wine-bugs February 2010

wine-bugs@winehq.org

1 participants
6508 discussions

[Bug 11420] New: service control manager API problem: name of named objects might differ (client vs. service process)
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=11420 Summary: service control manager API problem: name of named objects might differ (client vs. service process) Product: Wine Version: 0.9.54. Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: advapi32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Created an attachment (id=10552) --> (http://bugs.winehq.org/attachment.cgi?id=10552) patch which fixes client vs. service SCM API when using named objects Hello, there seems to be a misconception in usage of names for named objects in service control manager API. Some people might have experienced such situation: wine hangs for several seconds on startup (when starting services). I usually disabled autostart type services (except for builtin ones) and the problems went away. Lately while fixing Microsoft Visual Studio .NET installers (2002/2003/2005) this problem became more and more annoying because the installers and VS.NET depend on some services, namely "debug manager" (service has to be running). Consider the following snippet with "service_get_event_handle" trace message added by me to highlight the problem: 0x9: client (net.exe) 0x11: service process (mdm.exe) 0x12: dispatcher thread --- snip trace --- .. 0009:trace:advapi:OpenSCManagerW ((null),(null),0x000f003f) 0009:trace:advapi:sc_handle_alloc sc_handle type=0 -> 0x121450 0009:trace:advapi:OpenSCManagerW returning 0x121450 (access : 0x000f003f) 0009:trace:advapi:OpenServiceA 0x121450 "mdm" 983103 0009:trace:advapi:OpenServiceW 0x121450 L"mdm" 983103 0009:trace:advapi:sc_handle_alloc sc_handle type=1 -> 0x121488 0009:trace:advapi:OpenServiceW returning 0x121488 0009:trace:advapi:GetServiceDisplayNameA 0x121450 "mdm" 0x34ee48 0x34fe48 0009:trace:advapi:GetServiceDisplayNameW 0x121450 L"mdm" 0x1214b8 0x34ee08 The Machine Debug Manager service is starting. 0009:trace:advapi:StartServiceA (0x121488,0,(nil)) 0009:trace:advapi:StartServiceW 0x121488 0 (nil) 0009:trace:advapi:LockServiceDatabase 0x121450 0009:trace:advapi:LockServiceDatabase returning 0x3c 0009:trace:advapi:service_start_process service_get_event_handle(L"mdm"): 0x40 0011:trace:advapi:LookupPrivilegeValueW L"",L"SeDebugPrivilege",0x34fc80 0011:trace:advapi:LookupPrivilegeValueW L"" -> 00000000-00000014 0011:trace:advapi:AdjustTokenPrivileges 0011:trace:advapi:StartServiceCtrlDispatcherA 0x34fc9c 0011:trace:advapi:service_run_threads Starting 1 pipe listener threads. Services running as process 16 0012:trace:advapi:service_control_dispatcher 0x121ea8 L"Machine Debug Manager" 0012:trace:advapi:service_control_dispatcher service_get_event_handle(L"Machine Debug Manager"): 0x48 0009:trace:advapi:UnlockServiceDatabase 0x3c 0009:trace:advapi:StartServiceW returning 0 The Machine Debug Manager service failed to start. 0009:trace:advapi:CloseServiceHandle 0x121488 0009:trace:advapi:sc_handle_destroy_service destroying service 0x121488 0009:trace:advapi:CloseServiceHandle 0x121450 0009:trace:advapi:sc_handle_destroy_manager destroying SC Manager 0x121450 000d:trace:advapi:service_run_threads last user process exited, shutting down 0011:trace:advapi:service_run_threads last user process exited, shutting down --- snip trace --- When a service is started, named objects are created to internally communicate data/events from the service process (dispatcher) to client side SCM API and vice versa. Unfortunately there exist service proccesses which pass service names to dispatcher table (StartServiceCtrlDispatcher) not matching the service name on the client side (registry "Services" key). --- snip dlls/advapi32/service.c --- static DWORD WINAPI service_control_dispatcher(LPVOID arg) { service_data *service = arg; LPWSTR name; HANDLE pipe, event; TRACE("%p %s\n", service, debugstr_w(service->name)); /* create a pipe to talk to the rest of the world with */ name = service_get_pipe_name(service->name); <--------- *problem*! pipe = CreateNamedPipeW(name, PIPE_ACCESS_DUPLEX, PIPE_TYPE_BYTE|PIPE_WAIT, 1, 256, 256, 10000, NULL ); if (pipe==INVALID_HANDLE_VALUE) ERR("failed to create pipe for %s, error = %d\n", debugstr_w(service->name), GetLastError()); HeapFree(GetProcessHeap(), 0, name); /* let the process who started us know we've tried to create a pipe */ event = service_get_event_handle(service->name); <--------- *problem*! SetEvent(event); CloseHandle(event); .. --- snip dlls/advapi32/service.c --- As result, different named objects are created and signaled so the client side can't communicate with the service side using named pipe and events. This leads to infamous timeout problem (30 sec) and service startup failing. In case of Microsoft Debug Manager the service is called "mdm" (registry). The service process itself passes "Microsoft Debug Manager" as name to dispatch table (see first trace). Problem: how can both sides communicate a "common" name for named objects? The service dispatcher part on the service side has only dispatch table data available which doesn't help much. I thought about this problem and found a feasible solution ... well call it "hack" whatever. If the client side API passes the service name in one of the process startup parameter members before process creation, the service process' dispatcher routine is able to access this data and create the named objects used for communication. I (ab)used startupinfo "title" member because only wine itself make exclusive use of most service process startup parameters. If not considered safe, use reserved/undoc fields... With that patch applied the services in question start fine and can be controlled using "net" commands. No more hangs ;-) Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 22

