path: root/FAQ

                  =========================================
                  PTHREADS-WIN32 Frequently Asked Questions
                  =========================================

INDEX
-----

Q 1	Should I use Cygwin or Mingw32 as a development environment?

Q 2	Now that pthreads-win32 builds under Mingw32, why do I get
	memory access violations?

Q 3	How do I use pthread.dll for Win32 (Visual C++ 5.0)

Q 4	Cancelation doesn't work for me, why?

Q 5	Thread won't block after two calls to mutex_lock

Q 6	How do I generate pthreadGCE.dll and libpthreadw32.a for use with Mingw32?

=============================================================================

Q 1	Should I use Cygwin or Mingw32 as a development environment?
---

A 1
---
Important: see Q2 also.

In short, use Mingw32 with the MSVCRT library to build applications that use
the DLL. Cygwin's own internal support for POSIX threads is growing. Consult
that project's documentation for more information.

Date: Mon, 07 Dec 1998 15:11:37 +0100
From: Anders Norlander <anorland@hem2.passagen.se>
To: Ross Johnson <rpj@ise.canberra.edu.au>
Cc: pthreads-win32 <pthreads-win32@air.net.au>
Subject: Re: pthreads-win32: TryEnterCriticalSection patch (fwd)

Ross Johnson wrote:
> 
> Anders,
> 
> You said you're using GCC. Is that from cygwin32 or mingw32? What is your
> environment (so I can perhaps help other people out)? We have problems
> with cygwin32 et al that have been built on Win95. They're missing
> _{begin,end}threadex.

Ross,

I use mingw32 when compiling pthreads-win32, but unlike most people I
use MSVCRT as the C library instead of CRTDLL. For those that don't
feel like configuring and building the necessary components themselves,
Mumit Khan has released an add on for mingw32 to make it use MSVCRT40.
It is available at his ftp site, follow the minw32 links at
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/

For cygwin it is a completely different matter. I suppose
pthreads-win32 uses _beginthreadex and _endthreadex because the Win32
docs say that programs calling functions in the C library should not
use CreateThread and ExitThread. However, this applies only to
Microsoft's (and possibly others) multithreaded C libraries that need
to keep track of per thread data, it does not apply to cygwin.
This code solves the problem:

/* Check for old and new versions of cygwin */
#if defined(__CYGWIN32__) || defined(__CYGWIN__)
/* Macro uses args so we can cast start_proc to LPTHREAD_START_ROUTINE
   in order to avoid warnings because of return type */
#define _beginthreadex(security, stack_size, start_proc, arg, flags,
pid) \
CreateThread(security, stack_size, (LPTHREAD_START_ROUTINE) start_proc,
\
             arg, flags, pid)
#define _endthreadex ExitThread
#endif

I would be extremely careful using threads with cygwin, since it is
not (yet) threadsafe.

Regards,
Anders

------------------------------------------------------------------------------

Q 2	Now that pthreads-win32 builds under Mingw32, why do I get
---	memory access violations (segfaults)?

A 2
---
Note: issue resolved.
The latest Mingw32 package has thread-safe exception handling.
Make sure you also read A 6 below to get a fully working build.


The following email exchange describes the problem. Until this issue
is resolved people without the Microsoft compiler can obtain the current
MSVC prebuilt DLL (pthread.{dll,lib,h}) at:

ftp://sources.redhat.com/pub/pthreads-win32/dll-latest 

Date: Wed, 10 Feb 1999 13:21:01 -0000
From: "Ruland, Kevin" <Kevin.Ruland@anheuser-busch.com>
Reply-To: POSIX threads on Win32 <pthreads-win32@air.net.au>
To: 'POSIX threads on Win32' <pthreads-win32@air.net.au>
Subject: Mingw32 exceptions not thread safe.

Hello everyone.

I asked Mumit Khan, maintainer of egcs for mingw and assorted guru, about
the Known Problem listed below.  

> Known problems
> --------------
> 
> There is an unresolved bug which shows up as a segmentation fault
> (memory access violation) when the library is built using g++. Build
> the test program "eyal1.c" and run with an argument of "2" or
> greater. The argument is the number of threads to run, excluding the
> main thread, so the bug appears with 2 or more worker threads.
> 
> Kevin Ruland has traced the exception to the try/catch blocks in
> ptw32_threadStart().
> 

The official word is:

<Quote Mumit Khan [khan@xraylith.wisc.edu]>
EGCS-1.1.1 for win32 (either cygwin or crtdll/msvc runtimes) do not have
thread-safe exception support. 

