Github messages for voidlinux
 help / color / mirror / Atom feed
* [ISSUE] Restructure dependencies on elogind 
@ 2020-04-12 19:20 ericonr
  2020-04-12 20:18 ` st3r4g
                   ` (36 more replies)
  0 siblings, 37 replies; 38+ messages in thread
From: ericonr @ 2020-04-12 19:20 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1177 bytes --]

New issue by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930

Description:
As was agreed when writing the Wayland docs page, elogind is a requirement for a safe wayland session. Therefore, should we make elogind a dependency of the wayland package itself?

Packages like Sway currently have it as a build option, defaulting to a setuid binary instead. Cage, on the other hand, doesn't enable setuid, so it definitely requires elogind (at least in its current state). However, Cage currently doesn't list elogind as a dependency, so if you try to simply install it and run it (without looking at the docs), it will error out due to missing capabilities. Even running as root doesn't work, because then it can't drop privileges properly.

Making elogind a dependency for wayland seems like it would be the best solution for all cases. I would even make a case for an INSTALL message telling the user to enable the service and restart their device (or just log in and out, idk. Whenever I enable elogind I have to reboot for it to work).

Tagging @Duncaen, because you made the case for elogind being a requirement for Wayland.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
@ 2020-04-12 20:18 ` st3r4g
  2020-04-12 20:20 ` st3r4g
                   ` (35 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 20:18 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 432 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612670458

Comment:
It is indeed recommended in https://github.com/swaywm/sway/wiki/Running-Sway-without-systemd
I wouldn't add it to the `wayland` package, but to each compositor individually, because libwayland only requires `XDG_RUNTIME_DIR` to be set (both client and server need to know the socket path).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
  2020-04-12 20:18 ` st3r4g
@ 2020-04-12 20:20 ` st3r4g
  2020-04-12 20:30 ` st3r4g
                   ` (34 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 20:20 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 453 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612670458

Comment:
It is indeed recommended in https://github.com/swaywm/sway/wiki/Running-Sway-without-systemd
I wouldn't add it to the `wayland` package, but to each compositor individually, because `libwayland` only requires `XDG_RUNTIME_DIR` to be set (socket path must be in a standard location agreed by client and server).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
  2020-04-12 20:18 ` st3r4g
  2020-04-12 20:20 ` st3r4g
@ 2020-04-12 20:30 ` st3r4g
  2020-04-12 21:25 ` ericonr
                   ` (33 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 20:30 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 490 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612670458

Comment:
It is indeed recommended in https://github.com/swaywm/sway/wiki/Running-Sway-without-systemd
I wouldn't add it to the `wayland` package, but to each compositor individually, because `libwayland` only requires `XDG_RUNTIME_DIR` to be set (socket path must be in a standard location agreed by client and server in order to establish the connection).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (2 preceding siblings ...)
  2020-04-12 20:30 ` st3r4g
@ 2020-04-12 21:25 ` ericonr
  2020-04-12 21:56 ` travankor
                   ` (32 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-12 21:25 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 748 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612678163

Comment:
What other way of setting up `XDG_RUNTIME_DIR` would you suggest? I feel like adding elogind to all the compositors instead of to their common library (which is the part that actually reqiuires `XDG_RUNTIME_DIR`) is duplicated effort. In this case whoever packages another wayland compositor will have to know to add it as a dependency. Just now I realized we'd also have to update Wayfire.

And Wayland also requires privileges for DRM master. Is there any compositor that somehow bypasses this without setuid?If there isn't, elogind is necessary for wayland to be useful, so it should be a dependency.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (3 preceding siblings ...)
  2020-04-12 21:25 ` ericonr
@ 2020-04-12 21:56 ` travankor
  2020-04-12 21:57 ` st3r4g
                   ` (31 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-12 21:56 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612681958

Comment:
FWIW, the creator of sway does not use elogind (or systemd).

That said, wlroots-based compositors should have consistent behaviour wrt elogind or setuid.

>Cage, on the other hand, doesn't enable setuid, so it definitely requires elogind (at least in its current state).