[Bug 12079] New: VS.NET 7.x/. NET SDK installers fail if re-executed due to ACTION_StopServices being a stub
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=12079 Summary: VS.NET 7.x/.NET SDK installers fail if re-executed due to ACTION_StopServices being a stub Product: Wine Version: CVS/GIT Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: msi AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, .NET Framework SDK, VS.NET 7.x. etc. install "helper" service(s) which get automagically started. The installers die if executed again (update/change config) because the running service blocks file update. --- snip --- .. err:msi:ACTION_InstallFiles Failed to copy L"Z:\\mnt\\iso\\Program Files\\Common Files\\Microsoft Shared\\VS7Debug\\msdbg2.dll" to L"c:\\Program Files\\Common Files\\Microsoft Shared\\VS7Debug\\msdbg2.dll" (32) err:msi:ITERATE_Actions Execution halted, action L"InstallFinalize" returned 1603 .. --- snip --- James, if you find the time, please add "stop services" msi action implementation (dlls/msi/action.c:ACTION_StopServices). Although I have a patch, I constantly forget to apply it when tracking down installer problems. You could actually help to reduce my patch orgy :) Thanks. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 14

[Bug 12390] New: err:service: service_control_dispatcher failed to create pipe for L"C-Dilla_CdaC11BA", error = 231
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=12390 Summary: err:service:service_control_dispatcher failed to create pipe for L"C-Dilla_CdaC11BA", error = 231 Product: Wine Version: 0.9.58. Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P3 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: amaramrahul(a)gmail.com "Longman Dictionary of Contemporary English" (ISBN 0 582 50673 5) gets installed properly from CD-ROM using wine. However when I try to run it using wine, I get the below error in console: # wine "C:\Program Files\Longman\LDOCE\ldoce.exe" err:service:service_control_dispatcher failed to create pipe for L"C-Dilla_CdaC11BA", error = 231 err:service:service_control_dispatcher failed to create pipe for L"C-Dilla_CdaC11BA", error = 231 Then a pop-up appears with the message: "This machine must be re-booted before this product will run correctly. Reboot now?". Upon clicking "Yes"/"No", the program exits. I think this has to do something with copyright. Not sure about this though. -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 7

[Bug 14594] New: crypt32.CryptHashMessage needed for VS.NET 2005 (deployment project type creation)
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=14594 Summary: crypt32.CryptHashMessage needed for VS.NET 2005 (deployment project type creation) Product: Wine Version: CVS/GIT Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: crypt32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, using project wizard to create a "setup/deployment" project type (msi), it bails due to unimpl crypt32.CryptHashMessage --- snip --- .. 003f:trace:loaddll:load_native_dll Loaded L"C:\\Program Files\\Microsoft Visual Studio 8\\Common7\\Tools\\Deployment\\dpplg.dll" at 0x5d010000: native 003f:trace:loaddll:load_native_dll Loaded L"C:\\Program Files\\Microsoft Visual Studio 8\\Common7\\Tools\\Deployment\\1033\\dpplgui.dll" at 0x5d380000: native 003f:trace:loaddll:load_native_dll Loaded L"C:\\Program Files\\Microsoft Visual Studio 8\\Common7\\Tools\\Deployment\\dpedt.dll" at 0x5ce70000: native 003f:trace:loaddll:load_native_dll Loaded L"C:\\Program Files\\Microsoft Visual Studio 8\\VJ#\\bin\\vjsplg.dll" at 0x6ca90000: native 003f:trace:loaddll:load_native_dll Loaded L"C:\\Program Files\\Microsoft Visual Studio 8\\VJ#\\bin\\1033\\vjsplgui.dll" at 0x6cbc0000: native 003f:trace:seh:raise_exception code=80000100 flags=1 addr=0x7b8432e0 003f:trace:seh:raise_exception info[0]=60b7f840 003f:trace:seh:raise_exception info[1]=60b7fa01 wine: Call from 0x7b8432e0 to unimplemented function crypt32.dll.CryptHashMessage, aborting 003f:trace:seh:call_stack_handlers calling handler at 0x5cdea7c5 code=80000100 flags=1 003f:trace:seh:call_stack_handlers handler at 0x5cdea7c5 returned 1 003f:trace:seh:call_stack_handlers calling handler at 0x5cdf1732 code=80000100 flags=1 .. --- snip --- MSDN: http://msdn.microsoft.com/en-us/library/aa380203.aspx --- stub test --- 031:Call crypt32.CryptHashMessage(002fd208,00000001,00000001,002fd2e0,002fd2dc,002fd260,002fd248,002fd2f4,002fd2d8) ret=5cdd6009 0031:fixme:crypt:CryptHashMessage (0x2fd208, 1, 1, 0x2fd2e0, 0x2fd2dc, 0x2fd260, 0x2fd248, 0x2fd2f4): stub 0031:Ret crypt32.CryptHashMessage() retval=00000000 ret=5cdd6009 --- stub test -- Simple stub seems to satisfy it. For a better stub: MD5 is needed here, the supplied hash algorithm OID (pHashPara.HashAlgorithm) is "1.2.840.113549.2.5" Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 6

