Multiple processes with maximum CPU usage

[joelsleeba]

OS: Arch Kernel: 6.5.5-arch1-1 Hyprland: 0.30.0-1 xdg-desktop-portal-hyprland: 1.1.0-3

steps to reproduce:

1. Start Hyprland
2. Exit Hyprland
3. Start Hyprland

Now there was maximum CPU usage. I checked htop to find two proceses with PID 777 and 829, both of xdg-desktop-portal-hyprland utilizing over 100% of the CPU thread. I ran strace on the two processes. Details below

trace of 777

[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e390, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 0, NULL, FUTEX_BITSET_MATCH_ANY <unfinished ...>
[pid   829] futex(0x56437ec5e390, FUTEX_WAKE_PRIVATE, 2147483647 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAIT_PRIVATE, 2, NULL <unfinished ...>
[pid   829] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e398, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e394, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 0, NULL, FUTEX_BITSET_MATCH_ANY <unfinished ...>
[pid   829] futex(0x56437ec5e394, FUTEX_WAKE_PRIVATE, 2147483647 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAIT_PRIVATE, 2, NULL <unfinished ...>
[pid   829] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e398, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1^C[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1strace: Process 777 detached

trace of 829

[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e394, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 0, NULL, FUTEX_BITSET_MATCH_ANY <unfinished ...>
[pid   829] futex(0x56437ec5e394, FUTEX_WAKE_PRIVATE, 2147483647 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAIT_PRIVATE, 2, NULL <unfinished ...>
[pid   829] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e398, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e390, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 0, NULL, FUTEX_BITSET_MATCH_ANY <unfinished ...>
[pid   829] futex(0x56437ec5e390, FUTEX_WAKE_PRIVATE, 2147483647 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAIT_PRIVATE, 2, NULL <unfinished ...>
[pid   829] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   777] <... futex resumed>)        = -1 EAGAIN (Resource temporarily unavailable)
[pid   829] <... futex resumed>)        = 0
[pid   777] futex(0x56437ec5e398, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1 <unfinished ...>
[pid   777] <... futex resumed>)        = 0
[pid   829] <... poll resumed>)         = 1 ([{fd=6, revents=POLLIN|POLLHUP}])
[pid   777] futex(0x56437ec5e3c8, FUTEX_WAKE_PRIVATE, 1 <unfinished ...>
^C[pid   829] poll([{fd=3, events=POLLIN}, {fd=6, events=POLLIN}, {fd=7, events=POLLIN}], 3, -1strace: Process 829 detached

[vaxerski]

what if you kill xdph and restart it? Maybe it's an old session that is confused?

[thejch]

have this issue too. for me restarting xdph fixes it, but if i reboot this happens again

[darknebula05]

I am having a similar issue where if I close hyprland and restart it, there is an xdph process that is taking a large amount of cpu. Killing the process works fine and the issue goes away until I restart hyprland again.

[joelsleeba]

Yes. Same behavior. Once I kill one of those processes, the CPU usage is back to normal, but at the expense of xdg-..-hyprland. It happens almost every time I exit and relaunch Hyprland.

On Sun, 1 Oct 2023 at 09:36, darknebula05 @.***> wrote:

I am having a similar issue where if I close hyprland and restart it, there is an xdph process that is taking a large amount of cpu. Killing the process works fine and the issue goes away until I restart hyprland again.

— Reply to this email directly, view it on GitHub https://github.com/hyprwm/xdg-desktop-portal-hyprland/issues/103#issuecomment-1741949300, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANCQNW5IWQJYADKFK2BGOP3X5DT2TANCNFSM6AAAAAA5M5JRUM . You are receiving this because you authored the thread.Message ID: @.***>

[vaxerski]

patch.txt

Apply this patch, add -v to launching xdph and redirect its logs to somewhere (so e.g. -v > /tmp/xdphlog) and provide the log from an offending session, please :)

[joelsleeba]

[LOG] Initializing xdph...
[CRITICAL] Couldn't create the dbus connection ([org.freedesktop.DBus.Error.FileExists] Failed to request bus name (File exists))

This was the output. I don't know where to run xdph from. I added it to the hyprland config with exec-once

[vaxerski]

try the nuclear script from here

[joelsleeba]

I used the code snippet. But now I cannot seem to get a conflicting session and hence no cpu throttling. I wonder if it is because the script kills xdph beforehand.

[vaxerski]

possible

[houdinihacker]

Same for me (I'm using sway). Found this topic when my coolers went to the moon. I'll try to apply patch today and let you know

[vaxerski]

check with the above commit please :)

[pavanbhat1999]

Hope fix comes into release version soon.

I have the same issue in Nixos as well. Little bit modification in nuclear script is working fine for now.

[thejch]

Issue seems fixed on the latest commit

[vaxerski]

closing then

[bill88t]

Requesting reopen.

xdg-desktop-portal-hyprland 1.2.1-1 still exhibits this issue for me (arch, up-to-date). fine on boot -> logout -> login -> xdph takes up 20% cpu. systemctl --user restart fixes it.

I installed hyperland today, so I have no clue about previous versions.

[bill88t]

Nevermind, just tested it. 1.2.0 -> 1.2.1 reintroduces this issue.

[vaxerski]

I'll just make a nuclear solution tomorrow I guess, it's some damn thread in a deadlock most likely

[vaxerski]

done 667007fa4e370e7fbac5d9e6789b119f9456cf75