VCE test failed under Microsoft Visual Studio .NET 2003

VCE test failed under Microsoft Visual Studio .NET 2003

[Dongsheng Song]

I using Microsoft Visual Studio .NET 2003 compile the latest 
pthread-win32 code, VC test passed, VCE test fail:
...

cl /GX /TP /DPtW32NoCatchWarn /D__CLEANUP_CXX /O2 /Ob2 /W3 /WX /MD 
/nologo -I. s
emaphore4.c /Fesemaphore4.exe /link /INCREMENTAL:NO pthreadVCE1.lib 
ws2_32.lib
semaphore4.c
... Running VCE test: semaphore4.exe
Assertion failed: (-value == (MAX_COUNT - 1)), file semaphore4.c, line 122
NMAKE : fatal error U1077: “.\semaphore4.exe” : return code “0x1”
Stop.

Intel C++ 8.1 fix for pthread-win32 test code

[Dongsheng Song]

[-- Attachment #1: Type: text/plain, Size: 1972 bytes --]

D:\tmp\pthreads-win32\tests>cvs diff
cvs diff: Diffing .
Index: exit2.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/exit2.c,v
retrieving revision 1.11
diff -r1.11 exit2.c
49a50,51
 >
 >       return NULL;
Index: exit3.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/exit3.c,v
retrieving revision 1.8
diff -r1.8 exit3.c
47a48,49
 >
 >       return NULL;
Index: eyal1.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/eyal1.c,v
retrieving revision 1.17
diff -r1.17 eyal1.c
137c137
<       s += 2 * f * g / (h != 0 ? (h * h) : 1);
---
 >       s += 2 * f * g / ((int) h != 0 ? (h * h) : 1);
Index: mutex5.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/mutex5.c,v
retrieving revision 1.8
diff -r1.8 mutex5.c
50,53c50,53
<   assert(success = PTHREAD_MUTEX_DEFAULT == PTHREAD_MUTEX_NORMAL);
<   assert(success = PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_ERRORCHECK);
<   assert(success = PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_RECURSIVE);
<   assert(success = PTHREAD_MUTEX_RECURSIVE != PTHREAD_MUTEX_ERRORCHECK);
---
 >   assert(PTHREAD_MUTEX_DEFAULT == PTHREAD_MUTEX_NORMAL);
 >   assert(PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_ERRORCHECK);
 >   assert(PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_RECURSIVE);
 >   assert(PTHREAD_MUTEX_RECURSIVE != PTHREAD_MUTEX_ERRORCHECK);
55,57c55,57
<   assert(success = PTHREAD_MUTEX_NORMAL == PTHREAD_MUTEX_FAST_NP);
<   assert(success = PTHREAD_MUTEX_RECURSIVE == PTHREAD_MUTEX_RECURSIVE_NP);
<   assert(success = PTHREAD_MUTEX_ERRORCHECK == 
PTHREAD_MUTEX_ERRORCHECK_NP);
---
 >   assert(PTHREAD_MUTEX_NORMAL == PTHREAD_MUTEX_FAST_NP);
 >   assert(PTHREAD_MUTEX_RECURSIVE == PTHREAD_MUTEX_RECURSIVE_NP);
 >   assert(PTHREAD_MUTEX_ERRORCHECK == PTHREAD_MUTEX_ERRORCHECK_NP);


[-- Attachment #2: patch.txt --]
[-- Type: text/plain, Size: 1895 bytes --]

