DRI problems with ati/radeon on stable/7 r203425

[Robert Noland]

On Mon, 2010-02-08 at 09:26 -0800, David Wolfskill wrote:
> I had written to ports@ some months ago; at the time, my laptop was
> configured to boot stable/6 from slice 1, stable/7 from slice 2,
> stable/8 from slice 3, and head from slice 4; all ports (save compat6x &
> compat7x) were built under stable/6.
> 
> The reason for the message at the time was that I had found that
> sometimes -- though not always -- the laptop would seem to lock up
> instead of starting xdm (as it is configured to do).  I found that while
> I could reliably use DRI while running X under stable/6, the only way I
> could find to run X built under stable/6 reliably under the other OSen
> was to disable DRI in xorg.conf.
> 
> Given the recent activity in ports (with the upgrade to jpeg & every
> port that depends on it getting a bump over the weekend), I decided
> to "bite the bullet" and install all 800+ ports anew, building
> everything under stable/7, and bidding a rather fond farewall to
> stable/6.
> 
> By yesterday afternoon, I had finally managed to get enough of the
> ports built to have a usable X environment, which worked OK --
> though I noticed that I still had DRI disabled in xorg.conf (as an
> artifact of having last booted the laptop when the installed X.org
> had been built under stable/6).  So in prepartation for the next
> boot, I re-enabled DRI in xorg.conf.
> 
> Well, the time for the next boot has come, and from about 10 minutes
> before I started typing this note until just now, xdm would not start.
> Nor could I use ssh(1) to login to the machine, and attempting to login
> via serial got very sluggish response, to the point that it took a
> couple of minutes to echo the login I typed, and after I hit Enter, I
> got "Login timed out after 300 seconds" before the Password: prompt
> appeared.
> 
> And the xdm login banner finally showed up; I'd estimate that was around
> 20 minutes of idle time before that happened.
> 
> The ports tree is almost as up-to-date as my local SVN mirror, which was
> at r203541 as of Sat Feb  6 03:30:08 PST 2010.  (Local CVS mirror had
> been updated from cvsup4 at around Sat Feb  6 04:46:28 PST 2010.)
> 
> I had thought that the apparent "lock-ups" were because of an
> unaddressed issue in compat6x; I believe the recent evidence indicates
> otherwise.
> 
> I've attached copies of:
> 
> * /var/run/dmesg.boot
> * /var/log/Xorg.0.log
> * /var/log/xdm.log
> 
> and would appreciate suggestions for resolving the issue.

Are you under fairly severe memory pressure?

robert.

> (Please include me in replies; while I am subscribed to ports@, I am not
> currently subscribed to x11 at .)
> 
> Thanks.
> 
> Peace,
> david
-- 
Robert Noland <rnoland at FreeBSD.org>
FreeBSD