I think setuid was only recently added to cage.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (4 preceding siblings ...)
  2020-04-12 21:56 ` travankor
@ 2020-04-12 21:57 ` st3r4g
  2020-04-12 21:58 ` st3r4g
                   ` (30 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 21:57 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1256 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612682196

Comment:
> What other way of setting up `XDG_RUNTIME_DIR` would you suggest?
Well, if you want to follow the XDG spec pedantically there is no simple manual way that I know, because iirc it requires login aware stuff like creating the folder at first user login and deleting it after the last log out. But I don't think this actually matters so much for `libwayland`, it's just a way to have a standard path where to put/discover the socket. It's not even a strict `libwayland` requirement, it's just the only convention used in practice. Maybe someone in the future writes a stripped down elogind-like program that only sets `XDG_RUNTIME_DIR`, idk...
> And Wayland also requires privileges for DRM master
You can actually become DRM master without root privileges, if no one is already master (i.e. you launch from console). But then you can't drop master (don't ask me why... DRM API is like this). So technically it's possible to write such Wayland compositor, but it can't do VT switching.

I don't think adding it to each compositor is a lot of duplication, but anyway just my 2 cents let's see other opinions



^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (5 preceding siblings ...)
  2020-04-12 21:57 ` st3r4g
@ 2020-04-12 21:58 ` st3r4g
  2020-04-12 22:08 ` st3r4g
                   ` (29 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 21:58 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1260 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612682196

Comment:
> What other way of setting up `XDG_RUNTIME_DIR` would you suggest?

Well, if you want to follow the XDG spec pedantically there is no simple manual way that I know, because iirc it requires login aware stuff like creating the folder at first user login and deleting it after the last log out. But I don't think this actually matters so much for `libwayland`, it's just a way to have a standard path where to put/discover the socket. It's not even a strict `libwayland` requirement, it's just the only convention used in practice. Maybe someone in the future writes a stripped down elogind-like program that only sets `XDG_RUNTIME_DIR`, idk...
> And Wayland also requires privileges for DRM master

You can actually become DRM master without root privileges, if no one is already master (i.e. you launch from console). But then you can't drop master (don't ask me why... DRM API is like this). So technically it's possible to write such Wayland compositor, but it can't do VT switching.

I don't think adding it to each compositor is a lot of duplication, but anyway just my 2 cents let's see other opinions



^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (6 preceding siblings ...)
  2020-04-12 21:58 ` st3r4g
@ 2020-04-12 22:08 ` st3r4g
  2020-04-12 22:09 ` st3r4g
                   ` (28 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 22:08 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1319 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612682196

Comment:
> What other way of setting up `XDG_RUNTIME_DIR` would you suggest?

Well, if you want to follow the XDG spec pedantically there is no simple manual way that I know, because iirc it requires login aware stuff like creating the folder at first user login and deleting it after the last log out. But I don't think this actually matters so much for `libwayland`, it's just a way to have a standard path where to put/discover the socket. It's not even a strict `libwayland` requirement, it's just the only convention used in practice. Maybe someone in the future writes a stripped down elogind-like program that only sets `XDG_RUNTIME_DIR`, idk...
> And Wayland also requires privileges for DRM master

You can actually become DRM master without root privileges, if no one is already master (i.e. you launch from console). But then you can't drop master (don't ask me why... DRM API is like this). So technically it's possible to write such Wayland compositor, but it can't do VT switching.

I don't think adding it to each compositor is a lot of duplication (e.g. add it to `wlroots` to cover all wlroots compositor), but anyway just my 2 cents let's see other opinions



^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (7 preceding siblings ...)
  2020-04-12 22:08 ` st3r4g
@ 2020-04-12 22:09 ` st3r4g
  2020-04-12 22:30 ` st3r4g
                   ` (27 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 22:09 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1320 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612682196

Comment:
> What other way of setting up `XDG_RUNTIME_DIR` would you suggest?

Well, if you want to follow the XDG spec pedantically there is no simple manual way that I know, because iirc it requires login aware stuff like creating the folder at first user login and deleting it after the last log out. But I don't think this actually matters so much for `libwayland`, it's just a way to have a standard path where to put/discover the socket. It's not even a strict `libwayland` requirement, it's just the only convention used in practice. Maybe someone in the future writes a stripped down elogind-like program that only sets `XDG_RUNTIME_DIR`, idk...
> And Wayland also requires privileges for DRM master

You can actually become DRM master without root privileges, if no one is already master (i.e. you launch from console). But then you can't drop master (don't ask me why... DRM API is like this). So technically it's possible to write such Wayland compositor, but it can't do VT switching.

