April 2004 - Wine-Devel - list.winehq.org

winealsa fails the winmm test
by Francois Gouget April 27, 2004

April 27, 2004

I'm not having much luck with sound this week-end :-( First there was the crash in ALSA_WaveInit() (an old bug apparently), but now the winmm test either gets stuck or crashes. When it gets stuck it's because I get: trace:wave:wodUpdatePlayedTotal dwPlayedTotal=21987 dwWrittenTotal=22050 delay=63 trace:wave:wodPlayer_NotifyCompletions still playing 0x4069fd1c (22050/21987) trace:wave:wodPlayer waiting 2ms (7,2) trace:wave:wodUpdatePlayedTotal dwPlayedTotal=21987 dwWrittenTotal=22050 delay=63 ... So the problem is that everything has been played but because snd_pcm_delay() keeps returning 63, dwPlayedTotal never reaches the 22050 that would signal the completion of the first tone. Now it seems like snd_pcm_delay() should return 0 after a while so this looks like an Alsa bug (0.9.6). Is this a known bug? Is there a way to work around it? (besides commenting out the call to snd_pcm_delay()) And when I don't get into the above infinite loop it's because I've crashed before with an Alsa assertion: trace:wave:wodPlayer waiting 7ms (7,4294967295) trace:wave:wodUpdatePlayedTotal dwPlayedTotal=11781 dwWrittenTotal=17184 delay=5403 trace:wave:wodPlayer_WriteMaxFrags Writing wavehdr 0x4069fd1c.17184[22050] trace:wave:wodPlayer_WriteMaxFrags 1065 dwLength=2433 toWrite=110 0010:Call ntdll.RtlAcquireResourceShared(4037c168,00000001) ret=404e0751 0010:Ret ntdll.RtlAcquireResourceShared() retval=00000001 ret=404e0751 0010:Call ntdll.RtlReleaseResource(4037c168) ret=404e08ae 0010:Ret ntdll.RtlReleaseResource() retval=00000000 ret=404e08ae 0010:Call ntdll.RtlTryEnterCriticalSection(40340098) ret=404e0519 0010:Ret ntdll.RtlTryEnterCriticalSection() retval=00000001 ret=404e0519 0010:Call ntdll.RtlTryEnterCriticalSection(40340098) ret=404e0519 0010:Ret ntdll.RtlTryEnterCriticalSection() retval=00000001 ret=404e0519 0010:Call ntdll.RtlTryEnterCriticalSection(40340098) ret=404e0519 0010:Ret ntdll.RtlTryEnterCriticalSection() retval=00000001 ret=404e0519 wine-kthread: pcm.c:5899: snd_pcm_mmap_commit: Assertion `frames <= snd_pcm_mmap_avail(pcm)' failed. The interesting things are that wwo->write points to snd_pcm_mmap_writei(). This assertion would seem to imply that we are writing too much stuff at once? -- Francois Gouget fgouget(a)free.fr http://fgouget.free.fr/ The greatest programming project of all took six days; on the seventh day the programmer rested. We've been trying to debug the *&^%$#@ thing ever since. Moral: design before you implement.

