Bug 468430

What kind of cable is the monitor connected with?

Pasting the relevant part of the backtrace inline for searchability:


#5  QWeakPointer<QObject>::internalData() const (this=0x58) at /usr/include/qt/QtCore/qsharedpointer_impl.h:696
        view = 0x0
        screen = 0x556fb2436b50
        knownScreen = <optimized out>
        __for_range = @0x7ffd15385e18: {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        managedScreens = {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        screens = {q_hash = {{d = 0x7f928c5765c0 <QHashData::shared_null>, e = 0x7f928c5765c0 <QHashData::shared_null>}}}
#6  QPointer<QScreen>::data() const (this=0x58) at /usr/include/qt/QtCore/qpointer.h:77
        view = 0x0
        screen = 0x556fb2436b50
        knownScreen = <optimized out>
        __for_range = @0x7ffd15385e18: {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        managedScreens = {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        screens = {q_hash = {{d = 0x7f928c5765c0 <QHashData::shared_null>, e = 0x7f928c5765c0 <QHashData::shared_null>}}}
#7  QPointer<QScreen>::operator QScreen*() const (this=0x58) at /usr/include/qt/QtCore/qpointer.h:83
        view = 0x0
        screen = 0x556fb2436b50
        knownScreen = <optimized out>
        __for_range = @0x7ffd15385e18: {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        managedScreens = {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        screens = {q_hash = {{d = 0x7f928c5765c0 <QHashData::shared_null>, e = 0x7f928c5765c0 <QHashData::shared_null>}}}
#8  DesktopView::screenToFollow() const (this=0x0) at /usr/src/debug/plasma-workspace/plasma-workspace-5.27.4.1/shell/desktopview.cpp:124
        view = 0x0
        screen = 0x556fb2436b50
        knownScreen = <optimized out>
        __for_range = @0x7ffd15385e18: {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        managedScreens = {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        screens = {q_hash = {{d = 0x7f928c5765c0 <QHashData::shared_null>, e = 0x7f928c5765c0 <QHashData::shared_null>}}}
#9  ShellCorona::screenInvariants() const (this=0x556fb24b6fd0) at /usr/src/debug/plasma-workspace/plasma-workspace-5.27.4.1/shell/shellcorona.cpp:828
        view = 0x0
        screen = 0x556fb2436b50
--Type <RET> for more, q to quit, c to continue without paging--c
        knownScreen = <optimized out>
        __for_range = @0x7ffd15385e18: {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        managedScreens = {q_hash = {{d = 0x557207a20290, e = 0x557207a20290}}}
        screens = {q_hash = {{d = 0x7f928c5765c0 <QHashData::shared_null>, e = 0x7f928c5765c0 <QHashData::shared_null>}}}
#10 ShellCorona::screenInvariants() const (this=0x556fb24b6fd0) at /usr/src/debug/plasma-workspace/plasma-workspace-5.27.4.1/shell/shellcorona.cpp:808
#11 0x00007f928c4beb41 in QtPrivate::QSlotObjectBase::call(QObject*, void**) (a=0x7ffd15385f80, r=<optimized out>, this=0x556fb24e6a20, this=<optimized out>, r=<optimized out>, a=<optimized out>)
    at ../../include/QtCore/../../src/corelib/kernel/qobjectdefs_impl.h:398
        receiver = <optimized out>
        td = <optimized out>
        receiverInSameThread = <optimized out>
        senderData = {previous = 0x0, receiver = 0x556fb24b6fd0, sender = 0x556fb24b70f8, signal = 3}
        c = 0x556fb278bf70
        connections = {d = 0x556fb2786bc0}
        list = <optimized out>
        inSenderThread = true
        highestConnectionId = 1
        signalVector = 0x556fb298f610
        currentThreadId = 0x7f928761af80
        sp = <optimized out>
        signal_spy_set = 0x0
        empty_argv = {0x0}
        senderDeleted = false

*** Bug 466095 has been marked as a duplicate of this bug. ***

A few questions for you:
1. Does it happen if you temporarily disable the KScreen2 service in System Settings > Startup & Shutdown > Background Services
2. Does the screen being unplugged have a high refresh rate?
3. Does the screen being unplugged have AdaptiveSync?

