[Bug 1094733] New: Widgets bad rendering in plasmashell
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733 Bug ID: 1094733 Summary: Widgets bad rendering in plasmashell Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.0 Hardware: x86-64 OS: openSUSE Factory Status: NEW Severity: Normal Priority: P5 - None Component: KDE Workspace (Plasma) Assignee: opensuse-kde-bugs@opensuse.org Reporter: felipesabrahao@gmail.com QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Created attachment 771438 --> http://bugzilla.opensuse.org/attachment.cgi?id=771438&action=edit Gdb plasmashell -- System: openSUSE Leap 15.0 (currently updated) -- Bug Description: As shown in the pic (.png) the widgets in the panel and desktop crash and remain with a very weird render. -- Reproducibility: Sometimes, when I hover the mouse arrow over the panel (auto-hide enabled). -- Workaround: "pkill plasmashell" "kstart plasmashell" -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c1
--- Comment #1 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c2
--- Comment #2 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c3
--- Comment #3 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c5
--- Comment #5 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c6
--- Comment #6 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c7
--- Comment #7 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c8
Fabian Vogt
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c9
Michal Srb
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c10
--- Comment #10 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c11
--- Comment #11 from Felipe Abrahao
xf86-video-intel should not be used, so uninstall that again.
That it works with LIBGL_ALWAYS_SOFTWARE=1 means it's a bug in the graphics stack and unlikely Plasma itself, so reassigning.
Can you provide a backtrace for the software rendering crash?
-- Test "LIBGL_ALWAYS_SOFTWARE=1 plasmashell" Plasmashell crashed and, then, kwin_x11 crashed and after a while plasmashell was back again automatically. I'm attaching the "gdb -p 'plasmashell'" file until the plasmashell brief crash and kwin_x11 crash. I'm also attaching "gdb kwin_x11" file that I had use to restart kwin_x11 after the crash. -- Reproducibility: I have now pretty confidence after several tests that the issue only happens on my laptop running on battery. Thus, I'm attaching my tlp config file. The issue doesn't happen on Tumbleweed (Plasma 5.12, Qt 5.10, Kernel 4.15) and on Leap 42.3 (default) on the same laptop. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c12
--- Comment #12 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c13
--- Comment #13 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c14
--- Comment #14 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c15
--- Comment #15 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c16
Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c17
Wolfgang Bauer
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c18
--- Comment #18 from Felipe Abrahao
You need to type "bt" into gdb to actually produce a backtrace and provide that.
So far it's just "console output" that probably doesn't really help at all.
And btw, you should clear the "needinfo" flag after you provided the requested information, e.g. by ticking the "I am providing the requested information for ..." checkbox when you reply.
Thank you :) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c19
Felipe Abrahao
It could as well be application bug that manifests only with certain drivers. I've seen many of those...
Please also try to start plasma with:
MESA_DEBUG=incomplete_tex,incomplete_fbo,context plasmashell 2> plasmashell.log
Attach the resulting plasmashell.log after you reproduce the issue.
-- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c20
--- Comment #20 from Felipe Abrahao
(In reply to Michal Srb from comment #9)
It could as well be application bug that manifests only with certain drivers. I've seen many of those...
Please also try to start plasma with:
MESA_DEBUG=incomplete_tex,incomplete_fbo,context plasmashell 2> plasmashell.log
Attach the resulting plasmashell.log after you reproduce the issue.
-- Bug: The same initially reported rendering issue happened on "MESA_DEBUG=incomplete_tex,incomplete_fbo,context plasmashell 2>
plasmashell.log".