[Bug 14567] New: Shell32 IAutoCompleteDropDown needed for VS.NET 2005 and other tools which use MS DevEnv UI engine
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=14567 Summary: Shell32 IAutoCompleteDropDown needed for VS.NET 2005 and other tools which use MS DevEnv UI engine Product: Wine Version: CVS/GIT Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: shell32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, just visiting back VS.NET 2005 and friends... The annoying error message box, stating "The operation could not be completed" can be prevented by implementing IAutoCompleteDropDown in shell32. Due to this missing interface, the Devenv GUI is not really usable at all. Also the .NET CLR GUI debugger that ships with the .NET 2.0 SDK suffers from this bug. +shell,+ole --- snip --- .. trace:ole:CoCreateInstance (rclsid={00bb2763-6a77-11d0-a535-00c04fd7d062}, pUnkOuter=(nil), dwClsContext=00000001, riid={eac04bc0-3791-11d2-bb95-0060977b464c}, ppv=0x50757698) trace:ole:CoGetClassObject CLSID: {00bb2763-6a77-11d0-a535-00c04fd7d062}, IID: {00000001-0000-0000-c000-000000000046} trace:ole:WINE_StringFromCLSID 0x5005ec20->{00BB2763-6A77-11D0-A535-00C04FD7D062} trace:ole:apartment_getclassobject found L"shell32.dll" already loaded trace:ole:apartment_getclassobject calling DllGetClassObject 0x66426d20 trace:shell:HCR_RegOpenClassIDKey CLSID\{00bb2763-6a77-11d0-a535-00c04fd7d062} trace:shell:HCR_GetClassNameA -- AutoComplete trace:shell:DllGetClassObject CLSID: {00bb2763-6a77-11d0-a535-00c04fd7d062} (AutoComplete),IID: {00000001-0000-0000-c000-000000000046} (IID_IClassFactory) trace:shell:DllGetClassObject index[7] trace:shell:IDefClF_fnConstructor (0xc0c740)(null) trace:shell:IDefClF_fnQueryInterface (0xc0c740)->( {00000001-0000-0000-c000-000000000046} (IID_IClassFactory)) trace:shell:IDefClF_fnRelease (0xc0c740)->(count=2) trace:shell:DllGetClassObject -- pointer to class factory: 0xc0c740 trace:shell:IDefClF_fnCreateInstance 0xc0c740->((nil), {eac04bc0-3791-11d2-bb95-0060977b464c} (IID_IAutoComplete2),0x50757698) trace:shell:IAutoComplete_fnQueryInterface (0xc26b68)->( IID: {eac04bc0-3791-11d2-bb95-0060977b464c} (IID_IAutoComplete2),0x50757698) trace:shell:IAutoComplete2_fnAddRef (0xc26b68)->(count=1) trace:shell:IAutoComplete_fnAddRef (0xc26b68)->(1) trace:shell:IAutoComplete_fnQueryInterface -- Interface: (0x50757698)->(0xc26b6c) trace:shell:IAutoComplete_Constructor -- (0xc26b68)-> trace:shell:IDefClF_fnRelease (0xc0c740)->(count=1) trace:shell:IDefClF_fnRelease -- destroying IClassFactory(0xc0c740) trace:ole:CoCreateInstance (rclsid={00000323-0000-0000-c000-000000000046}, pUnkOuter=(nil), dwClsContext=00000001, riid={00000146-0000-0000-c000-000000000046}, ppv=0x55c538) trace:ole:CoCreateInstance Retrieved GIT (0xc1c540) trace:ole:StdGlobalInterfaceTable_RegisterInterfaceInGlobal iface=0xc1c540, pUnk=0x6512c8, riid={6d5140c1-7436-11ce-8034-00aa006009fa}, pdwCookie=0x0x320ff4 trace:ole:StdGlobalInterfaceTable_RegisterInterfaceInGlobal About to marshal the interface trace:ole:CoMarshalInterface (0xc26f88, {6d5140c1-7436-11ce-8034-00aa006009fa}, 0x6512c8, 3, (nil), MSHLFLAGS_TABLESTRONG) trace:ole:CoGetStandardMarshal ({6d5140c1-7436-11ce-8034-00aa006009fa},0x6512c8,3,(nil),1,0x320f5c) trace:ole:CoMarshalInterface Using standard marshaling trace:ole:CoMarshalInterface Calling IMarshal::MarshalInterace trace:ole:StdMarshalImpl_MarshalInterface (...,{6d5140c1-7436-11ce-8034-00aa006009fa},...) trace:ole:CoGetPSClsid () riid={6d5140c1-7436-11ce-8034-00aa006009fa}, pclsid=0x320e04 trace:ole:WINE_StringFromCLSID 0x50064ef8->{6D5140C1-7436-11CE-8034-00AA006009FA} trace:ole:__CLSIDFromString L"{b8da6310-e19b-11d0-933c-00a0c90dcaa9}" -> 0x320e04 trace:ole:CoGetPSClsid () Returning CLSID={b8da6310-e19b-11d0-933c-00a0c90dcaa9} trace:ole:CoGetClassObject CLSID: {b8da6310-e19b-11d0-933c-00a0c90dcaa9}, IID: {d5f569d0-593b-101a-b569-08002b2dbf7a} trace:ole:WINE_StringFromCLSID 0x320e04->{B8DA6310-E19B-11D0-933C-00A0C90DCAA9} trace:ole:apartment_getclassobject found L"C:\\windows\\system32\\actxprxy.dll" already loaded trace:ole:apartment_getclassobject calling DllGetClassObject 0x60a76650 trace:ole:NdrDllGetClassObject ({b8da6310-e19b-11d0-933c-00a0c90dcaa9}, {d5f569d0-593b-101a-b569-08002b2dbf7a}, 0x320e14, 0x60a7784c, {b8da6310-e19b-11d0-933c-00a0c90dcaa9}, 0x60a7785c) trace:ole:CStdPSFactory_QueryInterface (0x60a7785c)->QueryInterface({d5f569d0-593b-101a-b569-08002b2dbf7a},0x320e14) trace:ole:CStdPSFactory_CreateStub (0x60a7785c)->CreateStub({6d5140c1-7436-11ce-8034-00aa006009fa},0x6512c8,0x320e1c) trace:ole:FindProxyInfo found: ProxyInfo 0x60a77040 Index 0 trace:ole:CStdStubBuffer_Construct (0x6512c8,0x60a770e0,0x60a7785c,0x320e1c) IServiceProvider trace:ole:CStdStubBuffer_Construct iid={6d5140c1-7436-11ce-8034-00aa006009fa} trace:ole:CStdStubBuffer_Construct vtbl=0x60a770f0 trace:ole:CStdPSFactory_AddRef (0x60a7785c)->AddRef() trace:ole:CStdPSFactory_Release (0x60a7785c)->Release() trace:ole:stub_manager_int_addref before 1 trace:ole:get_stub_manager_from_object found 0xc1c688 for object 0x6512c8 trace:ole:marshal_object registering new ifstub on pre-existing manager trace:ole:NdrCStdStubBuffer_Release (0xc0fe28)->Release() trace:ole:CStdStubBuffer_Disconnect (0xc0fe28)->Disconnect() trace:ole:CStdPSFactory_Release (0x60a7785c)->Release() trace:ole:stub_manager_ext_addref added 1 refs to 0xc1c688 (oid 3), rc is now 4 trace:ole:RPC_RegisterInterface ({6d5140c1-7436-11ce-8034-00aa006009fa}) trace:ole:stub_manager_int_release after 1 trace:ole:CoMarshalInterface completed with hr 0x00000000 trace:ole:StdGlobalInterfaceTable_RegisterInterfaceInGlobal Cookie is 0xf103 trace:shell:IAutoComplete2_fnInit (0xc26b68) trace:shell:IAutoComplete_fnInit (0xc26b68)->(0x000201da, 0x55c4e4, (null), L"%s") trace:shell:IAutoComplete_fnInit ACO_AUTOAPPEND trace:shell:IAutoComplete2_fnSetOptions (0xc26b68) -> (0x71) trace:shell:HCR_RegOpenClassIDKey CLSID\{3cd141f4-3c6a-11d2-bcaa-00c04fd929db} trace:shell:HCR_GetClassNameA -- trace:shell:IAutoComplete2_fnQueryInterface (0xc26b68)->( {3cd141f4-3c6a-11d2-bcaa-00c04fd929db} (unknown),0x50757690) trace:shell:HCR_RegOpenClassIDKey CLSID\{3cd141f4-3c6a-11d2-bcaa-00c04fd929db} trace:shell:HCR_GetClassNameA -- trace:shell:IAutoComplete_fnQueryInterface (0xc26b68)->( IID: {3cd141f4-3c6a-11d2-bcaa-00c04fd929db} (unknown),0x50757690) trace:shell:IAutoComplete_fnQueryInterface -- Interface: E_NOINTERFACE trace:shell:IAutoComplete2_fnRelease (0xc26b68)->(count=2) trace:shell:IAutoComplete_fnRelease (0xc26b68)->(2) trace:ole:GetErrorInfo (0, 0x3210a4, (nil)) .. trace:ole:ICreateErrorInfoImpl_SetDescription (0xc24818): L"The operation could not be completed " .. --- snip --- I added/implemented the missing interface (include/shldisp.idl, dlls/shell32/autocomplete.c, dlls/shell32/debughlp.c) and verified it fixes the annoying error. To give the implementor/submitter full freedom, I'll omit a patch here ;-) Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 5

[Bug 14571] New: ole32.CoGetCallerTID needed for VS.NET 2005
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=14571 Summary: ole32.CoGetCallerTID needed for VS.NET 2005 Product: Wine Version: CVS/GIT Platform: PC OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: ole32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, keep the VS.NET 2005 IDE for ~10 secs open and it will crash due to this bug. --- snip --- 0026:Call user32.InSendMessage() ret=500891c3 0026:Ret user32.InSendMessage() retval=00000000 ret=500891c3 0026:Call KERNEL32.RaiseException(80000100,00000001,00000002,0032f618) ret=608a4d85 0026:trace:seh:raise_exception code=80000100 flags=1 addr=0x7b8432e0 0026:trace:seh:raise_exception info[0]=608a4e20 0026:trace:seh:raise_exception info[1]=608a4e2a wine: Call from 0x7b8432e0 to unimplemented function ole32.dll.CoGetCallerTID, aborting 0026:trace:seh:call_stack_handlers calling handler at 0x506a8ab0 code=80000100 flags=1 --- snip --- MSDN info here: http://msdn.microsoft.com/en-us/library/ms680683(VS.85).aspx Implementation is straight forward (minus remote), all the info is there. I've already verified with a patch. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 11

[Bug 18500] New: ntdll.NtQueryInformationProcess: provide simple ProcessDebugObjectHandle info class handling, returning "no debugger"
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=18500 Summary: ntdll.NtQueryInformationProcess: provide simple ProcessDebugObjectHandle info class handling, returning "no debugger" Product: Wine Version: 1.1.21 Platform: Other URL: http://www.exeinfo.go.pl/ OS/Version: other Status: UNCONFIRMED Severity: normal Priority: P2 Component: ntdll AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, this is a continuation of http://bugs.winehq.org/show_bug.cgi?id=12859#c7 The app queries for unimplemented "ProcessDebugObjectHandle" information class. --- snip --- ... 0021:Call ntdll.NtSetInformationThread(fffffffe,00000011,00000000,00000000) ret=004da0d5 0021:fixme:thread:NtSetInformationThread info class 17 not supported yet 0021:Ret ntdll.NtSetInformationThread() retval=c0000002 ret=004da0d5 0021:Call 0021:Call ntdll.NtQueryInformationProcess(ffffffff,0000001e,c0000002,00000004,00000000) ret=004da0e4 0021:fixme:ntdll:NtQueryInformationProcess (process=0xffffffff) Unimplemented information class: ProcessDebugObjectHandle 0021:Ret ntdll.NtQueryInformationProcess() retval=c0000003 ret=004da0e4 ... --- snip --- Wine already fakes info for such kind of classes, returning "no debugger". Have a look at dlls/ntdll/process.c:NtQueryInformationProcess, ProcessDebugPort how it's done and provide simple patch. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 7

[Bug 12859] New: HideThreadFromDebugger in NtSetInformationThread
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=12859 Summary: HideThreadFromDebugger in NtSetInformationThread Product: Wine Version: 0.9.60 Platform: Other OS/Version: other Status: UNCONFIRMED Severity: enhancement Priority: P2 Component: ntdll AssignedTo: wine-bugs(a)winehq.org ReportedBy: readams(a)readams.net Created an attachment (id=12555) --> (http://bugs.winehq.org/attachment.cgi?id=12555) Add HideThreadFromDebugger to winternl.h and to NtSetInformationThread This API exists in the windows NtSetInformationThread for some reason. The sensible thing to do in wine here seems to be to just ignore this call. Patch attached is against 0.9.60. -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 20

[Bug 21562] New: Multiple init/deinit of msvcrt (std fd's) causes pluto app installer to hang
by wine-bugs＠winehq.org 24 Jul '21

