Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 15 Oct 2010 11:38:15 +0200
From:      Daniel Bilik <daniel.bilik@neosystem.cz>
To:        freebsd-x11@freebsd.org
Subject:   Fw: Deignoring ports/x11-drivers/xf86-video-intel29
Message-ID:  <20101015113815.b9570be4.daniel.bilik@neosystem.cz>

next in thread | raw e-mail | index | archive | help
Hi.

Ping timeout from miwi@, so I'm forwarding here. Should there be any
disscussion regarding this, please keep me in Cc as I'm not subscribed to
the list. Thanks.

Begin forwarded message:

Date: Sat, 2 Oct 2010 01:56:01 +0200
From: Daniel Bilik <daniel.bilik@neosystem.cz>
To: miwi@freebsd.org
Subject: Deignoring ports/x11-drivers/xf86-video-intel29


Hi.

You seem to be the last one who made changes to
x11-drivers/xf86-video-intel29, so I'm writing to you. If you are not the
right person or you are short of time, you may want to forward to someone
else (or maybe to freebsd-x11@, but I'm not subscribed to that list). And
sorry to bother if this was already disscussed and is being worked on.
Anyway this is the story...

In recent days I've been struggling with miserable state of new Intel
graphics on FreeBSD (Pineview in my case). Using our "stable" (or better
stalled) version of the driver makes Xorg to hang during startup. With
(deIGNOREd) intel29, the Xorg works (without DRI), but it consumes
unreasonable amount of CPU... On switching to another workspace there is
several-seconds hang while Xorg re-draws app's windows there. It makes the
environment almost unusable if you are switching apps often. Sad and
annoying at the same time. :-/

After almost whole day of torturing Google, I've hit some notes about
so-called "legacy driver". It turned out that at the end of August, Chris
Wilson of Intel released his fork of the driver, based on 2.9 but GEM-free
and UMS. See this post:

http://www.mail-archive.com/intel-gfx@lists.freedesktop.org/msg01309.html

I've downloaded the code from here...

http://cgit.freedesktop.org/~ickle/xf86-video-intel/commit/?h=legacy

... and compiled it in place of x11-driver/xf86-video-intel29. XvMC is
disabled in this legacy driver but DRI seems to be working... somewhat.
It crashes Xorg after a while, but it may be due to my unusual setup
(large --fb with --panning). I don't need DRI, so I just disabled it, and
I've been happily using this legacy thing for several days now.

Maybe it's worth replacing xf86-video-intel29 with this version of the
driver and de-ignoring it. It won't break anything for users of
xf86-video-intel while giving a functional alternative to those of us
running newer Intel graphics.

Hope this helps (you or anybody else).

--
						Daniel Bilik
						neosystem.cz



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20101015113815.b9570be4.daniel.bilik>