Loading i915 module panics my system after patching with Intel GPU patch 14.4 for stable/9

[Konstantin Belousov]

On Fri, Apr 27, 2012 at 10:25:49PM +0200, Jose Garcia Juanino wrote:
> El jueves 26 de abril a las 21:02:19 CEST, Konstantin Belousov escribi?:
> > On Thu, Apr 26, 2012 at 08:49:35PM +0200, Jose Garcia Juanino wrote:
> > > 
> > > It is possible to execute a "startx" from the blank screen, but the
> > > results are a very weird X session, with no functional mouse nor keyboard.
> > 
> > Does the right image appear ? What do you mean by 'weird X session' ?
> > 
> > Non-functioning kbd/mouse has nothing to do with i195.ko.
> 
> Sorry, yesterday I was not very smart.
> 
> Actually, I have xf86-video-intel-2.17.0 and xorg-server-1.10.6,1
> (rebuilt today from a updated ports tree, with WITH_NEW_XORG=yes and
> WITH_KMS=yes in /etc/make.conf). Here I show you the results:
> 
> 1- startx with vesa driver.
> Everything works smoothly. Keyboard and mouse are fully functional. I
> can use the browser, open terminals, etc.
> 
> 
> 2- startx with intel driver
> X session takes a lot of time until it appears (1 minute). From a ssh
> session, I see the Xorg process to consume the 100% of cpu time (kernel
> time, not user time). But once the screen is showed up, it looks fine.
> Fonts are well displayed and depth color is 24 (vesa only supports 8).
> After a few minutes, CPU time drops until 0%, and it is possible to use
> the computer. But if I try to move the mouse, again the cpu time
> increase until 100% for 1 minute, and the system is unresponsive. Only I
> can use the computer by mean of the keyboard after cpu consume drops
> again.
> 
> In the meantime, I can see the following in syslog when the cpu is 100%:
> 
> Apr 27 21:33:36 riemann kernel: [drm:KMS:pid0:output_poll_execute] [CONNECTOR:11:VGA-1] status updated from 2 to 2
> Apr 27 21:34:16 riemann last message repeated 4 times
> Apr 27 21:36:16 riemann last message repeated 12 times
> 
> 
> Well, that is everything I can contribute. I am unable to decide wether
> this "weird" behavior is Xorg related or i915 kernel module related, but
> the scenario final (i915 kernel with KSM support plus recent Xorg) is
> not functional nor usable.
> 
> Best regards, I appreciate your time and efforts, and sorry for my poor
> english.

Try to comment out the call to 	drm_kms_helper_poll_init(dev); at
line 1032 of i915_dma.c. It might be that ddc reading consumes lots of
CPU, lets see.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-x11/attachments/20120428/a1e310a9/attachment.pgp