wine-devel

wine-devel@winehq.org

3 participants
84400 discussions

cleaning up pthreads implementation - ABI problem
by Peter Hunnisett 21 Nov '01

21 Nov '01

Howdy, I was just looking at cleaning up the pthreads implementation but I don't think that it's possible to have our pthreads implementation without providing binary compatibility with the pthread implementation we're hijacking. The reason is that the mutex and rwlock objects are supposed to be opaque to the programmer but in fact they are not opaque to the compiler. According to the standard, one doesn't have to call pthread_{mutex,rwlock}_init before using a {mutex,rwlock} as would be required with a totally opaque type. The standard allows for static initialization of the mutex and rwlock using PTHREAD_{MUTEX,RWLOCK}_INITIALIZER if you don't want any special attributes. libc does just this with its rwlock for gettext's global data. If you look at the ream of FIXME:pthread_rwlock_{rd,un}lock that sometimes come out starting wine you'll see that there's no call to pthread_rwlock_init. Of course, this stuff is safe because there's generally no writers wanting the pthread_rwlock_wrlock. So our current pthreads mutex implementation works in Linux by the skin of its teeth, becuase the first field in the mutex is reserved and initialized to 0 (in both the static and pthread_mutex_init cases), provided the version of LinuxThreads is new enough and a pointer is the same size as an int. Now I assume that this would mean it fails curiously for things that don't use LinuxThreads. This would also seem to mean, since we need binary compatibility, that things could even fail if someone upgrades their libc :( BTW, our check is presently basically only on __GLIBC__. Flavour me ignorant, and lazy, but I would assume that this not a Linux only symbol. So in fact the wine pthreads implementation might be already used incorrectly. Reader/Writer locks can be made to work the same way, as the mutex, since there are several fields which are initialized to NULL and we can then HeapAlloc our own structure off that field. We can then come up with weird and wonderful exceptions for the wonky cases that are sure to exist in some implementation of pthreads. Can anyone see a nice way out of this that I'm remaining ignorant to? I can't even think of a runtime way of detecting if our implementation is going to work. All in all I'm not too enthusiatic about the potential maintenance nightmare that this may provide but it would seem the only way is to have different mutex and rwlock structures depending on the pthreads implementation we're hijacking. Some code if people are more interested in details: >From glibc 2.2.4 /* Mutexes (not abstract because of PTHREAD_MUTEX_INITIALIZER). */ /* (The layout is unnatural to maintain binary compatibility with earlier releases of LinuxThreads.) */ typedef struct { int __m_reserved; /* Reserved for future use */ int __m_count; /* Depth of recursive locking */ _pthread_descr __m_owner; /* Owner thread (if recursive or errcheck) */ int __m_kind; /* Mutex kind: fast, recursive or errcheck */ struct _pthread_fastlock __m_lock; /* Underlying fast lock */ } pthread_mutex_t; #define PTHREAD_MUTEX_INITIALIZER \ {0, 0, 0, PTHREAD_MUTEX_TIMED_NP, __LOCK_INITIALIZER} What WINE uses (scheduler/pthread.c): /* assume that pthread_mutex_t has room for at least one pointer, * and hope that the users of pthread_mutex_t considers it opaque * (never checks what's in it) */ typedef struct { CRITICAL_SECTION *critsect; } *wine_mutex; rwlocks on the other hand are not so lucky as there's no unused bits: >From glibc-2.2.4 /* Read-write locks. */ typedef struct _pthread_rwlock_t { struct _pthread_fastlock __rw_lock; /* Lock to guarantee mutual exclusion */ int __rw_readers; /* Number of readers */ _pthread_descr __rw_writer; /* Identity of writer, or NULL if none */ _pthread_descr __rw_read_waiting; /* Threads waiting for reading */ _pthread_descr __rw_write_waiting; /* Threads waiting for writing */ int __rw_kind; /* Reader/Writer preference selection */ int __rw_pshared; /* Shared between processes or not */ } pthread_rwlock_t; # define PTHREAD_RWLOCK_INITIALIZER \ { __LOCK_INITIALIZER, 0, NULL, NULL, NULL, \ PTHREAD_RWLOCK_DEFAULT_NP, PTHREAD_PROCESS_PRIVATE } Ciao, Peter hunnise(a)yahoo.com __________________________________________________ Do You Yahoo!? Yahoo! GeoCities - quick and easy web site hosting, just $8.95/month. http://geocities.yahoo.com/ps/info1

