Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
Hi On Fri, Feb 27, 2015 at 4:09 AM, Zbigniew Jędrzejewski-Szmek wrote: > On Thu, Feb 26, 2015 at 01:37:25PM +0100, Jan Engelhardt wrote: >> >> On Wednesday 2015-02-25 21:10, Lennart Poettering wrote: >> >> >> The setup program would not set the font on tty16 upwards. >> >> >> There is a maximum of 63 VCs possible in Linux. (That number is >> >> >> hardcoded.) >> >> > >> >> >We deliberately do not support such high VTs in systemds. >> >> >> >> And what's the rationale? >> > >> >Initially it was that we had the "systemd" udev tag on all VCs, which >> >made them appear as .device units in systemd, and I really didn't want >> >63 of them. Then, some of the VC ioctls have a 16bit bitmap, which >> >makes the higher VCs not as accesible. >> >What's your usecase for this? >> >> No particular usecase. I just stumbled across this arbitrary number >> 15/16 in the source code while I was figuring out the font reload >> mess. The patches sent, >> vconsole: rerun setup on console takeovers as well >> vconsole: match on vtcon events, not fbcon ones, >> are still very much desired to have included. > One reason I see is automated tools which want to open many consoles > (for debug output or whatever), without stepping on the toes of the > user. They might start with tty20 or something. > > If the trick with checking for /dev/vcsXX works, I don't see why > we shouldn't support the full set provided by the kernel. One magic > number less is always good. It does not work. vcs are available if the VT is allocated, but v_state returns information whether someone actually has the VT _opened_. Thanks David ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Thu, Feb 26, 2015 at 01:37:25PM +0100, Jan Engelhardt wrote: > > On Wednesday 2015-02-25 21:10, Lennart Poettering wrote: > >> >> The setup program would not set the font on tty16 upwards. > >> >> There is a maximum of 63 VCs possible in Linux. (That number is > >> >> hardcoded.) > >> > > >> >We deliberately do not support such high VTs in systemds. > >> > >> And what's the rationale? > > > >Initially it was that we had the "systemd" udev tag on all VCs, which > >made them appear as .device units in systemd, and I really didn't want > >63 of them. Then, some of the VC ioctls have a 16bit bitmap, which > >makes the higher VCs not as accesible. > >What's your usecase for this? > > No particular usecase. I just stumbled across this arbitrary number > 15/16 in the source code while I was figuring out the font reload > mess. The patches sent, > vconsole: rerun setup on console takeovers as well > vconsole: match on vtcon events, not fbcon ones, > are still very much desired to have included. One reason I see is automated tools which want to open many consoles (for debug output or whatever), without stepping on the toes of the user. They might start with tty20 or something. If the trick with checking for /dev/vcsXX works, I don't see why we shouldn't support the full set provided by the kernel. One magic number less is always good. Zbyszek ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Wednesday 2015-02-25 21:10, Lennart Poettering wrote: >> >> The setup program would not set the font on tty16 upwards. >> >> There is a maximum of 63 VCs possible in Linux. (That number is >> >> hardcoded.) >> > >> >We deliberately do not support such high VTs in systemds. >> >> And what's the rationale? > >Initially it was that we had the "systemd" udev tag on all VCs, which >made them appear as .device units in systemd, and I really didn't want >63 of them. Then, some of the VC ioctls have a 16bit bitmap, which >makes the higher VCs not as accesible. >What's your usecase for this? No particular usecase. I just stumbled across this arbitrary number 15/16 in the source code while I was figuring out the font reload mess. The patches sent, vconsole: rerun setup on console takeovers as well vconsole: match on vtcon events, not fbcon ones, are still very much desired to have included. AFAICS, there is just _one_ ioctl with a bitmap, and that is VT_GETSTATE's v_state member. systemd-vconsole-setup uses VT_GETSTATE, but never inspect v_state. logind-core.c does - which raises the question why, when vconsole-setup can do without it. ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Tue, 24.02.15 19:53, Jan Engelhardt (jeng...@inai.de) wrote: > On Tuesday 2015-02-24 19:47, Lennart Poettering wrote: > > >On Tue, 24.02.15 17:49, Jan Engelhardt (jeng...@inai.de) wrote: > > > >> The setup program would not set the font on tty16 upwards. > >> There is a maximum of 63 VCs possible in Linux. (That number is > >> hardcoded.) > > > >We deliberately do not support such high VTs in systemds. > > And what's the rationale? Initially it was that we had the "systemd" udev tag on all VCs, which made them appear as .device units in systemd, and I really didn't want 63 of them. Then, some of the VC ioctls have a 16bit bitmap, which makes the higher VCs not as accesible. Mostly it's now though that the concept of having that many VCs is kinda crazy... What's your usecase for this? Lennart -- Lennart Poettering, Red Hat ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Tuesday 2015-02-24 19:53, Jan Engelhardt wrote: >On Tuesday 2015-02-24 19:47, Lennart Poettering wrote: > >>On Tue, 24.02.15 17:49, Jan Engelhardt (jeng...@inai.de) wrote: >> >>> The setup program would not set the font on tty16 upwards. >>> There is a maximum of 63 VCs possible in Linux. (That number is >>> hardcoded.) >> >>We deliberately do not support such high VTs in systemds. > >And what's the rationale? Right, VT_GETSTATE gives a bitmask which only contains 16 bits, that's the "vcs.v_active" just a few lines above the change that I did apparently. However, it also seems that we do not need to rely on v_state. /dev/vcsN is present if it is active, and is absent when deallocated -- at least when /dev is dynamically managed, which I figure might be a requirement for systemd. [suggestive patch in reply hereto] ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Tuesday 2015-02-24 19:47, Lennart Poettering wrote: >On Tue, 24.02.15 17:49, Jan Engelhardt (jeng...@inai.de) wrote: > >> The setup program would not set the font on tty16 upwards. >> There is a maximum of 63 VCs possible in Linux. (That number is >> hardcoded.) > >We deliberately do not support such high VTs in systemds. And what's the rationale? ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [PATCH 2/2] vconsole: set font on tty16..tty63 as well
On Tue, 24.02.15 17:49, Jan Engelhardt (jeng...@inai.de) wrote: > The setup program would not set the font on tty16 upwards. > There is a maximum of 63 VCs possible in Linux. (That number is > hardcoded.) We deliberately do not support such high VTs in systemds. If you want such high VTs, then please patch this downstream. Lennart -- Lennart Poettering, Red Hat ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel