[Bug 1184543] New: wayand crashes when trying to display keyboard layout in GNOME live system
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 Bug ID: 1184543 Summary: wayand crashes when trying to display keyboard layout in GNOME live system Classification: openSUSE Product: openSUSE Distribution Version: Leap 15.3 Hardware: x86-64 OS: Other Status: NEW Severity: Major Priority: P5 - None Component: X.Org Assignee: gfx-bugs@suse.de Reporter: Ulrich.Windl@rz.uni-regensburg.de QA Contact: gfx-bugs@suse.de Found By: --- Blocker: --- Whenever I try to display the keyboard layout in the GNOME live system (openSUSE-Leap-15.3-GNOME-Live-x86_64-Media.iso), something odd happens: Suddenly there are a few text characters in the upper left corner of the screen, and the graphical background is all gone. It looks as if the system has crashed, while keys pressed are displayed. As it turned out from journal, wayland had crashed: pr 08 21:04:44 linux.fritz.box systemd[2492]: Started Application launched by gnome-shell. Apr 08 21:04:44 linux.fritz.box gnome-shell[4182]: syntax error: line 1 of stdin Apr 08 21:04:44 linux.fritz.box gnome-shell[4182]: Errors encountered in stdin; not compiled. Apr 08 21:04:44 linux.fritz.box gnome-shell[4184]: The XKEYBOARD keymap compiler (xkbcomp) reports: Apr 08 21:04:44 linux.fritz.box gnome-shell[4184]: > Internal error: Could not resolve keysym XF86WWAN Apr 08 21:04:44 linux.fritz.box gnome-shell[4184]: > Internal error: Could not resolve keysym XF86RFKill Apr 08 21:04:44 linux.fritz.box gnome-shell[4184]: Errors from xkbcomp are not fatal to the X server Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: Modifiers specified, but DRI is too old Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: Modifiers specified, but DRI is too old Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Backtrace: Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 0: /usr/bin/Xwayland (xorg_backtrace+0x65) [0x55f1d24fd375] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 1: /usr/bin/Xwayland (0x55f1d2361000+0x1a0049) [0x55f1d2501049] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 2: /lib64/libpthread.so.0 (0x7f00e9e8e000+0x13f80) [0x7f00e9ea1f80] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 3: /usr/bin/Xwayland (0x55f1d2361000+0x4faec) [0x55f1d23b0aec] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 4: /usr/bin/Xwayland (0x55f1d2361000+0x46188) [0x55f1d23a7188] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 5: /usr/bin/Xwayland (BlockHandler+0x3a) [0x55f1d24cf9ca] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 6: /usr/bin/Xwayland (WaitForSomething+0xd7) [0x55f1d24faa97] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 7: /usr/bin/Xwayland (0x55f1d2361000+0x169de3) [0x55f1d24cade3] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 8: /usr/bin/Xwayland (0x55f1d2361000+0x16e03c) [0x55f1d24cf03c] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 9: /lib64/libc.so.6 (__libc_start_main+0xed) [0x7f00e9aef34d] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 10: /usr/bin/Xwayland (_start+0x2a) [0x55f1d23a623a] Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Segmentation fault at address 0x0 Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: Fatal server error: Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Caught signal 11 (Segmentation fault). Server aborting Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) Apr 08 21:04:44 linux.fritz.box gkbd-keyboard-d[4179]: gkbd-keyboard-display: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. Apr 08 21:04:44 linux.fritz.box gnome-shell[2593]: Connection to xwayland lost Apr 08 21:04:44 linux.fritz.box gsd-xsettings[2828]: gsd-xsettings: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. Apr 08 21:04:44 linux.fritz.box systemd[2492]: gnome-launched-gkbd-keyboard-display-4179.scope: Succeeded. ... When waiting long enough, the display manager is started again (by systemd, I guess). However it's not obvious to the user what had happened (I also had reset the computer the first time). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c1 Stefan Dirsch <sndirsch@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sndirsch@suse.com Component|X.Org |GNOME Assignee|gfx-bugs@suse.de |gnome-bugs@suse.de Summary|wayand crashes when trying |Wayland session restarts |to display keyboard layout |due to Xwayland crash when |in GNOME live system |trying to display keyboard | |layout | |(gkbd-keyboard-display) in | |GNOME live system QA Contact|gfx-bugs@suse.de |qa-bugs@suse.de --- Comment #1 from Stefan Dirsch <sndirsch@suse.com> --- I'm wondering why the GNOME keyboard configuration tool (gkbd-keyboard-display apparently) talks to Xwayland. Shouldn't it talk to Wayland instead inside of a Wayland session? Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session. Reassigning to GNOME component. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c2 Yifan Jiang <yfjiang@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |gnome-bugs@suse.de, | |yfjiang@suse.com Assignee|gnome-bugs@suse.de |alynx.zhou@suse.com --- Comment #2 from Yifan Jiang <yfjiang@suse.com> --- Hi Alynx, would you help on spending time on the issue please. Thanks. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c3 Daike Yu <yu.daike@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |yu.daike@suse.com --- Comment #3 from Daike Yu <yu.daike@suse.com> ---
Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 3: /usr/bin/Xwayland (0x55f1d2361000+0x4faec) [0x55f1d23b0aec]
This is line 274 of xwl_glamor_gbm_get_wl_buffer_for_pixmap():
(gdb) info line *0x4faec Line 274 of "xwayland-glamor-gbm.c" starts at address 0x4faec <xwl_glamor_gbm_get_wl_buffer_for_pixmap+204> and ends at 0x4faef <xwl_glamor_gbm_get_wl_buffer_for_pixmap+207>. (gdb) l xwl_glamor_gbm_get_wl_buffer_for_pixmap 256 } 257 258 static struct wl_buffer * 259 xwl_glamor_gbm_get_wl_buffer_for_pixmap(PixmapPtr pixmap, 260 Bool *created) 261 { 262 struct xwl_screen *xwl_screen = xwl_screen_get(pixmap->drawable.pScreen); 263 struct xwl_pixmap *xwl_pixmap = xwl_pixmap_get(pixmap); 264 struct xwl_gbm_private *xwl_gbm = xwl_gbm_get(xwl_screen); 265 unsigned short width = pixmap->drawable.width; (gdb) 266 unsigned short height = pixmap->drawable.height; 267 int prime_fd; 268 int num_planes; 269 uint32_t strides[4]; 270 uint32_t offsets[4]; 271 uint64_t modifier; 272 int i; 273 274 if (xwl_pixmap->buffer) { 275 /* Buffer already exists. Return it and inform caller if interested. */ (gdb) p &(((struct xwl_pixmap*)0)->buffer) $2 = (struct wl_buffer **) 0x0 (gdb)
which looks like https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/47/diffs?commit... too, as in boo#118072. Would you please also check if https://build.opensuse.org/package/show/home:yudaike:branches:SUSE:SLE-15-SP... makes any difference for you? Thanks. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c4 --- Comment #4 from Daike Yu <yu.daike@suse.com> --- boo#1184072 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 Frederic Crozat <fcrozat@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |fcrozat@suse.com -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c7 --- Comment #7 from Alynx Zhou <alynx.zhou@suse.com> --- (In reply to Daike Yu from comment #3)
Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 3: /usr/bin/Xwayland (0x55f1d2361000+0x4faec) [0x55f1d23b0aec]
This is line 274 of xwl_glamor_gbm_get_wl_buffer_for_pixmap():
(gdb) info line *0x4faec Line 274 of "xwayland-glamor-gbm.c" starts at address 0x4faec <xwl_glamor_gbm_get_wl_buffer_for_pixmap+204> and ends at 0x4faef <xwl_glamor_gbm_get_wl_buffer_for_pixmap+207>. (gdb) l xwl_glamor_gbm_get_wl_buffer_for_pixmap 256 } 257 258 static struct wl_buffer * 259 xwl_glamor_gbm_get_wl_buffer_for_pixmap(PixmapPtr pixmap, 260 Bool *created) 261 { 262 struct xwl_screen *xwl_screen = xwl_screen_get(pixmap->drawable.pScreen); 263 struct xwl_pixmap *xwl_pixmap = xwl_pixmap_get(pixmap); 264 struct xwl_gbm_private *xwl_gbm = xwl_gbm_get(xwl_screen); 265 unsigned short width = pixmap->drawable.width; (gdb) 266 unsigned short height = pixmap->drawable.height; 267 int prime_fd; 268 int num_planes; 269 uint32_t strides[4]; 270 uint32_t offsets[4]; 271 uint64_t modifier; 272 int i; 273 274 if (xwl_pixmap->buffer) { 275 /* Buffer already exists. Return it and inform caller if interested. */ (gdb) p &(((struct xwl_pixmap*)0)->buffer) $2 = (struct wl_buffer **) 0x0 (gdb)
which looks like https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/47/ diffs?commit_id=036794bebce72a3fa2f95996d2e537ff568e0ff1 too, as in boo#118072. Would you please also check if https://build.opensuse.org/package/show/home:yudaike:branches:SUSE:SLE-15- SP2:Update/xorg-x11-server makes any difference for you?
