February 2019 - Wine-Bugs - list.winehq.org

[Bug 10913] New: start.exe doesn't support optional process title argument
by wine-bugs＠winehq.org March 3, 2019

March 3, 2019

http://bugs.winehq.org/show_bug.cgi?id=10913 Summary: start.exe doesn't support optional process title argument Product: Wine Version: CVS/GIT Platform: Other URL: http://jedit.org OS/Version: other Status: NEW Keywords: download Severity: normal Priority: P2 Component: wine-programs AssignedTo: wine-bugs(a)winehq.org ReportedBy: dank(a)kegel.com Running real Java apps under the latest Sun win32 JRE on Wine seems like a fine stress test, even though in many cases users should run with the Linux JRE instead. The Sun JRE and jEdit install ok, but jedit has many startup problems. Here's the first one. Invoking it with the batch file provided by jEdit fails. Here's what's in the batch file: start "jEdit startup" "c:\windows\system32\javaw.exe" -Xms64M -Xmx192M -jar "C:\Program Files\jEdit\jedit.jar" -reuseview %* And here's what happens when you run it: $ cd ~/.wine/drive_c/Program Files/jEdit $ ~/wine-git/wine cmd /c jedit.bat fixme:exec:SHELL_execute flags ignored: 0x00000500 trace:process:CreateProcessW app (null) cmdline L"jEdit startup c:\\windows\\system32\\javaw.exe -Xms64M -Xmx192M -jar \"C:\\Program Files\\jEdit\\jedit.jar\" -reuseview" trace:process:find_exe_file looking for L"jEdit" trace:process:find_exe_file Trying native exe L"c:\\windows\\system32\\jEdit.exe" ... trace:process:find_exe_file Trying built-in exe L"c:\\windows\\system32\\jEdit startup.exe" ... trace:process:find_exe_file looking for L"jEdit startup c:\\windows\\system32\\javaw.exe -Xms64M -Xmx192M -jar \"C:\\Program Files\\jEdit\\jedit.jar\" -reuseview" Application could not be started, or no application associated with the specified file. ShellExecuteEx failed: File not found So it seems that start takes an optional process title argument. Sure enough, http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en… documents it. Who knew? -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching all bug changes.

1 20

[Bug 42508] New: start.exe does not detect its title argument when it should ( breaking .e.g URL opening in League of Legends)
by wine-bugs＠winehq.org March 3, 2019

March 3, 2019

https://bugs.winehq.org/show_bug.cgi?id=42508 Bug ID: 42508 Summary: start.exe does not detect its title argument when it should (breaking .e.g URL opening in League of Legends) Product: Wine Version: unspecified Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: programs Assignee: wine-bugs(a)winehq.org Reporter: hydratech(a)gmail.com Distribution: --- Created attachment 57397 --> https://bugs.winehq.org/attachment.cgi?id=57397 Trace showing League of Legends Client attempting to open a URL. When start.exe is invoked, its first argument in quotes that is not an argument of the application to be executed should be interpreted as the console title. Wine's start.exe fails to do so. -- How to reproduce -- Running the following line in Wine's cmd should open a URL in your browser: Z:\>start "" https://winehq.org/ the actual result is an application not found error. On Windows the same line opens the website as expected. -- Why this is a bug -- I know this was marked as fixed in bug 10913 but the reasoning was incorrect. The argument for stating the bug was fixed was, that when \"escaping your quotes\" in cmd to the title was parsed. This is however not consistent with with Windows behaviour. As explained at the end of that bug report, if you escape your quotes in Wine's cmd start.exe parses the argument as the console title, i.e.: Z:\>start \"hello\" cmd I will now clarify why this is incorrect by showing Windows' behaviour. The following command lines are not equivalent on Windows: C:\>start hello.exe vs. C:\>start "hello.exe" The latter will open a new cmd window titled hello.exe whereas the former attempts to execute an application named hello.exe. In other words cmd's start built-in parses its own command-line checking wether quotes where used. In fact escaping the quotes using back-slashes is invalid as it tries start a process named \hello.exe\. I have tested this on both Windows 7 and Windows XP. In other words, this is invalid: C:>start \"hello.exe\" To illustrate how unintuitive this Windows behaviour is: C:>start "C:\Program Files\Application\hello.exe" Few would guess this actually opens a console titled as such, instead of running an application. That's why I can understand the maintainers would think the bug was fixed. Wine will however correctly emulate unescaped quote-senstive behaviour with the echo built-in: Z:\>echo hoi hoi Z:\>echo "hoi" "hoi" We can see echo being "aware" of the quotes and correctly displaying them; behaviour identical to Windows' echo built-in. Looking at the wine source, start.exe relies on its "argv" parameter instead of using GetCommandLine() and thereby accessing the unmodified command-line. As a consequence the difference between a quoted argument and a non-quoted argument is lost as CreateProcess strips the quotes off the parameters when passing them to execve(). To quote the source of Wine's cmd.exe (wcmdmain.c): /* Can't use argc/argv as it will have stripped quotes from parameters * meaning cmd.exe /C echo "quoted string" is impossible */ It should now be evident that start.exe should either become a wrapper around cmd.exe's built-in start function, invoking cmd and let cmd itself implement start.exe's actual behaviour, or implement GetCommandLine() based argument parsing similar to cmd.exe's within start.exe, leading to duplicated command-line parsing code. Both are not ideal solutions. -- Why does this bug matter? -- There are actually Windows applications relying on this behaviour to open URLs in the user's browser. As an example, League of Legends executes the following shell command to open a URL: cmd.exe /c start "" "<url here>" In other words, League instructs start to launch an untitled shell window with a URL as application, effectively causing the shell to pass it to the browser. However, wine's start.exe interprets this as League trying to execute an empty string as executable name, with a URL as argument. I have attached the relevant output running League with WINEDEBUG=cmd,start,process,exec set. I will speculate that more applications will run into this exact same issue. League's client is based on CEF and it might very well be CEF which implements this style of external URL opening, furthermore there is documentation recommending the use of empty quotes as console title argument when invoking start to prevent it from accidentally parsing follow up arguments in quotes as the console title, think "C:\Program Files\etc...", establishing this as a pattern that might well have been adopted by other applications. See: https://ss64.com/nt/start.html I will also say that, given some instruction, I will be glad to help fixing this bug and write any necessary code. There is actually another issue with current start.exe's console title parsing which I will submit as a separate issue. That issue is however, significantly easier to fix. Finally this also means wine's current handling of command-line arguments might need fixing as: $ wine start '""' "http://winehq.og" will break in a fixed version of 'start.exe', while it should work. For this I will file yet another issue, although the necessity of this fix is debatable. -- 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 13