I don't think adding it to each compositor is a lot of duplication (e.g. add it to `wlroots` to cover all wlroots compositors), but anyway just my 2 cents let's see other opinions



^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (8 preceding siblings ...)
  2020-04-12 22:09 ` st3r4g
@ 2020-04-12 22:30 ` st3r4g
  2020-04-12 23:28 ` ericonr
                   ` (26 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-12 22:30 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1478 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612682196

Comment:
> What other way of setting up `XDG_RUNTIME_DIR` would you suggest?

Well, if you want to follow the XDG spec pedantically there is no simple manual way that I know, because iirc it requires login aware stuff like creating the folder at first user login and deleting it after the last log out. But I don't think this actually matters so much for `libwayland`, it's just a way to have a standard path where to put/discover the socket. It's not even a strict `libwayland` requirement, it's just the only convention used in practice. Maybe someone in the future writes a stripped down elogind-like program that only sets `XDG_RUNTIME_DIR`, idk...
> And Wayland also requires privileges for DRM master

You can actually become DRM master without root privileges, if no one is already master (i.e. you launch from console). But then you can't drop master (don't ask me why... DRM API is like this). So technically it's possible to write such Wayland compositor, but it can't do VT switching.
EDIT: actually not 100% sure about this, maybe it's possible to close and reopen the gpu file descriptor and recover the client buffers, I have to try it...

I don't think adding it to each compositor is a lot of duplication (e.g. add it to `wlroots` to cover all wlroots compositors), but anyway just my 2 cents let's see other opinions



^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (9 preceding siblings ...)
  2020-04-12 22:30 ` st3r4g
@ 2020-04-12 23:28 ` ericonr
  2020-04-15 19:33 ` st3r4g
                   ` (25 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-12 23:28 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 580 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-612691870

Comment:
@travankor at least for the sway package, setting setuid for it is determined by the template. Do you mean that Cage has only recently gained the capability of being setuid? Because the package definitely doesn't set it.

@st3r4g I actually really like the idea of adding elogind as a dependency of wlroots. GNOME and KDE already require elogind, so the only compositor that wouldn't require it is weston, AFAIK. Have you ever used it?

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (10 preceding siblings ...)
  2020-04-12 23:28 ` ericonr
@ 2020-04-15 19:33 ` st3r4g
  2020-04-15 19:34 ` st3r4g
                   ` (24 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-15 19:33 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 334 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614238386

Comment:
`weston` on void uses `weston-launch` by default, so a setuid binary. So again the user must set up `XDG_RUNTIME_DIR` in some way. `weston` is meant for testing and not for regular use though.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (11 preceding siblings ...)
  2020-04-15 19:33 ` st3r4g
@ 2020-04-15 19:34 ` st3r4g
  2020-04-16 12:22 ` travankor
                   ` (23 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-15 19:34 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 338 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614238386

Comment:
`weston` on void atm uses `weston-launch` by default, so a setuid binary. So again the user must set up `XDG_RUNTIME_DIR` in some way. `weston` is meant for testing and not for regular use though.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (12 preceding siblings ...)
  2020-04-15 19:34 ` st3r4g
@ 2020-04-16 12:22 ` travankor
  2020-04-16 12:41 ` xtraeme
                   ` (22 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-16 12:22 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 887 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614618827

Comment:
To be more explicit than my previous response, I don't think `wlroots`, a library, should have a runtime dependency on elogind. Implement it on a per-template basis; there only are a handful of wlroots compositors and you can use `xbps-query -RX wlroots` to find all of them.

Furthermore, even though I am currently using elogind, I don't think it's a good solution for several reasons. setuid is not a problem if privs are dropped correctly. The ideal solution would be a much, more lightweight session manager written in a safe language. So, I think an alternative to elogind should be supported in an official capacity by `xbps-src`.

`XDG_RUNTIME_DIR` can be setup like any other environment variable so I don't understand the issue.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (13 preceding siblings ...)
  2020-04-16 12:22 ` travankor
@ 2020-04-16 12:41 ` xtraeme
  2020-04-16 13:17 ` travankor
                   ` (21 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: xtraeme @ 2020-04-16 12:41 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 263 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614627965

Comment:
@travankor XDG_RUNTIME_DIR is not just a simple location... elogind does more than just setting this var. Read about it!

