May 2010 - Wine-Devel - list.winehq.org

Re: [12/13]docs/wineusr: unix paths are supported

by Alexandre Julliard

Austin English <austinenglish(a)gmail.com> writes: >> diff --git a/en/wineusr-running.sgml b/en/wineusr-running.sgml >> index f36383f..106be37 100644 >> --- a/en/wineusr-running.sgml >> +++ b/en/wineusr-running.sgml >> @@ -90,8 +90,8 @@ Usage: wine PROGRAM [ARGUMENTS...] Run the specified program >> in the <parameter>Path</parameter> environment variable, you can >> simply give the executable file name. However, if the executable >> is not in <parameter>Path</parameter>, you must give the full path to >> - the executable (in Windows format, not UNIX format!). For >> - example, given a <parameter>Path</parameter> of the following: >> + the executable. For example, given a <parameter>Path</parameter> >> + of the following: >> </para> >> <screen> >> Path="c:\windows;c:\windows\system;e:\;e:\test;f:\" >> @@ -112,7 +112,11 @@ Path="c:\windows;c:\windows\system;e:\;e:\test;f:\" >> </screen> >> <para> >> (note the backslash-escaped "\" !) >> + or: >> </para> >> + <screen> >> +<prompt>$</prompt> <userinput>wine ~/.wine/drive_c/myapps/foo.exe</userinput> >> + </screen> >> <para> >> For details on running text mode (CUI) executables, read the >> <link linkend="CUI-programs">section</link> below. >> -- > > A lot of programs break when using Unix paths. I don't think we should > encourage it. There's no difference between specifying the path in Windows or Unix format. The current directory may be an issue, but that's true in both cases. -- Alexandre Julliard julliard(a)winehq.org

May 4, 2010

Re: [12/13]docs/wineusr: unix paths are supported

by Austin English

2010/5/4 André Hentschel <nerv(a)dawncrow.de>: > --- > en/wineusr-running.sgml | 8 ++++++-- > 1 files changed, 6 insertions(+), 2 deletions(-) > > diff --git a/en/wineusr-running.sgml b/en/wineusr-running.sgml > index f36383f..106be37 100644 > --- a/en/wineusr-running.sgml > +++ b/en/wineusr-running.sgml > @@ -90,8 +90,8 @@ Usage: wine PROGRAM [ARGUMENTS...] Run the specified program > in the <parameter>Path</parameter> environment variable, you can > simply give the executable file name. However, if the executable > is not in <parameter>Path</parameter>, you must give the full path to > - the executable (in Windows format, not UNIX format!). For > - example, given a <parameter>Path</parameter> of the following: > + the executable. For example, given a <parameter>Path</parameter> > + of the following: > </para> > <screen> > Path="c:\windows;c:\windows\system;e:\;e:\test;f:\" > @@ -112,7 +112,11 @@ Path="c:\windows;c:\windows\system;e:\;e:\test;f:\" > </screen> > <para> > (note the backslash-escaped "\" !) > + or: > </para> > + <screen> > +<prompt>$</prompt> <userinput>wine ~/.wine/drive_c/myapps/foo.exe</userinput> > + </screen> > <para> > For details on running text mode (CUI) executables, read the > <link linkend="CUI-programs">section</link> below. > -- A lot of programs break when using Unix paths. I don't think we should encourage it. Perhaps instead put an example using wine start /unix. -- -Austin