Thanks.
I think this make sense, but it seems I cannot reproduce it on 15.3 in VM. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c8 --- Comment #8 from Yifan Jiang <yfjiang@suse.com> ---
I think this make sense, but it seems I cannot reproduce it on 15.3 in VM.
Just to be sure - is it GNOME Live you were using for testing? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c9 --- Comment #9 from Daike Yu <yu.daike@suse.com> --- (In reply to Yifan Jiang from comment #8)
I think this make sense, but it seems I cannot reproduce it on 15.3 in VM.
Just to be sure - is it GNOME Live you were using for testing?
In this case (In reply to Alynx Zhou from comment #7)
(In reply to Daike Yu from comment #3)
Apr 08 21:04:44 linux.fritz.box gnome-shell[2646]: (EE) 3: /usr/bin/Xwayland (0x55f1d2361000+0x4faec) [0x55f1d23b0aec]
This is line 274 of xwl_glamor_gbm_get_wl_buffer_for_pixmap():
(gdb) info line *0x4faec Line 274 of "xwayland-glamor-gbm.c" starts at address 0x4faec <xwl_glamor_gbm_get_wl_buffer_for_pixmap+204> and ends at 0x4faef <xwl_glamor_gbm_get_wl_buffer_for_pixmap+207>. (gdb) l xwl_glamor_gbm_get_wl_buffer_for_pixmap 256 } 257 258 static struct wl_buffer * 259 xwl_glamor_gbm_get_wl_buffer_for_pixmap(PixmapPtr pixmap, 260 Bool *created) 261 { 262 struct xwl_screen *xwl_screen = xwl_screen_get(pixmap->drawable.pScreen); 263 struct xwl_pixmap *xwl_pixmap = xwl_pixmap_get(pixmap); 264 struct xwl_gbm_private *xwl_gbm = xwl_gbm_get(xwl_screen); 265 unsigned short width = pixmap->drawable.width; (gdb) 266 unsigned short height = pixmap->drawable.height; 267 int prime_fd; 268 int num_planes; 269 uint32_t strides[4]; 270 uint32_t offsets[4]; 271 uint64_t modifier; 272 int i; 273 274 if (xwl_pixmap->buffer) { 275 /* Buffer already exists. Return it and inform caller if interested. */ (gdb) p &(((struct xwl_pixmap*)0)->buffer) $2 = (struct wl_buffer **) 0x0 (gdb)
which looks like https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/47/ diffs?commit_id=036794bebce72a3fa2f95996d2e537ff568e0ff1 too, as in boo#118072. Would you please also check if https://build.opensuse.org/package/show/home:yudaike:branches:SUSE:SLE-15- SP2:Update/xorg-x11-server makes any difference for you?
Thanks.
I think this make sense, but it seems I cannot reproduce it on 15.3 in VM.
xwl_pixmap_get() returns NULL if gbm_bo_create() fails, and the latter is called into GPU driver. So this is likely a driver/hardware specific issue... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c10 --- Comment #10 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> ---
(In reply to Yifan Jiang from comment #8) ... xwl_pixmap_get() returns NULL if gbm_bo_create() fails, and the latter is called into GPU driver. So this is likely a driver/hardware specific issue...
Some hardware info: Lenovo E15 Gen 2 (type 20T8, 20T9) with AMD Ryzen 4000 (Radeon Graphics) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c11 --- Comment #11 from Stefan Dirsch <sndirsch@suse.com> --- Thanks for the Xwayland patch. I suggest to submit the updated package to sle15-sp2/sle15-sp3. I don't see how the reporter could update Xwayland on a Live system for testing. Still it's sad to see that a Xwayland crash terminates the whole Wayland session!?! Could you explain? Again my questions from comment #2: "I'm wondering why the GNOME keyboard configuration tool (gkbd-keyboard-display apparently) talks to Xwayland. Shouldn't it talk to Wayland instead inside of a Wayland session? Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session." -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c12 --- Comment #12 from Daike Yu <yu.daike@suse.com> --- (In reply to Stefan Dirsch from comment #11)
Thanks for the Xwayland patch. I suggest to submit the updated package to sle15-sp2/sle15-sp3. I don't see how the reporter could update Xwayland on a Live system for testing.
Still it's sad to see that a Xwayland crash terminates the whole Wayland session!?! Could you explain? Again my questions from comment #2:
"I'm wondering why the GNOME keyboard configuration tool (gkbd-keyboard-display apparently) talks to Xwayland. Shouldn't it talk to Wayland instead inside of a Wayland session? Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session."
