Re: waveOutReset and Win16Mutex

2000-08-19 Thread Eric Pouech

James Abbatiello wrote:
 
 Continuing in my efforts to get MOD4WIN working, I have come across the
 following problem.  Its a 16bit application and it calls waveOutReset
 while holding the Win16Mutex (SysLevel 1).  Wine attempts to send a
 notify message back about this.  However, the notify gets sent from a
 different thread.  That thread needs to send the notify to 16bit code
 and thus needs to obtain the lock.  The original thread is still holding
 the lock and is waiting for the other thread to finish sending the
 notify before continuing.
 
 The following patch removes the problem for me.  I just arbitrarily
 stuck a ReleaseThunkLock/RestoreThunkLock around the wait call in the
 first thread.  Is this the right solution and the correct place to put
 it?  If so, do the other similar functions there need the same thing?

well, if this is needed, it shall not be done in wineoss/audio.c which
is meant to be pure 32 bit DLL, but probably in 16 bit MMSYSTEM.DLL.

In Windows, MMSYSTEM thunks up to WINMM (32 bit)hence releasing the
locks. currently in Wine, MMSYSTEM just 32 bit calls WINMM, which
can be the cause of the issues

can you send me a -debugmsg +mmsys,+wave,+msacm,+relay=mmsystem:winmm
just to check

A+
-- 
---
Eric Pouech (http://perso.wanadoo.fr/eric.pouech/)
"The future will be better tomorrow", Vice President Dan Quayle




Re: TSX* question

2000-08-19 Thread Peter Hunnisett


snip
 
  Yes, if you only consider systems using a recent glibc and a
  thread-safe Xlib. 
 
 True. But, I guess, we can assume that for Linux-based system,
 right?

Probably, yes.

There are people who run libc5 still. My personal computer has it still,
but I haven't used it in a bit as it's 4000 miles away. You were proposing 
just to make the check for glibc anyways, so that'll do.

snip

Ciao,
Peter Hunnisett
[EMAIL PROTECTED]






Re: wine .spec file autogenerator

2000-08-19 Thread Juergen Schmied


 However, I've noticed that it doesn't appear to correlate with the already
 existing shell32.spec file based on a dump of the WinNT4 shell32.dll (at least
 as far as ordinals go). Can anyone explain this?
No, the shell32.spec is not based on any dump of a native shell32.dll.

- the shell32 exports many functions by ordinal withou a name.
 Your tool likely can't cope with it - or it would need to use *.dbg 
and *.pdb files to get symboles from. (have not tryed it jet)

- our shell32.spec defines many functions from different versions off shell32.
(win95,win98,winnt, w2k). You have to carefully place some special functions
on special ordinals to avoid conflicts. All functions exported by ordinal have to
be on this ordinal. The leftover functions are exported by name.

Your tool might be good to create a initial implementation for a unknown dll but the
spec files for allready implemented dll's usually containing a lot of manual tweaking.

ciao

juergen