https://bugs.winehq.org/show_bug.cgi?id=49077
Bug ID: 49077 Summary: Unhandled exception: page fault on read access to 0x00000000 in 32-bit code (0x782795a4) Product: Wine Version: 5.7 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs@winehq.org Reporter: rich.coe2@gmail.com Distribution: ---
distro: opensuse tumbleweed version wine-32bit-5.7-3.1.x86_64
I upgraded from previous tumbleweed version to new version yesterday. I'm running AdobeDigitalEditions that I've been using almost daily. Today was first day to run it after the upgrade yesterday.
I'm getting Unhandled exception: page fault on read access to 0x00000000 in 32- bit code (0x782795a4) when I run ade after running for about a minute. It does not seem to be reacting to what I'm doing in the UI at the time.
It's always the same address, 0x782795a4.
I invoke ADE this way: WINEARCH=win32 WINEPREFIX=$HOME/.winedotnet wine c://Program\ Files//adobe//Adobe\ digital\ editions\ 4.5//digitaleditions.exe
I ran ADE under winedbg, and after a while it crashed. The backtrace is empty. It suggests to me a stack corruption followed by a return to value overwritten on the stack.
I am looking for any diagnostics to track down issue. I can run in debug, compile from source, or provide necessary traces.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #1 from RC rich.coe2@gmail.com --- Created attachment 67050 --> https://bugs.winehq.org/attachment.cgi?id=67050 backtrace and thread dump
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #2 from RC rich.coe2@gmail.com --- installed ade with ADE_4.5.2_Installer.exe
https://bugs.winehq.org/show_bug.cgi?id=49077
Olivier F. R. Dierick o.dierick@piezo-forte.be changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Unhandled exception: page |Adobe Digital Editions |fault on read access to |4.5.2 is unresponsive and |0x00000000 in 32-bit code |crashes after a minute |(0x782795a4) |after update to Wine 5.7 CC| |o.dierick@piezo-forte.be URL| |https://web.archive.org/web | |/20160812055622/http://www. | |adobe.com/solutions/ebook/d | |igital-editions/download.ht | |ml Keywords| |download, regression
--- Comment #3 from Olivier F. R. Dierick o.dierick@piezo-forte.be --- Hello,
1. Please read and follow the bug writing guidelines thoroughly: https://wiki.winehq.org/Bugs → Rewriting summary to conform to bug writing guidelines.
2. Adobe Digital Editions is freely available from the vendor website. The current available version is 4.5.11, though. Is the issue reproducible with that version?
(In reply to RC from comment #2)
installed ade with ADE_4.5.2_Installer.exe
Please provide sha1sum or sha256sum of the file for comparison.
Older versions can be found on the archive.org Wayback Machine. I'm putting the URL to the first occurrence of the 4.5.2 download. → Adding URL and download keyword.
3. Backtraces are usually not helpful. Please attach a normal (=without WINEDEBUG) terminal output. Instructions to get a log can be found there: https://wiki.winehq.org/FAQ#get_log
4. (In reply to RC from comment #0)
distro: opensuse tumbleweed version wine-32bit-5.7-3.1.x86_64
I upgraded from previous tumbleweed version to new version yesterday.
What was the previous version before the update? 5.7-1/5.7-2 or <=5.6?
5. When the same version of the application used to work and still works with previous Wine versions, it is a regression. → Adding regression keyword.
Performing a regression test would help speed up things for the wine developers. Regression testing instructions can be found there: https://wiki.winehq.org/Regression_Testing
Regards.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #4 from RC rich.coe2@gmail.com --- Created attachment 67052 --> https://bugs.winehq.org/attachment.cgi?id=67052 wine window output
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #5 from RC rich.coe2@gmail.com --- previously installed version: 2020-03-23 16:23:41|install|wine-32bit|5.3-2.2|x86_64||tw-oss|7e8d4428903e718110e0fd4cdf7e09066d283a7191bb5d5da4 113768038ec0dc|
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #6 from Olivier F. R. Dierick o.dierick@piezo-forte.be --- Hello,
--- quote terminal output --- 0038:err:eventlog:ReportEventW L"Application: digitaleditions.exe\nFramework Version: v4.0.30319\nDescription: The process was terminated due to an unhandled exception.\nException Info: exception code c0000005, exception address 782795A4\n" --- end quote ---
The error relates to .NET Framework.