2 1

Re: _dev_t for MSVCRT
by Marcus Meissner 21 Nov '01

21 Nov '01

On Tue, Nov 20, 2001 at 10:16:14PM +0100, Uwe Bonnes wrote: > Changelog: > include/msvcrt/sys/types: > _dev_t in a unsigned integer > -- > Uwe Bonnes bon(a)elektron.ikp.physik.tu-darmstadt.de > > Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt > --------- Tel. 06151 162516 -------- Fax. 06151 164321 ---------- > Index: wine/include/msvcrt/sys/types.h > =================================================================== > RCS file: /home/wine/wine/include/msvcrt/sys/types.h,v > retrieving revision 1.2 > diff -u -r1.2 types.h > --- wine/include/msvcrt/sys/types.h 2001/10/22 18:59:24 1.2 > +++ wine/include/msvcrt/sys/types.h 2001/11/20 19:44:21 > @@ -15,7 +15,7 @@ > #endif > > > -typedef unsigned short _dev_t; > +typedef unsigned _dev_t; Dann sollte da aber 'unsigned int' stehen ;) Ciao, Marcus

1 1

Re: (no subject) (msvcrt.getenv)
by Francois Gouget 20 Nov '01

20 Nov '01

On Tue, 20 Nov 2001, Uwe Bonnes wrote: > > Changelog: > dlls/msvcrt/environ.c: MSVCRT_getenv > Compare for the length of the key and return NULL in case of failure Good catch. Another thing to keep in mind: IIRC environment strings are not case sensitive on Windows. This means that normally getenv("PATH")==getenv("path") and that putenv("Path") would overwrite that same environment variable... This does not appear to be the case in the current code. -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ Dieu dit: "M-x Lumière". Et la lumière fut.

1 0

winsock2.h
by lawson_whitney＠juno.com 20 Nov '01

20 Nov '01

Good day! To try debugging techniques for winelib apps, I had in mind to make a winelib app with an intentional segment fault, and found that _my_ winelib app will no longer compile. (gcc-2.95.3, glibc-2.2.3, current wine CVS). Please, it needs unistd.h so it can comfortably log to syslog. running winemaker on it again didn't help. checking for wrc... /usr/local/bin/wrc updating cache ./config.cache creating ./config.status creating Make.rules creating Makefile creating launcher Configure finished. Do 'make' to build the project. [whit@giftie wlp]$ make gcc -c -g -O2 -fPIC -D_REENTRANT -DWINELIB -I. -I/usr/local/include/wine -o t oj2.o toj2.c In file included from /usr/local/include/wine/winsock2.h:30, from /usr/local/include/wine/windows.h:39, from toj2.h:29, from toj2.c:19: /usr/local/include/wine/winsock.h:886: conflicting types for `gethostname' /usr/include/unistd.h:734: previous declaration of `gethostname' make: *** [toj2.o] Error 1 [whit@giftie wlp]$ I would be glad of a bit of advice thanks, Lawson Constants aren't, and variables won't.

2 4

Civilization2 segment register troubles
by Jukka Heinonen 20 Nov '01

20 Nov '01

Hi. I noticed that Civilization 2 has several problems under Wine. First, the game wants to access memory using segment register value zero, which causes protection exception since that is not allowed in protected mode. Second problem is that the game uses segment register value 0xf8, which Wine does not like. If I drop null-pointer protection [call to DOSMEM_Init(TRUE)], emulate cmp and mov commands so that they work even if using segment register value zero [added stuff to memory/instr.c] and make Wine accept segment register value 0xf8 [by hacking INSTR_ReplaceSelector], Civilization 2 starts to work without any problems. There are several possible reasons why the above changes make the game work: Wine corrupts segment registers somewhere, Civilization 2 wants to use privileged commands and Windows emulates them or the game runs in VM86 mode under Windows. So, I would like to get some ideas what should be done with Wine in order to get the game working. Oh, the above modifications fix protection faults that happen when the game tries to access the following addresses: 0x22, 0xc8, 0xf6, 0xf07c and 0xf0aa. P.S. Civilization 2 seems to work almost perfectly but it is painfully slow. I am pretty certain that I know the reason for this, but I would like to get the game working under official Wine before trying to make it run faster. -- Jukka Heinonen <http://www.iki.fi/jhei/>

2 2

Re: Patches for FreeBSD 5.x compilation
by Alexandre Julliard 20 Nov '01

20 Nov '01

Steve Kargl <sgk(a)troutmask.apl.washington.edu> writes: > What I questioned was the criticism that FreeBSD was guilty > of namespace pollution. Alexandre isn't the only wine > developer who made this statement (check usenet). The header > files in question are implementation dependent and can > contain any variable name or #define or typedef. The fact is > that both wine and FreeBSD have struct thread in implementation > header files. AFAICT, neither is incorrect. The Wine headers that define struct thread are only used inside Wine; the FreeBSD headers are used by applications, no matter how much you argue they are internal headers. So the situation is very different. > But, to have > a wine developer emphatically state that FreeBSD needs to > fix its system header files is somewhat surprising. Why is > it surprising to me? Because this reminds me a company here > in Seattle. I couldn't care less whether FreeBSD fixes its headers or not; we will need to work around the problem in Wine in any case. I was simply expressing the opinion that changing this would make FreeBSD better. I'm certainly not demanding that the FreeBSD guys fix it, if they feel exporting struct thread is a good idea it's their choice. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: LibaudioIO Patch for Solaris
by eric pouech 20 Nov '01

