Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 22 Jan 2018 14:56:43 +0000
From:      bugzilla-noreply@freebsd.org
To:        x11@FreeBSD.org
Subject:   [Bug 222905] x11/libinput: Update to 1.9.1
Message-ID:  <bug-222905-7141-6UAOnufWmc@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-222905-7141@https.bugs.freebsd.org/bugzilla/>
References:  <bug-222905-7141@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D222905

Greg V <greg@unrelenting.technology> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #187854|0                           |1
        is obsolete|                            |

--- Comment #6 from Greg V <greg@unrelenting.technology> ---
Created attachment 189969
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=3D189969&action=
=3Dedit
libinput-1.9.901.patch

Update to 1.9.901 + add a new 'reopen' patch=E2=80=A6

So, here's the story about that patch. I'm working on Wayland compositor st=
uff
(https://github.com/myfreeweb/weston-rs + https://github.com/myfreeweb/logi=
nw).
I've been trying to figure out why in the heck does *only my compositor* (n=
ot
Weston proper) reject input devices *only on my laptop* (not on my desktop)=
=E2=80=A6

/dev/input devices were readable and writable for group video. My setuid
wrapper (loginw) was only setting the user when dropping privileges for the
compositor (i.e. wasn't setting groups). I was testing unprivileged access,=
 but
I thought I was still testing privileged access.

Anyway, it is very desirable to have the compositor unprivileged. The setuid
wrapper passes it file descriptors to input devices, it does not need to op=
en
them directly.

libudev-devd does open them =E2=80=94 to enumerate devices. When it is unpr=
ivileged, it
does not set the right properties, which leads to rejection (the "not tagge=
d as
supported input device" message).

libinput opens the device via the compositor-provided mechanism (fd passing)
only to compare to the device it got from libudev(-devd) enumeration. They =
have
the same path, but the libudev one lacks properties. The correct one gets
thrown away.

This patch uses the correct one instead of throwing it away :) which allows
compositors to be unprivileged.

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-222905-7141-6UAOnufWmc>