May 4, 2010

Stack corruption in dsound.c

by wylda＠volny.cz

Hi, for those of readers who don't know yet, i'm a "Hello word" programmer, so sorry for bad wording or speculation etc. I decided to look into a regression since wine 1.1.30, who nobody apparently not know what to do, otherwise it would be solved ;) The game freezes during sound initialization and i tracked it down by printf debugging to heap allocation in DirectSoundDevice_Create(). Game goes: ... DirectSoundDevice_Create() //HeapAlloc OK ... OpenAL Init ... DirectSoundDevice_Create() //HeapAlloc returns insane values ... fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5e70000 fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5a60000 fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5a40000 fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5950000 fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5940000 fixme:dbghelp:validate_addr64 Unsupported address ffffffffb5920000 Frozen This happens only when builtin OpenAL is used. In this case HeapAlloc gives values always beginning with 0x234vwxyz. Builtin OpenAL example (*** BAD - Game freezes ***): device = HeapAlloc(GetProcessHeap(),HEAP_ZERO_MEMORY,sizeof(DirectSoundDevice)); First call of DirectSoundDevice_Create(): 0x0014f908 = HeapAlloc(0x00110000, 0x00000008, 0x000003d4) Second call of DirectSoundDevice_Create(): 0x234228e0 = HeapAlloc(0x00110000, 0x00000008, 0x000003d4) Native OpenAL example(*** OK - Game runs***): First call of DirectSoundDevice_Create(): 0x0014fa30 = HeapAlloc(0x00110000, 0x00000008, 0x000003d4) Second call of DirectSoundDevice_Create(): 0x001566c0 = HeapAlloc(0x00110000, 0x00000008, 0x000003d4) So how do i find out, why HeapAlloc does not return sane values like in first call of DirectSoundDevice_Create()?

May 4, 2010

Re: gdi32: Fix underline and strikeout for clipped to the rectangle text

by Alexandre Julliard

Ilya Shpigor <shpigor(a)etersoft.ru> writes: > There is bug with underline and strikeout font in ExtTextOutW. > > The text is clipped if the ETO_CLIPPED flag have been set and rectangle > smaller then needed for text. But underline and strikeout lines is drawed > same as when text fit in the rectangle. > > The problem screenshot (wine below): > http://bugs.etersoft.ru/attachment.cgi?id=1678 > > I suggest to check if the y coordinate of line is smaller then text rectangle > bottom. That's of course not sufficient for correct clipping. -- Alexandre Julliard julliard(a)winehq.org

May 4, 2010

Re: What is the point of mincore() in reserve_area() in libs/wine/mmap.c?

by Charles Davis

On 5/4/10 10:17 AM, Johann "Myrkraverk" Oskarsson wrote: > Hi all, > > In the Solaris/BSD implementation of reserve_area() in libs/wine/mmap.c there is > > for (i = 0; i < size; i += pagesize) > if (mincore( (caddr_t)addr + i, pagesize, &vec ) != -1) break; > > &vec is never used again in that function. What is the point of mincore() ? It's trying to see if the address is valid. From the manpage: RETURN VALUES Upon successful completion, a value of 0 is returned. Otherwise, a value of -1 is returned and errno is set to indicate the error. ERRORS [EINVAL] addr is not a valid address for user mode. [EFAULT] an error occurred trying to copy to the output character array vec. vec is to make sure it failed because the address was invalid, not because it couldn't write the flags out. Chip

May 4, 2010

What is the point of mincore() in reserve_area() in libs/wine/mmap.c?

by Johann "Myrkraverk" Oskarsson

Hi all, In the Solaris/BSD implementation of reserve_area() in libs/wine/mmap.c there is for (i = 0; i < size; i += pagesize) if (mincore( (caddr_t)addr + i, pagesize, &vec ) != -1) break; &vec is never used again in that function. What is the point of mincore() ? I'm looking at this in reference to bug 22033. Johann

May 4, 2010

Valgrind's wine_cp_wcstombs warnings

by Joerg-Cyril.Hoehle＠t-systems.com

Hi, Dan Kegel wrote: >did you try the --track-origins option? That looks like an exact fit, thank you. Well, this option is not present in Ubuntu Intrepid's Valgrind. In Ubuntu Lucid, the option is present but Valgrind does not complain about wine_cp_wcstombs there. I conclude it was likely a false alarm due to a bug in old Valgrind. Regards, Jörg Höhle