2 1

2 Fatal Problems With Winesetuptk-0.73 (BOTH Sourceforge Release Formats)
by Kenny Love April 27, 2004

April 27, 2004

In the Bugzilla system, including the search page, there is no mention of Winesetuptk, so I can't tell whether the following bugs are new to you. These two problems are of considerable import to aspiring Wine users, particularly if they are running Red Hat Linux 6.2 or equivalent or older (for the first item, Red Hat Linux 7.x or equivalent might also be affected). 1. The rpm'ed Winesetuptk binary currently posted on the Sourceforge site, winesetuptk-0-73.i586.rpm, aborts on startup on a Red Hat Linux 6.2 system with the message "winesetuptk: /lib/libc.so.6: version `GLIBC_2.3' not found (required by winesetuptk)". I suspect this means that the binary (maybe the Tcl/Tk/ITcl) is dynamically linked to glibc 2.3, but Red Hat Linux 6.2 can only be easily upgraded to glibc 2.1.3-21; even Red Hat Linux 7.3 easily upgrades only to glibc 2.2.5-44, so this might prohibit most Red Hat Linux installations older than 8.0 from running this binary. 2. The scripts in the Winesetuptk tarball currently posted on the Sourceforge site, /winesetuptk-0-73.tar.bz2, hang. When run "as scripts directly" on a Red Hat Linux 6.2 system with Tcl 8.0.5-35 and ITcl 3.0.1-35 installed, after the program opens a window stating "Autodetection in Progress", an error notice appears stating that map is not a legal argument to the string function ("Error: bad option "map": must be compare, first, index,"...). After the notice's "Ok" button is pressed, the error notice box disappears, but the program quietly hangs, ignoring attempts to close the remaining windows. Bash's kill command, probably unknown to the novice, is required to exit the comatose program. Also, on the launching shell window appears the message "CBase::AutoConf:GetWinInstalls: unable to grep /dos/c/msdos.sys: child process exited abnormally ". My /dos/c/ directory contains my dual-boot Windows 3.1 system. In the tcl8.3.2 subdirectory, the "changes" file lists, between the "8.1 final" and "8.1.1" headers, a "new features" entry dated 5/3/99 mentioning adding Jeff Hobbs' patches introducing, among others, a map argument to the string function. The Sourceforge Winesetuptk-0.73 release note states that at version 0.34b, all was successfully converted from requiring Tcl 8.4 to using Tcl 8.0, so this map call appears to be a small overlooked mouse with a show-stopping roar. I lack Tcl programming experience, but the stack trace (see below) and 'grep'-ing the source tree for potential "string_ map" calls seems to implicate "winesetuptk-0.73/Enable8.0Compatibility.tcl", line 42 as the culprit; it looks like there's no if-clause to handle $aCmd == "map". Here's the stack trace: bad option "map": must be compare, first, index, last, length, match, range, tolower, toupper, trim, trimleft, trimright, wordend, or wordstart while executing "string_ map {{${WinDir}} windows} {${WinDir}/system}" ("eval" body line 1) invoked from within "eval string_ $aCmd [list $aStr] $args" (procedure "string" line 26) invoked from within "string map [list {${WinDir}} $WinDir] $SystemDir" (object "::ObjBase" method "::CBase::AutoConf:GetWinInstalls" body line 74) invoked from within "AutoConf:GetWinInstalls $Drives" (object "::ObjBase" method "::CBase::AutoConf" body line 39) invoked from within "$pObjBase AutoConf" ("GEN_NEW" arm line 6) invoked from within "switch [$pObjBase AccessCfgData GET_KEY_VALUE_BY_NAME Cfg WhatToDo] { GEN_NEW { CMsgDialog ObjWaitMsg . .wait_msg "WineSetup" "Autodetection..." (object "::ObjScrFileLoc" method "::CScrFileLoc::SetWineData" body line 2) invoked from within "SetWineData" (object "::ObjScrFileLoc" method "::CScrFileLoc::SetResult" body line 58) invoked from within "::ObjScrFileLoc SetResult NEXT" (in namespace inscope "::CBaseScr" script line 1) invoked from within "namespace inscope ::CBaseScr {::ObjScrFileLoc SetResult NEXT}" invoked from within ".f_scr.b_next invoke" ("uplevel" body line 1) invoked from within "uplevel #0 [list $w invoke]" (procedure "tkButtonUp" line 8) invoked from within "tkButtonUp .f_scr.b_next" (command bound to event) I'm going to try compiling the Winesetuptk-0.73 source from the tarball on my system; hopefully that will produce a >working< binary dynamically linked to glibc 2.1.3-21. Considering that most aspiring Wine users, (Remember: Many, perhaps most of those are new to Linux itself.), are currently steered by the Wine documentation to use the Winesetuptk program with its nice, friendly GUI, those who took the time-honored route of putting a Linux installation from second-hand CD's on an older, expendable PC (which often limits one to a 6.2/7.0 version or less because 2.4 kernels can thrash too much on old hardware to be useful) may be in for a big suprise when they try to graphically configure their wine.conf file. To summarize, the rpm binary release as-is probably won't start for Linux versions older than Red Hat Linux 8.0 or equivalent and the tarball release as-is hangs deviously when run "as scripts directly" on systems with Tcl 8.0, possibly 8.1.0. By the way, that rpm should have refused to install because the program required a higher version glibc than my Linux system possesses, shouldn't it? Thanks! Kenny A life spent making mistakes is not only more honorable, but more useful than a life spent doing nothing. G. B. Shaw

1 0

Alignment testing
by Hans Leidekker April 27, 2004

April 27, 2004

Hi, kernel/tests/generated.c currently shows one failure when the executable is compiled with MinGW. Investigation shows that MinGW and GCC differ when it comes to alignment of doubles (i.e. a 64 bit wide type). I have constructed the test below that shows the difference between MinGW and GCC. My question is, could someone compile this with MSVC >= 6, run it, and report the outcome here? Thanks, -Hans