24 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=21562 Summary: Multiple init/deinit of msvcrt (std fd's) causes pluto app installer to hang Product: Wine Version: 1.1.37 Platform: x86 URL: http://pluto.brain-killer.org/download/pluto_FE_w32_en _0036.exe OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: msvcrt AssignedTo: wine-bugs(a)winehq.org ReportedBy: focht(a)gmx.net Hello, an interesting one... WINEDEBUG=+tid,+seh,+loaddll,+msvcrt,+server,+relay wine ./pluto_FE_w32_en_0036.exe >>log.txt 2>&1 The app installer creates an additional worker (unpack) thread from its main thread: --- snip --- 0009:Call KERNEL32.CreateThread(00000000,00000000,0042b674,00000000,00000000,00436718) ret=0042c3e8 0009: *fd* 6 <- 22 0009: new_thread( access=001f03ff, attributes=00000000, suspend=1, request_fd=6 ) 0009: new_thread() = 0 { tid=001b, handle=0004 } ... 001b:Starting thread proc 0x42b674 (arg=(nil)) ... 0009:Ret KERNEL32.CreateThread() retval=00000004 ret=0042c3e8 --- snip --- The app installer main thread contains a loop to check if the (unpack) worker thread (handle = 0x4) is still alive. While doing so, it pumps messages and delays the next check, illustrated by following code snippet: --- snip --- 0009:Ret KERNEL32.Sleep() retval=00000000 ret=004167e4 0009:Call KERNEL32.GetExitCodeThread(00000004,0032fd9c) ret=0042c40f 0009: get_thread_info( handle=0004, tid_in=0000 ) 0009: get_thread_info() = 0 { pid=0008, tid=001b, teb=7ffd4000, affinity=ffffffffffffffff, creation_time=1caa28d3911cf76 (-0.0285120), exit_time=0, exit_code=259, priority=0, last=0 } 0009:Ret KERNEL32.GetExitCodeThread() retval=00000001 ret=0042c40f 0009:Call user32.PeekMessageA(0032fd0c,00000000,00000000,00000000,00000001) ret=004167d9 ... 0009:Ret user32.PeekMessageA() retval=00000000 ret=004167d9 0009:Call KERNEL32.Sleep(00000001) ret=004167e4 --- snip --- Unfortunately something goes wrong and the app installer refuses to terminate - because the worker checking loop suddenly became a forever loop: --- snip --- ... 0009:Call KERNEL32.Sleep(00000001) ret=004167e4 0009:Ret KERNEL32.Sleep() retval=00000000 ret=004167e4 0009:Call KERNEL32.GetExitCodeThread(00000004,0032fd9c) ret=0042c40f 0009: get_thread_info( handle=0004, tid_in=0000 ) 0009: get_thread_info() = INVALID_HANDLE { pid=0000, tid=0000, teb=00000000, affinity=00000000, creation_time=0, exit_time=0, exit_code=0, priority=0, last=0 } 0009:Ret KERNEL32.GetExitCodeThread() retval=00000000 ret=0042c40f 0009:Call user32.PeekMessageA(0032fd0c,00000000,00000000,00000000,00000001) ret=004167d9 0009: get_message( flags=00000001, get_win=00000000, get_first=00000000, get_last=ffffffff, hw_id=00000000, wake_mask=00000000, changed_mask=00000000 ) 0009: get_message() = PENDING { win=00000000, msg=00000000, wparam=00000000, lparam=00000000, type=0, time=00000000, active_hooks=80000000, total=0, data={} } ... 0009:Ret user32.PeekMessageA() retval=00000000 ret=004167d9 --- snip --- The installer code doesn't evaluate the return code of GetExitCodeThread() but instead accesses the exit code variable directly, comparing it to 0x103 -> STILL_ACTIVE. Because the thread once lived in this state, this variable is never touched by Wine in case the thread handle becomes invalid. GetExitCodeThread() fails but the variable still contains the last known "good" value -> STILL_ACTIVE (the app doesn't reinit it). The app still thinks the worker is alive and goes into forever loop. Now the culprit: the thread handle, delivered to caller upon returning from CreateThread() is _never_ closed by app explicitly -> CloseHandle() so GetExitCodeThread() should be successful in any case. We have to go back in time to unearth the problem ... lets look what happens to this thread handle (0x4). --- snip --- ... 001b:Call KERNEL32.FreeLibrary(10000000) ret=004242bb 001b:Call PE DLL (proc=0x10001148,module=0x10000000 L"unpack.dll",reason=PROCESS_DETACH,res=(nil)) 001b:Ret PE DLL (proc=0x10001148,module=0x10000000 L"unpack.dll",reason=PROCESS_DETACH,res=(nil)) retval=1 001b:Call PE DLL (proc=0x5de0cc9c,module=0x5de00000 L"crtdll.dll",reason=PROCESS_DETACH,res=(nil)) 001b:Ret PE DLL (proc=0x5de0cc9c,module=0x5de00000 L"crtdll.dll",reason=PROCESS_DETACH,res=(nil)) retval=1 001b:Call PE DLL (proc=0x5c7cd8b4,module=0x5c790000 L"msvcrt.dll",reason=PROCESS_DETACH,res=(nil)) ... 001b:trace:msvcrt:DllMain (0x5c790000, DLL_PROCESS_DETACH, (nil)) pid(8), tid(1b), tls(4) 001b:trace:msvcrt:msvcrt_free_mt_locks : uninitializing all mtlocks ... 001b:trace:msvcrt:MSVCRT__close :fd (3) handle (0xfffffff6) 001b:Call KERNEL32.CloseHandle(fffffff6) ret=5c7ad9fc 001b: close_handle( handle=0004 ) 001b: close_handle() = 0 001b:Ret KERNEL32.CloseHandle() retval=00000001 ret=5c7ad9fc 001b:trace:msvcrt:msvcrt_free_fd :fd (3) freed 001b:trace:msvcrt:MSVCRT__close :ok ... 