Bug#951117: libecore1: .xsession-errors eats all disk space
Control: forwarded -1 https://phab.enlightenment.org/T8621 Control: tags -1 upstream On Thu, Feb 27, 2020 at 02:14:52AM +0300, sergio wrote: > > Sorry but I don't think I can really help further. Maybe the upstream > folks > > would be able to track something down. > > OK, I'll report this bug on phab.enlightenment.org but really I so tired > of E bugs, that I'm thinking to switch to another WM (possibly tiling > with good float mode support). The only thing stopping me is time I need > for choice and configuration. Thanks for that. Sorry that you've had issues with your use-cases. I still can't seem to trigger this like you can, but I was able to get some backtraces today. I'll post these to the phab as well. ERR<36232>:eo ../src/lib/eo/eo.c:2192 efl_data_scope_get() Eo ID 0x403d1edf is not a valid object. Current thread: main. This ID has probably been deleted or this was never a valid object ID. (domain=0, current_domain=0, local_domain=0, available_domains=[0 1], generation=2df, id=f47, ref=1) /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 2054 @ eina_log_print_cb_stderr.part.0() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 1456 @ eina_log_print_unlocked() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 2259 @ eina_log_print() /usr/lib/x86_64-linux-gnu/libeo.so.1 | ??/?? : 2259 @ _eo_obj_pointer_get() /usr/lib/x86_64-linux-gnu/libeo.so.1 | ../src/lib/eo/eo.c : 2192 @ efl_data_scope_get() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/ecore_exe.c: 218 @ ecore_exe_pid_get() /usr/bin/enlightenment| ../src/bin/e_exec.c : 814 @ _e_exec_startup_id_pid_find() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_iterator.c: 150 @ eina_iterator_foreach() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_hash.c: 1241 @ eina_hash_foreach() /usr/bin/enlightenment| ../src/bin/e_exec.c : 298 @ e_exec_startup_id_pid_instance_find() /usr/bin/enlightenment| ../src/bin/e_client.c : 2202 @ _e_client_eval() /usr/bin/enlightenment| ../src/bin/e_client.c : 2540 @ e_client_idler_before() /usr/bin/enlightenment| ../src/bin/e_main.c : 1810 @ _e_main_cb_idle_before() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/ecore_idler.c : 35 @ _ecore_factorized_idle_process() /usr/lib/x86_64-linux-gnu/libeo.so.1 | ??/?? : 35 @ _event_callback_call() /usr/lib/x86_64-linux-gnu/libeo.so.1 | ??/?? : 35 @ efl_event_callback_call() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/ecore_main.c : 2413 @ _ecore_main_loop_iterate_internal() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/ecore_main.c : 1198 @ _ecore_main_loop_begin() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/efl_loop.c : 58 @ _efl_loop_begin() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/src/lib/ecore/efl_loop.eo.c : 28 @ efl_loop_begin() /usr/lib/x86_64-linux-gnu/libecore.so.1| ./obj-x86_64-linux-gnu/../src/lib/ecore/ecore_main.c : 1285 @ ecore_main_loop_begin() /usr/bin/enlightenment| ../src/bin/e_main.c : 1100 @ main() /lib/x86_64-linux-gnu/libc.so.6| /build/glibc-TrjWJf/glibc-2.29/csu/../csu/libc-start.c : 342 @ __libc_start_main() /usr/bin/enlightenment| ??/?? : 342 @ _start() ERR<36232>:eo ../src/lib/eo/eo.c:1814 efl_isa() Eo ID 0x403d1edf is not a valid object. Current thread: main. This ID has probably been deleted or this was never a valid object ID. (domain=0, current_domain=0, local_domain=0, available_domains=[0 1], generation=2df, id=f47, ref=1) /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 2054 @ eina_log_print_cb_stderr.part.0() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 1456 @ eina_log_print_unlocked() /usr/lib/x86_64-linux-gnu/libeina.so.1 | ./obj-x86_64-linux-gnu/../src/lib/eina/eina_log.c : 2259 @
Bug#951117: libecore1: .xsession-errors eats all disk space
https://phab.enlightenment.org/T8621 -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