(In reply to Nate Graham from comment #1)
> What kind of cable is the monitor connected with?
The screens where this issue is happening had DisplayPort-to-HDMI adapters (DisplayPort from GPU side HDMI to screen side). 
But now that you've mentioned it, I tried plugging/unplugging a screen that's connected to the GPU's only HDMI port (screen's also HDMI) and it didn't crash at all.

(In reply to Nate Graham from comment #3)
> A few questions for you:
> 1. Does it happen if you temporarily disable the KScreen2 service in System
> Settings > Startup & Shutdown > Background Services
Yes, it does.
These are the exact steps I followed: KScreen 2 was unticked but it was "Running". I pressed the Pause button and it changed to "Not running". After that I tried to plug/unplug the screens, the screen with the DisplayPort-to-HDMI adapter crashed again. The HDMI one didn't.
> 2. Does the screen being unplugged have a high refresh rate?
Nope. Both screens have 60Hz refresh rate.
> 3. Does the screen being unplugged have AdaptiveSync?
I'm assuming you're referring to FreeSync. Two of the screens that have DisplayPort-to-HDMI adapter, one supports FreeSync the other doesn't. Both showed this issue.

Thanks for the info!

*** Bug 467756 has been marked as a duplicate of this bug. ***

After getting this plasmashell segfault multiple times a day (after leaving my computer idle so KDE locks and blanks screens, then coming back), I have some observations:

- `ShellCorona::screenInvariants()` is the direct cause of the crash, calling `QScreen *screen = view->screenToFollow();` on a null pointer `const DesktopView *view = desktopForScreen(knownScreen);`. Fascinatingly, in Qt Creator, `ShellCorona::screenInvariants()` is only enabled in debug builds (disabled in release builds with NDEBUG set).
- `desktopForScreen` can return null in normal operation. All other calls to this function (aside from `ShellCorona::screenInvariants()`) check for nullptr return, except that `ShellCorona::color()` calls it once, checks for null pointer, and if not calls it *again* and relies on the function returning a non-null value the second time.
- Is the function supposed to return null after screen blanking and then moving my mouse? I don't know.

Screen layout:

