October 2021 - Wine-Bugs - list.winehq.org

[Bug 36731] New: Runes of Magic 'ClientUpdater.exe' crashes after a number of update cycles (mshtml environment setup contains stack buffer overflow)
by wine-bugs＠winehq.org Dec. 5, 2021

Dec. 5, 2021

https://bugs.winehq.org/show_bug.cgi?id=36731 Bug ID: 36731 Summary: Runes of Magic 'ClientUpdater.exe' crashes after a number of update cycles (mshtml environment setup contains stack buffer overflow) Product: Wine Version: 1.7.20 Hardware: x86 OS: Linux Status: NEW Severity: normal Priority: P2 Component: mshtml Assignee: wine-bugs(a)winehq.org Reporter: focht(a)gmx.net Hello folks, found during investigation of 'Runes of Magic' client updater. There is a crash after a number of update cycles. Unfortunately it's not easily traceable as it requires large downloads and many client restarts (= hours) to reach the crash point. I started the updater with a few debug channels (= reduced noise) which still allowed me to do post-mortem analysis. The launcher restarts itself after each update cycle. --- snip --- $ pwd /home/focht/.wine/drive_c/Program Files/Runes of Magic $ WINEDEBUG=+tid,+seh,+loaddll,+process,+mshtml wine ./launcher.exe ... <hours, multiple updater restarts> ... 004f:trace:loaddll:load_builtin_dll Loaded L"C:\\windows\\system32\\mshtml.dll" at 0x7c090000: builtin 004f:trace:mshtml:DllGetClassObject (CLSID_HTMLDocument {00000001-0000-0000-c000-000000000046} 0x3392a8) 004f:trace:mshtml:ClassFactory_AddRef (0x1c2c80) ref = 1 004f:trace:mshtml:HTMLDocument_Create ((nil) IID_IUnknown 0x1c19d4) 004f:trace:mshtml:load_gecko () 004f:trace:mshtml:check_version "Wine Gecko 2.24" 004f:trace:mshtml:load_xul (L"C:\\windows\\system32\\gecko\\2.24\\wine_gecko\\\\xul.dll") 004f:trace:seh:raise_exception code=c0000005 flags=0 addr=0x7c130001 ip=7c130001 tid=004f 004f:trace:seh:raise_exception info[0]=00000001 004f:trace:seh:raise_exception info[1]=8d43ade4 004f:trace:seh:raise_exception eax=00000001 ebx=006b0063 ecx=003389a0 edx=7bcda204 esi=00339330 edi=001c2de8 004f:trace:seh:raise_exception ebp=005c0070 esp=00338c00 cs=0023 ds=002b es=002b fs=0063 gs=006b flags=00010212 004f:trace:seh:call_stack_handlers calling handler at 0x4a97b0 code=c0000005 flags=0 004f:trace:seh:call_stack_handlers handler at 0x4a97b0 returned 1 004f:trace:seh:call_stack_handlers calling handler at 0x7bc9ecf7 code=c0000005 flags=0 wine: Unhandled page fault on write access to 0x8d43ade4 at address 0x7c130001 (thread 004f), starting debugger... --- snip --- I looked at the crash site and noticed it being in the middle of opcode sequence. --- snip --- 7C130000 45 INC EBP 7C130001 0889 4424108D OR BYTE PTR DS:[ECX+8D102444],CL 7C130007 8313 B5 ADC DWORD PTR DS:[EBX],-4B 7C13000A FA CLI --- snip --- Decoded with proper opcode start addresses: --- snip --- 7C12FFFF 8B45 08 MOV EAX,DWORD PTR SS:[EBP+8] 7C130002 894424 10 MOV DWORD PTR SS:[ESP+10],EAX 7C130006 8D83 13B5FAFF LEA EAX,[EBX+FFFAB513] --- snip --- Partial stack dump with 'esp' = 0x00338c00 from exception context: --- snip --- ... 00338BA8 C0000005 00338BAC 00000000 00338BB0 00000000 00338BB4 7C130001 00338BB8 00000002 00338BBC 00000001 00338BC0 8D43ADE4 äC 00338BC4 00650074 t e 00338BC8 0033006D m 3 00338BCC 005C0032 2 \ 00338BD0 00650067 g e 00338BD4 006B0063 c k 00338BD8 005C006F o \ 00338BDC 002E0032 2 . 