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (14 preceding siblings ...)
  2020-04-16 12:41 ` xtraeme
@ 2020-04-16 13:17 ` travankor
  2020-04-16 13:41 ` Duncaen
                   ` (20 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-16 13:17 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 246 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614646739

Comment:
You can set `XDG_RUNTIME_DIR=/run/user/$USER` in shell and it works. You don't need elogind for this.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (15 preceding siblings ...)
  2020-04-16 13:17 ` travankor
@ 2020-04-16 13:41 ` Duncaen
  2020-04-16 13:41 ` st3r4g
                   ` (19 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: Duncaen @ 2020-04-16 13:41 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 224 bytes --]

New comment by Duncaen on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614660047

Comment:
Its not about setting the environment variable, its about creating the directory.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (16 preceding siblings ...)
  2020-04-16 13:41 ` Duncaen
@ 2020-04-16 13:41 ` st3r4g
  2020-04-18  6:47 ` ericonr
                   ` (18 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-04-16 13:41 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 458 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-614660202

Comment:
As I said before, basically the point is that the [XDG spec](https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html) sets specific requirements tied to logins for this folder... but for how it is used by Wayland compositors it doesn't really matter, it just has to exist (with the right perms).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (17 preceding siblings ...)
  2020-04-16 13:41 ` st3r4g
@ 2020-04-18  6:47 ` ericonr
  2020-04-18  7:49 ` travankor
                   ` (17 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-18  6:47 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 822 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-615619830

Comment:
I think discussing this in matters of `XDG_RUNTIME_DIR` won't necessarily lead us anywhere. The most important point for this is the existence or not of setuid binaries.

In order to not break users' setups, elogind could be a (direct or not) dependency of Sway, and then setuid could be removed from Sway's build script (or made a build option at least). The way it currently works, `cage` isn't setuid and requires elogind to run, and `wayfire` even includes an [`INSTALL.msg`](https://github.com/void-linux/void-packages/blob/master/srcpkgs/wayfire/INSTALL.msg) telling the user to install elogind. It doesn't make sense that these packages don't directly depend on elogind.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (18 preceding siblings ...)
  2020-04-18  6:47 ` ericonr
@ 2020-04-18  7:49 ` travankor
  2020-04-18  7:50 ` travankor
                   ` (16 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-18  7:49 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 445 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-615701666

Comment:
I don't think there is a problem with removing the default setuid.

I made some notes on how to run `wlroots` compsitors without elogind here https://github.com/void-linux/void-docs/pull/195.

I guess people that don't want elogind would have to use
`ignorepkg=elogind` and setuid manually then?

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (19 preceding siblings ...)
  2020-04-18  7:49 ` travankor
@ 2020-04-18  7:50 ` travankor
  2020-04-18  7:55 ` travankor
                   ` (15 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-18  7:50 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 463 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-615701666

Comment:
I don't think there is a problem with removing the default setuid.

I made some notes on how to run `wlroots` compsitors without elogind here https://github.com/void-linux/void-docs/pull/195.

I guess people that don't want elogind would have to use
`ignorepkg=elogind` and setuid manually then (or build option)?

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (20 preceding siblings ...)
  2020-04-18  7:50 ` travankor
@ 2020-04-18  7:55 ` travankor
  2020-04-18 16:20 ` ericonr
                   ` (14 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-18  7:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 287 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-615710361

Comment:
Also does the `elogind` service actually work without dbus? My understanding is that elogind is dbus-activated so what's the service file for?

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (21 preceding siblings ...)
  2020-04-18  7:55 ` travankor
@ 2020-04-18 16:20 ` ericonr
  2020-04-21  7:44 ` ericonr
                   ` (13 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-18 16:20 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1087 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-615897918

Comment:
I saw your PR! The one issue I have with it in this context is that if you don't have elogind nor setuid on the binary nor CAP_SYS_ADMIN (capabilities require one more dependency, though), then you can't start the compositor, unless you add setuid by hand.

Making setuid a build option would work out well, I think. Depend on elogind or make the binary setuid. Given that to use a build option you have to build the package yourself, `ignorepkg` shouldn't even be necessary.

I will test elogind without dbus and get back to you.

Finally, I don't know if as a distro we want to push elogind as the solution for most problems. If you guys feel it shouldn't be, I'd advocate then for including the build_option snippet in all the compositors, so they become setuid binaries. I will take a look at PopCorn.voidlinux.org to see if I can figure out how many people use any of the wlroots based compositors and if they have elogind there too.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (22 preceding siblings ...)
  2020-04-18 16:20 ` ericonr
@ 2020-04-21  7:44 ` ericonr
  2020-04-22  2:45 ` travankor
                   ` (12 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-21  7:44 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 438 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-617012385

Comment:
@travankor it appears elogind usually can be started from the dbus without issue. The reason for enabling its service is that there can be issues derived from it not running when the system was started. It's basically "if you experience an issue with elogind, try to enable it at startup first".