Is native .NET Framework installed in the wineprefix?
If not, does 'winetricks -q dotnet40 corefonts' help?
Regards.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #7 from RC rich.coe2@gmail.com --- Yes, dotnet is installed. ADE is a dotnet application.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #8 from RC rich.coe2@gmail.com --- I've tested master, bisect'd from wine-5.3 to wine-5.7, wine-5.3, and wine-5.2 and have it always fail. I suspect that a library that wine depends on might be the culprit, but it could also be something else too.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #9 from Olivier F. R. Dierick o.dierick@piezo-forte.be --- Hello,
Thanks for the feedback.
What is the sha1sum of ADE_4.5.2_Installer.exe?
Regards.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #10 from RC rich.coe2@gmail.com --- c56bf9034f2390e149859f5bb6472afd /usr/local/c/ADE_4.5_Installer.exe 10a49fcf27e1dda3c539b73685fc8b85 /usr/local/c/ADE_4.5.2_Installer.exe
I had issues with 4.5, so I obtained 4.5.2. I've been using 4.5.2 since Dec'2016. In Dec'2019, I moved from running ADE in a VM, to running under wine after I got the dotnet to successfully install on new clean wine directory.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #11 from Olivier F. R. Dierick o.dierick@piezo-forte.be --- Hello,
I checked all archive.org pages marked '4.5.2' but the downloads are really 4.5.1 (aug 8 '16 → nov 12 '16) and 4.5.5 installers (nov 15 '16 onwards).
I tested both installers like this: - Create new 32-bit wineprefix; - Enable 1152 x 864 virtual desktop in winecfg; - Disable 'Manage file association' in winecfg; - Install 'winetricks dotnet40 corefonts'; - Copy ADE_4.5_installer.exe in $WINEPREFIX/drive_c, open terminal there and run the installer; - Disable every shortcut/file association checkboxes when asked in the installer; - When the application starts automatically, open the default ebook and browse for a few minutes.
I didn't have any issues, so I can't confirm.
My tests were done with Wine 5.7 (release tag) + commit c3fac6e36caab168974dd04a60ae1bbb1a0fd919 (nitpicked from GIT to workaround bug 49011). My system is Debian 8.
Regards.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #12 from RC rich.coe2@gmail.com --- Created attachment 67059 --> https://bugs.winehq.org/attachment.cgi?id=67059 traceback with symbols
I built wine with CFLAGS='-g -fno-inline -fno-omit-frame-pointer -fno-optimize-sibling-calls -O1'
Running ADE under winedbg, I got the attached traceback showing the failure in nouveau_dri, at the same failing address: =>0 0x782795a4 __driDriverGetExtensions_swrast+0xffffffff() in nouveau_dri.so (0x00000000)
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #13 from RC rich.coe2@gmail.com --- It appears that p_glDrawArrays is invalid in gl_info->gl_ops, called from wined3d_context_gl_draw_primitive_arrays().
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #14 from RC rich.coe2@gmail.com --- Drawing on the main context: 00f4:fixme:d3d:wined3d_context_gl_draw_primitive_arrays DPA (0x67291a0, 0x3b062b4, 0x7c3892e0) draw array
A popup occurs: 00f4:fixme:d3d:wined3d_context_gl_draw_primitive_arrays DPA (0x912d4b8, 0x3b062b4, 0x7c3892e0) draw array
Click popup down. Click on another entry: 00f4:fixme:d3d:wined3d_context_gl_draw_primitive_arrays DPA (0x912d4b8, 0x3b062b4, 0x7c3892e0) draw array 0114:fixme:thread:NtQueryInformationThread ThreadIsIoPending info class not supported yet 00f4:fixme:d3d:wined3d_context_gl_draw_primitive_arrays DPA (0x65f72f8, 0x3b062b4, 0x7c3892e0) draw array Unhandled exception: page fault on read access to 0x00000000 in 32-bit code (0x782795a4).
