- Wine-Devel - list.winehq.org

Re: RegSetValueEx problem?
by Alexandre Julliard April 11, 2001

April 11, 2001

davep <davep(a)cyw.uklinux.net> writes: > I have now been able to build and run WineMine under NT4, and it works > perfectly. On exit each parameter is saved to the registry as 16 bytes including > the training garbage (not visible in Regedit of course) and when restarted the > exact same 16 bytes are reloaded from the registry. > > It does therefore seem that there is something wrong with Wne's implementation. No, it depends if data[sizeof(data)] is null or not; the +1 on the size is done only if the last byte is null, on the assumption that the programmer forgot to count it (i.e. used strlen(data) instead of strlen(data)+1). It's the usual Microsoft approach of hiding bugs instead of fixing them, and it is definitely what NT does (or we wouldn't be doing such a monstrosity). Now what can happen is that the Microsoft compiler lays out the stack in a different way than gcc, and that the variable that is just above the array on the stack doesn't contain 0 in this case. In fact I suspect that if you run your Windows binary of WineMine under Wine it will work fine too. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: Traces: fs -> tid
by Alexandre Julliard April 11, 2001

April 11, 2001

Francois Gouget <fgouget(a)free.fr> writes: > Yes, but it would also be the simplest implementation (overdesign can > be ugly too). And if it's the one off hack that it current looks like it > is, why not. > I propose to just let Alexandre decide which implementation, if any, > he thinks would be worth including in CVS. I think we have now wasted enough bandwidth on this non-issue. The relay traces have always displayed the thread info, and will continue to do so. People who are morally offended by the extra 5 characters needed for the tid can hack their local tree, or learn to use sed/awk/perl. -- Alexandre Julliard julliard(a)winehq.com

1 0

Re: edit control patch
by Ian Pilcher April 11, 2001

April 11, 2001

Dan Engel wrote: > > This is my first attempt at submitting a patch--please someone let me know if there are "customs" I need to follow. > Word-wrap is nice. :-) -- ======================================================================== Ian Pilcher ian.pilcher(a)home.com ========================================================================

1 0

Re: RegSetValueEx problem?
by Alexandre Julliard April 11, 2001

April 11, 2001

gerard patel <gerard.patel(a)asi.fr> writes: > I have browsed server/registry.c and it seems that the registry > stores a terminating null for strings; so if you have in the registry > file a string like "0123456789", it get in memory a length of 11. > So I guess that the fix would be to fix the length by -1 somewhere > in the path server->ntdll->advapi No, the current behavior is correct for NT; what you want is a version check to get the Win95 behavior. Something like the patch below should do it (and of course WineMine should be fixed to work on NT too). Index: dlls/advapi32/registry.c =================================================================== RCS file: /opt/cvs-commit/wine/dlls/advapi32/registry.c,v retrieving revision 1.24 diff -u -r1.24 registry.c --- dlls/advapi32/registry.c 2001/03/23 19:12:01 1.24 +++ dlls/advapi32/registry.c 2001/04/09 20:49:35 @@ -640,8 +640,12 @@ { UNICODE_STRING nameW; - if (count && is_string(type)) + if (GetVersion() & 0x80000000) /* win95 */ { + if (type == REG_SZ) count = strlenW( (WCHAR *)data ) + sizeof(WCHAR); + } + else if (count && is_string(type)) + { LPCWSTR str = (LPCWSTR)data; /* if user forgot to count terminating null, add it (yes NT does this) */ if (str[count / sizeof(WCHAR) - 1] && !str[count / sizeof(WCHAR)]) @@ -663,7 +667,11 @@ WCHAR *dataW = NULL; NTSTATUS status; - if (count && is_string(type)) + if (GetVersion() & 0x80000000) /* win95 */ + { + if (type == REG_SZ) count = strlen(data) + 1; + } + else if (count && is_string(type)) { /* if user forgot to count terminating null, add it (yes NT does this) */ if (data[count-1] && !data[count]) count++; -- Alexandre Julliard julliard(a)winehq.com

2 1

VxD information
by Maurice van der Pot April 11, 2001

April 11, 2001

I'd like to get started on VxD support, but I'm unable to find any useful documentation on how it's implemented in Wine. I've been looking through the sources (e.g. vxd.c, wprocs.spec) but I can't find the connection between a call to int 2F and a call to the functions in vxd.c. Regards, Maurice van der Pot.