001b:trace:msvcrt:MSVCRT__close :fd (4) handle (0xfffffff5) 001b:Call KERNEL32.CloseHandle(fffffff5) ret=5c7ad9fc 001b: close_handle( handle=0008 ) 001b: close_handle() = 0 001b:Ret KERNEL32.CloseHandle() retval=00000001 ret=5c7ad9fc 001b:trace:msvcrt:msvcrt_free_fd :fd (4) freed 001b:trace:msvcrt:MSVCRT__close :ok ... 001b:trace:msvcrt:MSVCRT__close :fd (5) handle (0xfffffff4) 001b:Call KERNEL32.CloseHandle(fffffff4) ret=5c7ad9fc 001b: close_handle( handle=000c ) 001b: close_handle() = 0 001b:Ret KERNEL32.CloseHandle() retval=00000001 ret=5c7ad9fc 001b:trace:msvcrt:msvcrt_free_fd :fd (5) freed 001b:trace:msvcrt:MSVCRT__close :ok 001b:trace:msvcrt:MSVCRT__fcloseall :closed (3) handles ... 001b:trace:msvcrt:MSVCRT__close :fd (0) handle (0x8) 001b:Call KERNEL32.CloseHandle(00000008) ret=5c7ad9fc 001b: close_handle( handle=0008 ) 001b: close_handle() = INVALID_HANDLE 001b:Ret KERNEL32.CloseHandle() retval=00000000 ret=5c7ad9fc ... 001b:warn:msvcrt:MSVCRT__close :failed-last error (6) ... 001b:trace:msvcrt:MSVCRT__close :ok ... 001b:trace:msvcrt:MSVCRT__close :fd (1) handle (0xc) 001b:Call KERNEL32.CloseHandle(0000000c) ret=5c7ad9fc 001b: close_handle( handle=000c ) 001b: close_handle() = INVALID_HANDLE 001b:Ret KERNEL32.CloseHandle() retval=00000000 ret=5c7ad9fc ... 001b:warn:msvcrt:MSVCRT__close :failed-last error (6) ... 001b:trace:msvcrt:MSVCRT__close :ok ... 001b:trace:msvcrt:MSVCRT__close :fd (2) handle (0x50) 001b:Call KERNEL32.CloseHandle(00000050) ret=5c7ad9fc 001b: close_handle( handle=0050 ) 001b: close_handle() = 0 001b:Ret KERNEL32.CloseHandle() retval=00000001 ret=5c7ad9fc 001b:trace:msvcrt:msvcrt_free_fd :fd (2) freed 001b:Call KERNEL32.GetStdHandle(fffffff4) ret=5c7ac729 001b:Ret KERNEL32.GetStdHandle() retval=0000000c ret=5c7ac729 001b:trace:msvcrt:MSVCRT__close :ok ... 001b:trace:msvcrt:msvcrt_free_console :Closing console handles --- snip --- *eek* ... the thread handle gets closed as part of msvcrt cleanup! PROCESS_DETACH -> msvcrt_free_io() -> MSVCRT__fcloseall() This will close fd(3), fd(4), fd(5) -> stdin,stdout,stderr Corresponding msvcrt init: --- snip --- ... 001b:Call PE DLL (proc=0x5c7cd8b4,module=0x5c790000 L"msvcrt.dll",reason=PROCESS_ATTACH,res=(nil)) ... 001b:trace:msvcrt:DllMain (0x5c790000, DLL_PROCESS_ATTACH, (nil)) pid(8), tid(1b), tls(0) ... 001b:Call KERNEL32.GetStartupInfoA(0096e4b8) ret=5c7acdd6 001b:Ret KERNEL32.GetStartupInfoA() retval=00000011 ret=5c7acdd6 001b:Call KERNEL32.GetStdHandle(fffffff6) ret=5c7acf77 001b:Ret KERNEL32.GetStdHandle() retval=00000004 ret=5c7acf77 001b:Call KERNEL32.DuplicateHandle(ffffffff,00000004,ffffffff,5c7db2e0,00000000,00000001,00000002) ret=5c7acfc0 001b: dup_handle( src_process=ffffffff, src_handle=0004, dst_process=ffffffff, access=00000000, attributes=00000002, options=00000002 ) 001b: dup_handle() = 0 { handle=0008, self=1, closed=0 } 001b:Ret KERNEL32.DuplicateHandle() retval=00000001 ret=5c7acfc0 001b:Call KERNEL32.GetStdHandle(fffffff5) ret=5c7acff6 001b:Ret KERNEL32.GetStdHandle() retval=00000008 ret=5c7acff6 001b:Call KERNEL32.DuplicateHandle(ffffffff,00000008,ffffffff,5c7db304,00000000,00000001,00000002) ret=5c7ad03f 001b: dup_handle( src_process=ffffffff, src_handle=0008, dst_process=ffffffff, access=00000000, attributes=00000002, options=00000002 ) 001b: dup_handle() = 0 { handle=000c, self=1, closed=0 } 001b:Ret KERNEL32.DuplicateHandle() retval=00000001 ret=5c7ad03f 001b:Call KERNEL32.GetStdHandle(fffffff4) ret=5c7ad075 001b:Ret KERNEL32.GetStdHandle() retval=0000000c ret=5c7ad075 001b:Call KERNEL32.DuplicateHandle(ffffffff,0000000c,ffffffff,5c7db328,00000000,00000001,00000002) ret=5c7ad0be 001b: dup_handle( src_process=ffffffff, src_handle=000c, dst_process=ffffffff, access=00000000, attributes=00000002, options=00000002 ) 001b: dup_handle() = 0 { handle=0050, self=1, closed=0 } 001b:Ret KERNEL32.DuplicateHandle() retval=00000001 ret=5c7ad0be 001b:trace:msvcrt:msvcrt_init_io :handles (0x8)(0xc)(0x50) --- snip --- The culprit: msvcrt is actually loaded and unloaded _two_ times! Upon the first unload sequence all std/fd handles are teared down (including handle 0x4) - this makes the handle values available in server again. When the worker thread is created it gets the unused handle 0x4 assigned from wineserver. This is the handle which is then polled by installers GetExitCodeThread() loop. When msvcrt is loaded a second time, GetStdHandle( STD_XXX) will still deliver its original value from PEB (0x4, 0x8, 0xC). Upon second msvcrt unload, fd(3), fd(4), fd(5) are closed again but this time they close something completely different (indication are also fd close errors following), leaving the app installer in forever loop. Regards -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 7

