[kwin] [Bug 426891] Chrome on native Wayland causes a memory leak and excess CPU use in kwin_wayland
https://bugs.kde.org/show_bug.cgi?id=426891 Greg Varsanyi changed: What|Removed |Added Status|REPORTED|RESOLVED Resolution|--- |NOT A BUG --- Comment #3 from Greg Varsanyi --- (In reply to Greg Varsanyi from comment #2) > Not freeing memory after dropped decorations is probably still something > that needs to be fixed on the kwin side. > And a safeguard mechanism to handle/throttle request floods like this would > be great too. Looks like I was wrong about that, apparently xdg-decoration spec says: > [...] clients are responsible for preventing configure loops and > must make sure not to send multiple successive set_mode requests > with the same decoration mode. On the Chrome side they have fixed the excessive calls (should be in next release), this ticket can be closed, resolved as "not a bug" on the KDE side. -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 426891] Chrome on native Wayland causes a memory leak and excess CPU use in kwin_wayland
https://bugs.kde.org/show_bug.cgi?id=426891 --- Comment #2 from Greg Varsanyi --- (In reply to Vlad Zahorodnii from comment #1) > It seems like google chrome floods kwin with > xdg_toplevel_decoration_v1.set_mode requests, which forces kwin to re-create > decorations every frame. That sounds wrong, I have opened a bug for it: https://bugs.chromium.org/p/chromium/issues/detail?id=1131662 Not freeing memory after dropped decorations is probably still something that needs to be fixed on the kwin side. And a safeguard mechanism to handle/throttle request floods like this would be great too. -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 426891] New: Chrome on native Wayland causes a memory leak and excess CPU use in kwin_wayland
https://bugs.kde.org/show_bug.cgi?id=426891 Bug ID: 426891 Summary: Chrome on native Wayland causes a memory leak and excess CPU use in kwin_wayland Product: kwin Version: git master Platform: Neon Packages OS: Linux Status: REPORTED Severity: crash Priority: NOR Component: wayland-generic Assignee: kwin-bugs-n...@kde.org Reporter: gvarsa...@gmail.com Target Milestone: --- SUMMARY Latest dev versions of Chrome (currently at "87.0.4270.0 dev") finally support native Wayland via Ozone, but kwin_wayland seems to act out around it. STEPS TO REPRODUCE 1. Get Chrome dev package: https://www.google.com/chrome/dev/?platform=linux=devchannel 2. Run with Ozone/Wayland: google-chrome-unstable --enable-features=UseOzonePlatform --ozone-platform=wayland --disable-gpu OBSERVED RESULT kwin_wayland using 100% CPU, and building up memory usage EXPECTED RESULT normal CPU use, no memory leak SOFTWARE/OS VERSIONS Linux: 5.9rc3 KDE Plasma Version: Neon unstable packages (git master @ Sept 22, 2020, 5.19.90+) KDE Frameworks Version: 5.75 Qt Version: 5.15 ADDITIONAL INFORMATION Does not happen when run via X11 / XWayland, including --ozone=x11 Does not happen when running Firefox native Wayland version. -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 411361] Plasma/Wayland DisplayPort MST (or dock) reconnection doesn't work
https://bugs.kde.org/show_bug.cgi?id=411361 Greg Varsanyi changed: What|Removed |Added Status|REPORTED|RESOLVED Resolution|--- |DUPLICATE --- Comment #1 from Greg Varsanyi --- Looks like an identical ticket was created, and got the attention: https://bugs.kde.org/show_bug.cgi?id=419061 *** This bug has been marked as a duplicate of bug 419061 *** -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 419061] Wayland: Can't hotplug displays through docking stations
https://bugs.kde.org/show_bug.cgi?id=419061 Greg Varsanyi changed: What|Removed |Added CC||gvarsa...@gmail.com --- Comment #10 from Greg Varsanyi --- *** Bug 411361 has been marked as a duplicate of this bug. *** -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 411361] Plasma/Wayland DisplayPort MST (or dock) reconnection doesn't work
https://bugs.kde.org/show_bug.cgi?id=411361 Greg Varsanyi changed: What|Removed |Added Version|git |git master Component|libkscreen |wayland-generic Product|KScreen |kwin -- You are receiving this mail because: You are watching all bug changes.