I got a web site error when attaching the requested file plasmashell.log Once I reproduce the bug again I send the file. Sorry. -- Backtrace after the bug happened: (gdb) bt #0 0x00007f0d26b5108b in __GI___poll (fds=0x55cabb5e8fc0, nfds=10, timeout=6794) at ../sysdeps/unix/sysv/linux/poll.c:29 #1 0x00007f0d21441109 in g_main_context_poll (priority=<optimized out>, n_fds=10, fds=0x55cabb5e8fc0, timeout=<optimized out>, context=0x7f0d10004fc0) at gmain.c:4169 #2 g_main_context_iterate (context=context@entry=0x7f0d10004fc0, block=block@entry=1, dispatch=dispatch@entry=1, self=<optimized out>) at gmain.c:3863 #3 0x00007f0d2144121c in g_main_context_iteration (context=0x7f0d10004fc0, may_block=may_block@entry=1) at gmain.c:3929 #4 0x00007f0d2747fbef in QEventDispatcherGlib::processEvents ( this=0x55cab7f97290, flags=...) at kernel/qeventdispatcher_glib.cpp:423 #5 0x00007f0d2742809a in QEventLoop::exec (this=this@entry=0x7ffebe133e30, flags=..., flags@entry=...) at kernel/qeventloop.cpp:212 #6 0x00007f0d274309e4 in QCoreApplication::exec () at kernel/qcoreapplication.cpp:1289 #7 0x000055cab7932e20 in main (argc=<optimized out>, argv=<optimized out>) at /usr/src/debug/plasma5-workspace-5.12.5-lp150.1.1.x86_64/shell/main.cpp:172 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c21
--- Comment #21 from Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c22
--- Comment #22 from Felipe Abrahao
It could as well be application bug that manifests only with certain drivers. I've seen many of those...
Please also try to start plasma with:
MESA_DEBUG=incomplete_tex,incomplete_fbo,context plasmashell 2> plasmashell.log
Attach the resulting plasmashell.log after you reproduce the issue.
Attached :) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c23
Michal Srb
Attached :)
Great, thank you! It contains many errors like this: Mesa: User error: GL_INVALID_VALUE in glTexSubImage2D(xoffset 754 + width 260 > 952) It means that plasma called glTexSubImage2D trying to modify part of a texture, but that part is out of bounds for the current texture. That could explain the visual artifacts. Sadly, it is not easy to guess where is the offending call coming from. Could you try to run plasmashell in debugger (with the MESA_DEBUG variable set), put breakpoint on _mesa_error and generate core file when it is hit? The whole procedure may look like this:
MESA_DEBUG=incomplete_tex,incomplete_fbo,context gdb plasmashell ... (gdb) break _mesa_error Function "_mesa_error" not defined. Make breakpoint pending on future shared library load? (y or [n]) y (gdb) run ... Running until first error happens ... Thread 1 "plasmashell" hit Breakpoint 1, _mesa_error ... (gdb) generate-core-file Saved corefile core.1234 (gdb) quit
Then please attach the core file. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c24
Felipe Abrahao
xf86-video-intel should not be used, so uninstall that again.
That it works with LIBGL_ALWAYS_SOFTWARE=1 means it's a bug in the graphics stack and unlikely Plasma itself, so reassigning.
Can you provide a backtrace for the software rendering crash?
(gdb) bt #0 0x00007f2c1836908b in __GI___poll (fds=fds@entry=0x7f2c08ab2b88, nfds=nfds@entry=1, timeout=timeout@entry=-1) at ../sysdeps/unix/sysv/linux/poll.c:29 #1 0x00007f2c1d369387 in poll (__timeout=-1, __nfds=1, __fds=0x7f2c08ab2b88) at /usr/include/bits/poll2.h:46 #2 _xcb_conn_wait (c=c@entry=0x56166511c7b0, cond=cond@entry=0x56166511c7f0, vector=vector@entry=0x0, count=count@entry=0x0) at xcb_conn.c:479 #3 0x00007f2c1d36afaa in xcb_wait_for_event (c=0x56166511c7b0) at xcb_in.c:697 #4 0x00007f2c0ae29029 in QXcbEventReader::run (this=0x561665127e30) at qxcbconnection.cpp:1330 #5 0x00007f2c18a740ce in QThreadPrivate::start (arg=0x561665127e30) at thread/qthread_unix.cpp:368 #6 0x00007f2c17c22559 in start_thread (arg=0x7f2c08ab3700) at pthread_create.c:465 #7 0x00007f2c1837382f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c25
Felipe Abrahao
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c26
Fabian Vogt
Created attachment 772611 [details] Complete console gdb bt plasmashell always software rendering
Well, a backtrace is only helpful if it crashed. That's normal operation you uploaded. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c27
Felipe Abrahao
(In reply to Felipe Abrahao from comment #22)
Attached :)
Great, thank you! It contains many errors like this:
Mesa: User error: GL_INVALID_VALUE in glTexSubImage2D(xoffset 754 + width 260 > 952)
It means that plasma called glTexSubImage2D trying to modify part of a texture, but that part is out of bounds for the current texture. That could explain the visual artifacts.
