Bug#943664: xserver-xorg-video-radeon 19.1 (ppc64) crash at startup: undefined symbol: exaGetPixmapDriverPrivate

[Ralf P.]

Hi.

I assume that the error occurs on the PPC64 system because of
disabled HA (Hardware Acceleration) too:

According to 

   https://bugs.freedesktop.org/show_bug.cgi?id=99851

a "ring 0 test failed" message from drm means that HA gets disabled.


dmesg | grep -i drm

[    5.621228] [drm] radeon kernel modesetting enabled.
[    5.662906] [drm] initializing kernel modesetting (CEDAR 0x1002:0x68FA 0x174B:0xE127 0x00).
[    5.974683] [drm] GPU not posted. posting now...
[    6.017017] [drm] Detected VRAM RAM=1024M, BAR=256M
[    6.017018] [drm] RAM width 64bits DDR
[    6.082254] [drm] radeon: 1024M of VRAM memory ready
[    6.195713] [drm] radeon: 1024M of GTT memory ready.
[    6.220009] [drm] Loading CEDAR Microcode
[    6.335990] [drm] Internal thermal controller with fan control
[    6.364947] [drm] radeon: dpm initialized
[    6.396903] [drm] GART: num cpu pages 16384, num gpu pages 262144
[    6.458639] [drm] PCIE GART of 1024M enabled (table at 0x0000000000150000).
[    6.531588] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
[    6.576700] [drm] Driver supports precise vblank timestamp query.
[    6.641177] [drm] radeon: irq initialized.
[    6.834702] [drm:.r600_ring_test [radeon]] *ERROR* radeon: ring 0 test failed (scratch(0x8504)=0xCAFEDEAD)
[    6.863707] [drm] Radeon Display Connectors
[    6.880862] [drm] Connector 0:
[    6.880863] [drm]   DVI-I-1
[    6.880865] [drm]   HPD2
[    6.880867] [drm]   DDC: 0x6430 0x6430 0x6434 0x6434 0x6438 0x6438 0x643c 0x643c
[    6.880868] [drm]   Encoders:
[    6.880869] [drm]     DFP1: INTERNAL_UNIPHY1
[    6.880870] [drm]     CRT2: INTERNAL_KLDSCP_DAC2
[    6.880875] [drm] Connector 1:
[    6.902720] [drm]   DVI-I-2
[    6.913316] [drm]   HPD4
[    6.987101] [drm]   DDC: 0x6440 0x6440 0x6444 0x6444 0x6448 0x6448 0x644c 0x644c
[    7.001856] [drm]   Encoders:
[    7.007757] [drm]     DFP2: INTERNAL_UNIPHY
[    7.016093] [drm]     CRT1: INTERNAL_KLDSCP_DAC1
[    7.112647] [drm] fb mappable at 0x90360000
[    7.120638] [drm] vram apper at 0x90000000
[    7.128797] [drm] size 7733248
[    7.134875] [drm] fb depth is 24
[    7.141298] [drm]    pitch is 6400
[    7.212427] radeon 0000:0a:00.0: fb0: radeondrmfb frame buffer device
[    7.238821] [drm] Initialized radeon 2.50.0 20080528 for 0000:0a:00.0 on minor 0



I tested these kernel versions:
4.18.0-3-powerpc64       (official kernel from Debian PPC64 sid)
4.19.28-rt16		 (self compiled, with realtime patches)

Both result in "ring 0 test failed".


In my opinion it would be great if:

1) xserver-xorg-video-radeon_1%3a19.1.0-1 would not fail but fall back to
   frame buffer device like xserver-xorg-video-radeon_1%3a19.0.1-1

2) the "ring 0 test" would not fail
   According to the last post from the link above the failure was
   seen in 4.14.4 and 4.15-rc2 but not in kernel 4.17.2.
   Maybe I should get/compile a linux 4.17.2 kernel.

Thank you and best regards,
Ralf