    Native assembly file /usr/home/llvm/obj/test/tmp/feature-mc-packed.ll.s was not created; error output was: Assertion failed: (0 && "Invalid type to getClass!"), function getClass, file /usr/home/llvm/lib/Target/X86/X86ISelSimple.cpp, line 65.
Abort trap (core dumped)

I've filed a bug for the ICEs.

For the heck of it I tried upgrading to gcc 3.4.2 (from 3.3.3). It
didn't make a difference. So here are the failures for llvm-test. All
diffs are against the "native" output.

===================== MultiSource/Applications/sgefa

cbe failed differently from jit/llc. First cbe:

  84c84
  < One-Norm(A) ---------- 8.879153e+02.