Skip to content
Snippets Groups Projects

Blank screen on boot in quemu

    • Closed Issue created by Dorota Czaplejewicz @dorota.czaplejewicz

    Feature: #15 (closed)

    When (re)starting the systemd unit in QEMU (virt-manager), phosh only shows up the clock for a split second, and then stays lingering without anything happening on the screen.

    Works well with pkill phosh && /usr/lib/x86_64-linux-gnu/phosh.

    Edit: phosh probably doesn't show up at all based on new research, and that's just a glitch.

    Edited

    Designs

      Child items ...

      2. Activity

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        Somewhere near the end of phoc output:

        Missing Wayland requirements for glamor GBM backend
Failed to initialize glamor, falling back to sw
2019-06-06 10:04:57 - [types/wlr_relative_pointer_v1.c:173] relative_pointer_v1 manager bound to client 0x562c5ed7cac0
2019-06-06 10:04:57 - [types/wlr_surface.c:609] New wlr_surface 0x562c5ee61df0 (res 0x562c5ed50d60)
2019-06-06 10:04:57 - [xwayland/xwayland.c:350] sent SIGUSR1 to process 1265
2019-06-06 10:04:57 - [src/cursor.c:523] Denying request to set cursor from unfocused client

        pid 1265 is /usr/bin/phoc.

        Edited by Dorota Czaplejewicz
      • Dorota Czaplejewicz changed the description

        changed the description

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        Bad phosh output ends like:

        
[3615933.421] wl_output@13.done()
[3615933.437] wl_callback@14.done(2)
[3615933.450] wl_seat@15.name("seat0")
[3615933.462] wl_seat@15.capabilities(3)
[3615933.475]  -> wl_seat@15.get_pointer(new id wl_pointer@14)
[3615933.504]  -> zwp_pointer_gestures_v1@12.get_swipe_gesture(new id zwp_pointer_gesture_swipe_v1@3, wl_pointer@14)
[3615933.527]  -> zwp_pointer_gestures_v1@12.get_pinch_gesture(new id zwp_pointer_gesture_pinch_v1@22, wl_pointer@14)
[3615933.550]  -> wl_seat@15.get_keyboard(new id wl_keyboard@23)
[3615933.577] wl_callback@21.done(2)
[3615933.591]  -> wl_registry@2.bind(8, "xdg_wm_base", 1, new id [unknown]@21)

        good phosh output continues:

        (phosh:1452): dbind-WARNING **: 18:25:43.529: Error retrieving accessibility bus address: org.freedesktop.DBus.Error.ServiceUnknown: The name org.a11y.Bus was not provided by any .service files
[3615962.123]  -> wl_display@1.get_registry(new id wl_registry@24)
[3615962.166]  -> wl_display@1.sync(new id wl_callback@25)
[3615962.386] wl_display@1.delete_id(25)
[3615962.403] wl_keyboard@23.keymap(1, fd 9, 49311)
[3615964.317] wl_keyboard@23.repeat_info(25, 600)
[3615964.343] wl_registry@24.global(1, "wl_data_device_manager", 3)
[3615964.369] wl_registry@24.global(2, "wl_shm", 1)
[3615964.393] wl_registry@24.global(3, "zwp_linux_dmabuf_v1", 3)
[3615964.417] wl_registry@24.global(4, "zxdg_output_manager_v1", 2)
[3615964.441]  -> wl_registry@24.bind(4, "zxdg_output_manager_v1", 2, new id [unknown]@26)
      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        Backtrace of the bad output on thread 1:

        #0  0x00007ff74b6cf819 in poll () at /lib/x86_64-linux-gnu/libc.so.6
#1  0x00007ff74a397d97 in  () at /usr/lib/x86_64-linux-gnu/libxcb.so.1
#2  0x00007ff74a395b31 in xcb_connect_to_fd ()
    at /usr/lib/x86_64-linux-gnu/libxcb.so.1
#3  0x00007ff74a399984 in xcb_connect_to_display_with_auth_info ()
    at /usr/lib/x86_64-linux-gnu/libxcb.so.1
#4  0x00007ff74b363db2 in _XConnectXCB ()
    at /usr/lib/x86_64-linux-gnu/libX11.so.6
#5  0x00007ff74b354972 in XOpenDisplay ()
    at /usr/lib/x86_64-linux-gnu/libX11.so.6
#6  0x00007ff74978b074 in atspi_get_a11y_bus ()
    at /usr/lib/x86_64-linux-gnu/libatspi.so.0
#7  0x00007ff74aca760c in atk_bridge_adaptor_init ()
    at /usr/lib/x86_64-linux-gnu/libatk-bridge-2.0.so.0
#8  0x00007ff74c265694 in  () at /usr/lib/x86_64-linux-gnu/libgtk-3.so.0
#9  0x00007ff74c4105b9 in  () at /usr/lib/x86_64-linux-gnu/libgtk-3.so.0
#10 0x00007ff74bc99c7d in g_closure_invoke ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#11 0x00007ff74bcad345 in  ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#12 0x00007ff74bcb625e in g_signal_emit_valist ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#13 0x00007ff74bcb691f in g_signal_emit ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#14 0x00007ff74bc9e354 in  ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#15 0x00007ff74bca07e1 in g_object_notify ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#16 0x00007ff74c112b78 in  () at /usr/lib/x86_64-linux-gnu/libgdk-3.so.0
#17 0x00007ff74bc99c7d in g_closure_invoke ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#18 0x00007ff74bcacb9d in  ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#19 0x00007ff74bcb625e in g_signal_emit_valist ()
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#20 0x00007ff74bcb6df4 in g_signal_emit_by_name ()
--Type <RET> for more, q to quit, c to continue without paging--
    at /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#21 0x00007ff74c16dc85 in  () at /usr/lib/x86_64-linux-gnu/libgdk-3.so.0
#22 0x00007ff74c1129ed in gdk_display_manager_open_display ()
    at /usr/lib/x86_64-linux-gnu/libgdk-3.so.0
#23 0x00007ff74c4119d2 in  () at /usr/lib/x86_64-linux-gnu/libgtk-3.so.0
#24 0x00007ff74bbc4748 in g_option_context_parse ()
    at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
#25 0x000056272c6f163a in main ()