[Bug 15381] New: CoFreeUnusedLibraries will crash COM server or cause 0x800703E6/I_RpcReceive error 0x3e6
by wine-bugs＠winehq.org 20 Jul '21

20 Jul '21

http://bugs.winehq.org/show_bug.cgi?id=15381 Summary: CoFreeUnusedLibraries will crash COM server or cause 0x800703E6/I_RpcReceive error 0x3e6 Product: Wine Version: 1.0.0 Platform: Other OS/Version: other Status: UNCONFIRMED Severity: normal Priority: P2 Component: ole32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: dmda(a)yandex.ru Created an attachment (id=16230) --> (http://bugs.winehq.org/attachment.cgi?id=16230) demo illustrating the problem if CoFreeUnusedLibraries is called by COM server, it will break RPC channcel and lead to the errors Attached is client/server code that demonstrates the problem. If you comment out CoFreeUnusedLibraries call, it will start to work. Steps: -comple all 3 projects in the Workspace (Debug target is the only appropriate for the sake of demo) -make sure server and proxy are registered by running testserver.exe -RegServer, regsvr32 testserverPS.dll -run testproject.exe (the client) Expected behaviour: 1. it will show server's message "CServer is created", click ok 2. it will show client's message "press OK when ready", click ok 3. it will show server's message passed by the client "Hello world!", click ok 4. it will show client's message "call succeeded" and client will exit 5. after roughly 6sec it will show that the server is unloaded/ Actual behaviour: 1. it will show server's message "CServer is created", click ok 2. it will show client's message "press OK when ready", click ok 3. it will show server's message passed by the client "Hello world!", click ok 4. it will dump out trouble in the console and show client's message "call failed" You may call CoFreeUnusedLibraries anywhere in the appartment and it will break RPC on the server side. In some cases it will crash whole the server, like in the demo. Tested environment: Fedora Core 8, 32bit, x86, wine 1.0 installed through the updates -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email 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.

2 14

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-bugs February 2010