Note that context 0x65f72f8 is a new context, never seen before.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #15 from RC rich.coe2@gmail.com --- Created attachment 67064 --> https://bugs.winehq.org/attachment.cgi?id=67064 d3d trace showing context 0x65f87d0
WINEDBUG=+d3d trace showing use of context 0x65f87d0 before crash
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #16 from Henri Verbeet hverbeet@gmail.com --- (In reply to RC from comment #15)
Created attachment 67064 [details] d3d trace showing context 0x65f87d0
WINEDBUG=+d3d trace showing use of context 0x65f87d0 before crash
I'm not sure whether you're saying it is, but note that context 0x65f87d0 is not new:
00f8:trace:d3d:wined3d_swapchain_gl_create_context Creating a new context for swapchain 0x65dcc00, thread 248. 00f8:trace:d3d:wined3d_context_gl_init context_gl 0x65f87d0, swapchain 0x65dcc00. 00f8:trace:d3d:context_choose_pixel_format device 0x65fa338, dc 0x190069, color_format WINED3DFMT_B8G8R8A8_UNORM, ds_format WINED3DFMT_UNKNOWN, aux_buffers 0. 00f8:trace:d3d:context_choose_pixel_format Found iPixelFormat=5 for ColorFormat=WINED3DFMT_B8G8R8A8_UNORM, DepthStencilFormat=WINED3DFMT_UNKNOWN. 00f8:trace:d3d:wined3d_context_gl_enter Entering context 0x65f87d0, level 1. 00f8:trace:d3d:wined3d_context_gl_set_current Switching to D3D context 0x65f87d0, GL context 0x23001, device context 0x190069. 00f8:trace:d3d:wined3d_context_gl_init Setting up the screen 00f8:trace:d3d:device_context_add Adding context 0x65f87d0. 00f8:trace:d3d:wined3d_swapchain_gl_create_context Created context 0x65f87d0.
Note also that nouveau is known to have some issues with using multiple GL contexts. That's not generally an issue for wined3d because the command stream serialises GL calls, but would be for applications that use multiple D3D devices concurrently. A quick test would be to force usage of llvmpipe by running with "LIBGL_ALWAYS_SOFTWARE=1" set.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #17 from RC rich.coe2@gmail.com --- Sorry about that, the 'new' context in the comment above was a different testcase than the +d3d trace. ------- I suspect that recent changes in Mesa 20.0.[45] may be causing this issue. Previous version was mesa-20.0.3. - a bug in mesa-20.0.5 nouveau - a fix in mesa-20.0.5 where some previous behaviour worked is now broken
I'm updating to mesa-20.0.6 that just showed up in the repo. There's a couple of bug fixes in there.
If 20.0.6 fixes the issue, then this was a regression in 20.0.5, that's now fixed. If it's still broken, I will see if mesa-20.0.3 fixes the issue, and bisect if need be.
If not, I'll have find the gl commands and ordering that's causing the issue (hope not ...).
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #18 from RC rich.coe2@gmail.com --- Still broken with Mesa-20.0.6 :-(
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #19 from RC rich.coe2@gmail.com --- Created attachment 67102 --> https://bugs.winehq.org/attachment.cgi?id=67102 CLR runtime crash
I've upgraded ADE to 4.5.11. I'm still trying track down memory corruption in glDrawArrays(). I started running under valgrind and during startup, I'm getting an exception while increasing the value of the stack pointer? (odd).