[KScreen] [Bug 408964] Wayland fractional scaling - very pixelated by default
https://bugs.kde.org/show_bug.cgi?id=408964 Greg Varsanyi changed: What|Removed |Added Resolution|--- |FIXED Status|CONFIRMED |RESOLVED --- Comment #13 from Greg Varsanyi --- Just checked with latest Neon dev. Works like a charm! Thank you so much, this makes a huge deal of difference for everyone on Wayland and with non-integer scaled screens. Big-huge thanks! -- You are receiving this mail because: You are watching all bug changes.
[KScreen] [Bug 411361] Plasma/Wayland DisplayPort MST (or dock) reconnection doesn't work
https://bugs.kde.org/show_bug.cgi?id=411361 Greg Varsanyi changed: What|Removed |Added Summary|Plasma/Wayland doesn't |Plasma/Wayland DisplayPort |recognize external display |MST (or dock) reconnection |on reconnected USB-C dock |doesn't work -- You are receiving this mail because: You are watching all bug changes.
[KScreen] [Bug 411361] Plasma/Wayland doesn't recognize external display on reconnected USB-C dock
https://bugs.kde.org/show_bug.cgi?id=411361 Greg Varsanyi changed: What|Removed |Added CC||gvarsa...@gmail.com -- You are receiving this mail because: You are watching all bug changes.
[KScreen] [Bug 411361] New: Plasma/Wayland doesn't recognize external display on reconnected USB-C dock
https://bugs.kde.org/show_bug.cgi?id=411361 Bug ID: 411361 Summary: Plasma/Wayland doesn't recognize external display on reconnected USB-C dock Product: KScreen Version: git Platform: Neon Packages OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: libkscreen Assignee: se...@kde.org Reporter: gvarsa...@gmail.com Target Milestone: --- SUMMARY Plasma/Wayland doesn't recognize external display on reconnected USB-C dock. STEPS TO REPRODUCE 1. Use a type-c USB w/ alt mode laptop and a dock (like Dell WD15) 2. Log in to Plasma Wayland from SDDM with the dock connected, see external screen work right 3. Disconnect the dock, reconnect some time later OBSERVED RESULT External screen doesn't activate (but everything else does, e.g. sound, USB devices, ethernet work well on reconnect) EXPECTED RESULT Screen should be identified and turned on SOFTWARE/OS VERSIONS Operating System: KDE neon Unstable Edition KDE Plasma Version: 5.16.80 KDE Frameworks Version: 5.62.0 Qt Version: 5.12.3 Kernel Version: 5.2.0-050200rc7-generic OS Type: 64-bit Processors: 12 × Intel® Core™ i7-8750H CPU @ 2.20GHz Memory: 15.4 GiB of RAM ADDITIONAL INFORMATION - This works on both Plasma/X11 and Gnome/Wayland. - As far as I can tell, docks generally use DP MST to connect multiple displays, even HDMI connection is served via DP MST. So this might be an MST-specific issue (reconnecting works with direct HDMI on the laptop) - Tested with Intel gfx - I see the port appear on /sys/class/drm/card0-DP-6 $ cat /sys/class/drm/card0-DP-2/status connected $ cat /sys/class/drm/card0-DP-2/enabled disabled It seems that the display never gets enabled, even though it's found. While searching for a bug report on this, I only found a forum entry - same story, but with HP laptop + dock: https://forum.kde.org/viewtopic.php?f=111=156876 -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 408964] wayland fractinal scaling - very pixelated by default
