[Bug 36977] New: Magic: The Gathering – Duels of the Planeswalkers 2015 crashes after few frames of intro sequence
http://bugs.winehq.org/show_bug.cgi?id=36977 Bug ID: 36977 Summary: Magic: The Gathering – Duels of the Planeswalkers 2015 crashes after few frames of intro sequence Product: Wine Version: 1.7.21 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: blocker Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: otigkd(a)gmail.com Created attachment 49100 --> http://bugs.winehq.org/attachment.cgi?id=49100 crash log os: Gentoo x86 installed trough steam. a few seconds of loding/intro is played, then crashes -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=36977 Rosanne DiMesio <dimesio(a)earthlink.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Severity|blocker |normal --- Comment #1 from Rosanne DiMesio <dimesio(a)earthlink.net> --- Not a blocker.
From the log:
PE 1080000- 1140000 Deferred gameoverlayrenderer ------- Disable gameoverlayrenderer. This is probably a duplicate of bug 35361. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=36977 Anastasius Focht <focht(a)gmx.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW URL| |http://store.steampowered.c | |om/app/255420/ CC| |focht(a)gmx.net Component|-unknown |ole32 Summary|Magic: The Gathering – |Magic: The Gathering – |Duels of the Planeswalkers |Duels of the Planeswalkers |2015 crashes after few |2015 (Steam) crashes after |frames of intro sequence |few frames of intro | |sequence (multi-thread race | |during XAudio2 tear down) Ever confirmed|0 |1 --- Comment #2 from Anastasius Focht <focht(a)gmx.net> --- Hello folks, confirming, can be reproduced with the demo version as well. Steam game overlay is not the problem here. --- snip --- $ WINEDEBUG=+tid,+seh,+relay,+snoop,+dsound wine ./steam.exe -applaunch 255440 -no-dwrite ... 006d:CALL bink2w32._BinkOpenXAudio2(a)4(<unknown, check return>) ret=0108e6ea 006d:Call ole32.CoInitializeEx(00000000,00000000) ret=01098d54 ... 006d:Call ole32.CoCreateInstance(010cb2ac,00000000,00000001,010cb2bc,0827d5cc) ret=0108102c ... 006d:Call KERNEL32.LoadLibraryExW(0827d16e L"C:\\windows\\system32\\XAudio2_7.dll",00000000,00000008) ret=7e0c188c 006d:Ret KERNEL32.LoadLibraryExW() retval=08a10000 ret=7e0c188c ... 006d:CALL xaudio2.DllGetClassObject(<unknown, check return>) ret=7e0c3e99 006d:RET xaudio2.DllGetClassObject(0827d4e8,7e1b3eec,0827d4f8) retval=00000000 ret=7e0c3e99 ... 006d:Call KERNEL32.CreateThread(00000000,00000000,08a4a670,09599050,00000004,00000000) ret=08a4a372 006d:Ret KERNEL32.CreateThread() retval=0000026c ret=08a4a372 ... 0077:Starting thread proc 0x8a4a670 (arg=0x9599050) ... 006d:RET bink2w32._BinkOpenXAudio2(a)4(00000000) retval=01098bb0 ret=0108e6ea 006d:RET bink2w32._BinkSetSoundSystem(a)8(003b0363,00000000) retval=00000001 ret=008e5faf 006d:CALL bink2w32._BinkSetSoundTrack(a)8(<unknown, check return>) ret=008e5fc8 006d:RET bink2w32._BinkSetSoundTrack(a)8(00000001,11040438) retval=00000001 ret=008e5fc8 006d:CALL bink2w32._BinkSetIO(a)4(<unknown, check return>) ret=008e5ff2 006d:RET bink2w32._BinkSetIO(a)4(00b18900) retval=00b18900 ret=008e5ff2 ... 