> Now I can trigger the message Great. > but not a nonstop stream Indeed. > E prints it while opening and closing tilix, but not otherwise. If there are no other windows, closing tilix stops this messages, but in my session with other windows, closing tilix doesn't help, I need to close other windows, but not all of them > Moreover, I can only trigger it in the same session where I do all of > the config. Sorry, what config? > After logging out and back in, the message is not printed. Sure. > Also, it doesn't happen under Xephyr at all. I'm not able to reproduce it under Xephyr too. > Sorry but I don't think I can really help further. Maybe the upstream folks > would be able to track something down. OK, I'll report this bug on phab.enlightenment.org but really I so tired of E bugs, that I'm thinking to switch to another WM (possibly tiling with good float mode support). The only thing stopping me is time I need for choice and configuration. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Mon, Feb 24, 2020 at 01:59:31PM +0300, sergio wrote: > On 24/02/2020 09:25, Ross Vandegrift wrote: > Just tried, yes, I can trigger it on the clean user profile, with only > one file left: > > % cat .xsession > enlightenment_start > > Looks like you need to remove the shelf. > > > If it does trigger, there might be something else different about your > > system. > > Have you exactly Maximize Fullscreen'ed? It's not just Fullscreen. Now I can trigger the message, but not a nonstop stream. E prints it while opening and closing tilix, but not otherwise. Moreover, I can only trigger it in the same session where I do all of the config. After logging out and back in, the message is not printed. Also, it doesn't happen under Xephyr at all. Sorry but I don't think I can really help further. Maybe the upstream folks would be able to track something down. For the record, the full sequence I came up with is: 0. Install tilix and E. 1. Start a fresh E profile, go through the first run wizard. 2. Delete the default shelf. 3. Open Settings a. Windows -> Window Geometry -> Maximization i. Set Policy to Fullscreen ii. Enable "Allow manipulation of maximized windows" iii. Enable "Allow windows above fullscreen windows" iv. OK b. Input -> Key Bindings i. Add a binding for Window:State:Maximize Fullscreen ii. OK 4. Restart Enlightenment from the Enlightenment menu 5. Run tilix and maximize with the key binding from 3.b.i 6. Run tilix and close it immediately by exiting the terminal. Repeat 6 a few times, eventually it logs this a few times: ERR<2605>:eo ../src/lib/eo/eo.c:2192 efl_data_scope_get() Eo ID 0x40098cfa is not a valid object. Current thread: main. This ID has probably been deleted or this was never a valid object ID. (domain=0, current_domain=0, local_domain=0, available_domains=[0 1], generation=fa, id=263, ref=1) Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
On 24/02/2020 09:25, Ross Vandegrift wrote: > I still can't trigger the issue with your configuration. Have you > tried with a fresh E config? Just tried, yes, I can trigger it on the clean user profile, with only one file left: % cat .xsession enlightenment_start Looks like you need to remove the shelf. > If it does trigger, there might be something else different about your > system. Have you exactly Maximize Fullscreen'ed? It's not just Fullscreen. See screenshot attached. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Mon, Feb 24, 2020 at 02:04:38AM +0300, sergio wrote: > On 24/02/2020 01:11, Ross Vandegrift wrote: > > > Thanks for the details. Unfortunately, I still cannot reproduce this. > I've > > tried restarting Enlightenment and then opening tilix, making it > fullscreen, > > and then exiting. > > You should be able to reproduce this bug with the config that I've send you > for the next one. > > > How do you get a window on top of fullscreened tilix? > > Sorry for bad explanation, it should be vice-versa: tilix on top of > another fullscreened window. > > > > As soon as I do anything that would raise another window > > (Ctrl-Alt-down, Alt-Tab, etc), tilix exits fullscreen mode. > > My settings: > > Windows -> Window Geometry -> Maximization > Policy: Fullscreen > Direction: Both > Manipulation: > Allow manipulation of maximized windows: yes > Allow windows above fullscreen window: yes I still can't trigger the issue with your configuration. Have you tried with a fresh E config? IF that doesn't trigger the issue, then I think we just haven't found the right settings. If it does trigger, there might be something else different about your system. Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