As for your question: Presumably because gnome-shell exits on losing connection to Xwayland, and this is because, well, it does :\. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c13 --- Comment #13 from Alynx Zhou <alynx.zhou@suse.com> --- (In reply to Yifan Jiang from comment #8)
I think this make sense, but it seems I cannot reproduce it on 15.3 in VM.
Just to be sure - is it GNOME Live you were using for testing?
Nope, I am using installed 15.3 since another bug mentioned is 15.3. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c14 --- Comment #14 from Alynx Zhou <alynx.zhou@suse.com> --- (In reply to Stefan Dirsch from comment #11)
Thanks for the Xwayland patch. I suggest to submit the updated package to sle15-sp2/sle15-sp3. I don't see how the reporter could update Xwayland on a Live system for testing.
Still it's sad to see that a Xwayland crash terminates the whole Wayland session!?! Could you explain? Again my questions from comment #2:
"I'm wondering why the GNOME keyboard configuration tool (gkbd-keyboard-display apparently) talks to Xwayland. Shouldn't it talk to Wayland instead inside of a Wayland session?
See <https://gitlab.gnome.org/GNOME/libgnomekbd/-/blob/master/libgnomekbd/gkbd-keyboard-display.c#L60>, it's a little bit old program and is set to only run in X11 (or XWayland on Wayland), so it won't be a native Wayland client.
Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session."
I think gnome-shell needs to keep a XWayland process running with it for all time now. There will be a XWayland run as needed feature in upstream but not in current version we are using. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c15 --- Comment #15 from Alynx Zhou <alynx.zhou@suse.com> --- Sorry, my mistake, on-demand XWayland is already implemented (See <https://gitlab.gnome.org/GNOME/mutter/commit/141373f0ba137152d412f1b6839fb20c1150f762>), but as a experimental feature and is disabled by default (See <https://gitlab.gnome.org/GNOME/mutter/-/blob/gnome-3-34/src/core/main.c#L729-752>, the commit called this and it by default returns META_DISPLAY_POLICY_MANDATORY, so GNOME Shell will just exit if XWayland exits as coded). -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c16 --- Comment #16 from Stefan Dirsch <sndirsch@suse.com> --- (In reply to Alynx Zhou from comment #14)
See <https://gitlab.gnome.org/GNOME/libgnomekbd/-/blob/master/libgnomekbd/gkbd- keyboard-display.c#L60>, it's a little bit old program and is set to only run in X11 (or XWayland on Wayland), so it won't be a native Wayland client.
Hmm. Ok. So on a GNOME wayland session we rely on a keyboard configuration tool, which only works with X11? Wow! Good to know ...
Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session."
I think gnome-shell needs to keep a XWayland process running with it for all time now. There will be a XWayland run as needed feature in upstream but not in current version we are using.
Well, you could restart Xwayland if it would ever crash (losing the connected X clients of course), but apparently the desktop is already becoming unusable with Xwayland no longer running, so gnome-shell decides to just terminate. :-( -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c17 --- Comment #17 from Stefan Dirsch <sndirsch@suse.com> --- (In reply to Alynx Zhou from comment #15)
Sorry, my mistake, on-demand XWayland is already implemented (See <https://gitlab.gnome.org/GNOME/mutter/commit/ 141373f0ba137152d412f1b6839fb20c1150f762>), but as a experimental feature and is disabled by default (See <https://gitlab.gnome.org/GNOME/mutter/-/blob/gnome-3-34/src/core/main. c#L729-752>, the commit called this and it by default returns META_DISPLAY_POLICY_MANDATORY, so GNOME Shell will just exit if XWayland exits as coded).
Thanks! Sounds good. Would be a candidate for Tumbleweed then. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c18 --- Comment #18 from Stefan Dirsch <sndirsch@suse.com> --- (In reply to Stefan Dirsch from comment #17)
(In reply to Alynx Zhou from comment #15)
Sorry, my mistake, on-demand XWayland is already implemented (See <https://gitlab.gnome.org/GNOME/mutter/commit/ 141373f0ba137152d412f1b6839fb20c1150f762>), but as a experimental feature and is disabled by default (See <https://gitlab.gnome.org/GNOME/mutter/-/blob/gnome-3-34/src/core/main. c#L729-752>, the commit called this and it by default returns META_DISPLAY_POLICY_MANDATORY, so GNOME Shell will just exit if XWayland exits as coded).
Thanks! Sounds good. Would be a candidate for Tumbleweed then.