006d:RET bink2w32._BinkOpen(a)8(0827d884,02004400) retval=00000000 ret=008e6015 0077:trace:dsound:IDirectSoundBufferImpl_GetCurrentPosition (0x11100780,0xf2be944,0xf2be93c) ... 0077:trace:dsound:IDirectSoundBufferImpl_GetCurrentPosition playpos = 0, writepos = 1764, buflen=52920 (0x11100780, time=8160876) ... 006d:CALL bink2w32._BinkGetError(a)0(<unknown, check return>) ret=008e601e 006d:RET bink2w32._BinkGetError(a)0() retval=010c0618 ret=008e601e 006d:CALL bink2w32._BinkClose(a)4(<unknown, check return>) ret=004e2c51 006d:RET bink2w32._BinkClose(a)4(00000000) retval=0827d88c ret=004e2c51 006d:Call KERNEL32.GetLastError() ret=005abf07 006d:Ret KERNEL32.GetLastError() retval=000000cb ret=005abf07 ... 0077:trace:dsound:IDirectSoundBufferImpl_GetCurrentPosition (0x11100780,0xf2be944,0xf2be93c) ... 006d:Ret PE DLL (proc=0xf5ee4a,module=0xe90000 L"fmodex.dll",reason=THREAD_DETACH,res=(nil)) retval=1 006d:Call PE DLL (proc=0x7e1a0258,module=0x7e0a0000 L"ole32.dll",reason=THREAD_DETACH,res=(nil)) 006d:CALL xaudio2.DllCanUnloadNow(<unknown, check return>) ret=7e0c3384 006d:RET xaudio2.DllCanUnloadNow() retval=00000000 ret=7e0c3384 006d:Call KERNEL32.FreeLibrary(08a10000) ret=7e0c1c03 006d:Call PE DLL (proc=0x8a41ce5,module=0x8a10000 L"XAudio2_7.dll",reason=PROCESS_DETACH,res=(nil)) ... 006d:Ret PE DLL (proc=0x8a41ce5,module=0x8a10000 L"XAudio2_7.dll",reason=PROCESS_DETACH,res=(nil)) retval=1 ... 006d:Ret KERNEL32.FreeLibrary() retval=00000001 ret=7e0c1c03 ... 0077:Call KERNEL32.GetTickCount() ret=7d03db34 0077:Ret KERNEL32.GetTickCount() retval=007c866c ret=7d03db34 0077:trace:dsound:IDirectSoundBufferImpl_GetCurrentPosition playpos = 0, writepos = 1764, buflen=52920 (0x11100780, time=8160876) 0077:Call KERNEL32.WaitForMultipleObjects(00000003,09599068,00000000,0000000a) ret=08a4a5e0 0077:Ret KERNEL32.WaitForMultipleObjects() retval=00000102 ret=08a4a5e0 0077:trace:seh:raise_exception code=c0000005 flags=0 addr=0x8a4a5e0 ip=08a4a5e0 tid=0077 0077:trace:seh:raise_exception info[0]=00000000 0077:trace:seh:raise_exception info[1]=08a4a5e0 0077:trace:seh:raise_exception eax=00000102 ebx=7bcd1000 ecx=09599068 edx=00000000 esi=0f2bfb40 edi=003d0f00 0077:trace:seh:raise_exception ebp=0f2be988 esp=0f2be97c cs=0023 ds=002b es=002b fs=0063 gs=006b flags=00010202 0077:trace:seh:call_stack_handlers calling handler at 0x7bc9dbe3 code=c0000005 flags=0 ... Unhandled exception: page fault on read access to 0x08a4a5e0 in 32-bit code (0x08a4a5e0). Register dump: CS:0023 SS:002b DS:002b ES:002b FS:0063 GS:006b EIP:08a4a5e0 ESP:0f2be97c EBP:0f2be988 EFLAGS:00010202( R- -- I - - - ) EAX:00000102 EBX:7bcd1000 ECX:09599068 EDX:00000000 ESI:0f2bfb40 EDI:003d0f00 ... Backtrace: =>0 0x08a4a5e0 (0x0f2be988) 1 0x08a494f0 (0x0f2bea4c) 2 0x08a4a6ab (0x0f2bea58) 3 0x7bc857e8 call_thread_func_wrapper+0xb() in ntdll (0x0f2bea68) 4 0x7bc85831 call_thread_func+0x3e(entry=0x8a4a670, arg=0x9599050, frame=0xf2beb68) [/home/focht/projects/wine/wine.repo/src/dlls/ntdll/signal_i386.c:2637] in ntdll (0x0f2beb48) 5 0x7bc857c6 call_thread_entry_point+0x11() in ntdll (0x0f2beb68) 6 0x7bc8cbce start_thread+0x11a(info=0x7ff6cfb8) [/home/focht/projects/wine/wine.repo/src/dlls/ntdll/thread.c:429] in ntdll (0x0f2bf3a8) 7 0xf75e29da start_thread+0xc9() in libpthread.so.0 (0x0f2bf468) ... 