Also I don't know how bad or what it means when wine complains: preloader: Warning: failed to reserve range 00110000-68000000 preloader: Warning: failed to reserve range 7f000000-82000000 All web searches/FAQ talks about the 0 range issue, which did not change this from occurring.
Also seeing heap error (nil) and attempts to find where it is coming from have failed. 0048:err:heap:HEAP_GetPtr Invalid heap (nil)!
This is apparently badness in the dotnet run time: Unhandled exception: 0xe0434352 in 32-bit code (0x7b030449).
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #20 from RC rich.coe2@gmail.com ---
A quick test would be to force usage of llvmpipe by running with "LIBGL_ALWAYS_SOFTWARE=1" set.
I ran with with this env variable set, and glibc finds an error: *** buffer overflow detected ***
I'll add the messages logged.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #21 from RC rich.coe2@gmail.com --- Created attachment 67103 --> https://bugs.winehq.org/attachment.cgi?id=67103 libgl always software wine output
glDrawArrays() with GL software enabled.
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #22 from Henri Verbeet hverbeet@gmail.com --- (In reply to RC from comment #21)
LIBGL_ALWAYS_SOFTARE=1
You seem to have a small typo there.
https://bugs.winehq.org/show_bug.cgi?id=49077
RC rich.coe2@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Attachment #67103|0 |1 is obsolete| |
--- Comment #23 from RC rich.coe2@gmail.com --- Created attachment 67104 --> https://bugs.winehq.org/attachment.cgi?id=67104 libgl always software wine output (corrected)
corrected error with LIBGL_SOFTWARE_ALWAYS
https://bugs.winehq.org/show_bug.cgi?id=49077
Anastasius Focht focht@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- URL|https://web.archive.org/web |https://web.archive.org/web |/20160812055622/http://www. |/20160329143657if_/https:// |adobe.com/solutions/ebook/d |adedownload.adobe.com/pub/a |igital-editions/download.ht |dobe/digitaleditions/ADE_4. |ml |5_Installer.exe Summary|Adobe Digital Editions |Adobe Digital Editions 4.5 |4.5.2 is unresponsive and |is unresponsive and crashes |crashes after a minute |after a minute after update |after update to Wine 5.7 |to Wine 5.7 (Mesa 20.0.x / | |Nouveau) CC| |focht@gmx.net Ever confirmed|0 |1 Status|UNCONFIRMED |NEEDINFO
--- Comment #24 from Anastasius Focht focht@gmx.net --- Hello RC,
please retest with recent Wine 6.x release.
Also use an Adobe ADE 4.5.y version that is actually still obtainable either from Adobe or use archive.org snapshots.
https://web.archive.org/web/*/https://adedownload.adobe.com/pub/adobe/digita...
Sadly Adobe doesn't use unique filenames for ADE patch releases but simply overwrites the existing installer hence it's hard to reproduce unless archive.org has snapshots from that time frame. There are no snapshots for ADE 4.5.2. Snapshots exist for ADE 4.5.1, 4.5.9, 4.5.10, 4.5.10 and later, see archive.org link.
The app works fine here, like Olivier stated in comment #11. This indicates you likely have a broken graphics driver setup with -> upstream. Tested with dual GPU system (Prime), Intel Haswell/Mesa 20.2.3 and NVIDIA 460.x proprietary graphics driver. I'm not going to mess up my system with inferior Nouveau driver just for the sake of testing.
$ wine --version wine-6.2-212-g4de079bb724
Regards
https://bugs.winehq.org/show_bug.cgi?id=49077
--- Comment #25 from RC rich.coe2@gmail.com --- I will give it a try. Thanks for the update.
https://bugs.winehq.org/show_bug.cgi?id=49077
Zebediah Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |NOTOURBUG Status|NEEDINFO |RESOLVED Keywords|regression |
--- Comment #26 from Zebediah Figura z.figura12@gmail.com --- Based on comment 8, this is not a regression. Rather, it's almost certainly a matter of nouveau handling multithreaded GL poorly.