Index: exit2.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/exit2.c,v
retrieving revision 1.11
diff -r1.11 exit2.c
49a50,51
> 	
> 	return NULL;
Index: exit3.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/exit3.c,v
retrieving revision 1.8
diff -r1.8 exit3.c
47a48,49
> 	
> 	return NULL;
Index: eyal1.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/eyal1.c,v
retrieving revision 1.17
diff -r1.17 eyal1.c
137c137
<       s += 2 * f * g / (h != 0 ? (h * h) : 1);
---
>       s += 2 * f * g / ((int) h != 0 ? (h * h) : 1);
Index: mutex5.c
===================================================================
RCS file: /cvs/pthreads-win32/pthreads/tests/mutex5.c,v
retrieving revision 1.8
diff -r1.8 mutex5.c
50,53c50,53
<   assert(success = PTHREAD_MUTEX_DEFAULT == PTHREAD_MUTEX_NORMAL);
<   assert(success = PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_ERRORCHECK);
<   assert(success = PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_RECURSIVE);
<   assert(success = PTHREAD_MUTEX_RECURSIVE != PTHREAD_MUTEX_ERRORCHECK);
---
>   assert(PTHREAD_MUTEX_DEFAULT == PTHREAD_MUTEX_NORMAL);
>   assert(PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_ERRORCHECK);
>   assert(PTHREAD_MUTEX_DEFAULT != PTHREAD_MUTEX_RECURSIVE);
>   assert(PTHREAD_MUTEX_RECURSIVE != PTHREAD_MUTEX_ERRORCHECK);
55,57c55,57
<   assert(success = PTHREAD_MUTEX_NORMAL == PTHREAD_MUTEX_FAST_NP);
<   assert(success = PTHREAD_MUTEX_RECURSIVE == PTHREAD_MUTEX_RECURSIVE_NP);
<   assert(success = PTHREAD_MUTEX_ERRORCHECK == PTHREAD_MUTEX_ERRORCHECK_NP);
---
>   assert(PTHREAD_MUTEX_NORMAL == PTHREAD_MUTEX_FAST_NP);
>   assert(PTHREAD_MUTEX_RECURSIVE == PTHREAD_MUTEX_RECURSIVE_NP);
>   assert(PTHREAD_MUTEX_ERRORCHECK == PTHREAD_MUTEX_ERRORCHECK_NP);

Re: VCE test failed under Microsoft Visual Studio .NET 2003

[Ross Johnson]

Hi,

Thanks for the test suite patches re Intel C++ 8.1.

VCE test failure is a known problem for VC6.0 and VC7.0. I will add
VS .NET 2003 to the list. See the BUGS file for more on this, but note
that the workaround described there doesn't work with recent snapshots,
although please try it if you have the time - it may provide clues.

As far as anyone can tell, this isn't a problem in the library, but is
assumed to be a problem in the VC++ compiler series related to exception
frame unwinding, and possibly also related to optimisation levels. Try
tracing the exception and it simply seems to disappear.

The GCE version tests pass, which supports the view that it's not a bug
in the library. Did the Intel C++ 8.1 version pass?

It is generally recommended, for POSIX portability, that people use the
VC version rather than VCE version of the pthread dll - even for C++
applications. See the README file and FAQ file for discussion on this.

Regards.
Ross

On Thu, 2005-03-03 at 11:04 +0800, Dongsheng Song wrote:
> I using Microsoft Visual Studio .NET 2003 compile the latest 
> pthread-win32 code, VC test passed, VCE test fail:
> ...
> 
> cl /GX /TP /DPtW32NoCatchWarn /D__CLEANUP_CXX /O2 /Ob2 /W3 /WX /MD 
> /nologo -I. s
> emaphore4.c /Fesemaphore4.exe /link /INCREMENTAL:NO pthreadVCE1.lib 
> ws2_32.lib
> semaphore4.c
> ... Running VCE test: semaphore4.exe
> Assertion failed: (-value == (MAX_COUNT - 1)), file semaphore4.c, line 122
> NMAKE : fatal error U1077: “.\semaphore4.exe” : return code “0x1”
> Stop.
> 
> 

Re: VCE test failed under Microsoft Visual Studio .NET 2003

[Dongsheng Song]

Visual C++ 2005 Express Edition Beta2 passed VC & VCE test(cvs 2005-03-08).

U:\xchg>cl
Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50110.28 for 80x86
Copyright (C) Microsoft Corporation.  All rights reserved.

usage: cl [ option... ] filename... [ /link linkoption... ]


