V4L2 driver : Kernel bug: VPSS_FVID2: queue timeout

Hi,

This is solved in latest EZSDK release + few patches posted on community. Please refer to following threads for details.

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/186865/675787.aspx#675787

Regards,

Hardik Shah

Hi Hardik,

Thanks for quick help.
Can you exactly tell me whether the fix for FVID2 timeout is only in Linux driver (A8) or it is in firmware binary at M3 ?

Since I am using Android, I want to take minimal changes to fix timeout error, keeping ezsdk_5_03 as reference. Porting ezsdk_5_04 on Android will be a big task at this point of time, so please tell me the exact source for this fix.

Regards,
Sweta

Hi,

Changes are there only on A8 side video drivers. You can take those patches and port it to your version of PSP

Regards,

Hardik Shah

Hi Hardik,

As I understand, first two patches are to be applied for my issue.

I applied following patches to the base PSP.

0001-TI81xx-Capture-IR-Kernel-dump-is-seen-on-starting-ca.patch

0002-TI81xx-Video-Proxy-Same-containers-are-used-for-Queu.patch

After applying these patches, the frequency of occurrence of issue is reduced. But I got following error now.

VPSS_FVID2: contrl event 0x5 timeout
VPSS_FVID2: queue timeout

What can be the reason for the same ?

Regards,

Sweta

Hi,

You should be applying all patches. What made you think that only first two patches are important?

Regards,

Hardik Shah

Hi Hardik,

From the description of patches, I understood 1 & 2 relevant for my use-case and issue.

Any ways, to be on same page, I applied all the 5 patches and still I am getting FVID timeout error unfortunately.

I also nailed down, that the issue is at HDVPSS side. I am getting assertion, when kernel says FVID timeout.

Please see log below for the reference.

N:VPSS  P:2 #:00370 T:0000000bae8595fd S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<OMX_BASE_DIO_Control> @line<452> with error<0:ErrorNone>
N:Video P:1 #:00393 T:0000000b2528b61d S:StopInstance: HDVICP_1
N:VPSS  P:2 #:00371 T:0000000bae8607a9 S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Entering<OMX_BASE_DIO_Control> @line<446>
N:Video P:1 #:00394 T:0000000b25ad4853 S:StartInstance: HDVICP_1
N:VPSS  P:2 #:00372 T:0000000bae899d5d S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<OMX_BASE_DIO_Control> @line<452> with error<0:ErrorNone>
N:Video P:1 #:00395 T:0000000b25db7b1b S:StopInstance: HDVICP_1
N:VPSS  P:2 #:00373 T:0000000bae8b179b S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<_OMX_BASE_EventNotifyToClient> @line<1171> with error<0:ErrorNone>
N:Video P:1 #:00396 T:0000000b2666606d S:StartInstance: HDVICP_1
N:VPSS  P:2 #:00374 T:0000000bae8b8565 S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<OMX_BASE_PROCESS_CmdCompleteEvent> @line<930> with error<0:ErrorNone>
N:Video P:1 #:00397 T:0000000b2694ef81 S:StopInstance: HDVICP_1
N:VPSS  P:2 #:00375 T:0000000bae8bf3e3 S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<OMX_BASE_CmdCompleteEventHandler> @line<496> with error<0:ErrorNone>
N:Video P:1 #:00398 T:0000000b273ada37 S:StartInstance: HDVICP_1
N:VPSS  P:2 #:00376 T:0000000bae8c62c5 S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Entering<OMX_BASE_CmdEventHandler> @line<434>
N:Video P:1 #:00399 T:0000000b27698fa5 S:StopInstance: HDVICP_1
N:VPSS  P:2 #:00377 T:0000000bae8cc643 S:Module<OMX.TI.VPSSM3.VFPC.INDTXSCWB> Leaving<OMX_BASE_CmdEventHandler> @line<466> with error<0:ErrorNone>
N:Video P:1 #:00400 T:0000000b27f21dc1 S:StartInstance: HDVICP_1
N:VPSS  P:2 #:00378 T:00000114d138b64f S: 2376362: Assertion @ Line: 520 in drivers/capture/src/vpsdrv_captureApi.c: queStatus == FVID2_SOK : failed !!!
N:Video P:1 #:00401 T:0000000b28204d53 S:StopInstance: HDVICP_1
N:Video P:1 #:00402 T:0000000b28908e7b S:StartInstance: HDVICP_1
N:Video P:1 #:00403 T:0000000b28bf2913 S:StopInstance: HDVICP_1
N:Video P:1 #:00404 T:0000000b29663067 S:StartInstance: HDVICP_1
N:Video P:1 #:00405 T:0000000b2994ee3d S:StopInstance: HDVICP_1
N:Video P:1 #:00406 T:0000000b2a06d1c9 S:StartInstance: HDVICP_1
N:Video P:1 #:00407 T:0000000b2a3552af S:StopInstance: HDVICP_1

Can you tell me what is the reason for this assertion ?

Just to reiterate, I am using firmware binary of ezsdk_5_03_01_15 and linux from the same ezsdk, on top of this kernel, I applied all 5 patches.

Regards,
Sweta

Hi,

These patches are tested with SDK5.04 so this will be really difficult to debug. You also may be corrupting M3 memory. I am not very sure. Above error status says that you can queuing buffers more than M3 queue can receive. This queue is 16 elements deep. So I think something is getting corrupted.

Regards,

Hardik Shah

Hi Hardik,

1) By corruption of M3 memory, if you mean to say that my buffer addresses (for queue/dequeue) are corrupted, then this is not the case.

I have already checked for above and addresses are same when this issue occurs.

2) I have also ensured that I am queueing only 4 buffers, not more than that.

Let me know if there is any other possibility.

It would be very helpful if you can help me understand, whether the issue is at M3 or A8 ?

Since I get assertion from M3, I would like to double check, if TI has fixed this issue in firmware binaries of SDK4, due to which you are not getting this error.

If you can say that the issue is solely on ARM, then I can take latest linux kernel as reference with firmware binary of ezsdk_5_03_01_15 to check solution of issue as something else (some fix beyond these 5 patches which is there in latest Linux kernel).

Thanks,
Sweta

Hi,

Did it solve after that?

I also have the same problem.

(SDK5.04 + 5 patches =>VPSS_FVID2: queue timeout)

Regards,

Hatta

Hi,

No. It is not resolved for me yet, although frequency of queue timeout has been reduced.

Regards,
Sweta

Hi,

I am using EZSDK_05.05.02.

While running video playback at high bit rate (44.7mbps), after 12+ hours of run, I got the same error VPSS_FVID2 queue timeout.

Do I need to apply the patches on EZSDK 05.05.02 too?

Salim