💾 Archived View for gemini.dimakrasner.com › puppy-slow-shutdown.gmi captured on 2023-06-16 at 16:21:18. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2022-04-28)
-=-=-=-=-=-=-
For some reason, poweroff in Puppy takes a very very long time.
Puppy replaces the poweroff executable (normally, a symlink to busybox) with a script. This way, it can hook the shutdown process and do some extra steps, like saving files in RAM back to the flash drive before the system shuts down. All these goodies happen inside /etc/rc.d/rc.shutdown:
/etc/rc.d/rc.shutdown <= case $script in poweroff) if [ $can_shutdown -eq 1 ]; then /bin/busybox poweroff else clear > /dev/console dialog --ok-label "Restart" --msgbox "IT'S NOW SAFE TO TURN OFF YOUR COMPUTER" 5 43 /bin/busybox reboot fi ;; reboot) /bin/busybox reboot ;; esac
But you can't just bash -x /etc/rc.d/rc.shutdown to see why it takes so long, because it has this magic block:
if [ ! -f /tmp/debugshutdown ] ; then clear exec 1> /dev/null 2>&1 fi
Let's try to create this magic file then, and see what /etc/rc.d/rc.shutdown does:
~$ touch /tmp/debugshutdown ~$ bash -x /etc/rc.d/rc.shutdown > /tmp/log 2>&1
Oh no! The log is empty and the redirection doesn't work. How can it be? The answer is in line 9:
[ -f /tmp/debugshutdown ] && exec &>/dev/console
/tmp/debugshutdown lets you choose between output you can't grep or redirect to a file, or no output at all. This makes it useless as a debugging tool, as the name suggests.
Once line 9 is deleted (only for the sake of debugging), it's easy to see that rc.shutdown hangs when it tries to stop seatd:
~$ bash -x /etc/rc.d/rc.shutdown . . . + /etc/init.d/seatd stop
This happens because the init script waits for start-stop-daemon, which takes a very long time:
~$ ps aux | grep seatd root 3053 0.0 0.0 2860 1096 ? S 17:01 0:00 /usr/sbin/seatd root 4110 0.0 0.1 6572 2236 pts/1 S+ 17:01 0:00 grep seatd ~$ bash -x /etc/init.d/seatd stop . . . + do_stop + start-stop-daemon --stop --quiet --retry=TERM/30/KILL/5 --pidfile /var/run/seatd.pid --name seatd + RETVAL=1 + '[' 1 = 2 ']' + start-stop-daemon --stop --quiet --oknodo --retry=0/30/KILL/5 --exec /usr/sbin/seatd
Now, let's see why start-stop-daemon takes so long. It's not a shell script, so we'll use the venerable strace to see what it does. I'm often surprised when people rush to find a debugger, install debugging symbols for dependencies, recompile the application with -g, find a way to attach to a running process without killing it, etc', when a tool like strace is such a perfect fit for syscall-oriented problems, like processes that get stuck waiting for something. It's so easy to use, and very effective.
Back to debugging! It looks like start-stop-daemon is stuck in a loop:
~$ strace start-stop-daemon --stop --quiet --oknodo --retry=0/30/KILL/5 --exec /usr/sbin/seatd
.
.
.
4384 readlink("/proc/4363/exe", "/usr/sbin/seatd", 256) = 15
4384 newfstatat(AT_FDCWD, "/usr/sbin/seatd", {st_mode=S_IFREG|0755, st_size=47504, ...}, 0) = 0
.
.
.
4384 kill(4363, 0) = 0
4384 pselect6(0, NULL, NULL, NULL, {tv_sec=0, tv_nsec=33284000}, {sigmask=NULL, sigsetsize=8}) = 0 (Timeout)
.
.
.
It uses polling to wait for seatd to terminate:
If signal is 0, then no actual signal is sent, but error checking is still performed.
But why is seatd still running? It looks like there is some bug here. I'm not sure where it is, but I see two likely options:
This doesn't matter really, because Puppy doesn't have logind. A wlroots-based compositor that talks to seatd via libseat eventually falls back to libseat's "builtin" backend, where libseat itself takes care of seat management without relying on seatd or logind. Puppy doesn't need the seatd daemon in first place, so the solution is to disable it: this solves the slow shutdown problem and reduces the number of running processes, which is always good.