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (23 preceding siblings ...)
  2020-04-21  7:44 ` ericonr
@ 2020-04-22  2:45 ` travankor
  2020-04-22 14:57 ` travankor
                   ` (11 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-22  2:45 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 429 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-617515709

Comment:
@ericonr I created a PR to hopefully resolve this issue. The consus (not sure about this honestly?) seems like they want elogind, so I switched that on by default and made everything available from a build style.

I will update void-docs based on whatever is agreed upon and merged.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (24 preceding siblings ...)
  2020-04-22  2:45 ` travankor
@ 2020-04-22 14:57 ` travankor
  2020-04-22 17:22 ` ericonr
                   ` (10 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-04-22 14:57 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 603 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-617831371

Comment:
>it appears elogind usually can be started from the dbus without issue. The reason for enabling its service is that there can be issues derived from it not running when the system was started. It's basically "if you experience an issue with elogind, try to enable it at startup first".

So, do you still need dbus?

This should be documented preferably. Also issues like https://github.com/void-linux/void-packages/issues/20780 should be documented, too.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (25 preceding siblings ...)
  2020-04-22 14:57 ` travankor
@ 2020-04-22 17:22 ` ericonr
  2020-04-22 17:26 ` ericonr
                   ` (9 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-22 17:22 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 299 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-617917163

Comment:
@travankor from what I understand, you always need dbus. Enabling the elogind service is optional, but it's the best route to be sure stuff works correctly.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (26 preceding siblings ...)
  2020-04-22 17:22 ` ericonr
@ 2020-04-22 17:26 ` ericonr
  2020-05-30 11:51 ` travankor
                   ` (8 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-04-22 17:26 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 426 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-617917163

Comment:
@travankor from what I understand, you always need dbus. Enabling the elogind service is optional, but it's the best route to be sure stuff works correctly.

Regarding documentation, I missed the `dbus` part. I mentioned this in a docs issue, I will get to it when I have the time.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (27 preceding siblings ...)
  2020-04-22 17:26 ` ericonr
@ 2020-05-30 11:51 ` travankor
  2020-05-30 12:52 ` st3r4g
                   ` (7 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-05-30 11:51 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 266 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-636320249

Comment:
Discussion in [wlroots](https://github.com/swaywm/wlroots/issues/2232) for removing some setuid parts (potentially all?).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (28 preceding siblings ...)
  2020-05-30 11:51 ` travankor
@ 2020-05-30 12:52 ` st3r4g
  2020-08-14  2:38 ` travankor
                   ` (6 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-05-30 12:52 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 234 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-636326897

Comment:
Interesting. I don't get why they're saying that CAP_SYS_ADMIN is needed for handling input.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (29 preceding siblings ...)
  2020-05-30 12:52 ` st3r4g
@ 2020-08-14  2:38 ` travankor
  2020-12-03  9:52 ` SevereOverfl0w
                   ` (5 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: travankor @ 2020-08-14  2:38 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 251 bytes --]

New comment by travankor on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-673851224

Comment:
https://git.sr.ht/~kennylevinsen/seatd

This was mentioned in IRC, but just posting here for visibility.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (30 preceding siblings ...)
  2020-08-14  2:38 ` travankor
@ 2020-12-03  9:52 ` SevereOverfl0w
  2020-12-03 11:51 ` st3r4g
                   ` (4 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: SevereOverfl0w @ 2020-12-03  9:52 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 356 bytes --]

New comment by SevereOverfl0w on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-737804558

Comment:
There's also rundird being worked on which does roughly the same thing as pam_rundir.  I think a package will be incoming for that soon enough too.  Along with seatd, it's plenty to run wayland compositors.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (31 preceding siblings ...)
  2020-12-03  9:52 ` SevereOverfl0w
@ 2020-12-03 11:51 ` st3r4g
  2020-12-03 20:49 ` ericonr
                   ` (3 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: st3r4g @ 2020-12-03 11:51 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 312 bytes --]

New comment by st3r4g on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-737894093

Comment:
Now that `wlroots` uses `libseat`, it doesn't make sense to force an elogind dependency anymore. It is up to the user to choose the backend he prefers (elogind or seatd).

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (32 preceding siblings ...)
  2020-12-03 11:51 ` st3r4g
@ 2020-12-03 20:49 ` ericonr
  2020-12-19  3:30 ` ericonr
                   ` (2 subsequent siblings)
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-12-03 20:49 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 232 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-738298592

Comment:
I will leave this open to remind myself to check if the docs have all the necessary info.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (33 preceding siblings ...)
  2020-12-03 20:49 ` ericonr
@ 2020-12-19  3:30 ` ericonr
  2020-12-19  3:30 ` [ISSUE] [CLOSED] " ericonr
  2020-12-19  3:31 ` ericonr
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-12-19  3:30 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 236 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-748412605

Comment:
I'd appreaciate input in https://github.com/void-linux/void-docs/pull/534 . Closing this now.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: [ISSUE] [CLOSED] Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (34 preceding siblings ...)
  2020-12-19  3:30 ` ericonr
@ 2020-12-19  3:30 ` ericonr
  2020-12-19  3:31 ` ericonr
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-12-19  3:30 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1180 bytes --]

Closed issue by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930

Description:
As was agreed when writing the Wayland docs page, elogind is a requirement for a safe wayland session. Therefore, should we make elogind a dependency of the wayland package itself?

Packages like Sway currently have it as a build option, defaulting to a setuid binary instead. Cage, on the other hand, doesn't enable setuid, so it definitely requires elogind (at least in its current state). However, Cage currently doesn't list elogind as a dependency, so if you try to simply install it and run it (without looking at the docs), it will error out due to missing capabilities. Even running as root doesn't work, because then it can't drop privileges properly.

Making elogind a dependency for wayland seems like it would be the best solution for all cases. I would even make a case for an INSTALL message telling the user to enable the service and restart their device (or just log in and out, idk. Whenever I enable elogind I have to reboot for it to work).

Tagging @Duncaen, because you made the case for elogind being a requirement for Wayland.

^ permalink raw reply	[flat|nested] 38+ messages in thread

* Re: Restructure dependencies on elogind 
  2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
                   ` (35 preceding siblings ...)
  2020-12-19  3:30 ` [ISSUE] [CLOSED] " ericonr
@ 2020-12-19  3:31 ` ericonr
  36 siblings, 0 replies; 38+ messages in thread
From: ericonr @ 2020-12-19  3:31 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 235 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/20930#issuecomment-748412605

Comment:
I'd appreciate input in https://github.com/void-linux/void-docs/pull/534 . Closing this now.

^ permalink raw reply	[flat|nested] 38+ messages in thread

end of thread, other threads:[~2020-12-19  3:31 UTC | newest]

Thread overview: 38+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-04-12 19:20 [ISSUE] Restructure dependencies on elogind ericonr
2020-04-12 20:18 ` st3r4g
2020-04-12 20:20 ` st3r4g
2020-04-12 20:30 ` st3r4g
2020-04-12 21:25 ` ericonr
2020-04-12 21:56 ` travankor
2020-04-12 21:57 ` st3r4g
2020-04-12 21:58 ` st3r4g
2020-04-12 22:08 ` st3r4g
2020-04-12 22:09 ` st3r4g
2020-04-12 22:30 ` st3r4g
2020-04-12 23:28 ` ericonr
2020-04-15 19:33 ` st3r4g
2020-04-15 19:34 ` st3r4g
2020-04-16 12:22 ` travankor
2020-04-16 12:41 ` xtraeme
2020-04-16 13:17 ` travankor
2020-04-16 13:41 ` Duncaen
2020-04-16 13:41 ` st3r4g
2020-04-18  6:47 ` ericonr
2020-04-18  7:49 ` travankor
2020-04-18  7:50 ` travankor
2020-04-18  7:55 ` travankor
2020-04-18 16:20 ` ericonr
2020-04-21  7:44 ` ericonr
2020-04-22  2:45 ` travankor
2020-04-22 14:57 ` travankor
2020-04-22 17:22 ` ericonr
2020-04-22 17:26 ` ericonr
2020-05-30 11:51 ` travankor
2020-05-30 12:52 ` st3r4g
2020-08-14  2:38 ` travankor
2020-12-03  9:52 ` SevereOverfl0w
2020-12-03 11:51 ` st3r4g
2020-12-03 20:49 ` ericonr
2020-12-19  3:30 ` ericonr
2020-12-19  3:30 ` [ISSUE] [CLOSED] " ericonr
2020-12-19  3:31 ` ericonr

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).