firefox52 quits on netbsd-9 w/i915drmkmsfb

["John D. Baker" <jdbaker%consolidated.net@localhost>]

I switched back to my intel-based workstation as it's slightly faster
and has more RAM than the radeon-based one.

All was working fine for a few days and after some time away came back
to find that firefox52 was no-longer running.  There was no core file,
either.  The only tell-tail was the last line in ".xsession-errors":

  i965: Failed to submit batchbuffer: Input/output error

Restarting firefox52, it recovers the open tabs and tries to repaint the
one that was last open, but then it just quits--no core, just the same
message at the end of the ".xsession-errors" file.

I have been noticing a lot of messages on the console (xconsole) about:

[...]
[ 149770[ 149776.8299129] kern info: [drm] stuck on render ring
[ 149776.8299129] kern info: [drm] GPU HANG: ecode 4:0:0x87f7fff8, reason: Ring hung, action: reset
[ 149776.8299129] kern error: [drm:(/x/netbsd-9/src/sys/external/bsd/drm2/dist/drm/i915/i915_gem.c:3196)i915_context_is_banned] *ERROR* gpu hanging too fast, banning!
[ 149776.8299129] drm/i915: Resetting chip after gpu hang
[ 149776.8299129] kern error: [drm:(/x/netbsd-9/src/sys/external/bsd/drm2/dist/drm/i915/i915_irq.c:2703)i915_handle_error] *ERROR* Error state:
[ 149776.8299129] GPU HANG: ecode 4:0:0xf7d85fff, reason: Ring hung, action: reset
[ 149776.8299129] Time: 1572840829 s 821141 us
[ 149776.8299129] Kernel: 900000000
[ 149776.8299129] Reset count: 0
[ 149776.8299129] Suspend count: 0
[ 149776.8299129] PCI ID: 0x2e32
[ 149776.8299129] IOMMU enabled?: -1
[ 149776.8299129] EIR: 0x00000000
[ 149776.8299129] IER: 0x02028053
[ 149776.8299129] PGTBL_ER: 0x00000000
[ 149776.8299129] FORCEWAKE: 0x00000000
[ 149776.8299129] DERRMR: 0x00000000
[ 149776.8299129] CCID: 0x00000000
[ 149776.8299129] Missed interrupts: 0x00000000
[ 149776.8299129]   fence[0] = 00000000
[ 149776.8299129]   fence[1] = fbc00000abd09d
[ 149776.8299129]   fence[2] = 11460000113301d
[ 149776.8299129]   fence[3] = 89f2000089db07d
[ 149776.8299129]   fence[4] = 00000000
[ 149776.8299129]   fence[5] = 00000000
[ 14977 fence[8] = 70360000702309d
[ 149776.8299129]   fence[9] = 00000000
[ 149776.8299129]   fence[10] = 95c0000090c109d
[ 149776.8299129]   fence[11] = 00000000
[ 149776.8299129]   fence[12] = 103d0000103409d
[ 149776.8299129]   fence[13] = 00000000
[ 149776.8299129]   fence[14] = 70dc000070d309d
[ 149776.8299129]   fence[15] = 00000000
[ 149776.8299129]   INSTDONE_0: 0xffffffff
[ 149776.8299129]   INSTDONE_1: 0xbfffffff
[ 149776.8299129]   INSTDONE_2: 0x00000000
[ 149776.8299129]   INSTDONE_3: 0x00000000
[ 149776.8299129] render command stream:
[ 149776.8299129]   START: 0x00003000
[ 149776.8299129]   HEAD:  0x0e209600
[ 149776.8299129]   TAIL:  0x00009640
[ 149776.8299129]   CTL:   0x0001f001
[ 149776.8299129]   HWS:   0x00001000
[ 149776.8299129]   ACTHD: 0x00000000 00a96138
[ 149776.8299129]   IPEIR: 0x00000000
[ 149776.8299129]   IPEHR: 0x0827a000
[ 149776.8299129]   INSTDONE: 0xffffffff
[ 149776.8299129]   BBADDR: 0x00000000 00a96133
[ 149776.8299129]   BB_STATE :  0000002c

And then what looks like a series of addresses/offsets with apparent
contents.

I'm using the stock 2019 intel X driver, but due to SNA being broken
on my particular hardware, I must force UXA acceleration to get a
working display.

Anyone else having issues like this?

-- 
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]consolidated[flyspeck]net  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