FULL details of symptoms with copious troubleshooting attempts are in my Manjaro Forum thread https://forum.manjaro.org/t/kwin-regression-in-stable-update-2018-04-14/44938?u=kdemeoz Summary: Dragging any window from any VD in KDE’s Desktop Grid to any other VD, crashes Desktop Grid & corrupts KWin for the rest of the session [when the pre-conditions are met]. Pre-condition Settings: More than one Activity, with specific windows isolated to specific Activities. Multiple VDs. Compositing set to either OpenGL variant, but not XRender. “Fall Apart” & “Desktop Grid” Desktop Effects enabled [others allowed too, but this pair are minimal requirement otherwise crash is impossible]. KDE Plasma 5.12.4 [with identical settings, crash never occurs in prior versions]. What Should Happen [& used to]: Up until this last update, no defect was present, such that dragging windows between VDs in Desktop Grid view uneventfully worked correctly, ie, the target window changed VD as intended, thence the DG remained happily on-screen until User clicked to return to chosen VD. All windows remained isolated to their User-nominated Activity. On any individual VD, if User chose to close any window, the “Fall Apart” Desktop Effect occurred correctly. What Now Happens [but did not used to]: Since this update, dragging windows between VDs in Desktop Grid view now crashes Desktop Grid once the window drop finishes, & corrupts KWin for the rest of the session [compositing on each VD is damaged, visual artefacts & gaps can occur, & all windows now populate all Activities (ruining one of the chief benefits of Activities)]. Workaround -- Initial: Prior to the regression, the “Fall Apart” Desktop Effect was NOT activated when dragging windows between VDs in DG view, & hence DG stayed stable. Since the regression the “Fall Apart” Desktop Effect IS [unnecessarily] activated when dragging windows between VDs in DG view, & hence DG crashes, & the KWin damage ensues. Hence, disabling the “Fall Apart” Desktop Effect allows the proper “traditional Manjaro KDE” DG behaviour to prevail, but the penalty is necessarily loss of the the “Fall Apart” Desktop Effect eye-candy when working in any individual VD. Workaround -- Current: Re-enable Fall Apart, downgrade [only] kwin from 5.12.4 to 5.12.3, logout/in... now all is good again. Comments: Many other Manjaro forum posters do NOT experience this failure, but for me it has a 0% occurrence before 5.12.4 & 100% occurrence with 5.12.4. It applies to BOTH my PCs. It occurs even in a temporary TEST user account created *after* the system upgrade to 5.12.4, hence [i assume] implying the root cause is not corruption of my normal user account config settings. A SegFault coredump accompanies [or causes] each instance of the failure. Eg, here i triggered the crash three times: [root@Lappy-XPS15 testing]# journalctl -p err -b --since "2018-04-18 13:55" -- Logs begin at Thu 2018-04-05 22:45:56 AEST, end at Wed 2018-04-18 14:04:01 AEST. -- Apr 18 13:55:50 Lappy-XPS15 kwin_x11[27082]: Splitting quads is allowed only in pre-paint calls! Apr 18 13:56:16 Lappy-XPS15 systemd-coredump[31100]: Process 27082 (kwin_x11) of user 1002 dumped core. Stack trace of thread 27082: #0 0x00007f5a61524860 n/a (n/a) Apr 18 14:00:13 Lappy-XPS15 kwin_x11[31096]: Splitting quads is allowed only in pre-paint calls! Apr 18 14:00:33 Lappy-XPS15 systemd-coredump[31258]: Process 31096 (kwin_x11) of user 1002 dumped core. Stack trace of thread 31096: #0 0x00007fcab3804860 n/a (n/a) Apr 18 14:01:57 Lappy-XPS15 kwin_x11[31254]: Splitting quads is allowed only in pre-paint calls! Apr 18 14:02:33 Lappy-XPS15 systemd-coredump[31308]: Process 31254 (kwin_x11) of user 1002 dumped core. Stack trace of thread 31254: #0 0x00007f68bd618860 n/a (n/a) [root@Lappy-XPS15 testing]#
Please provide a backtrace with full debug symbols of the crash. This is a mandatory prerequisite to be able to investigate.