Sadly, it is not easy to guess where is the offending call coming from. Could you try to run plasmashell in debugger (with the MESA_DEBUG variable set), put breakpoint on _mesa_error and generate core file when it is hit? The whole procedure may look like this:
MESA_DEBUG=incomplete_tex,incomplete_fbo,context gdb plasmashell ... (gdb) break _mesa_error Function "_mesa_error" not defined. Make breakpoint pending on future shared library load? (y or [n]) y (gdb) run ... Running until first error happens ... Thread 1 "plasmashell" hit Breakpoint 1, _mesa_error ... (gdb) generate-core-file Saved corefile core.1234 (gdb) quit
Then please attach the core file.
Here it is :) Since it is too big to be uploaded, I'm sending the dropbox link (~3Gb): https://www.dropbox.com/s/v720oq66h7bccld/mesadebug%20generated%20core%20fil... In the case it was useful anyway I'm sending the gdb bt too below: (gdb) . . . QXcbConnection: XCB error: 2 (BadValue), sequence: 21826, resource id: 83886719, major code: 142 (Unknown), minor code: 3 QXcbConnection: XCB error: 2 (BadValue), sequence: 21827, resource id: 83889928, major code: 142 (Unknown), minor code: 3 Thread 1 "plasmashell" hit Breakpoint 1, _mesa_error ( ctx=ctx@entry=0x555556424f80, error=1281, fmtString=0x7fff477860b8 "%s(xoffset %d + width %d > %u)") at main/errors.c:268 268 { (gdb) generate-core-file Saved corefile core.21051 (gdb) bt #0 _mesa_error (ctx=ctx@entry=0x555556424f80, error=1281, fmtString=0x7fff477860b8 "%s(xoffset %d + width %d > %u)") at main/errors.c:268 #1 0x00007fff471e2d69 in error_check_subtexture_dimensions ( ctx=ctx@entry=0x555556424f80, dims=dims@entry=2, destImage=destImage@entry=0x555557d64120, xoffset=xoffset@entry=1322, yoffset=yoffset@entry=86, zoffset=zoffset@entry=0, subWidth=34, subHeight=1, subDepth=1, func=0x7fff47786549 "glTexSubImage2D") at main/teximage.c:1199 #2 0x00007fff471e6169 in texsubimage_error_check (ctx=ctx@entry=0x555556424f80, dimensions=dimensions@entry=2, texObj=texObj@entry=0x555557d6d310, target=target@entry=3553, level=level@entry=0, xoffset=xoffset@entry=1322, yoffset=86, zoffset=0, width=34, height=1, depth=1, format=32993, type=5121, pixels=0x7fffffffc450, callerName=0x7fff47786549 "glTexSubImage2D", dsa=<optimized out>) at main/teximage.c:2249 #3 0x00007fff471e6706 in texsubimage_error_check (dsa=false, callerName=<optimized out>, pixels=0x7fffffffc450, type=5121, format=32993, depth=1, height=1, width=34, zoffset=0, yoffset=86, xoffset=1322, level=0, ---Type <return> to continue, or q <return> to quit--- target=3553, texObj=0x555557d6d310, dimensions=2, ctx=0x555556424f80) at main/teximage.c:2199 #4 texsubimage_err (ctx=0x555556424f80, dims=2, target=3553, level=0, xoffset=1322, yoffset=86, zoffset=0, width=34, height=1, depth=1, format=32993, type=5121, pixels=0x7fffffffc450, callerName=0x7fff47786549 "glTexSubImage2D") at main/teximage.c:3358 #5 0x00007fff471e9b88 in _mesa_TexSubImage2D (target=<optimized out>, level=<optimized out>, xoffset=<optimized out>, yoffset=<optimized out>, width=<optimized out>, height=<optimized out>, format=32993, type=5121, pixels=0x7fffffffc450) at main/teximage.c:3594 #6 0x00007ffff57abb3d in QOpenGLFunctions::glTexSubImage2D ( pixels=<optimized out>, type=5121, format=<optimized out>, height=1, width=34, yoffset=<optimized out>, xoffset=<optimized out>, level=0, target=3553, this=<optimized out>) at /usr/include/qt5/QtGui/qopenglfunctions.h:1074 #7 QSGAtlasTexture::Atlas::uploadBgra (this=0x555555d2b490, texture=<optimized out>) ---Type <return> to continue, or q <return> to quit--- at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/util/qsgatlastexture.cpp:305 #8 0x00007ffff57ac85d in QSGAtlasTexture::Atlas::bind (this=0x555555d2b490, filtering=QSGTexture::Nearest) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/util/qsgatlastexture.cpp:408 #9 0x00007ffff57a909a in QSGOpaqueTextureMaterialShader::updateState ( this=0x555555d23020, state=..., newEffect=<optimized out>, oldEffect=0x555559b46b80) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/util/qsgtexturematerial.cpp:116 #10 0x00007ffff57986d8 in QSGBatchRenderer::Renderer::renderMergedBatch ( this=0x555558dcf680, batch=0x555558357ad0) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp:2326 #11 0x00007ffff5799a25 in QSGBatchRenderer::Renderer::renderBatches ( ---Type <return> to continue, or q <return> to quit--- this=this@entry=0x555558dcf680) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp:2569 #12 0x00007ffff579eecc in QSGBatchRenderer::Renderer::render (this=<optimized out>) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/coreapi/qsgbatchrenderer.cpp:2763 #13 0x00007ffff578fedd in QSGRenderer::renderScene (this=0x555558dcf680, bindable=...) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/coreapi/qsgrenderer.cpp:243 #14 0x00007ffff579036b in QSGRenderer::renderScene (this=<optimized out>, fboId=<optimized out>) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/coreapi/qsgrenderer.cpp:196 #15 0x00007ffff57c8e20 in QSGDefaultRenderContext::renderNextFrame ( this=0x555555fffc80, renderer=0x555558dcf680, fboId=<optimized out>) ---Type <return> to continue, or q <return> to quit--- at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/qsgdefaultrendercontext.cpp:181 #16 0x00007ffff5822b78 in QQuickWindowPrivate::renderSceneGraph ( this=this@entry=0x555558493c60, size=...) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/items/qquickwindow.cpp:476 #17 0x00007ffff57bbab4 in QSGGuiThreadRenderLoop::renderWindow ( this=this@entry=0x555556157600, window=<optimized out>) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/qsgrenderloop.cpp:417 #18 0x00007ffff57bc9da in QSGGuiThreadRenderLoop::exposureChanged ( this=0x555556157600, window=<optimized out>) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/scenegraph/qsgrenderloop.cpp:465 #19 0x00007ffff220d6db in QWindow::event (this=this@entry=0x55555849c120, ev=ev@entry=0x7fffffffd760) at kernel/qwindow.cpp:2267 ---Type <return> to continue, or q <return> to quit--- #20 0x00007ffff582bf45 in QQuickWindow::event (this=this@entry=0x55555849c120, e=e@entry=0x7fffffffd760) at /usr/src/debug/libqt5-qtdeclarative-5.9.4-lp150.1.6.x86_64/src/quick/items/qquickwindow.cpp:1609 #21 0x00007ffff799cbc3 in PlasmaQuick::Dialog::event ( this=this@entry=0x55555849c120, event=event@entry=0x7fffffffd760) at /usr/src/debug/plasma-framework-5.45.0-lp150.2.1.x86_64/src/plasmaquick/dialog.cpp:1301 #22 0x00007fff41b3b015 in ToolTipDialog::event (this=0x55555849c120, e=0x7fffffffd760) at /usr/src/debug/plasma-framework-5.45.0-lp150.2.1.x86_64/src/declarativeimports/core/tooltipdialog.cpp:99 #23 0x00007ffff2e13e8c in QApplicationPrivate::notify_helper ( this=this@entry=0x55555582ea20, receiver=receiver@entry=0x55555849c120, e=e@entry=0x7fffffffd760) at kernel/qapplication.cpp:3723 #24 0x00007ffff2e1b244 in QApplication::notify (this=0x7fffffffdb20, ---Type <return> to continue, or q <return> to quit--- receiver=0x55555849c120, e=0x7fffffffd760) at kernel/qapplication.cpp:3482 #25 0x00007ffff1c30a88 in QCoreApplication::notifyInternal2 ( receiver=receiver@entry=0x55555849c120, event=event@entry=0x7fffffffd760) at kernel/qcoreapplication.cpp:1016 #26 0x00007ffff2202fe9 in QCoreApplication::sendSpontaneousEvent ( event=0x7fffffffd760, receiver=0x55555849c120) at ../../include/QtCore/../../src/corelib/kernel/qcoreapplication.h:236 #27 QGuiApplicationPrivate::processExposeEvent (e=0x7fffdc025160) at kernel/qguiapplication.cpp:2875 #28 0x00007ffff2203b4d in QGuiApplicationPrivate::processWindowSystemEvent ( e=e@entry=0x7fffdc025160) at kernel/qguiapplication.cpp:1798 #29 0x00007ffff21dc65b in QWindowSystemInterface::sendWindowSystemEvents ( flags=...) at kernel/qwindowsysteminterface.cpp:946 #30 0x00007fffe3e780f0 in userEventSourceDispatch ( source=source@entry=0x555555884700) at