On 24/02/2020 01:11, Ross Vandegrift wrote: > Thanks for the details. Unfortunately, I still cannot reproduce this. I've > tried restarting Enlightenment and then opening tilix, making it fullscreen, > and then exiting. You should be able to reproduce this bug with the config that I've send you for the next one. > How do you get a window on top of fullscreened tilix? Sorry for bad explanation, it should be vice-versa: tilix on top of another fullscreened window. > As soon as I do anything that would raise another window > (Ctrl-Alt-down, Alt-Tab, etc), tilix exits fullscreen mode. My settings: Windows -> Window Geometry -> Maximization Policy: Fullscreen Direction: Both Manipulation: Allow manipulation of maximized windows: yes Allow windows above fullscreen window: yes -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Sun, Feb 23, 2020 at 10:37:13PM +0300, sergio wrote: > More details: > > instead of fullscreening the tilix it can be closed when it's above > another fullscreen window > > I'm unable to reproduce it just after session start, but after restart > it reproduces frequently. > > So the sequence is: > > 1. restart enlightment > 2. open tilix > 3. put tilix above another fullscreen window or fullscreen itself > 4. close it > 5. repeat the sequence several times untill errors will be massively > printed into ~/.xsession-errors > > I've checked this on NUC, and it's reproducable. Thanks for the details. Unfortunately, I still cannot reproduce this. I've tried restarting Enlightenment and then opening tilix, making it fullscreen, and then exiting. Even after two dozen cycles, no constant stream of logs is triggered. How do you get a window on top of fullscreened tilix? As soon as I do anything that would raise another window (Ctrl-Alt-down, Alt-Tab, etc), tilix exits fullscreen mode. I've also tried testing with and without fullscreen compositing enabled, it doesn't seem to make a difference. Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
More details: instead of fullscreening the tilix it can be closed when it's above another fullscreen window I'm unable to reproduce it just after session start, but after restart it reproduces frequently. So the sequence is: 1. restart enlightment 2. open tilix 3. put tilix above another fullscreen window or fullscreen itself 4. close it 5. repeat the sequence several times untill errors will be massively printed into ~/.xsession-errors I've checked this on NUC, and it's reproducable. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
I know how to trigger it: Alt-Esc, Alt-PgDown, Ctrl-d Open tilix window, fullscreen it and close. It happens not every time, but quite often. ALT + Escape is Launch/Command x-terminal-emulator ALT + Next is Maximize Fullscreen x-terminal-emulator is /usr/bin/tilix.wrapper -- -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
Now (2020-02-22 12:10 GMT+3) I just woke up and found that there are no free scree space due to the same error. 1. The session was started at 2020-02-20 10:50 GMT+3, so it takes about 2 days, moreover host was rebooted before the session start, uprecords shows 2 days 21 hour (2020-02-19 15:25) 2. This bug is not related with user actions. Before going to sleep there was no errors records as I monitor .xsession-errors. GUI apps opened permanently: urxvt, riot-web (electron), GTK3: firefox, thunderbird, tilix, pavucontrol, deluge, GTK2: liferea. I have no hourly cron scripts and daily scripst are the same as on NUC, that doesn't have this error even session there lives longer. Not sure if this could be the reason, but my desktop has ATI FirePro 2460 with 4 miniDP and 3 monitors connected. (NUC is intel with only one monitor) -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On 21/02/2020 19:50, Ross Vandegrift wrote: > Which rendering backend do you use? OpenGL with "Tear-free updates" and "Texture from pixmap" on both hosts. Swapping method is "Auto" on the first host and "Triple buffered" on NUC >> I've just realized that I've another E runned permanently on intel NUC, >> there no above errors, but a lot of >> >> ERR<1162>:evas_main ../src/lib/evas/canvas/evas_object_smart.c:145 >> evas_object_smart_data_get() calling smart object API on non-smart object! > Do you logs contain any backtraces along with these messages? No. > While investigating this bug, I found these too - they are trigered > when interacting with EFL windows. The only EFL window I use on both hosts it's right-click favorites menu, opening it produces no errors. But opening settings gives: ERR<2144776>:evas_main ../src/lib/evas/canvas/evas_object_smart.c:145 evas_object_smart_data_get() calling smart object API on non-smart object! ERR<2144776>:evas-gl_x11 ../src/modules/evas/engines/gl_x11/evas_engine.c:1967 _native_bind_cb() eglCreateImage() for Pixmap 0x0x16fbe89 failed: 0x3003 Looks like closing tilix window gives: ERR<2144776>:evas-gl_x11 ../src/modules/evas/engines/gl_x11/evas_engine.c:1967 _native_bind_cb() eglCreateImage() for Pixmap 0x0x1661d26 failed: 0x3003 but not every time. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Fri, Feb 21, 2020 at 11:10:35AM +0300, sergio wrote: > > > Thanks, I've setup a session to stay open and idle for tonight. Let's > > see what happens if I let it sit for a while. So far, no results - E logged zero messages overnight while I wasn't interacting with it. Which rendering backend do you use? You can check in Settings -> Look -> Compositor -> Rendering. > I've just realized that I've another E runned permanently on intel NUC, > there no above errors, but a lot of > > ERR<1162>:evas_main ../src/lib/evas/canvas/evas_object_smart.c:145 > evas_object_smart_data_get() calling smart object API on non-smart object! Do you logs contain any backtraces along with these messages? While investigating this bug, I found these too - they are trigered when interacting with EFL windows. When I asked upstream, I was told that there should be a backtrace along with the message if libunwindsupport was built in. I don't get any backtrace, but haven't had time to investigate further. Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
> Thanks, I've setup a session to stay open and idle for tonight. Let's > see what happens if I let it sit for a while. I've just realized that I've another E runned permanently on intel NUC, there no above errors, but a lot of ERR<1162>:evas_main ../src/lib/evas/canvas/evas_object_smart.c:145 evas_object_smart_data_get() calling smart object API on non-smart object! $ ls -l .xsession-errors 554K 2020-02-21 .xsession-errors $ grep 'ERR<1162>:evas_main ../src/lib/evas/canvas/evas_object_smart.c:145 evas_object_smart_data_get() calling smart object API on non-smart object!' .xsession-errors | wc -l 3805 -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Thu, Feb 20, 2020 at 11:51:04AM +0300, sergio wrote: > On 17/02/2020 09:03, Ross Vandegrift wrote: > > Okay great- about how long it was from the time the session started to the > > time > > you saw the log messages? > > I'm monitoring .xsession-errors and can't say anything definite. > Sometimes it takes several hours, sometimes I don't see it for a day or > two. Anyway I see no obvious triggers. I have a couple apps runned > permanently and have no idea what it can be. Thanks, I've setup a session to stay open and idle for tonight. Let's see what happens if I let it sit for a while. Thanks, Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
On 17/02/2020 09:03, Ross Vandegrift wrote: > Okay great- about how long it was from the time the session started to the > time > you saw the log messages? I'm monitoring .xsession-errors and can't say anything definite. Sometimes it takes several hours, sometimes I don't see it for a day or two. Anyway I see no obvious triggers. I have a couple apps runned permanently and have no idea what it can be. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On Sun, Feb 16, 2020 at 10:04:59PM +0300, sergio wrote: > > Does this end if you restart enlightenemnt? (Main -> Enlightenment -> > > Restart) > > I've just captured this condition before the end of the free space and > can answer the question: yes, enlightenemnt restart stops this errors. Okay great- about how long it was from the time the session started to the time you saw the log messages? Thanks, Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
> Does this end if you restart enlightenemnt? (Main -> Enlightenment -> > Restart) I've just captured this condition before the end of the free space and can answer the question: yes, enlightenemnt restart stops this errors. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
On 12/02/2020 00:44, Ross Vandegrift wrote: > I haven't seen this, so lowering the priority to normal. Let's try to > figure out the trigger. I'm sure there no trigger, just something fails inside E. > How do you start enlightenment? With enlightenment_start in ~/.xsession via xdm. > Does your desktop suspend after inactivity? No suspend, noway! I have only xscreensaver and xorg.conf.d/00-flags.conf that turns off my monitors after 40 minutes of inactivity. > Do you leave any other EFL apps running? I don't use EFL apps at all. Moreover my E settings are very minimalistic: most modules are disables. I have only system/dbus, core/notifications, core/settings/panel, core/windows_switcher and all settings modules. > Does this end if you restart enlightenemnt? (Main -> Enlightenment -> > Restart) I don't know even how to check it. I notice this error only when free space ends, and I must kill / restart x session to release .xsession-errors file. > This bug was opened against libecore1, not enlightenment. I've > reassigned. I was misled efl_ functions. -- sergio.