0x08a4a5e0: addb %al,0x0(%eax) Modules: Module Address Debug info Name (168 modules) PE 360000- 372000 Deferred xinput9_1_0 PE 400000- e8d000 Deferred dotp_d15 PE e90000- 1007000 Deferred fmodex PE 1010000- 107c000 Deferred fmod_event PE 1080000- 10e6000 Deferred bink2w32 PE 3a10000- 3ae3000 Deferred steam PE 3c20000- 3fe8000 Deferred d3dx9_39 PE 10000000-101ff000 Deferred d3dx9_43 PE 30000000-302c1000 Deferred steam2 PE 38000000-38893000 Deferred steamclient PE 3b400000-3b41e000 Deferred steam_api PE 3f000000-3f0b2000 Deferred tier0_s PE 3f600000-3f64f000 Deferred vstdlib_s ... Threads: process tid prio (all id:s are in hex) ... 00000058 (D) C:\Program Files\Steam\SteamApps\common\Magic 2015 Demo\DotP_D15.exe 00000077 15 <== 00000076 15 ... --- snip --- XAudio2 is brought up by 'bink2w32._BinkOpenXAudio2' API. RAD's Bink loads the 'XAudio2_7.dll' inproc-server which in turn creates a worker thread. A short time later, 'XAudio2_7.dll' inproc-server gets unloaded/released ('xaudio2.DllCanUnloadNow' -> S_OK, refcount = 0) after Bink failed to load some non-existing file (by design). The thread that owns the MTA exits, which triggers apartment destruction. Debugger session (not the same run as trace log): --- snip --- Wine-dbg>bt Backtrace: =>0 0x7e0c3368 apartment_freeunusedlibraries+0x2d(apt=0x10ae24c0, delay=0) [/home/focht/projects/wine/wine.repo/src/dlls/ole32/compobj.c:1082] in ole32 (0x0740e608) 1 0x7e0c386d apartment_release+0x3df(apt=0x10ae24c0) [/home/focht/projects/wine/wine.repo/src/dlls/ole32/compobj.c:1194] in ole32 (0x0740e6a8) ... 7 0x7bc58f1b LdrShutdownThread+0xee() [/home/focht/projects/wine/wine.repo/src/dlls/ntdll/loader.c:2592] in ntdll (0x0740e998) 8 0x7bc8d268 exit_thread+0xb1(status=0) [/home/focht/projects/wine/wine.repo/src/dlls/ntdll/thread.c:378] in ntdll (0x0740ea58) 9 0x7bc86071 call_thread_func_wrapper+0x14() in ntdll (0x0740ea68) Wine-dbg>p *entry {entry={next=0x10d29d48, prev=0x10ae251c}, dll=0xcce9770, unload_time=0, multi_threaded=0x1} Wine-dbg>p *entry->dll {refs=0x1, library_name="C:\windows\system32\XAudio2_7.dll", library=0x7520000, DllGetClassObject=0x7551150, DllCanUnloadNow=0x7550aa0, entry={next=0x7e1f3ba0, prev=0x5da2ac4}} --- snip --- The XAudio2 worker thread is still alive at this point. Upon return from 'WaitForMultipleObjects' due to 'WAIT_TIMEOUT' (see log), the thread resumes execution on a page that is no longer there because dll was unmapped earlier, causing the crash. Wine already tries to handle races with object destruction vs. thread shutdown. http://source.winehq.org/git/wine.git/blob/efdaa4a17a4a2582123e83da80a03f401... --- snip --- 1076 static void apartment_freeunusedlibraries(struct apartment *apt, DWORD delay) 1077 { 1078 struct apartment_loaded_dll *entry, *next; 1079 EnterCriticalSection(&apt->cs); 1080 LIST_FOR_EACH_ENTRY_SAFE(entry, next, &apt->loaded_dlls, struct apartment_loaded_dll, entry) 1081 { 1082 if (entry->dll->DllCanUnloadNow && (entry->dll->DllCanUnloadNow() == S_OK)) 1083 { 1084 DWORD real_delay = delay; 1085 1086 if (real_delay == INFINITE) 1087 { 1088 /* DLLs that return multi-threaded objects aren't unloaded 1089 * straight away to cope for programs that have races between 1090 * last object destruction and threads in the DLLs that haven't 1091 * finished, despite DllCanUnloadNow returning S_OK */ 1092 if (entry->multi_threaded) 1093 real_delay = 10 * 60 * 1000; /* 10 minutes */ 1094 else 1095 real_delay = 0; 1096 } 1097 1098 if (!real_delay || (entry->unload_time && ((int)(GetTickCount() - entry->unload_time) > 0))) 1099 { 1100 list_remove(&entry->entry); 1101 COMPOBJ_DllList_ReleaseRef(entry->dll, TRUE); 1102 HeapFree(GetProcessHeap(), 0, entry); 1103 } 1104 else 1105 { 1106 entry->unload_time = GetTickCount() + real_delay; 1107 if (!entry->unload_time) entry->unload_time = 1; 1108 } 1109 } 1110 else if (entry->unload_time) 1111 entry->unload_time = 0; 1112 } 1113 LeaveCriticalSection(&apt->cs); 1114 } --- snip --- 'apartment_release' passes zero delay, causing immediate unload. A small delay prevents this but this is obviously a hack to work around broken behaviour. This can also be worked around by disabling 'XAudio2_7.dll' via 'winecfg'. Regards -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=36977 Anastasius Focht <focht(a)gmx.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Summary|Magic: The Gathering – |Magic: The Gathering - |Duels of the Planeswalkers |Duels of the Planeswalkers |2015 (Steam) crashes after |2015 (Steam) crashes after |few frames of intro |intro sequence |sequence (multi-thread race |(multi-thread race during |during XAudio2 tear down) |XAudio2 tear down) -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=36977 Anastasius Focht <focht(a)gmx.net> changed: What |Removed |Added ---------------------------------------------------------------------------- Keywords| |download Summary|Magic: The Gathering - |Magic: The Gathering - |Duels of the Planeswalkers |Duels of the Planeswalkers |2015 (Steam) crashes after |2015 (Steam) crashes after |intro sequence |intro sequence |(multi-thread race during |(multi-thread race during |XAudio2 tear down) |native XAudio2 tear down) --- Comment #3 from Anastasius Focht <focht(a)gmx.net> --- Hello folks, refining summary to state it's happening with native XAudio2 - which is still preferred over builtin. Wine has builtin XAudio2 stub dlls since early 2015 which could be forced in loader config to avoid the crash. This doesn't change the fact that this is still a valid bug, revealing a Wine deficiency. $ wine --version wine-1.7.48-74-gb3fe96a Regards -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=36977 Andrew Eikum <aeikum(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |aeikum(a)codeweavers.com -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=36977 Zhiyi Zhang <zzhang(a)codeweavers.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |zzhang(a)codeweavers.com --- Comment #4 from Zhiyi Zhang <zzhang(a)codeweavers.com> --- I think this is a known issue with xaudio 2.7 https://blogs.msdn.microsoft.com/chuckw/2015/10/09/known-issues-xaudio-2-7/ -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
participants (1)
-
wine-bugs＠winehq.org