For Cygwin, it'll happen when Cygwin runtime has mature thread safety and
pthread is fully integrated. Then it's just a matter of rebuilding GCC (or
just libgcc in this) with thread safe EH support.

For Mingw crtdll/msvc, someone needs to write the thread-wrapper for win32
threads. Anyone who knows win32 threads should be able to do this without
much trouble at all. It's low on my priority list, so unless someone else
volunteers, it'll have to wait.
<\Quote>

Kevin

------------------------------------------------------------------------------

Q 3	How do I use pthread.dll for Win32 (Visual C++ 5.0)
---	

A 3
---
>
> I'm a "rookie" when it comes to your pthread implementation.  I'm currently
> desperately trying to install the prebuilt .dll file into my MSVC compiler.
> Could you please provide me with explicit instructions on how to do this (or
> direct me to a resource(s) where I can acquire such information)?
>
> Thank you,
>

You should have a .dll, .lib, .def, and three .h files.

The .dll can go in any directory listed in your PATH environment
variable, so putting it into C:\WINDOWS should work.

The .lib file can go in any directory listed in your LIB environment
variable.

The .h files can go in any directory listed in your INCLUDE
environment variable.

Or you might prefer to put the .lib and .h files into a new directory
and add its path to LIB and INCLUDE. You can probably do this easiest
by editing the file:-

C:\Program Files\DevStudio\vc\bin\vcvars32.bat

The .def file isn't used by anything in the pre-compiled version but 
is included for information.

Cheers.
Ross

------------------------------------------------------------------------------

Q 4	Cancelation doesn't work for me, why?
---

A 4
---
> I'm investigating a problem regarding thread cancelation. The thread I want
> to cancel has PTHREAD_CANCEL_ASYNCHRONOUS, however, this piece of code
> blocks on the join():
>
>               if ((retv = Pthread_cancel( recvThread )) == 0)
>               {
>                       retv = Pthread_join( recvThread, 0 );
>               }
>
> Pthread_* are just macro's; they call pthread_*.
>
> The thread recvThread seems to block on a select() call. It doesn't get
> cancelled.
>
> Two questions:
>
> 1) is this normal behaviour?
>
> 2) if not, how does the cancel mechanism work? I'm not very familliar to
> win32 programming, so I don't really understand how the *Event() family of
> calls work.

Async cancelation should be in versions post snapshot-1999-11-02
of pthreads-win32 (currently only for x86 architectures).

The answer to your first question is, normal POSIX behaviour would  
be to asynchronously cancel the thread. However, even that doesn't
guarantee cancelation as the standard only says it should be
cancelled as soon as possible.

However ...

Snapshot 99-11-02 or earlier only partially supports asynchronous cancellation.
Snapshots since then simulate async cancelation by poking the address of
a cancelation routine into the PC of the threads context. This requires
the thread to be resumed in some way for the cancelation to actually
proceed. This is not true async cancelation, but it is as close as we've
been able to get to it.

If the thread you're trying to cancel is blocked (for instance, it could be
waiting for data from the network), it will only get cancelled when it unblocks
(when the data arrives). Unfortunately, there is no way to do so from
outside the thread.

Using deferred cancelation would normally be the way to go, however,
even though the POSIX threads standard lists a number of C library
functions that are defined as deferred cancelation points, there is
no hookup between those which are provided by Windows and the
pthreads-win32 library.

Incidently, it's worth noting for code portability that the POSIX
threads standard list doesn't include "select" because (as I read in
Butenhof) it isn't part of POSIX.

Effectively, the only cancelation points that pthreads-win32 can
recognise are those the library implements itself, ie.
        
        pthread_testcancel
        pthread_cond_wait
        pthread_cond_timedwait
	pthread_join
        sem_wait
        pthread_delay_np

Pthreads-win32 also provides two functions that allow you to create
cancelation points within your application, but only for cases where
a thread is going to block on a Win32 handle. These are:

        pthreadCancelableWait(HANDLE waitHandle) /* Infinite wait */
 
        pthreadCancelableTimedWait(HANDLE waitHandle, DWORD timeout)