Bug#951117: libecore1: .xsession-errors eats all disk space
Control: severity -1 normal Control: reassign -1 enlightenment 0.23.1-4 Control: retitle -1 eo logs in .xsession-errors eat all disk space On Tue, Feb 11, 2020 at 01:31:28PM +0300, sergio wrote: > Package: libecore1 > Version: 1.23.3-6 > Severity: critical > Justification: breaks unrelated software > > Dear Maintainer, > > it's critical as it eats all disk space causing other programs fail and > lose data. I haven't seen this, so lowering the priority to normal. Let's try to figure out the trigger. > I don't know how to reproduce it. It happens after several days of > running enlightenment. (My desktop is always on.) > > At the beginning all works fine, and ~/.xsession-errors takes less than > 100Kb. But some time later I find it taking all free space (dozen > gigabytes in my case). How do you start enlightenment? Does your desktop suspend after inactivity? Do you leave any other EFL apps running? > It repeats the following two lines endlessly: > > ERR<32091>:eo ../src/lib/eo/eo.c:2192 efl_data_scope_get() Eo ID > 0x402d3617 is not a valid object. Current thread: main. This ID has > probably been deleted or this was never a valid object ID. (domain=0, > current_domain=0, local_domain=0, available_domains=[0 1], > generation=217, id=b4d, ref=1) > ERR<32091>:eo ../src/lib/eo/eo.c:1814 efl_isa() Eo ID 0x402d3617 is not a > valid object. Current thread: main. This ID has probably been deleted or this > was never a valid object ID. (domain=0, current_domain=0, local_domain=0, > available_domains=[0 1], generation=217, id=b4d, ref=1) Does this end if you restart enlightenemnt? (Main -> Enlightenment -> Restart) > With the previous version of E all worked fine, began to happen with the > update to 0.23.1-4 This bug was opened against libecore1, not enlightenment. I've reassigned. Ross
Bug#951117: libecore1: .xsession-errors eats all disk space
Package: libecore1 Version: 1.23.3-6 Severity: critical Justification: breaks unrelated software Dear Maintainer, it's critical as it eats all disk space causing other programs fail and lose data. I don't know how to reproduce it. It happens after several days of running enlightenment. (My desktop is always on.) At the beginning all works fine, and ~/.xsession-errors takes less than 100Kb. But some time later I find it taking all free space (dozen gigabytes in my case). It repeats the following two lines endlessly: ERR<32091>:eo ../src/lib/eo/eo.c:2192 efl_data_scope_get() Eo ID 0x402d3617 is not a valid object. Current thread: main. This ID has probably been deleted or this was never a valid object ID. (domain=0, current_domain=0, local_domain=0, available_domains=[0 1], generation=217, id=b4d, ref=1) ERR<32091>:eo ../src/lib/eo/eo.c:1814 efl_isa() Eo ID 0x402d3617 is not a valid object. Current thread: main. This ID has probably been deleted or this was never a valid object ID. (domain=0, current_domain=0, local_domain=0, available_domains=[0 1], generation=217, id=b4d, ref=1) If you need I can share the full xsession-errors, zstandard comressed it takes 1.4M With the previous version of E all worked fine, began to happen with the update to 0.23.1-4 -- System Information: Debian Release: sid APT prefers unstable APT policy: (800, 'unstable'), (200, 'experimental') Architecture: amd64 (x86_64) Kernel: Linux 5.4.0-3-amd64 (SMP w/8 CPU cores) Locale: LANG=C.UTF-8, LC_CTYPE=C.UTF-8 (charmap=UTF-8), LANGUAGE=C.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: systemd (via /run/systemd/system)