Kodi Community Forum

I have a number of Zotac HD-ID11 boxes running Dharma which were built on top of a minimal Ubuntu server 10.10 64-bit install. All have 1GB RAM except for one which has 2GB. The boxes drive several TVs and one projector at 1920x1080 resolution.

These boxes are used to front-end a tvheadend HTSP server to play satellite and cable TV feeds to TVs in a pub and in general this configuration works fine.

However the boxes occasionally crash while running PAL resolution media with the errors:

Code:
40:10 T:139825078195984 M:1861103616 WARNING: CDVDMessageQueue(audio)::Get - asked for new data packet, with nothing available

40:11 T:139825101264656 M:1861103616 WARNING: CDVDMessageQueue(video)::Get - asked for new data packet, with nothing available

40:16 T:139825078195984 M:1859190784 WARNING: CALSADirectSound::CALSADirectSound - device is not able to pause playback, will flush and prefix with 0 frames

40:16 T:139825101264656 M:1859190784   ERROR:  (VDPAU) Error: The system does not have enough resources to complete the requested operation at this time.(23) at VDPAU.cpp:1111

40:16 T:139825101264656 M:1859190784  NOTICE: Attempting recovery

40:16 T:139825101264656 M:1859190784  NOTICE:  (VDPAU) FiniVDPAUOutput

40:16 T:139825101264656 M:1866842112  NOTICE: vdp_device = 0x00000001 vdp_st = 0x00000000

40:16 T:139825101264656 M:1866842112  NOTICE:  (VDPAU) screenWidth:544 vidWidth:544

40:16 T:139825101264656 M:1866842112  NOTICE:  (VDPAU) screenHeight:576 vidHeight:576

40:16 T:139825101264656 M:1864970240  NOTICE:  (VDPAU) Total Output Surfaces Available: 4 of a max (tmp: 4 const: 4)

40:16 T:139825101264656 M:1864970240  NOTICE:  (VDPAU) Creating the video mixer

40:16 T:139825101264656 M:1864970240   ERROR:  (VDPAU) Error: The system does not have enough resources to complete the requested operation at this time.(23) at VDPAU.cpp:1111

40:16 T:139825101264656 M:1864970240  NOTICE: Attempting recovery

40:16 T:139825101264656 M:1864970240  NOTICE:  (VDPAU) FiniVDPAUOutput

40:16 T:139825101264656 M:1866510336  NOTICE: vdp_device = 0x00000001 vdp_st = 0x00000000

I have spent countless hours trying to get to the root cause of this, but nothing I have tried has resolved the issue. I've tried 270 (beta) 256 and several 260 (stable) versions of the NVidia drivers, tried adjusting the swiotlb= kernel boot parameter, set the VideoRam option under the Devices section in X, and the NVreg_RemapLimit in /etc/modprobe.d/nvidia to no avail.

The HD-ID11 does not have a BIOS option to change the amount of shared graphics memory, and I have not been able to find any way to adjust it from within Linux. The system reports 256MB from lspci and 512MB is reported by nvidia-settings, but I've found this is normal due to NVidia TurboCache.

A number of posts I've seen seem to indicate that 256MB is enough to drive the display at 1920x1080 (1080p) resolution.

The boxes typically have 300~600MB memory free, but I've occasionally seen as little as 150MB free. However the one box with 2GB has also been prone to crashing with over 1GB free.

Below is a rundown of the system configuration, I hope there is a simple solution to this issue. If there is any other information that could be of help, please let me know.

Thanks,

Tom
--

Code:
monitor@monitor06:~$ export DISPLAY=:0 && sudo nvidia-settings -q VideoRam

[sudo] password for monitor: 

  [color=red]Attribute 'VideoRam' (monitor06:0.0): 524288.[/color]

    'VideoRam' is an integer attribute.

    'VideoRam' is a read-only attribute.

    'VideoRam' can use the following target types: X Screen, GPU.

Code:
monitor@monitor06:~$ sudo lspci -v -s 03:00.0

03:00.0 VGA compatible controller: nVidia Corporation GT218 [ION] (rev a2) (prog-if 00 [VGA controller])

    Subsystem: PC Partner Limited Device 3100

    Physical Slot: 34

    Flags: bus master, fast devsel, latency 0, IRQ 18

    Memory at fd000000 (32-bit, non-prefetchable) [size=16M]

    [color=red]Memory at d0000000 (64-bit, prefetchable) [size=256M][/color]

    Memory at ce000000 (64-bit, prefetchable) [size=32M]

    I/O ports at ec00 [size=128]

    [virtual] Expansion ROM at fcf80000 [disabled] [size=512K]

    Capabilities: [60] Power Management version 3

    Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+

    Capabilities: [78] Express Endpoint, MSI 00

    Capabilities: [b4] Vendor Specific Information: Len=14 <?>

    Capabilities: [100] Virtual Channel

    Capabilities: [128] Power Budgeting <?>

    Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>

    Kernel driver in use: nvidia

    Kernel modules: nvidia, nouveau, nvidiafb

Code:
monitor@monitor06:~$ uname -a

Linux monitor06 2.6.35-22-server #33-Ubuntu SMP Sun Sep 19 20:48:58 UTC 2010 x86_64 GNU/Linux

Code:
monitor@monitor06:~$ cat /proc/driver/nvidia/version

NVRM version: NVIDIA UNIX x86_64 Kernel Module  260.19.44  Sun Feb 27 22:41:03 PST 2011

GCC version:  gcc version 4.4.5 (Ubuntu/Linaro 4.4.4-14ubuntu5)

Bump

544x576 is strange. This problem might be related to tvheadend. I have seen another case where a channel had a wrong resolution using tvheadend.

I've opened ticket 11450

It's a good point, since it should be PAL resolution, which is 576x720. However it looks normal on the screen.

I've posted the problem to the tvheadend forum.

funkknob

funkknob

FernetMenta

funkknob

funkknob