https://bugs.kde.org/show_bug.cgi?id=408964 --- Comment #3 from Greg Varsanyi --- I appreciate your care for this bit, thank you. Of course, I'm glad to help if you have something in store to try. -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 408964] wayland fractinal scaling - very pixelated by default
https://bugs.kde.org/show_bug.cgi?id=408964 --- Comment #1 from Greg Varsanyi --- sorry, disregard the attached screen shot - Spectacle seems to have taken it downsampled to 1x, so it doesn't really show the problem. -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 408964] New: wayland fractinal scaling - very pixelated by default
https://bugs.kde.org/show_bug.cgi?id=408964 Bug ID: 408964 Summary: wayland fractinal scaling - very pixelated by default Product: kwin Version: git master Platform: Neon Packages OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: wayland-generic Assignee: kwin-bugs-n...@kde.org Reporter: gvarsa...@gmail.com Target Milestone: --- Created attachment 121044 --> https://bugs.kde.org/attachment.cgi?id=121044=edit screen shot of the info app - as taken by spectacle at 1.5x scaling SUMMARY wayland fractional scaling works for the most part, but produces *very* pixelated visuals. Interestlingly, the image of any window becomes practically perfect while moving a window with the "Wobbly Windows" effect on. STEPS TO REPRODUCE 1. Use a hi-res monitor, set 2x scaling, turn on wobbly windows effect 2. Change scaling to 1.5x (`kscreen-doctor output.1.scale.1.5`), see the pixelated fonts (no anti-alias? wrong re-sampling algorithm?) 3. Start moving a window around OBSERVED RESULT "Wobbly Windows" probably uses a re-sampling algorithm that makes the pixelated looks go away EXPECTED RESULT Fancy looks without having to move the window :) SOFTWARE/OS VERSIONS Linux: KDE Neon Dev/Unstable w/ kernel 5.1.6 KDE Plasma Version: 5.16.80 KDE Frameworks Version: 5.60.0 Qt Version: 5.12.3 -- You are receiving this mail because: You are watching all bug changes.
[kwin] [Bug 407199] cannot start Wayland session on neon dev unstable. coredumpctl indicates kwin_wayland crash.
https://bugs.kde.org/show_bug.cgi?id=407199 --- Comment #2 from Greg Varsanyi --- Same symptoms, same versions, on kernel 5.1. Core dump looks similar too: Thread 7 (Thread 0x7f72823a7700 (LWP 11724)): #0 0x7f729b430bf9 in __GI___poll (fds=0x7f72780046c0, nfds=2, timeout=-1) at ../sysdeps/unix/sysv/linux/poll.c:29 resultvar = 18446744073709551100 sc_cancel_oldtype = 0 #1 0x7f72925685c9 in () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0 #2 0x7f72925686dc in g_main_context_iteration () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0 #3 0x7f729bf9b13f in QEventDispatcherGlib::processEvents(QFlags) () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #4 0x7f729bf3c64a in QEventLoop::exec(QFlags) () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #5 0x7f729bd6441a in QThread::exec() () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #6 0x7f729bd65bc2 in () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #7 0x7f729d3cf6db in start_thread (arg=0x7f72823a7700) at pthread_create.c:463 pd = 0x7f72823a7700 now = unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140129787868928, 440595788140762970, 140129787866752, 0, 94798741986608, 140721842154032, -506571411667953830, -506525279646263462}, mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0, cleanup = 0x0, canceltype = 0}}} not_first_call = #8 0x7f729b43d88f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95 Thread 6 (Thread 0x7f726bfff700 (LWP 11747)): #0 0x7f729b430bf9 in __GI___poll (fds=0x7f7260002930, nfds=1, timeout=-1) at ../sysdeps/unix/sysv/linux/poll.c:29 resultvar = 18446744073709551100 sc_cancel_oldtype = 0 #1 0x7f72925685c9 in () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0 #2 0x7f72925686dc in g_main_context_iteration () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0 #3 0x7f729bf9b13f in QEventDispatcherGlib::processEvents(QFlags) () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #4 0x7f729bf3c64a in QEventLoop::exec(QFlags) () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #5 0x7f729bd6441a in QThread::exec() () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #6 0x7f729691c2e5 in () at /usr/lib/x86_64-linux-gnu/libQt5Qml.so.5 #7 0x7f729bd65bc2 in () at /usr/lib/x86_64-linux-gnu/libQt5Core.so.5 #8 0x7f729d3cf6db in start_thread (arg=0x7f726bfff700) at pthread_create.c:463 pd = 0x7f726bfff700 now = unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140129414936320, 440595788140762970, 140129414934144, 0, 140129549196400, 140721842151680, -506205845526554790, -506525279646263462}, mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0, cleanup = 0x0, canceltype = 0}}} not_first_call = #9 0x7f729b43d88f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95 Thread 5 (Thread 0x7f72719ca700 (LWP 11725)): #0 0x7f729d3d59f3 in futex_wait_cancelable (private=, expected=0, futex_word=0x56380d6b7048) at ../sysdeps/unix/sysv/linux/futex-internal.h:88 __ret = -512 oldtype = 0 err = spin = 0 buffer = {__routine = 0x7f729d3d5690 <__condvar_cleanup_waiting>, __arg = 0x7f72719c9cc0, __canceltype = 46, __prev = 0x0} cbuffer = {wseq = 0, cond = 0x56380d6b7020, mutex = 0x56380d6b6ff8, private = 0} err = g = 0 flags = signals = result = 0 seq = 0 #1 0x7f729d3d59f3 in __pthread_cond_wait_common (abstime=0x0, mutex=0x56380d6b6ff8, cond=0x56380d6b7020) at pthread_cond_wait.c:502 spin = 0 buffer = {__routine = 0x7f729d3d5690 <__condvar_cleanup_waiting>, __arg = 0x7f72719c9cc0, __canceltype = 46, __prev = 0x0} cbuffer = {wseq = 0, cond = 0x56380d6b7020, mutex = 0x56380d6b6ff8, private = 0} err = g = 0 flags = signals = ---Type to continue, or q to quit--- result = 0 seq = 0 #2 0x7f729d3d59f3 in __pthread_cond_wait (cond=0x56380d6b7020, mutex=0x56380d6b6ff8) at pthread_cond_wait.c:655 #3 0x7f7272eaadcb in () at /usr/lib/x86_64-linux-gnu/dri/i965_dri.so #4 0x7f7272eaaaf7 in () at /usr/lib/x86_64-linux-gnu/dri/i965_dri.so #5 0x7f729d3cf6db in start_thread (arg=0x7f72719ca700) at pthread_create.c:463 pd = 0x7f72719ca700 now = unwind_buf = {cancel_jmp_buf = {{jmp_buf = {140129509091072, 440595788140762970, 140129509088896, 0, 94798741939920, 140721842148656, -506183603501541542, -506525279646263462}, mask_was_saved = 0}}, priv = {pad = {0x0, 0x0, 0x0, 0x0}, data = {prev = 0x0, cleanup = 0x0, canceltype = 0}}} not_first_call = #6 0x7f729b43d88f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95 Thread 4 (Thread 0x7f7282ba8700 (LWP 11723)): #0 0x7f729b430bf9 in __GI___poll (fds=0x7f7274003ce0, nfds=2, timeout=-1) at ../sysdeps/unix/sysv/lin
[kwin] [Bug 407199] cannot start Wayland session on neon dev unstable. coredumpctl indicates kwin_wayland crash.
https://bugs.kde.org/show_bug.cgi?id=407199 Greg Varsanyi changed: What|Removed |Added CC||gvarsa...@gmail.com -- You are receiving this mail because: You are watching all bug changes.