[Bug 46741] New: Cemu says "This program has encountered a serious problem " after opening Mario Maker
by wine-bugs＠winehq.org March 2, 2019

March 2, 2019

https://bugs.winehq.org/show_bug.cgi?id=46741 Bug ID: 46741 Summary: Cemu says "This program has encountered a serious problem" after opening Mario Maker Product: Wine Version: 4.0-rc5 Hardware: x86 OS: Mac OS X Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: 2021duffykod(a)isd47.org I'm using Cemu version 1.15.2 and I have open GL 4.1 on my Mac. Also I'm opening the file "Block.rpx" from the mario maker rom. -- 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 46707] New: Microsoft Word 2019 freeze the whole screen on nouveau driver with CSMT enabled
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

https://bugs.winehq.org/show_bug.cgi?id=46707 Bug ID: 46707 Summary: Microsoft Word 2019 freeze the whole screen on nouveau driver with CSMT enabled Product: Wine Version: 4.2 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: minor Priority: P2 Component: directx-d3d Assignee: wine-bugs(a)winehq.org Reporter: peathot(a)hotmail.com Distribution: --- Created attachment 63673 --> https://bugs.winehq.org/attachment.cgi?id=63673 Console output with CSMT enabled When using Microsoft Word 2019 (version 1902, build 11328.20100, installed via Office 365) for a while, the program, followed by the whole screen, will freeze. The mouse pointer is still movable, and I can SSH into the machine and kill WINWORD.EXE. After a moment, the screen will become usable again. By disabling CSMT in registry (HKCU/Software/Wine/Direct3D, csmt=dword:00000000), the problem doesn't occurred. The console output for both when CSMT is enabled and disabled is attached. Also, the kernel messages when CSMT is enabled is attached (there's no kernel messages when CSMT is disabled). This is Wine 4.2 on Xubuntu 18.04. The machine is Mac Mini early 2009 with Nvidia Geforce 9400m [1]. The graphic stack is mesa + nouveau (i.e. Nvidia proprietary driver is not installed). The MS Office is installed using the instructions from [2]. The first registry workaround is also used. Actually, the kernel messages suggests that it's nouveau fault. From my research, it seems like nouveau doesn't like OpenGL calls from multiple thread [3][4]. If it's actually nouveau's fault, I guess we could disable CSMT by default if we detect nouveau driver? [1] Yes, it's a bit old. But this same machine can run this software on Windows 10. So, it's not the problem. [2] https://appdb.winehq.org/objectManager.php?sClass=version&iId=35527&iTestin… [3] https://bugs.freedesktop.org/show_bug.cgi?id=92438#c39 [4] https://bugreports.qt.io/browse/QTBUG-73715 -- 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

[Bug 46695] New: Linking atl fails on ppc64el with undefined reference to SetWindowLongPtrW
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

https://bugs.winehq.org/show_bug.cgi?id=46695 Bug ID: 46695 Summary: Linking atl fails on ppc64el with undefined reference to SetWindowLongPtrW Product: Wine Version: 4.2 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: tpearson(a)raptorengineering.com Distribution: --- When building on ppc64el, linking atl fails with an undefined reference to SetWindowLongPtrW. I suspect this is due to the existing PowerPC code assuming ppc platforms are 32 bit only; SetWindowLongPtrW is a 64 bit call. -- 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 31644] New: glMapBufferRange is extremely slow while playing Resident Evil 5 ( happens only on nvidia binary driver )
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

http://bugs.winehq.org/show_bug.cgi?id=31644 Bug #: 31644 Summary: glMapBufferRange is extremely slow while playing Resident Evil 5 ( happens only on nvidia binary driver ) Product: Wine Version: 1.5.12 Platform: x86 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d AssignedTo: wine-bugs(a)winehq.org ReportedBy: tizbac2(a)gmail.com Classification: Unclassified Calls to glMapBufferRange are extremely slow on nvidia drivers with resident evil 5, if this should be reported to nvidia instead just close it over 90% of frame time is taken by glMapBufferRange -- 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.

