[Intel-gfx] [GM965] X fails to come back from suspend and gdm stop hangs

[Robert Lowery]

Hi Folks,

I am running Ubuntu Intrepid (1.5.2 X server, 2.6.27 kernel), but recent
git master drm and intel driver and am finding X does not come back up
after a suspend.

Before switching drm and intel driver, everything was fine with default
Intrepid packages, but I wanted to try out XvMC support.

Running "sudo /etc/init.d/gdm stop" also fails to stop X, I'm forced to
kill -9.

When I connect to it with gdb, the stack backtrace is

#0  0x00007f8734cdda17 in ioctl () from /lib/libc.so.6
#1  0x00007f87338b9c43 in drmIoctl (fd=11, request=1074029637,
    arg=0x7fff3eeaff70) at xf86drm.c:183
#2  0x00007f87338b9ccb in drmCommandWrite (fd=11,
    drmCommandIndex=<value optimized out>, data=0x7fff3eeaff70,
    size=18446744073709551615) at xf86drm.c:2343
#3  0x00007f873341576e in _fence_wait_internal (bufmgr_fake=0x20f1690,
    seq=6245892) at intel_bufmgr_fake.c:373
#4  0x00007f8733416738 in drm_intel_fake_reloc_and_validate_buffer (
    bo=0x5a322d0) at intel_bufmgr_fake.c:1123
#5  0x00007f8733416a42 in drm_intel_fake_bo_exec (bo=0x5a322d0, used=8,
    cliprects=0x0, num_cliprects=0, DR4=-1) at intel_bufmgr_fake.c:1354
#6  0x00007f8733638dbf in intel_batch_flush (pScrn=0x20e26a0,
    flushed=<value optimized out>) at i830_batchbuffer.c:191
#7  0x00007f8733638661 in I830Sync (pScrn=0x20e26a0) at i830_accel.c:195
#8  0x00007f87329c2f6c in exaWaitSync () from
/usr/lib/xorg/modules//libexa.so
#9  0x00007f87329c4121 in ExaDoPrepareAccess ()
   from /usr/lib/xorg/modules//libexa.so
#10 0x00007f87329cba95 in ExaCheckPutImage ()
   from /usr/lib/xorg/modules//libexa.so
#11 0x00007f87329c5452 in ?? () from /usr/lib/xorg/modules//libexa.so
#12 0x0000000000533a3f in ?? ()
#13 0x0000000000507bf9 in ?? ()

Am I doing something wrong?

Thanks for any advice.

-Rob