Re: Broken X (DRM errors)

["John D. Baker" <jdbaker%mylinuxisp.com@localhost>]

On Tue, Mar 29, 2011 at 10:05 PM, Julio Merino <jmmv%julipedia.org@localhost> 
wrote:

> Since I upgraded yesterday to a freshly-built -current, I can't get X
> to run any more.  This is on a MacBook Pro with an ATI Radeon Mobility
> X1600.
>
> As soon as I start the X server up, the screen goes off and the
> machine gets extremely sluggish.  I can still access it through the
> network, but killing the X server has no effect.  Other "unrelated"
> processes, like sshd and tmux, can be seen chewing incredible amounts
> of CPU (presumably due to a loop in the kernel; see below).
>
> The dmesg right before the lock up shows:
>
> Mar 29 21:43:08 desky /netbsd: info: [drm] Setting GART location based
> on new memory map
> Mar 29 21:43:08 desky /netbsd: info: [drm] Loading R520 Microcode
> Mar 29 21:43:08 desky /netbsd: info: [drm] Num pipes: 1
> Mar 29 21:43:08 desky /netbsd: info: [drm] writeback test failed
> Mar 29 21:43:08 desky /netbsd: radeondrm0: interrupting at ioapic0 pin 16
> Mar 29 21:43:09 desky /netbsd: info: [drm] wait idle failed status :
> 0x90010140 0x00050000
> Mar 29 21:43:53 desky syslogd[219]: last message repeated 63 times
>
> Any ideas as to what change broke this?

I've seen and am still seeing this behavior on an HP Pavilion a1235c
with ATI RC410.  It's only Xorg that's eating CPU, but on at least
one attempt, it rendered the machine comatose, requiring a power-cycle
to regain control.

Following the advice of other messages in this thread, I've posted
various log and trace files:

  http://bobdbob.com/~jdbaker/nbsd-debug/X-radeon.ktrace
  http://bobdbob.com/~jdbaker/nbsd-debug/X-radeon-Xorg.log
  http://bobdbob.com/~jdbaker/nbsd-debug/X-radeon-xorg.conf
  http://bobdbob.com/~jdbaker/nbsd-debug/X-radeon.dmesg
  http://bobdbob.com/~jdbaker/nbsd-debug/X-radeonCPU.txt

The last is several successive dumps of 'top' showing the growing CPU
usage to around 113% (I saw 121% on a previous attempt).  The dmesg
file shows the stream of error messages through a forced reboot (X is
unkillable) followed by the usual kernel messages on boot for reference.


Also, when using the "vesa" driver, it is very, very slow to initialize.
Upon running 'startx', I see the wscons vt screen for about 30 seconds
followed by X-server black screen for about another 30 seconds before
the classic "grey stipple" root pattern appears, followed shortly by my
first client application (typically 'rxvt-unicode').

During X-server initialization, CPU utilization was observed to reach
86% but then dropped to a more respectable range between 2-8% during
normal operation.

  http://bobdbob.com/~jdbaker/nbsd-debug/X-vesa.ktrace
  http://bobdbob.com/~jdbaker/nbsd-debug/X-vesa-Xorg.log

Do these suggest anything?  It'd be nice to use the radeon driver
eventually.

Thanks?
--
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645