May 4, 2010

Re: Valgrind's wine_cp_wcstombs warnings

by Jordan Ayers

> Hi, > > running the MCI tests with Valgrind generates a lot of output like follows > > ==13170== Use of uninitialised value of size 4 > ==13170== at 0x4035369: wine_cp_wcstombs (wctomb.c:147) > ...[every line from 148 to 161] > ==13170== Use of uninitialised value of size 4 > ==13170== at 0x403550A: wine_cp_wcstombs (wctomb.c:162) > > I suspect the A<->W mapping is still not fully correct (I know at > least one more case, but I'm not sure it gets triggered by the tests) > but perhaps it's also the W->UNIX.UTF-8 conversions internal to Wine > which cause these hits. How do I tell? > > Is there any way to get more precise reporting for the above messages > (e.g. a backtrace)? The present output is not helpful at identifying > the origin of the uninitialised read. > > Perhaps Valgrind is broken? Why does it mention "size 4" when all > access in the offending function is either "char *" or "short *" > according to the source > http://source.winehq.org/source/libs/wine/wctomb.c#L145 ? > > Thanks for your help, > Jörg Höhle Joerg, If you're on a 32-bit system, your pointers would be where the 'size 4' is coming from. It looks like Valgrind thinks a pointer being dereferenced isn't initialized. (maybe src?) Jordan Ayers

May 4, 2010

Re: [3/3] rundll32: Recognize entry points passed as ordinal numbers. (try 2)

by Alexandre Julliard

Andrew Nguyen <anguyen(a)codeweavers.com> writes: > @@ -119,30 +119,43 @@ static FARPROC16 get_entry_point16( HINSTANCE16 inst, LPCWSTR entry ) > static void *get_entry_point32( HMODULE module, LPCWSTR entry, BOOL *unicode ) > { > void *ret; > - DWORD len = WideCharToMultiByte( CP_ACP, 0, entry, -1, NULL, 0, NULL, NULL ); > - char *entryA = HeapAlloc( GetProcessHeap(), 0, len + 1 ); > - > - if (!entryA) > - return NULL; > > - WideCharToMultiByte( CP_ACP, 0, entry, -1, entryA, len, NULL, NULL ); > + /* determine if the entry point is an ordinal */ > + if (entry[0] == '#') > + { > + int ordinal = atoiW( entry + 1 ); > + if (ordinal <= 0) > + return NULL; You don't set the unicode flag in this case. -- Alexandre Julliard julliard(a)winehq.org

May 4, 2010

Trying again

by Joakim Tjernlund

This didn't show up in the archives, so trying again. >From 7882df972ec827f0e7b7b622c01be2644e54d122 Mon Sep 17 00:00:00 2001 From: Joakim Tjernlund <Joakim.Tjernlund(a)transmode.se> Date: Mon, 3 May 2010 20:30:32 +0200 Subject: [PATCH] RegisterWaitForSingleObject() fix special tmo == 0 case. At http://msdn.microsoft.com/en-us/library/aa332406%28VS.71%29.aspx one can read: "If the timeOutInterval parameter is not zero (0) and the executeOnlyOnce parameter is false, the timer is reset every time the event is signaled or the time-out interval elapses." timeOutInterval == 0 makes the above false and the timer should NOT be reset, fix that. Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund(a)transmode.se> --- dlls/ntdll/threadpool.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/dlls/ntdll/threadpool.c b/dlls/ntdll/threadpool.c index 44afb01..f6c9fb4 100644 --- a/dlls/ntdll/threadpool.c +++ b/dlls/ntdll/threadpool.c @@ -377,7 +377,8 @@ static DWORD CALLBACK wait_thread_proc(LPVOID Arg) wait_work_item->Callback( wait_work_item->Context, TimerOrWaitFired ); wait_work_item->CallbackInProgress = FALSE; - if (wait_work_item->Flags & WT_EXECUTEONLYONCE) + if (!wait_work_item->Milliseconds || + wait_work_item->Flags & WT_EXECUTEONLYONCE) break; } else -- 1.6.4.4

May 4, 2010