        Seems X-related?

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        @guido.gunther I'm not sure exactly what it is, but it goes away when I set DISPLAY="" when running phosh. Based on the stack trace, it's GTK trying to access ATK via X to Xwayland, and there's something stalling in this path.

        I haven't noticed any bad effects when removing DISPLAY, maybe we should drop it altogther instead of trying to find the root cause?

      • Guido Gunther
        Guido Gunther @guido.gunther ·
        Owner

        See the bt above is phosh? What does phoc do at that point? Could that be an xwayland startup race?

        I'd be o.k. with your suggestion if we could do away with xwayland altogether but since parts of g-s-d still need X11 (Apps_Issues#42) we might hit this with something else later on. And we also want to allow people to reliably run X11 apps when they want to so I think we should debug further.

        Could you try to not lazy start xwayland (see wlr_xwayland_create()) and see if that changes a thing? Did you check what xwayland does in the hanging case?

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        The bt is phosh indeed. I didn't check bt of phoc, but the output is the same as in the first post. I'll try to check the startup race suggestion. I'm not sure what xwayland was doing, gdb-ing it didn't give me any insight.

        we might hit this with something else later on

        I think that's actually a good thing, if it nudges us towards fixing gsd. I also thought we don't want to ship anything X. I'm only complaining because this is difficult to debug, with standard outputs going somewhere I can't find them, things not being started in the debugger, and so on. Hope you understand :)

      • Guido Gunther
        Guido Gunther @guido.gunther ·
        Owner

        Hope you understand :)

        Sure I do. but I also think making things debuggable is an important task by itself (same as testability). If we ship several thousand devices there will be bugs and when we can't say: "please do x, y and z" and send the output of "whatever.log" we will be having a very hard time to reproduce issues.

        Some of these things might just be documentation issues

        developer.puri.sm#79 (closed) developer.puri.sm#38 (closed)

        but others might not be. Ideally we'd have everything e.g. in the user's journal so we can correlate timestamps.

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        The entry in /usr/share/phosh/rootston.ini seems to solve it:

        [core]
xwayland=immediate

        When it comes to logging, I couldn't find any way whatsoever to get the output directly the last time I tried, so I resorted to a wrapper scripts, both for phoc and for phosh. That made me believe that maybe I'm not the most qualified to help make logging easy.

      • Dorota Czaplejewicz mentioned in commit dorota.czaplejewicz/phosh@43e8be47

        mentioned in commit dorota.czaplejewicz/phosh@43e8be47

      • Dorota Czaplejewicz mentioned in merge request phosh!191 (closed)

        mentioned in merge request phosh!191 (closed)

      • Dorota Czaplejewicz changed the description

        changed the description

      • L5 Robot Underling mentioned in issue #15 (closed)

        mentioned in issue #15 (closed)

      • Dorota Czaplejewicz mentioned in issue #11 (closed)

        mentioned in issue #11 (closed)

      • Dorota Czaplejewicz
        Dorota Czaplejewicz @dorota.czaplejewicz ·
        Author Maintainer

        I filed phosh!191 (closed) which solves this.

      • Guido Gunther
        Guido Gunther @guido.gunther ·
        Owner

        The cause is that we don't end up receiving the SIGUSR1 from our child. I have yet to figure out why that is.

      • Guido Gunther mentioned in merge request !14 (merged)

        mentioned in merge request !14 (merged)

      • Dorota Czaplejewicz closed via merge request !14 (merged)

        closed via merge request !14 (merged)

      • Dorota Czaplejewicz closed via commit 937c1f78

        closed via commit 937c1f78

      • Guido Gunther closed via commit b9ba9bd4

        closed via commit b9ba9bd4

      Please register or sign in to reply