Now, the process list with a fullscreen Xwayland, JWM, PCManFM (with desktop icons), Blueman and the traditional set of tray icons, under a patched dwl-based compositor, looks like this:
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 1 1.8 0.0 3792 1320 ? Ss 17:23 0:01 /bin/busybox init root 2 0.0 0.0 0 0 ? S 17:23 0:00 [kthreadd] root 3 0.0 0.0 0 0 ? I< 17:23 0:00 [rcu_gp] root 4 0.0 0.0 0 0 ? I< 17:23 0:00 [rcu_par_gp] root 5 0.0 0.0 0 0 ? I< 17:23 0:00 [netns] root 6 0.0 0.0 0 0 ? I 17:23 0:00 [kworker/0:0-events] root 7 0.0 0.0 0 0 ? I< 17:23 0:00 [kworker/0:0H-events_highpri] root 8 1.2 0.0 0 0 ? I 17:23 0:00 [kworker/u2:0-events_unbound] root 9 0.0 0.0 0 0 ? I< 17:23 0:00 [mm_percpu_wq] root 10 0.0 0.0 0 0 ? S 17:23 0:00 [rcu_tasks_rude_] root 11 0.0 0.0 0 0 ? S 17:23 0:00 [rcu_tasks_trace] root 12 0.3 0.0 0 0 ? S 17:23 0:00 [ksoftirqd/0] root 13 3.1 0.0 0 0 ? I 17:23 0:01 [rcu_sched] root 14 0.0 0.0 0 0 ? S 17:23 0:00 [migration/0] root 15 0.0 0.0 0 0 ? S 17:23 0:00 [idle_inject/0] root 16 0.7 0.0 0 0 ? I 17:23 0:00 [kworker/0:1-events] root 17 0.0 0.0 0 0 ? S 17:23 0:00 [cpuhp/0] root 18 0.0 0.0 0 0 ? S 17:23 0:00 [kdevtmpfs] root 19 0.0 0.0 0 0 ? I< 17:23 0:00 [inet_frag_wq] root 20 0.0 0.0 0 0 ? S 17:23 0:00 [kauditd] root 21 0.0 0.0 0 0 ? S 17:23 0:00 [khungtaskd] root 22 0.0 0.0 0 0 ? S 17:23 0:00 [oom_reaper] root 23 0.0 0.0 0 0 ? I< 17:23 0:00 [writeback] root 24 0.0 0.0 0 0 ? S 17:23 0:00 [kcompactd0] root 25 0.0 0.0 0 0 ? SN 17:23 0:00 [ksmd] root 26 0.0 0.0 0 0 ? SN 17:23 0:00 [khugepaged] root 78 0.0 0.0 0 0 ? I< 17:23 0:00 [kintegrityd] root 79 0.0 0.0 0 0 ? I< 17:23 0:00 [kblockd] root 80 0.0 0.0 0 0 ? I< 17:23 0:00 [blkcg_punt_bio] root 81 0.0 0.0 0 0 ? I< 17:23 0:00 [tpm_dev_wq] root 82 0.0 0.0 0 0 ? I< 17:23 0:00 [tifm] root 83 0.0 0.0 0 0 ? I< 17:23 0:00 [ata_sff] root 84 0.0 0.0 0 0 ? I< 17:23 0:00 [md] root 85 0.0 0.0 0 0 ? I< 17:23 0:00 [edac-poller] root 86 0.0 0.0 0 0 ? I< 17:23 0:00 [devfreq_wq] root 87 0.0 0.0 0 0 ? S 17:23 0:00 [watchdogd] root 88 0.3 0.0 0 0 ? I 17:23 0:00 [kworker/u2:1-loop0] root 89 2.0 0.0 0 0 ? I 17:23 0:01 [kworker/u2:2-events_unbound] root 90 0.2 0.0 0 0 ? I< 17:23 0:00 [kworker/0:1H-kblockd] root 91 0.0 0.0 0 0 ? S 17:23 0:00 [kswapd0] root 92 0.0 0.0 0 0 ? S 17:23 0:00 [ecryptfs-kthrea] root 94 0.0 0.0 0 0 ? I< 17:23 0:00 [kthrotld] root 95 0.0 0.0 0 0 ? I< 17:23 0:00 [acpi_thermal_pm] root 96 0.0 0.0 0 0 ? I< 17:23 0:00 [nvme-wq] root 97 0.0 0.0 0 0 ? I< 17:23 0:00 [nvme-reset-wq] root 98 0.0 0.0 0 0 ? I< 17:23 0:00 [nvme-delete-wq] root 99 0.0 0.0 0 0 ? S 17:23 0:00 [scsi_eh_0] root 100 0.0 0.0 0 0 ? I< 17:23 0:00 [scsi_tmf_0] root 101 0.0 0.0 0 0 ? S 17:23 0:00 [scsi_eh_1] root 102 0.0 0.0 0 0 ? I< 17:23 0:00 [scsi_tmf_1] root 103 0.0 0.0 0 0 ? I< 17:23 0:00 [vfio-irqfd-clea] root 104 0.0 0.0 0 0 ? I< 17:23 0:00 [kvub300c] root 105 0.0 0.0 0 0 ? I< 17:23 0:00 [kvub300p] root 106 0.0 0.0 0 0 ? I< 17:23 0:00 [kvub300d] root 107 0.0 0.0 0 0 ? I< 17:23 0:00 [mld] root 108 0.0 0.0 0 0 ? I< 17:23 0:00 [ipv6_addrconf] root 109 0.0 0.0 0 0 ? I 17:23 0:00 [kworker/u2:3] root 118 0.0 0.0 0 0 ? I< 17:23 0:00 [kstrp] root 119 0.0 0.0 0 0 ? I< 17:23 0:00 [zswap-shrink] root 120 0.0 0.0 0 0 ? I< 17:23 0:00 [kworker/u3:0] root 125 0.0 0.0 0 0 ? I< 17:23 