1 20

[Bug 38488] New: Wrong taces in functions *_invalidate_location
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

https://bugs.winehq.org/show_bug.cgi?id=38488 Bug ID: 38488 Summary: Wrong taces in functions *_invalidate_location Product: Wine Version: 1.7.41 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d Assignee: wine-bugs(a)winehq.org Reporter: isakov-sl(a)bk.ru Distribution: --- I got a plenty of messages like fixme:d3d:wined3d_debug_location Unrecognized location flag(s) 0xfffffc00. fixme:d3d:wined3d_debug_location Unrecognized location flag(s) 0xfffffc00. fixme:d3d:wined3d_debug_location Unrecognized location flag(s) 0xfffffc00. fixme:d3d:wined3d_debug_location Unrecognized location flag(s) 0xfffffc00. It rises because the function wined3d_volume_invalidate_location() is used with reversed parameter For example dlls/wined3d/device.c: wined3d_volume_invalidate_location(dst_volume, ~WINED3D_LOCATION_TEXTURE_RGB); The function assumed location and not ~location. ----- void wined3d_volume_invalidate_location(struct wined3d_volume *volume, DWORD location) { TRACE("Volume %p, clearing %s.\n", volume, wined3d_debug_location(location)); volume->locations &= ~location; TRACE("new location flags are %s.\n", wined3d_debug_location(volume->locations)); } ----- Here wined3d_debug_location() will message like above. Same about wined3d_resource_invalidate_location(struct wined3d_resource *resource, DWORD location) -- 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 6

[Bug 45149] New: regedit crashes (with no visible crash dialog) in winetricks dotnet35sp1
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

https://bugs.winehq.org/show_bug.cgi?id=45149 Bug ID: 45149 Summary: regedit crashes (with no visible crash dialog) in winetricks dotnet35sp1 Product: Wine Version: 3.7 Hardware: x86 OS: Linux Status: NEW Keywords: download Severity: normal Priority: P2 Component: kernel32 Assignee: wine-bugs(a)winehq.org Reporter: austinenglish(a)gmail.com CC: z.figura12(a)gmail.com Distribution: Gentoo Originally reported at https://github.com/Winetricks/winetricks/issues/810 winetricks hangs when running regedit (though this doesn't happen anywhere else). If you run ps aux, you can see that winedbg --auto is running, but there's no crash dialog (it was not diabled). Talking with Zebediah about it on IRC, he pointed out that staging has a patch that allows the dialog to at least show: https://github.com/wine-staging/wine-staging/blob/master/patches/kernel32-D… I couldn't find any other bug bug referencing it. Of course, there's a second bug, in that running regedit crashes (when setting the windows version).. -- 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 11

[Bug 18496] New: BREW simulator always shows a black display
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

http://bugs.winehq.org/show_bug.cgi?id=18496 Summary: BREW simulator always shows a black display Product: Wine Version: 1.1.21 Platform: Other OS/Version: other Status: UNCONFIRMED Severity: enhancement Priority: P2 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: jon.dufresne(a)gmail.com Tested with wine-1.1.21 I am attempting to use the Qualcomm BREW simulator under wine. After installing, I attempted to run the example applets. The application launches and I am able to select an applet directory, but the simulated device's screen always shows black and not the usual list of icons that represent available applets. The BREW simulator installer is found here: https://brewx.qualcomm.com/bws/content/gi/products/sdk/3.1/en/3.1.5.179/ins… Steps to reproduce: 1. Download and install BREW SDK simulator 2. Run simulator 3. Use menu to selec "File->Change Applet Dir..." 4. Navigate to c:/Program Files/BREW 3.1.5/sdk/examples/ 5. Hit "OK" 6. Notice screen is black Expected results: Under windows, the BREW simulator will show a listing of the available BREW applets (one for each *.mif file in the directory chosen). This list appears as a series of icons inside the device's display area. -- 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.

1 17

[Bug 15703] New: LEGO Star Wars controller options
by wine-bugs＠winehq.org March 1, 2019

March 1, 2019

http://bugs.winehq.org/show_bug.cgi?id=15703 Summary: LEGO Star Wars controller options Product: Wine Version: 1.1.6 Platform: Other OS/Version: other Status: UNCONFIRMED Severity: enhancement Priority: P2 Component: -unknown AssignedTo: wine-bugs(a)winehq.org ReportedBy: infernux(a)web.de Going to "Controller Options" freezes the game, pressing "Esc" does not make it playable again. I don't know what other information / logs would be helpful, if you need anything specific just say so. This only affects the _first_ of the two games - but the 2nd seems to have a different input system anyway (requires xinpit1_2.dll for example). -- 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.

1 17