00338BE0 00340032 2 4 00338BE4 0077005C \ w 00338BE8 006E0069 i n 00338BEC 00000005 00338BF0 00650067 g e 00338BF4 006B0063 c k 00338BF8 005C006F o \ 00338BFC 7C130000 00338C00 00338F74 ; UNICODE "C:\windows\system32\gecko\2.24\wine_gecko\" 00338C04 7C1FF480 00338C08 7C19E234 ; ASCII "load_xul" 00338C0C 7C19CF48 ; ASCII "(%s)" 00338C10 7BCEC8C1 ; ASCII "L"C:\\windows\\system32\\gecko\\2.24\\wine_gecko\\\\xul.dll"" 00338C14 7C19D4E8 ; ASCII "Wine Gecko 2.24" 00338C18 7C19E250 ; ASCII "check_version" 00338C1C 7C19D4C0 ; ASCII "%s" 00338C20 7BCEC8AF ; ASCII ""Wine Gecko 2.24"" ... --- snip --- Yes, looks pretty much like a stack buffer overflow. A string buffer overwrote 'ebp', 'ebx' values (register save on stack for '__x86_get_pc_thunk_bx') and parts of the return address. The NULL terminator cancelled out the lower 16 bits of the return address. The culprit: 'load_xul' -> 'set_environment' Source: http://source.winehq.org/git/wine.git/blob/0be56d27d2d4b22367313fa4c6f1e658… --- snip --- 439 static void set_environment(LPCWSTR gre_path) 440 { 441 WCHAR path_env[MAX_PATH], buf[20]; 442 int len, debug_level = 0; 443 444 static const WCHAR pathW[] = {'P','A','T','H',0}; 445 static const WCHAR warnW[] = {'w','a','r','n',0}; 446 static const WCHAR xpcom_debug_breakW[] = 447 {'X','P','C','O','M','_','D','E','B','U','G','_','B','R','E','A','K',0}; 448 static const WCHAR nspr_log_modulesW[] = 449 {'N','S','P','R','_','L','O','G','_','M','O','D','U','L','E','S',0}; 450 static const WCHAR debug_formatW[] = {'a','l','l',':','%','d',0}; 451 452 /* We have to modify PATH as XPCOM loads other DLLs from this directory. */ 453 GetEnvironmentVariableW(pathW, path_env, sizeof(path_env)/sizeof(WCHAR)); 454 len = strlenW(path_env); 455 path_env[len++] = ';'; 456 strcpyW(path_env+len, gre_path); 457 SetEnvironmentVariableW(pathW, path_env); 458 459 SetEnvironmentVariableW(xpcom_debug_breakW, warnW); 460 461 if(TRACE_ON(gecko)) 462 debug_level = 5; 463 else if(WARN_ON(gecko)) 464 debug_level = 3; 465 else if(ERR_ON(gecko)) 466 debug_level = 2; 467 468 sprintfW(buf, debug_formatW, debug_level); 469 SetEnvironmentVariableW(nspr_log_modulesW, buf); 470 } --- snip --- 'path_env' must have overflowed ... but how? I used a JIT debugger to examine the process environment block at the time of the crash since 'GetEnvironmentVariableW' reads from 'NtCurrentTeb()->Peb->ProcessParameters->Environment'. --- snip ---- Address UNICODE dump ... 00231EC0 m32\cmd.exe.PATH 00231EE0 =C:\windows\syst 00231F00 em32;C:\windows; 00231F20 C:\windows\syste 00231F40 m32\wbem;C:\wind 00231F60 ows\system32\gec 00231F80 ko\2.24\wine_gec 00231FA0 ko\;C:\windows\s 00231FC0 ystem32\gecko\2. 00231FE0 24\wine_gecko\;C 00232000 :\windows\system 00232020 32\gecko\2.24\wi 00232040 ne_gecko\;C:\win 00232060 dows\system32\ge 00232080 cko\2.24\wine_ge 002320A0 cko\;C:\windows\ 002320C0 system32\gecko\2 002320E0 .24\wine_gecko\. 00232100 TEMP=C:\users\fo 00232120 cht\Temp.TMP=C:\ 00232140 users\focht\Temp 00232160 .windir=C:\windo 00232180 ws.ALLUSERSPROFI 002321A0 LE=C:\users\Publ 002321C0 ic.APPDATA=C:\us ... --- snip --- At the time 'gre_path' path was appended, the string from 'PATH' environment variable had already grown near 'MAX_PATH' (260 characters) buffer limit. 'PATH' is of course not limited to 'MAX_PATH' since it contains a list of paths. A better option would be to query with 'GetEnvironmentVariableW( value, NULL, 0)' first and allocate the needed buffer from heap, including length for 'gre_path'. Even with these things corrected there is still a general problem: at one point it will overflow/being blocked from appending to 'PATH'. Each newly created updater process inherits the process environment from parent (client updater restarts itself each time). A more sophisticated thing to do would be to search for existing value and not append if already present. Wine Mono 'mscoree' component has a similar potential stack buffer overflow: http://source.winehq.org/git/wine.git/blob/8cdcf470016f0655dfc8810f9d4d2f2d… 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.