[Falkon] [Bug 394049] New: Windows version doesn't scale on high DPI
https://bugs.kde.org/show_bug.cgi?id=394049 Bug ID: 394049 Summary: Windows version doesn't scale on high DPI Product: Falkon Version: unspecified Platform: Other OS: MS Windows Status: UNCONFIRMED Severity: normal Priority: NOR Component: general Assignee: now...@gmail.com Reporter: gvarsa...@gmail.com Target Milestone: --- Created attachment 112535 --> https://bugs.kde.org/attachment.cgi?id=112535=edit chrome vs falkon on a 2x scale screen High DPI handled poorly on Windows. On a high-resolution (24" 4k, 200% scale) and low-resolution (27" 1440p, 100% scale), when open on the high res monitor: - address bar and other toolbars are clipped, half sized - click areas are off: one needs to click cca 20px below the address bar to get the address bar activated. Similar issues with other toolbars, and event to the web area - web area doesn't scale (sits on 1x scale regardless to which monitor it is on) - when moved to the low DPI screen, icons get very pixelated and look oversized -- You are receiving this mail because: You are watching all bug changes.
[neon] [Bug 362767] High DPI is messed up after Qt 5.6 update
https://bugs.kde.org/show_bug.cgi?id=362767 Greg Varsanyi <gvarsa...@gmail.com> changed: What|Removed |Added Resolution|WAITINGFORINFO |FIXED Status|NEEDSINFO |RESOLVED --- Comment #11 from Greg Varsanyi <gvarsa...@gmail.com> --- A lot has changed since I opened this ticket. At this point (Qt 5.7) defaults work well mostly, no need for extra trickery with env variables and such. closing as fixed. -- You are receiving this mail because: You are watching all bug changes.
[neon] [Bug 362767] High DPI is messed up after Qt 5.6 update
https://bugs.kde.org/show_bug.cgi?id=362767 --- Comment #10 from Greg Varsanyi <gvarsa...@gmail.com> --- (In reply to Weng Xuetian from comment #9) > IIRC, though QT_DEVICE_PIXEL_RATIO is still there, it's not doing the exact > same thing as it is in 5.5 . For now I'm using a combination of > QT_AUTO_SCREEN_SCALE_FACTOR and QT_SCREEN_SCALE_FACTORS to achieve the same > effect. what combination do you use? and how do you make it system wide exactly? > And kwin/plasma shouldn't be affected since they are not using it anyway, if > you found kwin/plasma is scaled, probably your distro need to compile > kwin/plasma against Qt 5.6 to make some qt5.6 compile time specific check > work. plasma and kwin seem to use the font size, around the idea of `1 line height in rendered px = perceivedDPI / 96 * referenceFontBoundaryHeight` where perceivedDPI seems to come from QT_FONT_DPI / fonts.conf DPI value / X.org global DPI -- how do I check what Qt version are plasmashell or kwin_x11 compiled against? for apps it's shown on the about/version panel, but `plasmashell -v` only gives me plasma version. -- You are receiving this mail because: You are watching all bug changes.
[neon] [Bug 362767] High DPI is messed up after Qt 5.6 update
https://bugs.kde.org/show_bug.cgi?id=362767 --- Comment #8 from Greg Varsanyi <gvarsa...@gmail.com> --- (In reply to David Edmundson from comment #5) > For Case #1: > - qt now scales up font DPI for some reason, you can counter that by > explicitly also setting > QT_FONT_DPI to be half/double (not sure which way round it is) > > For Case #3 > > Since Plasma 5.5 we do have code to disable internal high DPI managing on > Qt5.6 > > However, it requires kwin/plasma also be *compiled* with Qt5.6 > > #if (QT_VERSION >= QT_VERSION_CHECK(5, 6, 0)) > QCoreApplication::setAttribute(Qt::AA_DisableHighDpiScaling); > #endif playing with QT_FONT_DPI gets me the closest. a combination of QT_FONT_DPI=96 and QT_DEVICE_PIXEL_RATIO=2 (and not setting/unsetting any new Qt 5.6 env vars) gives me consistent look and feel across the *apps*, but not plasmashell and kwin - they go half sized. it would seem that plasmashell and kwin read the font dpi value and apply that literally as pixels. that can also be mitigated by starting these two (and only these two) with QT_FONT_DPI=192. So, using a deprecated and a falsified env var - with some exception handling - mostly gets us there. -- i wonder why did you guys decided to disable the native scaling. -- You are receiving this mail because: You are watching all bug changes.
