Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 07:17
    AleksandrNY commented #144
  • 07:16
    AleksandrNY commented #144
  • 07:15
    AleksandrNY commented #144
  • Nov 24 21:47
    tuaris commented #144
  • Nov 24 13:15
    AleksandrNY commented #144
  • Nov 23 18:42

    zeising on master

    Normalize the Makefiles and do … (compare)

  • Nov 23 18:42
    zeising closed #14
  • Nov 23 11:10
    bzfbd commented #14
  • Nov 19 03:14
    shkhln commented #248
  • Nov 16 09:26

    hselasky on drm-v4.16-fbsd12.0

    Buildfix for 12-stable after r3… (compare)

  • Nov 09 10:31
    sssme closed #254
  • Nov 09 10:17
    ykla closed #253
  • Nov 09 09:21
    sssme commented #253
  • Nov 09 08:55
    zeising commented #253
  • Nov 09 08:45
    sssme commented #254
  • Nov 09 08:45
    sssme commented #254
  • Nov 09 08:45
    sssme reopened #254
  • Nov 09 08:45
    sssme closed #254
  • Nov 09 08:40
    sssme commented #253
  • Nov 09 08:30
    zeising commented #253
myfreeweb
@myfreeweb
hm same problems with these rolled back. could this be the gpu drivers' fault
myfreeweb
@myfreeweb
the nvme missed interrupt still sometimes happens :/
Jan Beich
@jbeich
@myfreeweb, r367935 works fine here. Maybe check systat -v 1 output. Non-instant hangs maybe not related to graphics e.g., network card.
Nevermind, I don't use 5.5-wip.
myfreeweb
@myfreeweb
yeah yeah looks like those are fine, used 5.4 for several hours today just fine again, looks like something in 5.5 caused this very strange breakage at a distance
either having the MSI disable function stubbed (but that's only supposed to happen on unload? OTOH related to interrupts and i've had nvme interrupt issues) or my weak symbol workaround (also idk how that could even). testing more now with fixes
fbsdgitter
@fbsdgitter
[irc] <grinzie> Has anyone tried the new river Wayland compositor?
[irc] <grinzie> It is the first Wayland compositor I have seen with a client-server model similar to herbstluftwm, allowing for command line control of the window manager.
fbsdgitter
@fbsdgitter
[irc] <debdrup> It's tiling, so maybe I'll try it once I switch to Wayland.
[irc] <debdrup> That'll also give it time to not be subject to breaking changes.
[irc] <grinzie> Yes, at the moment it seems not to play nicely with seatd.
[irc] <debdrup> I don't even know what that is. :)
[irc] <debdrup> I assume it's something related to the IRIX seat feature that eventually made it into Xorg?
[irc] <grinzie> It is a new seat manager for wlroots
Jan Beich
@jbeich
grinzie: can you elaborate on "not to play nicely with seatd"?
fbsdgitter
@fbsdgitter
[irc] <debdrup> Ah, so it is related to the IRIX idea of multi-seat configuration, gotcha.
Jan Beich
@jbeich
debdrup: seatd is a replacement for systemd-logind where seats are groups of input, see https://wayland-book.com/seat.html
fbsdgitter
@fbsdgitter
[irc] <grinzie> To run with seatd service I had to setuid. With seatd service, it launched in the controlling terminal (ttyv1) and I have to switch virtual terminals about every 15-30 seconds...
[irc] <grinzie> I mean witout service, setuid required
[irc] <debdrup> jbeich: Oh, interesting - thanks!
Jan Beich
@jbeich
grinzie: river doesn't support setuid as there's no code to drop permissions.
fbsdgitter
@fbsdgitter
[irc] <grinzie> hikari works fine with seatd, and I noticed the binary is setuid, so I tried it
[irc] <grinzie> Is there a way to control which terminal seatd service starts with?
Jan Beich
@jbeich
grinzie: seatd itself doesn't use any particular terminal but when a compositor starts it tries to silence duplicate input on console
fbsdgitter
@fbsdgitter
[irc] <grinzie> I was able to run river with setuid (with config.sh from dist), but trying to use keybindings to launch terminal causes crash and I lose tty control
[irc] <grinzie> I launched seatd service in terminal. Would autostarting it solve the tty problem?
[irc] <grinzie> No, because I still start river in terminal (for the moment)
Jan Beich
@jbeich
grinzie: launching terminal via keybinding works fine for me. Basically, "service seatd onestart; river -c /path/to/config.sh" then Win+Return
no need for setuid on "river"
fbsdgitter
@fbsdgitter
[irc] <grinzie> OK, nice to have confirmation it works, let me try again
[irc] <grinzie> Wait... The same thing worked for me, the problem for me in that scenario (not setuid on river, seatd service) is that the screen freezes periodically and I have to switch ttys back and forth to unfreeze it
Jan Beich
@jbeich
grinzie: instead of using contrib/config.sh better write config from scratch. Maybe add "riverctl spawn alacritty" as the first line, so you can kill even if no bindings work.
fbsdgitter
@fbsdgitter
[irc] <grinzie> How long did you let river run with seatd service? Did you launch other applications? Did the screen freeze?
[irc] <grinzie> I actually changed config.sh to use sakura
Jan Beich
@jbeich
grinzie: do you see the screen freezes with hikari when it doesn't use setuid when seatd is running?
fbsdgitter
@fbsdgitter
[irc] <grinzie> I totally understand and don't expect you to. Let me try that
Jan Beich
@jbeich
wlroots direct session creates a separate ttyv%d and switches to it in order to consumer undesired input on console. However, libseat uses the same ttyv%d where a compositor is started and partially disables input via K_RAW.
I don't like either behavior: wlroots direct session sometimes fail to switch back to a usable console when compositor is killed/exited while libseat way prints junk on a console that maybe under use.
Alas, FreeBSD doesn't support K_OFF.
fbsdgitter
@fbsdgitter
[irc] <grinzie> jbeich: without setuid on hikari binary, I get this error: 00:00:00.000 [ERROR] [libseat] [libseat/libseat.c:73] libseat_open_seat: No backend was able to open a seat 00:00:00.000 [ERROR] [backend/session/libseat.c:102] Unable to create seat: Function not implemented 00:00:00.002 [ERROR] [backend/session/direct-freebsd.c:191] Could not open /dev/ttyv0 to find a free vt: Permission denied 00:00:00.004 [ERROR] [backend/session/session.c:113] Failed to
Jan Beich
@jbeich
grinzie: did you start seatd before starting hikari without setuid?
fbsdgitter
@fbsdgitter
[irc] <grinzie> No...
[irc] <grinzie_> jbeich: hikari+seatd service is working, terminal is not freezing
fbsdgitter
@fbsdgitter
[irc] <grinzie_> wait...
fbsdgitter
@fbsdgitter
[irc] <grinzie> jbeich: hikari works as before, tty does not freeze
fbsdgitter
@fbsdgitter
[irc] <grinzie> I will keep playing with it
[irc] <grinzie> river is not freezing at the moment either...
[irc] <grinzie> :priv-t somafm.org
myfreeweb
@myfreeweb
@jbeich we really should implement K_OFF.. sucks that these modes are currently handled in like the individual keyboard drivers >_<
Jan Beich
@jbeich
Can someone test alacritty 0.6.0 to give mikael@ more confidence?