0:00 [charger_manager] root 950 0.4 0.2 14384 4584 ? Ss 17:23 0:00 udevd --daemon --resolve-names=early message+ 2449 0.0 0.1 8276 3464 ? Ss 17:23 0:00 dbus-daemon --system root 2469 0.0 0.0 3856 152 ? Ss 17:23 0:00 syslogd -m 0 root 2472 0.0 0.0 3800 1872 ? Ss 17:23 0:00 klogd root 2680 0.1 0.2 8304 4892 tty1 Ss+ 17:23 0:00 -sh root 2681 0.0 0.0 2816 1948 tty2 Ss+ 17:23 0:00 getty 38400 tty2 root 2710 0.1 0.2 10572 5024 ? S 17:23 0:00 /usr/sbin/bluetoothd root 2740 0.0 0.2 16780 4328 ? Ss 17:23 0:00 /usr/sbin/connmand --nodnsproxy root 2748 0.2 0.3 16520 6604 ? S 17:23 0:00 /sbin/wpa_supplicant -u -s -O /run/wpa_supplicant root 2777 0.0 0.1 7588 4004 tty1 S+ 17:23 0:00 /bin/sh /bin/xwin root 2789 0.0 0.1 7384 3504 tty1 S+ 17:23 0:00 /bin/sh /bin/startxwayland /root/.xinitrc root 2856 0.0 0.0 2828 1040 ? S 17:23 0:00 bpfilter_umh root 2924 14.1 4.5 193012 92704 tty1 Sl+ 17:23 0:05 dwl-kiosk -s Xwayland :0 -auth /root/.Xauthority -dpi 96 root 3020 0.1 0.2 26060 5200 ? Sl 17:23 0:00 pup-volume-monitor daemon 3048 0.0 0.0 6252 348 ? Ss 17:23 0:00 /usr/sbin/nscd root 3066 2.8 4.5 194444 91816 tty1 S+ 17:23 0:00 Xwayland :0 -auth /root/.Xauthority -dpi 96 root 3132 0.0 0.0 6456 1644 tty1 S+ 17:23 0:00 swayidle root 3133 2.5 1.1 41080 23784 tty1 S+ 17:23 0:00 jwm root 3154 0.0 0.1 8292 2056 tty1 S+ 17:23 0:00 dbus-launch --exit-with-x11 --sh-syntax root 3155 0.0 0.1 8052 3232 ? Ss 17:23 0:00 /usr/bin/dbus-daemon --syslog --fork --print-pid 5 --print-address 7 --session spot 3163 0.0 0.0 8292 2020 tty1 S+ 17:23 0:00 dbus-launch --exit-with-x11 spot 3164 0.0 0.1 8072 3196 ? Ss 17:23 0:00 /usr/bin/dbus-daemon --syslog --fork --print-pid 5 --print-address 7 --session spot 3173 0.8 0.7 384856 15064 ? Sl 17:23 0:00 pulseaudio --start --log-target=syslog root 3182 0.0 0.2 25824 5824 ? Sl 17:23 0:00 /usr/libexec/dconf-service spot 3206 0.0 0.2 25824 5716 ? Sl 17:23 0:00 /usr/libexec/dconf-service root 3353 0.1 0.2 9152 4724 ? Ss 17:23 0:00 xload -nolabel -fg #56718A -hl white -bg #ccc spot 3360 7.1 2.3 111668 46684 ? Sl 17:23 0:01 /usr/bin/python3 /usr/bin/blueman-applet.bin spot 3367 0.8 3.7 219836 76060 ? Sl 17:23 0:00 /usr/bin/connman-ui-gtk.bin root 3375 0.0 0.0 2828 1136 ? S 17:23 0:00 fixmenusd fixmenus; jwm -reload; killall -HUP labwc root 3386 2.0 3.8 229392 78540 ? Sl 17:23 0:00 pcmanfm --desktop root 3397 0.5 3.6 219076 73520 ? Sl 17:23 0:00 firewallstatus root 3400 0.6 3.6 219136 73224 ? S 17:23 0:00 freememapplet_tray spot 3409 0.6 3.7 484824 76564 ? S 17:23 0:00 /usr/bin/pa-applet.bin root 3431 0.3 0.0 3856 156 ? Ss 17:23 0:00 acpid root 3961 4.2 1.8 81680 36472 ? Sl 17:24 0:00 /usr/bin/python3 /usr/libexec/blueman-mechanism root 3966 0.1 0.3 38320 7736 ? Sl 17:24 0:00 /usr/libexec/polkitd --no-debug spot 3971 0.3 0.3 18200 7224 ? S 17:24 0:00 /usr/lib/bluetooth/obexd spot 3973 2.1 2.1 110524 42624 ? Sl 17:24 0:00 /usr/bin/python3 /usr/bin/blueman-tray
This is very clean and very good, considering the compatibility with legacy Puppy stuff that uses X11 tools like setxkbmap, the ability to run Wayland and X11 applications side-by-side, the functionality to RAM/CPU consumption ratio, and how snappy it feels. The Wayland port of Puppy is starting to look like a bright future for Puppy, and a leap forward.