qeventdispatcher_glib.cpp:77 #31 0x00007fffebc47f57 in g_main_dispatch (context=0x7fffdc004fc0) at gmain.c:3142 ---Type <return> to continue, or q <return> to quit--- #32 g_main_context_dispatch (context=context@entry=0x7fffdc004fc0) at gmain.c:3795 #33 0x00007fffebc48190 in g_main_context_iterate ( context=context@entry=0x7fffdc004fc0, block=block@entry=1, dispatch=dispatch@entry=1, self=<optimized out>) at gmain.c:3868 #34 0x00007fffebc4821c in g_main_context_iteration (context=0x7fffdc004fc0, may_block=may_block@entry=1) at gmain.c:3929 #35 0x00007ffff1c86bef in QEventDispatcherGlib::processEvents ( this=0x555555886560, flags=...) at kernel/qeventdispatcher_glib.cpp:423 #36 0x00007ffff1c2f09a in QEventLoop::exec (this=this@entry=0x7fffffffda30, flags=..., flags@entry=...) at kernel/qeventloop.cpp:212 #37 0x00007ffff1c379e4 in QCoreApplication::exec () at kernel/qcoreapplication.cpp:1289 #38 0x0000555555574e20 in main (argc=<optimized out>, argv=<optimized out>) at /usr/src/debug/plasma5-workspace-5.12.5-lp150.1.1.x86_64/shell/main.cpp:172 (gdb) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c28
Michal Srb
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c31
--- Comment #31 from Felipe Abrahao
(In reply to Stefan Dirsch from comment #29)
Any news on that one?
Doesn't seem so. :-(
Hi, Sorry for the delay. I didn't have time to test it with the tool from https://download.opensuse.org/repositories/home:/michalsrb:/X11/openSUSE_Lea... I have been using the workaround "LIBGL_ALWAYS_SOFTWARE=1 plasmashell" and plasmashell has only crashed once or twice during all this time (when swap is full and/or i/o usage is also heavy, e.g., when baloo_file_extractor is running). The system is usable, but not entirely reliable, since I need to redo "LIBGL_ALWAYS_SOFTWARE=1 plasmashell" once in a while when plasmashell crashes. The only additional test I did during this time was with XOrg. I have updated to the latest version of XOrg from the Leap 15's repositories. The problem occurs, but much less frequent. I would like to test with Qt 5.9.6 just in case, but latest lts Qt 5.9.6 is getting unresolvable dependency issues with KDE 5.12 from the Leap 15's repositories. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c32
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733
http://bugzilla.opensuse.org/show_bug.cgi?id=1094733#c33
Felipe Abrahao
Thank you for the coredump and sorry for the long delay.
It is QSGAtlasTexture who is calling the glTexSubImage2D. It is trying to upload 32x32 "TEX" icon to position 1322x86. It believes that the current texture has size 2048x2048, but in reality it has size 956x509. It wants to have texture with name 12 bound and that one really is bound at the moment. All texture parameters match what the QSGAtlasTexture would set up.
I did not find any other QSGAtlasTexture or QSGPlainTexture that would have size 956x509 or were using texture the name 12.
So we know that the right texture is bound, but unfortunately I still have no idea how the texture got the dimensions 956x509.
Would you be willing to collect more data using the apitrace tool? It is not packaged in openSUSE, but you can either build it from sources or get it from my OBS home project: https://download.opensuse.org/repositories/home:/michalsrb:/X11/ openSUSE_Leap_15.0/
Please run it like this: export APITRACE_BACKTRACE="glViewport glTex* glGenTex* glDeleteTex* glBindTex*" apitrace trace plasmashell
Let the plasmashell run until it shows the distortion, then terminate it. It will produce file plasmashell.trace. Please attach it.
Sorry for the long delay. In fact, I should say this happened for a good reason. I'm happy to testify that, after I have updated to KDE 5.12.6 and MESA 18.1.7, I have not experienced any problem in my full-time working under any sort of usage, heavy load or possible configuration. -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com