2 6

[Bug 49999] New: It should be possible to submit results for all versions of the current stable branch
by WineHQ Bugzilla Dec. 5, 2021

Dec. 5, 2021

https://bugs.winehq.org/show_bug.cgi?id=49999 Bug ID: 49999 Summary: It should be possible to submit results for all versions of the current stable branch Product: WineHQ Apps Database Version: unspecified Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: appdb-unknown Assignee: wine-bugs(a)winehq.org Reporter: elk_aide(a)startmail.com Distribution: --- Lots of Linux Distributions and FreeBSD only ship the stable releases of wine in their repositories. They often stick to the first minor release (In Debian and FreeBSD this currently is 5.0 for example). I think it should be possible to submit tests with the distribution's own most recent wine package, as that probably is the main way wine is aquired. Therefore the AppDB list should include the "outdated" minor releases of the current stable branch for selection aswell. This is currently not the case (only 5.0.2 is available for selection currently). Cheers! -- 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.

1 10

[Bug 47867] New: comctl32:animate: create_animate() fails in Japanese on Windows 10 1809
by WineHQ Bugzilla Dec. 5, 2021

Dec. 5, 2021

https://bugs.winehq.org/show_bug.cgi?id=47867 Bug ID: 47867 Summary: comctl32:animate: create_animate() fails in Japanese on Windows 10 1809 Product: Wine Version: 4.17 Hardware: x86 OS: Windows Status: NEW Severity: normal Priority: P2 Component: comctl32 Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com comctl32:animate has the following failures on the new w1064v1809_ja VM (Windows 10 1809 + Japanese locale): animate.c:106: Test failed: GetUpdateRect: There should be a region that needs to be updated animate.c:106: Test failed: GetUpdateRect: There should be a region that needs to be updated Interestingly this does not fail on the fg-win7u64-1spie9-ja VM so this is specific to the Japanese + Windows 10 combination. The test was introduced in this commit: commit bc357819296594aca300fdf30bb0cda46e9f0be0 Author: Bruno Jesus <00cpxxx(a)gmail.com> AuthorDate: Fri Mar 18 14:29:48 2016 +0530 Commit: Alexandre Julliard <julliard(a)winehq.org> CommitDate: Tue Mar 22 00:36:30 2016 +0900 comctl32/tests: Add initial tests for the animation control. Signed-off-by: Bruno Jesus <00cpxxx(a)gmail.com> Signed-off-by: Alexandre Julliard <julliard(a)winehq.org> -- 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.

1 3

[Bug 51899] New: 16-bit Collins Dictionary has black background.
by WineHQ Bugzilla Dec. 4, 2021

Dec. 4, 2021

https://bugs.winehq.org/show_bug.cgi?id=51899 Bug ID: 51899 Summary: 16-bit Collins Dictionary has black background. Product: Wine Version: 6.19 Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: gdi32 Assignee: wine-bugs(a)winehq.org Reporter: zzhang(a)codeweavers.com Distribution: --- This is a split from bug 51802. Bisect points to 6f095cea53787f03a159f0464b9881ace3724e6b is the first bad commit commit 6f095cea53787f03a159f0464b9881ace3724e6b Author: Jacek Caban <jacek(a)codeweavers.com> Date: Tue Aug 3 12:56:49 2021 +0200 gdi32: Use NtGdiPatBlt for PatBlt implementation. Signed-off-by: Jacek Caban <jacek(a)codeweavers.com> Signed-off-by: Huw Davies <huw(a)codeweavers.com> Signed-off-by: Alexandre Julliard <julliard(a)winehq.org> -- 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.

1 35

[Bug 31358] New: Bobcad v25 hangs on startup
by wine-bugs＠winehq.org Dec. 4, 2021

Dec. 4, 2021

http://bugs.winehq.org/show_bug.cgi?id=31358 Bug #: 31358 Summary: Bobcad v25 hangs on startup Product: Wine Version: 1.5.9 Platform: x86 OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: dank(a)kegel.com Classification: Unclassified Tried installing free trial 13f91536f3a429ccb65b45b7bf72adb927c7d294 BobCAD_V25_Demo_32bit.exe from http://www.bobcad.com/products/cad Seemed to install fine, and put up a UI, but the main window of the UI was never drawn, and the UI was nonresponsive. -- 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 24664] New: The letter 'R' is not visible in Shapes On A Plane
by wine-bugs＠winehq.org Dec. 4, 2021

Dec. 4, 2021

http://bugs.winehq.org/show_bug.cgi?id=24664 Summary: The letter 'R' is not visible in Shapes On A Plane Product: Wine Version: 1.3.4 Platform: x86 URL: https://www.digipen.edu/?id=1170&proj=433 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: andrew.millington(a)gmail.com Nearly all the letter 'R' are not visible in text such as the options screen. -- 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 12