nyanpasu64@ryzen ~/.ssh> kscreen-doctor -o
Output: 68 DP-1 enabled connected priority 1 DisplayPort Modes: 74:2560x1440@60*! 75:2560x1440@75 76:1920x1440@75 77:1920x1440@60 78:1856x1392@75 79:1856x1392@60 80:1792x1344@75 81:1792x1344@60 82:2048x1152@60 83:2048x1152@60 84:2048x1152@60 85:1920x1200@60 86:1920x1200@60 87:1920x1080@60 88:1920x1080@60 89:1920x1080@60 90:1920x1080@60 91:1920x1080@50 92:1920x1080@60 93:1920x1080@60 94:1600x1200@60 95:1600x1200@75 96:1600x1200@70 97:1600x1200@65 98:1600x1200@60 99:1680x1050@60 100:1680x1050@60 101:1400x1050@75 102:1400x1050@60 103:1600x900@60 104:1600x900@60 105:1600x900@60 106:1600x900@60 107:1280x1024@75 108:1280x1024@70 109:1280x1024@60 110:1440x900@60 111:1400x900@60 112:1400x900@60 113:1280x960@60 114:1440x810@60 115:1440x810@60 116:1368x768@60 117:1368x768@60 118:1280x800@60 119:1280x800@60 120:1280x800@60 121:1280x800@60 122:1280x800@60 123:1152x864@75 124:1280x720@60 125:1280x720@60 126:1280x720@60 127:1280x720@60 128:1280x720@50 129:1280x720@60 130:1280x720@60 131:1024x768@75 132:1024x768@60 133:1024x768@75 134:1024x768@70 135:1024x768@60 136:960x720@75 137:960x720@60 138:928x696@75 139:928x696@60 140:896x672@75 141:896x672@60 142:1024x576@60 143:1024x576@60 144:1024x576@60 145:1024x576@60 146:960x600@60 147:960x600@60 148:832x624@75 149:960x540@60 150:960x540@60 151:960x540@60 152:960x540@60 153:800x600@75 154:800x600@70 155:800x600@65 156:800x600@60 157:800x600@72 158:800x600@75 159:800x600@60 160:800x600@56 161:840x525@60 162:840x525@60 163:864x486@60 164:864x486@60 165:720x576@50 166:700x525@75 167:700x525@60 168:800x450@60 169:800x450@60 170:720x480@60 171:720x480@60 172:640x512@75 173:640x512@60 174:700x450@60 175:700x450@60 176:640x480@60 177:640x480@75 178:640x480@73 179:640x480@67 180:640x480@60 181:640x480@60 182:720x405@60 183:720x405@59 184:720x400@70 185:684x384@60 186:684x384@60 187:640x400@60 188:640x400@60 189:576x432@75 190:640x360@60 191:640x360@60 192:640x360@60 193:640x360@59 194:512x384@75 195:512x384@70 196:512x384@60 197:512x288@60 198:512x288@60 199:416x312@75 200:480x270@60 201:480x270@60 202:400x300@72 203:400x300@75 204:400x300@60 205:400x300@56 206:432x243@60 207:432x243@60 208:320x240@73 209:320x240@75 210:320x240@60 211:360x202@60 212:360x202@59 213:320x180@60 214:320x180@59 Geometry: 0,0 2560x1440 Scale: 1 Rotation: 1 Overscan: 0 Vrr: incapable RgbRange: unknown
Output: 69 DP-2 disabled disconnected priority 0 DisplayPort Modes: Geometry: 0,0 0x0 Scale: 1 Rotation: 1 Overscan: 0 Vrr: incapable RgbRange: unknown
Output: 70 HDMI-1 enabled connected priority 2 HDMI Modes: 88:1920x1080@60 90:1920x1080@60*! 93:1920x1080@60 99:1680x1050@60 100:1680x1050@60 101:1400x1050@75 102:1400x1050@60 105:1600x900@60 106:1600x900@60 107:1280x1024@75 109:1280x1024@60 111:1400x900@60 112:1400x900@60 113:1280x960@60 115:1440x810@60 116:1368x768@60 117:1368x768@60 119:1280x800@60 120:1280x800@60 121:1280x800@60 122:1280x800@60 123:1152x864@75 124:1280x720@60 125:1280x720@60 126:1280x720@60 130:1280x720@60 131:1024x768@75 132:1024x768@60 133:1024x768@75 134:1024x768@70 135:1024x768@60 136:960x720@75 137:960x720@60 138:928x696@75 139:928x696@60 140:896x672@75 141:896x672@60 142:1024x576@60 143:1024x576@60 144:1024x576@60 145:1024x576@60 146:960x600@60 147:960x600@60 148:832x624@75 149:960x540@60 150:960x540@60 151:960x540@60 152:960x540@60 153:800x600@75 154:800x600@70 155:800x600@65 156:800x600@60 157:800x600@72 158:800x600@75 159:800x600@60 160:800x600@56 161:840x525@60 162:840x525@60 163:864x486@60 164:864x486@60 166:700x525@75 167:700x525@60 168:800x450@60 169:800x450@60 172:640x512@75 173:640x512@60 174:700x450@60 175:700x450@60 176:640x480@60 177:640x480@75 178:640x480@73 181:640x480@60 182:720x405@60 183:720x405@59 184:720x400@70 185:684x384@60 186:684x384@60 187:640x400@60 188:640x400@60 189:576x432@75 190:640x360@60 191:640x360@60 192:640x360@60 193:640x360@59 194:512x384@75 195:512x384@70 196:512x384@60 197:512x288@60 198:512x288@60 199:416x312@75 200:480x270@60 201:480x270@60 202:400x300@72 203:400x300@75 204:400x300@60 205:400x300@56 206:432x243@60 207:432x243@60 208:320x240@73 209:320x240@75 210:320x240@60 211:360x202@60 212:360x202@59 213:320x180@60 214:320x180@59 215:1680x1050@60 216:1440x900@60 217:1280x800@60 218:1280x720@60 1913:640x480@60 Geometry: 2560,0 1080x1920 Scale: 1 Rotation: 2 Overscan: 0 Vrr: incapable RgbRange: unknown
Output: 71 HDMI-2 disabled disconnected priority 0 HDMI Modes: Geometry: 0,0 0x0 Scale: 1 Rotation: 1 Overscan: 0 Vrr: incapable RgbRange: unknown
Output: 72 DVI-D-1 disabled disconnected priority 0 DVI Modes: Geometry: 0,0 0x0 Scale: 1 Rotation: 1 Overscan: 0 Vrr: incapable RgbRange: unknown