Regards.
Ross

------------------------------------------------------------------------------
 
Q 5	Thread won't block after two calls to mutex_lock
---
 
A 5
---
> i was testing this pthread for win32 in my prog.
> when i checked if it was blocking mutex_lock calls, i was surprised when it
> didnt lock
> 
> pthread_mutex_t DBlock;
> 
> pthread_mutex_init( &DBlock, NULL );
> pthread_mutex_lock( &DBlock );
> pthread_mutex_lock( &DBlock );
> 
> ^^ these two calls didnt block

POSIX leaves the result "undefined" for a thread that tries
to recursively lock the same mutex (one that it owns already).
That means the actual semantics are left up to the
implementation, but should not be relied upon for code that
will be ported to different POSIX threads implementations.

In the pthreads-win32 implementation a thread won't deadlock
itself by relocking the mutex. Subsequent calls to
pthread_mutex_lock() as in your example above increment
the lock count but the thread continues on. Consequently,
the thread must ensure that it unlocks the mutex once for
each lock operation. That is, pthreads-win32 mutexes are
always recursive.

You may want to look at the other synchronisation devices
available in the library, such as condition variables or
read-write locks.

Ross

------------------------------------------------------------------------------
 
Q 6	How do I generate pthreadGCE.dll and libpthreadw32.a for use with Mingw32?
---
 
A 6
---
Once you've followed Thomas Pfaff's instructions below to fix
Mingw32, then you can simply run "make" to build the library and dll.


From - Sat Dec  9 22:56:10 2000
From: "Thomas Pfaff" <tpfaff@gmx.net>
To: <mingw-users@lists.sourceforge.net>, <pthreads-win32@sources.redhat.com>
Subject: mingw32 DLLs, threads and exceptions HOWTO
Date: Thu, 7 Dec 2000 11:12:43 +0100

Dear all,

this is a summary that should help users to have thread safe exception
handling over DLL exported functions.
If you don't care about c++ exceptions you can stop reading here.

The first time i struggled with c++ exceptions was when i tried to throw an
exception in a dll exported function where the exception handler resides in
the program module.
Instead of catching the exception the program stopped with an abnormal
termination.
The reason was that the exception code is in libgcc.a. Since this is a
static library the code and some static variables are both in the dll and in
the program module, each module runs in its own context.
It was Franco Bez that pointed me in the right direction, that is convert
libgcc.a into a dll.

That done i tried to build the pthreads-win32 library, but some tests failed
with an access violation. Due to the fact that the dll was not build
was -mthreads support, eh_context_static instead of eh_context_specific (the
mthreads version) was used for exception handling.
I did a rebuild of the gcc dll with -mthreads, now all tests are passed
(except a nonportable exception test that relies on a MSVC feature).

To build the gcc dll i did the following steps.

1. create a temporary directory libgcc
2. copy libgcc.a from gcc-2.95.2\lib\gcc-lib\i386-mingw32\gcc-2.95.2 to that
directory
3. ar -x libgcc.a
4. create a directory tmp and move __main.o, _exit.o and __dummy.o in that
directory
5. build the dll
gcc -shared -mthreads -o gcc.dll *.o
strip gcc.dll
Move this dll into your gcc\bin directory
6. Move _chkstk.o and frame.o to the tmp directory, otherwise you break the
builtin alloca.
7. Build the import library libgcc.a
dllwrap --export-all --dllname=gcc.dll --output-def=libgcc.def --output-lib=
libgcc.a *.o
ar -q libgcc.a tmp/*.o
strip --strip-debug libgcc.a
ranlib libgcc.a
8. save your old libgcc.a, copy the new libgcc.a into
gcc-2.95.2\lib\gcc-lib\i386-mingw32\gcc-2.95.2

I am using gcc-2.95.2-1 with Mumits patched binutils-19990818-1and msvcrt
runtime-2000-03-27.
I don't know if this is still required with the current binutils and gcc
since i have seen no sources until now.

I believe that these steps are at least necessary if you are trying to use
the pthreads-win32 library (which is required if you want to use gtk+ on
win32).
They will make mingw32 a real replacement for MSVC (at least for me).

What is left:

1. Include the mingwm10.dll function into the gcc.dll to have only one dll
left.
2. make -mthreads and -fnative-struct default compiler options.
3. convert libstdc++ to a dll by adding the declspec dllexport and dllimport
to every class definition.

Regards,
    Thomas

------------------------------------------------------------------------------