3 4

GetDriveTypeW crash
by Robert Reif April 27, 2004

April 27, 2004

I'm having problems with several programs crashing in GetDriveTypeW. I recently did a fresh install from source with tools/wineinstall so it may be a config problem. Any ideas where to start? Unhandled exception: page fault on read access to 0x00000000 in 32-bit code (0x404944b0). In 32-bit mode. 0x404944b0 (GetDriveTypeW+0xac [volume.c:110] in KERNEL32.DLL): movw 0x0(%edi),%ax 111 { Wine-dbg>bt Backtrace: =>0 0x404944b0 (GetDriveTypeW+0xac(root=0x0) [volume.c:110] in KERNEL32.DLL) (ebp=4081e8b8) 1 0x40493294 (GetVolumeInformationW+0x300(root=0x0, label=0x0, label_len=0x0, serial=0x0, filename_len=0x4081fb34, flags=0x4081fb38, fsname=0x0, fsname_len=0x0) [volume.c:683] in KERNEL32.DLL) (ebp=4081f7b8) 2 0x40493551 (GetVolumeInformationA+0x91(root=0x0, label=0x0, label_len=0x0, serial=0x0, filename_len=0x4081fb34, flags=0x4081fb38, fsname=0x0, fsname_len=0x0) [volume.c:780] in KERNEL32.DLL) (ebp=4081f7fc) 3 0x01004012 (PPView97.exe..text+0x3012 in PPView97.exe) (ebp=4081fb48) 4 0x01004c0c (PPView97.exe..text+0x3c0c in PPView97.exe) (ebp=4081fc8c) 5 0x0100577f (PPView97.exe..text+0x477f in PPView97.exe) (ebp=4081fdb0) 6 0x01005970 (PPView97.exe..text+0x4970 in PPView97.exe) (ebp=4081fec4) 7 0x01005aa8 (PPView97.exe..text+0x4aa8 in PPView97.exe) (ebp=4081ff2c) 8 0x404773ca (start_process+0xf2(arg=0x0) [process.c:785] in KERNEL32.DLL) (ebp=4081fff4) 9 0x4002921d (wine_switch_to_stack+0x11 in libwine.so.1) (ebp=00000000)

2 1

Re: Make dlls-kernel-tests-file.c work on Windows 95 and XP.
by Alexandre Julliard April 27, 2004

April 27, 2004

Jakob Eriksson <jakov(a)vmlinux.org> writes: > +static int test_win9x_based() > +{ > + OSVERSIONINFOEX ver; > + > + ver.dwOSVersionInfoSize = sizeof(OSVERSIONINFO); > + ok( GetVersionEx ((OSVERSIONINFO *) &ver), " --- Can't get OS version.\n" ); > + if ((4 == ver.dwMajorVersion && 0 == ver.dwMinorVersion && 1381 == ver.dwBuildNumber) > + || ver.dwMajorVersion >= 5) > + { > + return 0; > + } > + > + return 1; You should not do version checks in tests, the checks should be based on the features we are testing. Otherwise the results are going to depend on the Windows version that Wine is configured for, which is the wrong thing to do since the behavior of the Wine code in that case does not depend on the version. -- Alexandre Julliard julliard(a)winehq.org

2 3

Hmm.. something strange in winetest reporting
by Jakob Eriksson April 26, 2004

April 26, 2004

Check out http://test.winehq.org/data/20040425-1000/ OSR2 shell32:string. It says "0", but if you click, there seems to be some kind of error. Is it the test output somehow erratic, or is the log parser doing something bad? regards, Jakob

2 1

Re: msvcrt: compatibility fixes
by Alexandre Julliard April 26, 2004

April 26, 2004

"Dimitrie O. Paun" <dpaun(a)rogers.com> writes: > ChangeLog > Fix the _WCHAR_T_DEFINED/_WCHAR_T_DEFINED sentries for > compatibility with the MSVCRT ones. Unfortunately _WCHAR_T_DEFINED is also used by gcc in stddef.h. -- Alexandre Julliard julliard(a)winehq.org

2 3

Re: String constification
by Dmitry Timoshkov April 26, 2004

April 26, 2004

"Francois Gouget" <fgouget(a)free.fr> wrote: > We find the same issues with an added twist: now that we can use > literals we can write things like: > > const char* str = "String literal"; > > However this is slightly different from: > > static const char* str[] = "String literal"; > > The difference is that in the first case sizeof(str) returns 4 (on 32 > bit machines<g>) while in the latter we get 15. Some of our code relies > on this so care must be taken when converting the latter to the former. My understanding of this is that in the first case you declare a const pointer and hope that a compiler is smart enough to place a pointed by it object into a read only section, while in the second case you explicitly say to a compiler to place the whole object into a read only section. I prefer to not rely on a compiler's good will. -- Dmitry.

4 5

Re: RPCRT4: make the on-wire protocol Windows compatible
by Steven Edwards April 26, 2004

April 26, 2004

Hiya Filip, You should leave this part out. I dont think Alexandre will remove it for you. +#ifndef __REACTOS__ __TRY { - if (func) func(&msg); + if (func) func(msg); } __EXCEPT(rpc_filter) { /* failure packet was created in rpc_filter */ } __ENDTRY +#else + if (func) func(msg); +#endif Thanks Steven __________________________________ Do you Yahoo!? Yahoo! Photos: High-quality 4x6 digital prints for 25� http://photos.yahoo.com/ph/print_splash

1 0

Re: Remove debugmsg from guides
by Ivan Leo Murray-Smith April 26, 2004

April 26, 2004

> Please don't do that, we've been over this a thousand times, > let's not rehash it. Do either or, otherwise you end up > confusing Alexandre's script, etc. Inlining is prefered > for a lot of reasons, so if you're using a decent mail client > (and you seem to be), just inline it and don't look back :) I'm not using an email client, my ESP only allows that if I connect via their dial-up service. That's why I can't have text/plain attachments. Ivan.

1 0