Re: diagnosing freezes (DRI?)

On Thu, 2009-04-16 at 11:16 -0400, John Baldwin wrote:
> [ moved to current_at_ ]
> 
> On Wednesday 15 April 2009 11:58:38 pm deeptech71_at_gmail.com wrote:
> > I can reliably (~40%) reproduce a freeze, which I think is related.
> > 
> > Using the GENERIC debug kernel built from SVN HEAD:
> > 
> > # cd /usr/obj/usr/src/sys/GENERIC/
> > # kgdb kernel.debug /var/crash/vmcore.0
> > GNU gdb 6.1.1 [FreeBSD]
> > Copyright 2004 Free Software Foundation, Inc.
> > GDB is free software, covered by the GNU General Public License, and you are
> > welcome to change it and/or distribute copies of it under certain 
> > conditions.
> > Type "show copying" to see the conditions. 
> > 
> > There is absolutely no warranty for GDB.  Type "show warranty" for 
> > details.
> > This GDB was configured as "i386-marcel-freebsd"... 
> 
> [ snipped lots of stuff ]
> 
> > Kernel page fault with the following non-sleepable locks held: 
> > 
> > exclusive sleep mutex drmdev (drmdev) r = 0 (0xc373f860) locked _at_ 
> > /usr/src/sys/modules/drm/drm/../../../dev/drm/drm_drv.c:777 
> > 
> > KDB: stack backtrace: 
> > 
> > calltrap() at calltrap+0x6 
> > 
> > --- trap 0xc, eip = 0xc0b611b6, esp = 0xd67d4a98, ebp = 0xd67d4b48 --- 
> > 
> > slow_copyin(c373f800,c4103300,c42e64e0,d67d4b64,0,...) at 
> > slow_copyin+0x6
> > radeon_cp_texture(c373f800,c42e64e0,c4103300,309,c0c26218,...) at 
> > radeon_cp_texture+0x199 
> > 
> > drm_ioctl(c39d4e00,c018644e,c42e64e0,3,c40afaf0,...) at drm_ioctl+0x356 
> > 
> 
> The drm code is doing a copyin() while holding a mutex (which is not allowed).

Ok, the quick and dirty fix for this is
http://people.freebsd.org/~rnoland/drm_radeon_state-copyin-fix.patch

I think there may be other places of concern though and a more proper
fix is needed.

robert.

-- 
Robert Noland <rnoland_at_FreeBSD.org>
FreeBSD