20 Nov '01

Robert Lunnon wrote: > > Attached find files for libaudioio patch for audio on solaris 7 > > winmm.diff - Only the makefile patch is significant wavemap, just contains > some extra debugging traces > > libaudioio.tar, libaudioio goes under winmm > > base.diff, Patches to configure.in and Makefile.in to better support Solaris > and Libaudioio > > config.h.in added HAVE_SOLARIS and HAVE_LIBAUDIOIO and added a way for force > a 32 bit mmap. a few comments : - why do you add the libaudioio lib to winmm ? it should only be linked against the libaudioio (or whatever) driver... winmm should be independant of any driver... - in the libaudioio subdir: + you shouldn't use C++ comment. wine is a C project + it would a good idea to remove any reference to OSS in libaudioio... + I don't think that libaudioio supports midi... so, the preferred way to handle this would be to create a driver which doesn't export the midMessage and modMessage, and to get rid of midi.c and midipatch.c (same for mixMessage/mixer.c and auxMessage/aux.c) (it seems midi.c and mixer.c are protected by non existent HAVE_LIBAUDIO_MIDI and HAVE_LIBAUDIOIO_MIXER which are defined nowhere) + so the driver should only be made of the wave.c part and oss.c (which should be renamed to IMO) + anyway, I think that the libaudioio subdir should be named against the name of the DLL, which is audioio (eventually, since it's a wine only DLL, and to prevent DLL name collisions, I'd call it wineaudioio(.drv) - I don't see any reasons to define HAVE_SOLARIS... this is a wrong practice for system dependancies anyway A+ -- --------------- Eric Pouech (http://perso.wanadoo.fr/eric.pouech/) "The future will be better tomorrow", Vice President Dan Quayle

1 0

Re: Patches for FreeBSD 5.x compilation
by Alexandre Julliard 20 Nov '01

20 Nov '01

Steve Kargl <sgk(a)troutmask.apl.washington.edu> writes: > Can you please explain how this is namespace pollution on the > part of the OS? AFAIK, <sys/user.h> is not specified by any > Standard (ISO C89, C90, or C99). <sys/user.h> can include > any (unprotected) variable name. Yes of course it can, but the fact that this isn't explicitly forbidden by some standard doesn't mean it's good practice or that it doesn't pollute the namespace. sys/user.h is also legally allowed to do "#define if else"; that doesn't make it a good idea. > If an application includes > <sys/user.h>, then the application should be prepared to > deal with conflicts. Great, then you need to fix all applications every time someone feels like exporting a new symbol from user.h. Of course user.h is a pretty obscure header so it's no big deal, but it would be easier to avoid exporting stuff under such common names as "struct thread". -- Alexandre Julliard julliard(a)winehq.com

4 4

interesting comment about WINE from MS
by Igor 20 Nov '01

20 Nov '01

As far as WINE and Mono are concerned, I'm just not sure whether these organizations will end up being able to produce satisfactory products. The real problem is this: it is not good enough for WINE to run some Windows apps. If the general public is going to consume it, it must run all Windows apps. Windows 2000/XP has a significant level of POSIX compliancy, but many developers just don?t write their apps with these standards in mind. You'll find pieces of code everywhere in third party apps which run statements like if (WINVER==4.0 && GrokTheBIOSForSomeACPIGoo()), which have not been tested to see whether they will run on non-Windows/x86 platforms. In fact, Microsoft has had to put in significant time and effort to insure that older Win32 applications originally written to run in Win95 will run in NT." original article is here http://www.osnews.com/story.php?news_id=314

1 0

Implementation of TB_REPLACEBITMAP
by Marco Bizzarri 20 Nov '01

20 Nov '01

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all. I'm trying to provide an implementation of TB_REPLACEBITMAP Included is the patch I've done to record the bitmaps which are provided to the toolbar and to replace them when the message arrives. Even though the number of buttons displayed is correct, the output is otherwise garbled. I would appreciate any suggestion on this topic. The platform is a RH Linux 6.2. wine is from CVS (20/11/2001) - -- Marco Bizzarri - Responsabile Tecnico - Icube S.r.l. Sede: Via Ridolfi 15 - 56124 Pisa (PI), Italia E-mail: m.bizzarri(a)icube.it WWW: www.icube.it Tel: (+39) 050 97 02 07 Fax: (+39) 050 31 36 588 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://quantumlab.net/pine_privacy_guard/ iD8DBQE7+oUXXhfyAQQVoaIRAudsAJ9QEbPYChk0YpTonozvkjUge4HtjACgl8cg A87v4vPHKOeeCgj6LpW18dw= =J2rq -----END PGP SIGNATURE-----

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

wine-devel