1 0

Re: Traces: fs -> tid
by Alexandre Julliard April 10, 2001

April 10, 2001

Francois Gouget <fgouget(a)free.fr> writes: > But maybe we could ask for traces generated with +relay,+tid instead > of just +relay. > Also when you're working on a specific application for which you have > determined that threads don't play a role, then you would have the > option of omitting the +tid to get simpler traces. I disagree, I think the thread information is important even if you are not chasing a thread problem, simply to tell you which calls are part of the same sequence. It's not possible to make sense of any relay trace if the calls made from different threads cannot be distinguished. If the tid really disturbs you, you can always pipe the output into a sed 's/tid=.*//' or something similar. -- Alexandre Julliard julliard(a)winehq.com

3 2

Traces: fs -> tid
by Francois Gouget April 10, 2001

April 10, 2001

I noticed that fs was replaced by tid in relay (and snoop & co too now). I propose a slightly different version based on the following premises: * when suspecting a threading problem it could be nice to tag each trace with the id of the thread that emits it * when there's no thread issue, then fs and tid are unnecessary and only clutter the traces by making lines longer and harder to read * I find that lines are more readable if the thread id comes first as it's always the same length and doesn't mess up the alignment of the other fields So this patch adds a debug channel called 'tid' which activates printing the tid as the first field of all traces. Actually, it might make sense to merge 'tid' with the 'thread' debug channel. 'Screenshot': 08063178:Call user32.DispatchMessageA(405b6e50) ret=00402135 08063178:trace:relay:WINPROC_CallWndProc (wndproc=0x401c74,hwnd=00000138,msg=WM_SYSCOMMAND,wp=0000f060,lp=00000000) 08063178:Call user32.DefWindowProcA(00000138,00000112,0000f060,00000000) ret=0040206c Comments? Changelog: * dlls/ntdll/ntdll.spec Declare the tid debug channel * dlls/ntdll/debugtools.c, relay32/relay386.c, relay32/snoop.c Output the current thread id as the first field if tid is turned on -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ La terre est une bêta...

2 3

Re: [PATCH] Add Callback for wavemapper -> widOpen
by Alexandre Julliard April 9, 2001

April 9, 2001

"James Hatheway" <james(a)macadamian.com> writes: > I implemented the callback in a very similar way as the Wave out callback > is implemented. You can see this in the functions wodOpen(), wodOpenHelper(), > and WAVEMAP_DstCallback() , look at the data passed in the wom structure > (in dlls/winmm/wavemap/wavemap.c). I don't see any other way that I will > be able to callback into the app when the WineOSS layer notifies the wavemap > layer. The way it is now (wineoss directly calling the app) > causes my app to crash, and doesn't work that way on waveout. But these other functions are not using the WINE_MLD internal winmm structure, so why do you need it in widOpen()? Couldn't you instead do exactly what wodOpen() does? -- Alexandre Julliard julliard(a)winehq.com

3 5

Re: RegSetValueEx problem?
by gerard patel April 8, 2001

April 8, 2001

At 10:54 AM 08/04/2001 +0100, you wrote: > >I've found time to do some more digging, but I'm still stumped. <snip> I have browsed server/registry.c and it seems that the registry stores a terminating null for strings; so if you have in the registry file a string like "0123456789", it get in memory a length of 11. So I guess that the fix would be to fix the length by -1 somewhere in the path server->ntdll->advapi I have tried this but I am not sure at all this is correct : --- registry.c.orig Mon Mar 5 20:34:21 2001 +++ registry.c Sun Apr 8 18:11:39 2001 @@ -799,14 +799,17 @@ struct key_value *value; int index; size_t ret = 0; + size_t l; if ((value = find_value( key, name, &index ))) { *type = value->type; - *len = value->len; - if (value->data && offset < value->len) + l = value->len; + if ((value->type == REG_SZ) && l) l--; + *len = l; + if (value->data && offset < l) { - if (maxlen > value->len - offset) maxlen = value->len - offset; + if (maxlen > l - offset) maxlen = l - offset; memcpy( data, (char *)value->data + offset, maxlen ); ret = maxlen; } Gerard

2 1

Wrc question
by gerard patel April 8, 2001

April 8, 2001

I have looked at wrc code and there is a strange 100K limit on the size of mem allocation : assert(size > 0); assert(size < 102400); res = malloc(size); if(res == NULL) { Baffling code - hardcoded limit hidden in a C file :-/ I removed the assert and wrc was able to handle a 1 megabyte bitmap. Anyone has an idea about the meaning of this limit ? Gerard

2 1