I have been trying to achieve your request for many hours [i have never created a backtrace before & don't know how to do it]. Lots of online research gave me some clues using gdb but nothing worked that i tried. I alternatively tried running a strace, & prior to it i forced the crash, but the 3602 lines of resultant text does not contain "segfault" or "error", so i doubt it is useful for you, & anyway i doubt your site will allow me to paste that many lines. There's now two options: 1. I abandon this, & you reject my Bug Report. 2. You kindly please set out step by step instructions for how i should obtain the backtrace.
We have instructions at https://community.kde.org/Guidelines_and_HOWTOs/Debugging/How_to_create_useful_crash_reports But I am not sure if Manjaro offers debug symbols; I know Arch doesn't. I suggest to ask in a forum of your distribution.
Thanks but i was already aware of that linked page. In turn, relevant for Arch/Manjaro [given as you correctly state by default they do not have debug symbols] i have thrice read the associated page [1] https://wiki.archlinux.org/index.php/Debug_-_Getting_Traces, & its associated page [2] https://wiki.archlinux.org/index.php/Core_dump#Examining_a_core_dump. I have also now read [3] `man coredumpctl` & [4] `coredumpctl --help`. From [2] i was able to see: ``` [kdemeoz@Lappy-XPS15 ~]$ coredumpctl list TIME PID UID GID SIG COREFILE EXE <<snipped for brevity>> Thu 2018-04-19 14:06:13 AEST 9260 1000 1000 6 missing /usr/bin/kwin_x11 Thu 2018-04-19 14:10:14 AEST 1229 1000 1000 6 missing /usr/bin/kwin_x11 Sat 2018-04-21 14:31:30 AEST 29658 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 17:00:26 AEST 32096 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 17:00:27 AEST 31811 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 17:14:33 AEST 31946 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 19:18:53 AEST 6444 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 19:19:27 AEST 29534 1000 1000 6 truncated /usr/bin/kwin_x11 Sat 2018-04-21 19:37:48 AEST 4066 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 14:24:41 AEST 23291 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 14:27:18 AEST 1285 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 22:43:22 AEST 19574 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 22:45:19 AEST 21322 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 22:45:20 AEST 21672 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 22:45:22 AEST 21542 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 23:53:39 AEST 1389 1000 1000 6 truncated /usr/bin/kwin_x11 Sun 2018-04-22 23:53:43 AEST 2817 1000 1000 6 truncated /usr/bin/kwin_x11 Mon 2018-04-23 00:40:35 AEST 5255 1000 1000 6 truncated /usr/bin/kwin_x11 Mon 2018-04-23 00:41:03 AEST 7165 1000 1000 6 truncated /usr/bin/kwin_x11 lines 13-50/50 (END) ``` ...& then via [3] "Example 2. Invoke gdb on the last core dump": ``` [kdemeoz@Lappy-XPS15 ~]$ coredumpctl gdb PID: 7165 (kwin_x11) UID: 1000 (kdemeoz) GID: 1000 (kdemeoz) Signal: 6 (ABRT) Timestamp: Mon 2018-04-23 00:40:50 AEST (11h ago) Command Line: /usr/bin/kwin_x11 --crashes 1 Executable: /usr/bin/kwin_x11 Control Group: /user.slice/user-1000.slice/session-c8.scope Unit: session-c8.scope Slice: user-1000.slice Session: c8 Owner UID: 1000 (kdemeoz) Boot ID: 4d57b850825045aaa73d73195fd923ef Machine ID: 3054f65cf0df4fa190579b2f767ba3aa Hostname: Lappy-XPS15 Storage: /var/lib/systemd/coredump/core.kwin_x11.1000.4d57b850825045aaa73d73195fd923ef.7165.1524408050000000.lz4 (truncated) Message: Process 7165 (kwin_x11) of user 1000 dumped core. Stack trace of thread 7165: #0 0x00007f1382d3c860 n/a (n/a) <<snipped for brevity>> Reading symbols from /usr/bin/kwin_x11...