BTW, I'm just about splitting off a xwayland package for Tumbleweed, so Xwayland should be more up-to-date in the future. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c19 --- Comment #19 from Alynx Zhou <alynx.zhou@suse.com> --- (In reply to Stefan Dirsch from comment #16)
(In reply to Alynx Zhou from comment #14)
See <https://gitlab.gnome.org/GNOME/libgnomekbd/-/blob/master/libgnomekbd/gkbd- keyboard-display.c#L60>, it's a little bit old program and is set to only run in X11 (or XWayland on Wayland), so it won't be a native Wayland client.
Hmm. Ok. So on a GNOME wayland session we rely on a keyboard configuration tool, which only works with X11? Wow! Good to know ...
Maybe it can be updated to a native Wayland client, if it does not depends on any X11 specific APIs.
Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session."
I think gnome-shell needs to keep a XWayland process running with it for all time now. There will be a XWayland run as needed feature in upstream but not in current version we are using.
Well, you could restart Xwayland if it would ever crash (losing the connected X clients of course), but apparently the desktop is already becoming unusable with Xwayland no longer running, so gnome-shell decides to just terminate. :-(
It will try to restart XWayland if the experimental feature is enabled, but it's disabled by default. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c20 --- Comment #20 from Alynx Zhou <alynx.zhou@suse.com> --- (In reply to Stefan Dirsch from comment #18)
(In reply to Stefan Dirsch from comment #17)
(In reply to Alynx Zhou from comment #15)
Sorry, my mistake, on-demand XWayland is already implemented (See <https://gitlab.gnome.org/GNOME/mutter/commit/ 141373f0ba137152d412f1b6839fb20c1150f762>), but as a experimental feature and is disabled by default (See <https://gitlab.gnome.org/GNOME/mutter/-/blob/gnome-3-34/src/core/main. c#L729-752>, the commit called this and it by default returns META_DISPLAY_POLICY_MANDATORY, so GNOME Shell will just exit if XWayland exits as coded).
Thanks! Sounds good. Would be a candidate for Tumbleweed then.
BTW, I'm just about splitting off a xwayland package for Tumbleweed, so Xwayland should be more up-to-date in the future.
I've seen news that XWayland upstream already make a it a standalone project and have the first release. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c21 --- Comment #21 from Stefan Dirsch <sndirsch@suse.com> --- (In reply to Alynx Zhou from comment #20)
I've seen news that XWayland upstream already make a it a standalone project and have the first release.
Yes, that's what I'm using. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 Cliff Zhao <qzhao@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |qzhao@suse.com -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c23 --- Comment #23 from Frederic Crozat <fcrozat@suse.com> --- (In reply to Stefan Dirsch from comment #16)
(In reply to Alynx Zhou from comment #14)
See <https://gitlab.gnome.org/GNOME/libgnomekbd/-/blob/master/libgnomekbd/gkbd- keyboard-display.c#L60>, it's a little bit old program and is set to only run in X11 (or XWayland on Wayland), so it won't be a native Wayland client.
Hmm. Ok. So on a GNOME wayland session we rely on a keyboard configuration tool, which only works with X11? Wow! Good to know ...
GNOME doesn't rely on it, it appears one specific application is pulling this (and I'm not sure why it wasn't updated to not rely on X11..).
Also I don't see why a Wayland session should crash just because of Xwayland is crashing. I would expect Xclients terminating but nothing more, i.e. not the whole Wayland session."
I think gnome-shell needs to keep a XWayland process running with it for all time now. There will be a XWayland run as needed feature in upstream but not in current version we are using.
Well, you could restart Xwayland if it would ever crash (losing the connected X clients of course), but apparently the desktop is already becoming unusable with Xwayland no longer running, so gnome-shell decides to just terminate. :-(
Latest GNOME release (3.36 or 3.38) are no longer requiring xwayland to be running, allowing it to be started on demand. But for 3.34 and older version, we have to keep xwayland running all the time. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c33 --- Comment #33 from Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de> --- Maybe it's just wrong expectations, but the GNOME live image downloaded yesterday (assuming it corresponds to RC1) just crashes the very same way (and trying to start YaST's Online Update from within the Live system results in the very same effect). A guess of the build time stamp is 2021-04-26 19:17:50... -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=1184543 http://bugzilla.opensuse.org/show_bug.cgi?id=1184543#c34 --- Comment #34 from OBSbugzilla Bot <bwiedemann+obsbugzillabot@suse.com> --- This is an autogenerated message for OBS integration: This bug (1184543) was mentioned in https://build.opensuse.org/request/show/1034269 Factory / xorg-x11-server -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠suse.com