netbsd-9 radeon GPU lockup, X stuck in "radfen/N"

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

After updating a machine to 9.0_BETA, it ran fine for about a week and
a half, but has now had to be rebooted twice in a 12-hour period (one
by power-cycle) to regain control of the video console/graphics display.

It seems to happen when the machine has been left idle and 'xscreensaver'
activates.  Keyboard/mouse activity do not wake it up and attempting to
switch virtual terminals aith Ctrl-Alt-Fn doesn't work.

I can SSH into the machine and it's running, but 'top' shows 'X' always
in "radfen/N".  It will sometimes transition to "RUN/N", but always
returns to "radfen/N":


load averages:  0.21,  0.19,  0.20;               up 0+14:01:21        10:05:01
192 threads: 9 idle, 179 sleeping, 2 zombie, 2 on CPU
CPU states:  0.0% user,  0.0% nice,  3.4% system,  0.5% interrupt, 96.0% idle
Memory: 1457M Act, 892M Inact, 261M Wired, 199M Exec, 1287M File, 145M Free
Swap: 4096M Total, 4096M Free

  PID   LID USERNAME PRI STATE      TIME   WCPU    CPU NAME      COMMAND
  699     1 root      85 radfen/1  82:31  0.00%  0.00% -         X
[...]

Most recently, I was able to stop X with SIGTERM and when the console
reappeared, it was streaming messages like:


[...]
[ 50615.096841] radeon0: autoconfiguration error: error: ring 0 stalled for more than 6308000msec
[ 50615.096841] radeon0: autoconfiguration error: warn: GPU lockup (current fence id 0x00000000002505b3 last fence id 0x00000000002507ad on ring 0)
[ 50615.597914] radeon0: autoconfiguration error: error: ring 0 stalled for more than 6308000msec
[ 50615.597914] radeon0: autoconfiguration error: warn: GPU lockup (current fence id 0x00000000002505b3 last fence id 0x00000000002507ad on ring 0)
[ 50616.098981] radeon0: autoconfiguration error: error: ring 0 stalled for more than 6309000msec
[ 50616.098981] radeon0: autoconfiguration error: warn: GPU lockup (current fence id 0x00000000002505b3 last fence id 0x00000000002507ad on ring 0)
[ 50616.600071] radeon0: autoconfiguration error: error: ring 0 stalled for more than 6309000msec
[ 50616.600071] radeon0: autoconfiguration error: warn: GPU lockup (current fence id 0x00000000002505b3 last fence id 0x00000000002507ad on ring 0)
[ 50617.101137] radeon0: autoconfiguration error: error: ring 0 stalled for more than 6310000msec
[ 50617.101137] radeon0: autoconfiguration error: warn: GPU lockup (current fence id 0x00000000002505b3 last fence id 0x00000000002507ad on ring 0)
[...]

The video card is:

[...]
radeon0 at pci1 dev 0 function 0: ATI Technologies product 7183 (rev. 0x00)
ATI Technologies product 71a3 (miscellaneous display) at pci1 dev 0 function 1 n
ot configured
[...]
kern info: [drm] register mmio base: 0xfeaf0000
kern info: [drm] register mmio size: 65536
kern info: ATOM BIOS: X1550
kern info: [drm] Generation 2 PCI interface, using max accessible memory
radeon0: info: VRAM: 256M 0x0000000000000000 - 0x000000000FFFFFFF (256M used)
radeon0: info: GTT: 512M 0x0000000010000000 - 0x000000002FFFFFFF
kern info: [drm] Detected VRAM RAM=100M, BAR=256M
kern info: [drm] RAM width 128bits DDR
Zone  kernel: Available graphics memory: 901668 kiB
kern info: [drm] radeon: 256M of VRAM memory ready
kern info: [drm] radeon: 512M of GTT memory ready.
kern info: [drm] GART: num cpu pages 131072, num gpu pages 131072
kern info: [drm] radeon: 1 quad pipes, 1 z pipes initialized.
kern info: [drm] PCIE GART of 512M enabled (table at 0x0000000000040000).
radeon0: info: WB enabled
radeon0: info: fence driver on ring 0 use gpu addr 0x0000000010000000 and cpu addr 0x0xfffff7cf436b5000
kern info: [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
kern info: [drm] Driver supports precise vblank timestamp query.
radeon0: info: radeon: MSI limited to 32-bit
radeon0: interrupting at ioapic0 pin 18 (radeon0)
kern info: [drm] radeon: irq initialized.
kern info: [drm] Loading R500 Microcode
kern info: [drm] radeon: ring at 0x0000000010001000
kern info: [drm] ring test succeeded in 6 usecs
kern info: [drm] ib test succeeded in 0 usecs
kern info: [drm] Radeon Display Connectors
kern info: [drm] Connector 0:
kern info: [drm]   VGA-1
kern info: [drm]   DDC: 0x7e40 0x7e40 0x7e44 0x7e44 0x7e48 0x7e48 0x7e4c 0x7e4c
kern info: [drm]   Encoders:
kern info: [drm]     CRT1: INTERNAL_KLDSCP_DAC1
kern info: [drm] Connector 1:
kern info: [drm]   SVIDEO-1
kern info: [drm]   Encoders:
kern info: [drm]     TV1: INTERNAL_KLDSCP_DAC2
kern info: [drm] Connector 2:
kern info: [drm]   DVI-I-1
kern info: [drm]   HPD2
kern info: [drm]   DDC: 0x7e50 0x7e50 0x7e54 0x7e54 0x7e58 0x7e58 0x7e5c 0x7e5c
kern info: [drm]   Encoders:
kern info: [drm]     CRT2: INTERNAL_KLDSCP_DAC2
kern info: [drm]     DFP3: INTERNAL_LVTM1
radeondrmkmsfb0 at radeon0
radeondrmkmsfb0: framebuffer at 0xffffb5005b226000, size 1280x1024, depth 32, stride 5120
wsdisplay0 at radeondrmkmsfb0 kbdmux 1: console (default, vt100 emulation), using wskbd0
wsmux1: connecting to wsdisplay0


Anyone else seeing a problem like this?

More data to gather?

Thanks.

-- 
|/"\ 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