(no debugging symbols found)...done. BFD: warning: /var/tmp/coredump-7lm5u3 is truncated: expected core file size >= 2484768768, found: 2147483648 [New LWP 7165] [New LWP 7369] [New LWP 7340] [New LWP 7169] [New LWP 7171] [New LWP 7172] [New LWP 7173] [New LWP 7188] [New LWP 7325] [New LWP 7170] Failed to read a valid object file image from memory. Core was generated by `/usr/bin/kwin_x11 --crashes 1'. Program terminated with signal SIGABRT, Aborted. #0 0x00007f1382d3c860 in ?? () [Current thread is 1 (LWP 7165)] (gdb) bt #0 0x00007f1382d3c860 in ?? () Backtrace stopped: Cannot access memory at address 0x7fff46ccd2a0 (gdb) q [kdemeoz@Lappy-XPS15 ~]$ ``` It seems that i cannot make further progress unless i am able to overcome the missing debugging symbols, BUT they in turn IMO seem pointless given that `bt` is unable to access the data it needs anyway. Wrt [1], though i can understand specific parts of it, i'm afraid that i still can't quite grasp a firm picture of the total & exact steps i would need to perform for my situation. I regret therefore, that despite this bug being 100% reproducible on both my PCs, my inability to conform with your mandate to supply a Backtrace leaves me no practical alternative but to abandon my putative bug report.
Because my Lappy had been in Suspend state since the crash targeted in my preceding post, just in case that was the cause of the inability to read memory today, just a few minutes ago i deliberately triggered another crash & then immediately ran the commands again. As you can see, it has made no practical beneficial change: [kdemeoz@Lappy-XPS15 ~]$ coredumpctl list TIME PID UID GID SIG COREFILE EXE <<snipped for brevity>> Mon 2018-04-23 13:30:35 AEST 16571 1000 1000 6 truncated /usr/bin/kwin_x11 lines 13-51/51 (END) ``` [kdemeoz@Lappy-XPS15 ~]$ coredumpctl gdb PID: 16571 (kwin_x11) UID: 1000 (kdemeoz) GID: 1000 (kdemeoz) Signal: 6 (ABRT) Timestamp: Mon 2018-04-23 13:30:16 AEST (1min 6s ago) Command Line: /usr/bin/kwin_x11 -session 10118bffdc0000152440805000000052280082_1524409043_409279 Executable: /usr/bin/kwin_x11 Control Group: /user.slice/user-1000.slice/session-c12.scope Unit: session-c12.scope Slice: user-1000.slice Session: c12 Owner UID: 1000 (kdemeoz) Boot ID: 4d57b850825045aaa73d73195fd923ef Machine ID: 3054f65cf0df4fa190579b2f767ba3aa Hostname: Lappy-XPS15 Storage: /var/lib/systemd/coredump/core.kwin_x11.1000.4d57b850825045aaa73d73195fd923ef.16571.1524454216000000.lz4 (truncated) Message: Process 16571 (kwin_x11) of user 1000 dumped core. Stack trace of thread 16571: #0 0x00007f2ba5d1a860 n/a (n/a) <<snipped for brevity>> Reading symbols from /usr/bin/kwin_x11...(no debugging symbols found)...done. BFD: warning: /var/tmp/coredump-vCrWX5 is truncated: expected core file size >= 2427691008, found: 2147483648 [New LWP 16571] [New LWP 16619] [New LWP 16634] [New LWP 17974] [New LWP 16632] [New LWP 16603] [New LWP 21043] Cannot access memory at address 0x7f2ba64d00c8 Cannot access memory at address 0x7f2ba64d00c0 Failed to read a valid object file image from memory. Core was generated by `/usr/bin/kwin_x11 -session 10118bffdc0000152440805000000052280082_1524409043_40'. Program terminated with signal SIGABRT, Aborted. #0 0x00007f2ba5d1a860 in ?? () [Current thread is 1 (LWP 16571)] (gdb) bt #0 0x00007f2ba5d1a860 in ?? () Backtrace stopped: Cannot access memory at address 0x7ffc18e72d20 (gdb) q [kdemeoz@Lappy-XPS15 ~]$ ```
kwin was aborted via SIGABRT. The Konsole output when running kwin in Konsole could reveal the abort message.