On Sat, 5 Mar 2005 12:20:31 +0800, Dongsheng Song
<dongsheng.song@gmail.com> wrote:
> Yes,  the Intel C++ 8.1 version passed VC & VCE test.
> 
> 
> On Fri, 04 Mar 2005 18:30:44 +1100, Ross Johnson
> <rpj@callisto.canberra.edu.au> wrote:
> > Hi,
> >
> > Thanks for the test suite patches re Intel C++ 8.1.
> >
> > VCE test failure is a known problem for VC6.0 and VC7.0. I will add
> > VS .NET 2003 to the list. See the BUGS file for more on this, but note
> > that the workaround described there doesn't work with recent snapshots,
> > although please try it if you have the time - it may provide clues.
> >
> > As far as anyone can tell, this isn't a problem in the library, but is
> > assumed to be a problem in the VC++ compiler series related to exception
> > frame unwinding, and possibly also related to optimisation levels. Try
> > tracing the exception and it simply seems to disappear.
> >
> > The GCE version tests pass, which supports the view that it's not a bug
> > in the library. Did the Intel C++ 8.1 version pass?
> >
> > It is generally recommended, for POSIX portability, that people use the
> > VC version rather than VCE version of the pthread dll - even for C++
> > applications. See the README file and FAQ file for discussion on this.
> >
> > Regards.
> > Ross
> >
> > On Thu, 2005-03-03 at 11:04 +0800, Dongsheng Song wrote:
> > > I using Microsoft Visual Studio .NET 2003 compile the latest
> > > pthread-win32 code, VC test passed, VCE test fail:
> > > ...
> > >
> > > cl /GX /TP /DPtW32NoCatchWarn /D__CLEANUP_CXX /O2 /Ob2 /W3 /WX /MD
> > > /nologo -I. s
> > > emaphore4.c /Fesemaphore4.exe /link /INCREMENTAL:NO pthreadVCE1.lib
> > > ws2_32.lib
> > > semaphore4.c
> > > ... Running VCE test: semaphore4.exe
> > > Assertion failed: (-value == (MAX_COUNT - 1)), file semaphore4.c, line 122
> > > NMAKE : fatal error U1077: ".\semaphore4.exe" : return code "0x1"
> > > Stop.
> > >
> > >
> >
> >
>

Re: VCE test failed under Microsoft Visual Studio .NET 2003

[Ross Johnson]

On Tue, 2005-03-08 at 20:08 +0800, Dongsheng Song wrote:
> Visual C++ 2005 Express Edition Beta2 passed VC & VCE test(cvs 2005-03-08).
> 

Great! I'll add it to the note in the BUGS file. Unfortunately I still
won't be able to generate a pre-built dll for VCE. I'm stuck way back at
VC6.0, which probably suggests that people might generally be better off
building their own dlls from source, if they don't already.

Thanks.
Ross

> U:\xchg>cl
> Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50110.28 for 80x86
> Copyright (C) Microsoft Corporation.  All rights reserved.
> 
> usage: cl [ option... ] filename... [ /link linkoption... ]
> 
> 
> On Sat, 5 Mar 2005 12:20:31 +0800, Dongsheng Song
> <dongsheng.song@gmail.com> wrote:
> > Yes,  the Intel C++ 8.1 version passed VC & VCE test.
> > 
> > 
> > On Fri, 04 Mar 2005 18:30:44 +1100, Ross Johnson
> > <rpj@callisto.canberra.edu.au> wrote:
> > > Hi,
> > >
> > > Thanks for the test suite patches re Intel C++ 8.1.
> > >
> > > VCE test failure is a known problem for VC6.0 and VC7.0. I will add
> > > VS .NET 2003 to the list. See the BUGS file for more on this, but note
> > > that the workaround described there doesn't work with recent snapshots,
> > > although please try it if you have the time - it may provide clues.
> > >
> > > As far as anyone can tell, this isn't a problem in the library, but is
> > > assumed to be a problem in the VC++ compiler series related to exception
> > > frame unwinding, and possibly also related to optimisation levels. Try
> > > tracing the exception and it simply seems to disappear.
> > >
> > > The GCE version tests pass, which supports the view that it's not a bug
> > > in the library. Did the Intel C++ 8.1 version pass?
> > >
> > > It is generally recommended, for POSIX portability, that people use the
> > > VC version rather than VCE version of the pthread dll - even for C++
> > > applications. See the README file and FAQ file for discussion on this.
> > >
> > > Regards.
> > > Ross
> > >
> > > On Thu, 2005-03-03 at 11:04 +0800, Dongsheng Song wrote:
> > > > I using Microsoft Visual Studio .NET 2003 compile the latest
> > > > pthread-win32 code, VC test passed, VCE test fail:
> > > > ...
> > > >
> > > > cl /GX /TP /DPtW32NoCatchWarn /D__CLEANUP_CXX /O2 /Ob2 /W3 /WX /MD
> > > > /nologo -I. s
> > > > emaphore4.c /Fesemaphore4.exe /link /INCREMENTAL:NO pthreadVCE1.lib
> > > > ws2_32.lib
> > > > semaphore4.c
> > > > ... Running VCE test: semaphore4.exe
> > > > Assertion failed: (-value == (MAX_COUNT - 1)), file semaphore4.c, line 122
> > > > NMAKE : fatal error U1077: ".\semaphore4.exe" : return code "0x1"
> > > > Stop.
> > > >
> > > >
> > >
> > >
> >
>