(please ignore `1913:640x480@60`, that's a custom mode added by a xrandr manager I'm developing at https://codeberg.org/nyanpasu64/auto-modeline)

DP-1 is my main display, HDMI-1 is my second display hooked up vertically, and DP-2 is a DP-to-VGA converter connected to no monitor at the time of the plasmashell crash. The crash occurred after the screen locked, blanked, and woke up.

I've managed to reproduce this issue using:

- Check "Manually block sleep and screen locking" (not sure if the issue reproduces without this)
- Lock the screen
- Wait for monitors to go black with a mouse cursor, then turn off
- Move the mouse to wake up the monitors (and crash plasmashell), then log in and run coredumpctl to find a crashed plasmashell.

Should be fixed by https://invent.kde.org/plasma/plasma-workspace/-/merge_requests/2841

Confirmed fixed for me with both https://invent.kde.org/plasma/plasma-workspace/-/merge_requests/2841 AND https://invent.kde.org/plasma/plasma-workspace/-/merge_requests/2842

The first MR made the initial backtrace disappear, but there was a second one behind, fixed by the second MR.

*** Bug 465892 has been marked as a duplicate of this bug. ***

*** Bug 465935 has been marked as a duplicate of this bug. ***

*** Bug 476135 has been marked as a duplicate of this bug. ***

*** Bug 479589 has been marked as a duplicate of this bug. ***

*** Bug 474868 has been marked as a duplicate of this bug. ***

We have two new reports from this with the exact same backtrace from people using a version of Plasma newer than 5.27.5; one from Debian testing and one from Neon. Re-opening.

It would be amazing if anyone previously affected by this could test Plasma 6 on X11 and see if they can still reproduce the issue.

I'm not sure I can get 6 running on debian whilst keeping 5 too.
Would it help to patch the debian's 5.27 with more debugs or experiments? I think I can patch and rebuild the .deb package for plasma-workspace.

Unfortunately the changes are so vast that patching 5.27 is not really possible.

*** Bug 480283 has been marked as a duplicate of this bug. ***

Demoting this from being a 15-minute bug since the X11 session is no longer default in Plasma 6.

I'm also still interested in learning whether anyone using X11 on Plasma 6 can still reproduce the issue.

I can't reproduce the issue.
I have a multimonitor setup with three screens two of which support FreeSync and I used to get this issue a couple of months ago (even posted a couple of comments here #c0 #c4) where one screen disconnecting would crash plasmashell. I can't reproduce the crash anymore, I tried removing cables, cutting power to monitors and it didn't crash. 
My system configuration has changed since when I last had the crash. I've removed one of the GPUs and recent kernel upgrades made the graphics driver much more stable than it used to be; it used to have crashes which were unrelated to Plasma. 

SOFTWARE/OS VERSIONS
Operating System: Arch Linux 
KDE Plasma Version: 6.0.1
KDE Frameworks Version: 6.0.0
Qt Version: 6.6.2
Graphics Platform: X11
Graphics Processor: AMD Radeon RX 6700 XT

Fantastic, thanks so much for re-testing. Let's call it fixed, then.