(In reply to Christoph Feck from comment #6) > The Konsole output when running kwin in > Konsole could reveal the abort message. Pls excuse my ignorance, but i do not understand how to actually do this. Whilst i am very familiar with launching discrete programs in Konsole to observe any error messages that might be produced, i cannot see how to also do this with a basic "system" package like kwin, given that it automatically launches & runs each time i boot & log into my desktop. How could i then start it in Konsole given it is already running?
kwin_x11 --replace
(In reply to Martin Flöser from comment #8) > kwin_x11 --replace Oh, that's a cool trick, thanks. I temporarily re-enabled Fall Apart, then ran that command of yours in Konsole, then went to Desktop Grid view, dragged a window from one VD to another, & of course the crash then occurred again... [kdemeoz@Lappy-XPS15 ~]$ kwin_x11 --replace Module 'org.kde.kwin.decoration' does not contain a module identifier directive - it cannot be protected from external registrations. QXcbConnection: XCB error: 3 (BadWindow), sequence: 2050, resource id: 39845891, major code: 18 (ChangeProperty), minor code: 0 OpenGL vendor string: Intel Open Source Technology Center OpenGL renderer string: Mesa DRI Intel(R) Ironlake Mobile OpenGL version string: 2.1 Mesa 18.0.0 OpenGL shading language version string: 1.20 Driver: Intel GPU class: i965 OpenGL version: 2.1 GLSL version: 1.20 Mesa version: 18.0 X server version: 1.19.6 Linux kernel version: 4.16.2 Requires strict binding: yes GLSL shaders: yes Texture NPOT support: yes Virtual Machine: no kf5.kcoreaddons.desktopparser: Property type "Url" is not a known QVariant type. Found while parsing property definition for "X-KWin-Video-Url" in "/usr/share/kservicetypes5/kwineffect.desktop" QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2227, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2228, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2229, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2231, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2233, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2234, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2237, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2238, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2239, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2241, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2243, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2244, resource id: 0, major code: 14 (GetGeometry), minor code: 0 trying to show an empty dialog trying to show an empty dialog Splitting quads is allowed only in pre-paint calls! Application::crashHandler() called with signal 6; recent crashes: 1 KCrash: crashing... crashRecursionCounter = 2 KCrash: Application Name = kwin_x11 path = /usr/bin pid = 14594 KCrash: Arguments: /usr/bin/kwin_x11 --replace QXcbConnection: XCB error: 3 (BadWindow), sequence: 172, resource id: 130023432, major code: 20 (GetProperty), minor code: 0 Module 'org.kde.kwin.decoration' does not contain a module identifier directive - it cannot be protected from external registrations. OpenGL vendor string: Intel Open Source Technology Center OpenGL renderer string: Mesa DRI Intel(R) Ironlake Mobile OpenGL version string: 2.1 Mesa 18.0.0 OpenGL shading language version string: 1.20 Driver: Intel GPU class: i965 OpenGL version: 2.1 GLSL version: 1.20 Mesa version: 18.0 X server version: 1.19.6 Linux kernel version: 4.16.2 Requires strict binding: yes GLSL shaders: yes Texture NPOT support: yes Virtual Machine: no kf5.kcoreaddons.desktopparser: Property type "Url" is not a known QVariant type. Found while parsing property definition for "X-KWin-Video-Url" in "/usr/share/kservicetypes5/kwineffect.desktop" QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2278, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2279, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2280, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2282, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2284, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2285, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2288, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2289, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2290, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2292, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2294, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 