[neon] [Bug 362767] High DPI is messed up after Qt 5.6 update
https://bugs.kde.org/show_bug.cgi?id=362767 --- Comment #2 from Greg Varsanyi <gvarsa...@gmail.com> --- (In reply to Harald Sitter from comment #1) > Which edition of neon are you on? I can poke some rebuilds so we'll know for > sure. dev/stable btw, new packages seem to be compiled against qt 5.6 but all issues remain the same. Unless there is a tricky combination of ENV vars I haven't figured out, I'm afraid this is a development issue where deprecated logic/code has to be updated all across the code base to play better with Qt 5.6 high dpi logic. -- You are receiving this mail because: You are watching all bug changes.
[neon] [Bug 362767] New: High DPI is messed up after Qt 5.6 update
https://bugs.kde.org/show_bug.cgi?id=362767 Bug ID: 362767 Summary: High DPI is messed up after Qt 5.6 update Product: neon Version: unspecified Platform: Other OS: Linux Status: UNCONFIRMED Severity: normal Priority: NOR Component: general Assignee: n...@kde.org Reporter: gvarsa...@gmail.com Screen scaling (kcm_kscreen) still sets the deprecated QT_DEVICE_PIXEL_RATIO, the new QT_AUTO_SCREEN_SCALE_FACTOR confuses kwin_x11 and plasmashell really bad. I wonder if this has to do with packages still being compiled against Qt 5.5.1 Reproducible: Always Steps to Reproduce: 1. Grab a high DPI screen 2. Set screen scaling (Display Configuration > Screen Scaling > x2) 3. Log out and back in or 1. Update to latest with Qt 5.6 on a system with high DPI configured. Actual Results: case #1 (using QT_DEVICE_PIXEL_RATIO=2) double (4x instead of 2x scaled) plasma parts and apps case #2 (unset QT_DEVICE_PIXEL_RATIO) some items are not sized up (fonts define minimum size), icons are too small (1x instead of 2x) case #3 (using QT_AUTO_SCREEN_SCALE_FACTOR=1) kwin_x11 provides small (1x) decorations, has placement issues and changes the height of all windows to the minimum possible Expected Results: Scale to 2x consistantly In an ideal word: make use of Qt 5.6 automatic screen scale factor I realize this is partly of a Plasma development issue, but setting the environment variables and compiling against Qt 5.6 might be a packaging thing. -- You are receiving this mail because: You are watching all bug changes.
[plasmashell] [Bug 360333] New system tray: icons do not scale with panel height, stay very small
https://bugs.kde.org/show_bug.cgi?id=360333 --- Comment #8 from Greg Varsanyi <gvarsa...@gmail.com> --- (In reply to Marco Martin from comment #6) > this bug is pretty much mutually exclusive with > https://bugs.kde.org/show_bug.cgi?id=353834 > I fear icons are either going to be too big for some people or too small for > some other people (and i don't want to add a configure icon size ui) This maybe an issue VDG could be involved with. I'm not sure what you mean by "too big for some", but systray icons did indeed look bigger than the clock font next to them (since both got the same height, but fonts have some padding b/c of unused accent space above and below). Having that aligned would be a win. But now it's not following the height of the panel at all. Fixed height systray icons mean they will look wrong on every resized taskbar panel - proportionally with how much it is resized. Bringing back the proportionate design with make it much more usable for many of us, making it aligned with the text-based plasmoids on the panel would just make it look awesome. -- You are receiving this mail because: You are watching all bug changes.