[Bug 40111] New: mpr:mpr regression
by wine-bugs＠winehq.org Dec. 4, 2021

Dec. 4, 2021

https://bugs.winehq.org/show_bug.cgi?id=40111 Bug ID: 40111 Summary: mpr:mpr regression Product: Wine Version: 1.9.3 Hardware: x86 OS: Linux Status: NEW Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com Distribution: --- Since the commit below, all Windows XP machines (even non-VMs) except the TestBot have a couple of systematic failures. It's not clear why the TestBot VM does not get a failure. It could be because it's the only one to be a Pro version, or because of a subtle network configuration difference. Failure: mpr.c:283: Test failed: Unexpected return: 53 mpr.c:303: Test failed: Unexpected return: 53 mpr:mpr results on test.winehq.org: https://test.winehq.org/data/tests/mpr:mpr.html commit d2ce22ba239d6ea4955e58784cf3e912920bb93b Author: Pierre Schweitzer <pierre(a)reactos.org> Date: Sun Jan 31 22:17:27 2016 +0100 mpr/tests: Add tests for WNetUseConnectionA(). Signed-off-by: Pierre Schweitzer <pierre(a)reactos.org> Signed-off-by: Alexandre Julliard <julliard(a)winehq.org> -- 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 8

[Bug 47938] New: kernel32:heap fails on fg-win7u64-* since 9 September 2019
by WineHQ Bugzilla Dec. 4, 2021

Dec. 4, 2021

https://bugs.winehq.org/show_bug.cgi?id=47938 Bug ID: 47938 Summary: kernel32:heap fails on fg-win7u64-* since 9 September 2019 Product: Wine Version: 4.17 Hardware: x86 OS: Linux Status: NEW Severity: normal Priority: P2 Component: kernel32 Assignee: wine-bugs(a)winehq.org Reporter: z.figura12(a)gmail.com Distribution: --- <http://test.winehq.org/data/tests/kernel32:heap.html> Very consistently we see this: heap.c:1209: Test failed: GetPhysicallyInstalledSystemMemory unexpectedly failed heap.c:1210: Test failed: expected total_memory != 0 heap.c:1215: Test failed: expected total_memory >= memstatus.ullTotalPhys / 1024 I'm more than a little inclined to blame the virtualization software. -- 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.

1 5

[Bug 20987] New: msi tests can't be run in parallel
by wine-bugs＠winehq.org Dec. 4, 2021

Dec. 4, 2021

http://bugs.winehq.org/show_bug.cgi?id=20987 Summary: msi tests can't be run in parallel Product: Wine Version: 1.1.34 Platform: PC OS/Version: Linux Status: NEW Severity: enhancement Priority: P2 Component: testcases AssignedTo: wine-bugs(a)winehq.org ReportedBy: dank(a)kegel.com We spend an awful lot of time waiting for tests to finish. It'd be awesome if "make -j 100 test" worked. As a first step, let's make the msi tests work in parallel. "make -j100" in the msi directory explodes at the moment because many of the tests use the same filenames. -- 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 48047] New: urlmon:protocol - The https test fails on XP since the winehq.org upgrade
by WineHQ Bugzilla Dec. 4, 2021

Dec. 4, 2021

https://bugs.winehq.org/show_bug.cgi?id=48047 Bug ID: 48047 Summary: urlmon:protocol - The https test fails on XP since the winehq.org upgrade Product: Wine Version: unspecified Hardware: x86 OS: Windows Status: NEW Severity: normal Priority: P2 Component: urlmon Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com Created attachment 65588 --> https://bugs.winehq.org/attachment.cgi?id=65588 Screen capture of test.winehq showing the failures starting on Sep 09 The https protocol test in urlmon:protocol has been failing on Windows XP and 2003 since the winehq.org upgrade: protocol.c:3591: Testing https protocol (from urlmon)... protocol.c:1190: Test failed: hrResult = 800c0008, expected: 00000000 protocol.c:3435: Test failed: wait timed out protocol.c:3437: Test failed: expected Switch urlmon:protocol:0444 done (258) in 120s Test failed: timed out Note: 800c0008 is INET_E_DOWNLOAD_FAILURE. See https://test.winehq.org/data/tests/urlmon:protocol.html On Windows XP trying to open https://www.winehq.org in Internet Explorer (8.0.6001.18702) fails with no clear reason. Many other sites fail in the same way while trying to access google.com clearly indicates there is a certificate problem and asks whether to trust it anyway. As is Windows XP is unusable to surf the web. It's still possible to install Firefox 52.9.0 and then accessing https://www.winehq.org works. But then Firefox uses its own rot certificates and may not even use the Windows libraries for encryption. Windows Update offers no meaningful update (only Microsoft Security Essentials and Windows Live Essentials). So it's possible Windows XP and 2003 are either missing the required root certificates or do not support support whichever standard is used in the current certificates. In that case this test should be skipped on these platforms. -- 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.

1 4