9 (BadDrawable), sequence: 2295, resource id: 0, major code: 14 (GetGeometry), minor code: 0 QXcbConnection: XCB error: 3 (BadWindow), sequence: 2436, resource id: 130023427, major code: 18 (ChangeProperty), minor code: 0 trying to show an empty dialog Aborted (core dumped) [kdemeoz@Lappy-XPS15 ~]$ qdbus org.kde.kwin /KWin supportInformation Service 'org.kde.kwin' does not exist. [kdemeoz@Lappy-XPS15 ~]$ coredumpctl listQXcbConnection: XCB error: 3 (BadWindow), sequence: 25490, resource id: 132120589, major code: 18 (ChangeProperty), minor code: 0 <<in a separate Konsole:>> [kdemeoz@Lappy-XPS15 ~]$ coredumpctl list TIME PID UID GID SIG COREFILE EXE <<snipped>> Tue 2018-04-24 12:31:24 AEST 28591 1000 1000 6 truncated /usr/bin/kwin_x11 Tue 2018-04-24 12:32:11 AEST 30784 1000 1000 6 truncated /usr/bin/kwin_x11 Tue 2018-04-24 15:33:11 AEST 14594 1000 1000 6 truncated /usr/bin/kwin_x11 [kdemeoz@Lappy-XPS15 ~]$ [kdemeoz@Lappy-XPS15 ~]$ coredumpctl gdb PID: 14594 (kwin_x11) UID: 1000 (kdemeoz) GID: 1000 (kdemeoz) Signal: 6 (ABRT) Timestamp: Tue 2018-04-24 15:32:50 AEST (2min 17s ago) Command Line: kwin_x11 --replace Executable: /usr/bin/kwin_x11 Control Group: /user.slice/user-1000.slice/session-c16.scope Unit: session-c16.scope Slice: user-1000.slice Session: c16 Owner UID: 1000 (kdemeoz) Boot ID: 4d57b850825045aaa73d73195fd923ef Machine ID: 3054f65cf0df4fa190579b2f767ba3aa Hostname: Lappy-XPS15 Storage: /var/lib/systemd/coredump/core.kwin_x11.1000.4d57b850825045aaa73d73195fd923ef.14594.1524547970000000.lz4 (truncated) Message: Process 14594 (kwin_x11) of user 1000 dumped core. Stack trace of thread 14594: #0 0x00007fb8d51e0860 n/a (n/a) <<snipped>> Reading symbols from /usr/bin/kwin_x11...(no debugging symbols found)...done. BFD: warning: /var/tmp/coredump-ip6mEG is truncated: expected core file size >= 2417315840, found: 2147483648 [New LWP 14594] [New LWP 14596] [New LWP 14597] [New LWP 14603] [New LWP 14614] [New LWP 14619] [New LWP 14620] [New LWP 14735] Cannot access memory at address 0x7fb8d59960c8 Cannot access memory at address 0x7fb8d59960c0 Failed to read a valid object file image from memory. Core was generated by `kwin_x11 --replace'. Program terminated with signal SIGABRT, Aborted. #0 0x00007fb8d51e0860 in ?? () [Current thread is 1 (LWP 14594)] (gdb) bt #0 0x00007fb8d51e0860 in ?? () Backtrace stopped: Cannot access memory at address 0x7ffd24db89e0 (gdb) q [kdemeoz@Lappy-XPS15 ~]$
Thanks for the update. The abort message is "Splitting quads is allowed only in pre-paint calls!", see also bug 368847.
I do hope that sooner or later the root cause might be eliminated. until then however, i continue to run with *Fall Apart* disabled, as this way everything works very nicely. One day it would be nice to be able to use FA again...
Setting as duplicate *** This bug has been marked as a duplicate of bug 368847 ***
(In reply to Martin Flöser from comment #12) > Setting as duplicate > > *** This bug has been marked as a duplicate of bug 368847 *** Sorry but i disagree with you. I have NO PROBLEMS using Desktop Grid [ZERO CRASHES], **unless** specifically the Fall Apart Desktop Effect is enabled. When it is, DG & KWin crash 100% of the time i drag windows in DG. Conversely when FA is disabled, DG & KWin crash 0% of the time i drag windows in DG. The key here is Fall Apart. By you marking my report as a duplicate, you lose this important information. The other report focuses only on DG. My report gives you additional information, which i would have thought was important. By the way, since my original report i have changed my Workaround. I am again using the latest KWin [ie, was 5.12.4 but is now 5.12.5], but i now leave FA permanently disabled. This combination provides me with perfect DG behaviour... though it is a pity that i have permanently lost the use of the nice FA effect.
The problem is the splitting of the quads. We have several effects triggering it. If we fix it for one, we automatically fix it for all.