Video converter with many cameras (Tegra APIs)

Autonomous Machines Jetson & Embedded Systems Jetson TX2
1

Dear all,

Now I’m facing problem when terminate application converting frames from UYVY to YUV420 on 2 or 3 cameras. More details:

  1. If I use only v4l2 to capture frame (memory buff type : V4L2_MEMORY_MMAP) and convert to RGB to display with OpenCV everything would work well even 2 or 3 cameras, there is no warning message on debug terminal.

  2. If I use v4l2 to capture frame (memory buff type: V4L2_MEMORY_DMABUF) and feed to Video converter to convert to YUV420:
    a. Run only one camera, it will show below warning when start

nvmap_alloc_handle: PID 1775: HX500: WARNING: All NvMap Allocations must have a tag to identify the subsystem allocating memory.Please pass the tag to the API call NvRmMemHanldeAllocAttr() or relevant.

  and when stop

[   78.792273] tegra-vi4 15700000.vi: Status:  5 channel:00 frame:0000
[   78.798654] tegra-vi4 15700000.vi:          timestamp sof 88469448320 eof 88469628512 data 0x00000001
[   78.808015] tegra-vi4 15700000.vi:          capture_id 210 stream  0 vchan  0

b. If run 2 or 3 cameras same time then OS easily would be hang/stuck once application has been terminated. I create each thread for each camera capturing and converting.
Hang/stuck log

[20701.927991] [VuNguyen] ov10635_power_off 1918
[20739.188519] [VuNguyen] ov10635_power_on 1881
[20739.223924] nvmap_alloc_handle: PID 4761: HX500: WARNING: All NvMap Allocations must have a tag to identify the subsystem allocating memory.Please pass the tag to the API call NvRmMemHanldeAllocAttr() or relevant. 
[20739.244232] [VuNguyen] ov10635_power_on 1881
[20739.270743] [vunguyen] ov10635_s_stream [+] 1
[20739.275235] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[20739.280183] [vunguyen] ov10635_s_stream [+] 1
[20739.280185] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[20742.253805] tegra-vi4 15700000.vi: Status:  5 channel:00 frame:0000
[20742.253809] [vunguyen] ov10635_s_stream [+] 0
[20742.264527] tegra-vi4 15700000.vi:          timestamp sof 20751904736992 eof 20751904870656 data 0x00000001
[20742.274322] tegra-vi4 15700000.vi:          capture_id 1835 stream  0 vchan  0
[20742.281615] tegra-vi4 15700000.vi: Status:  5 channel:01 frame:0000
[20742.287922] tegra-vi4 15700000.vi:          timestamp sof 20751896158208 eof 20751906639072 data 0x00000002
[20742.297707] tegra-vi4 15700000.vi:          capture_id 3552 stream  2 vchan  0
[20742.304981] Unable to handle kernel paging request at virtual address ffffffbe429f0f40
[20742.304993] [vunguyen] ov10635_s_stream [+] 0
[20742.305823] [VuNguyen] ov10635_power_off 1918
[20742.306180] [VuNguyen] ov10635_power_off 1918
[20742.326057] pgd = ffffffc19efab000
[20742.329487] [ffffffbe429f0f40] *pgd=0000000000000000, *pud=0000000000000000
[20742.336519] Internal error: Oops: 96000005 [#1] PREEMPT SMP
[20742.342084] Modules linked in:
[20742.345150] CPU: 5 PID: 3161 Comm: kworker/5:1 Not tainted 4.4.38 #135
[20742.351666] Hardware name: quill (DT)
[20742.355329] Workqueue: events tegra_channel_status_worker
[20742.360730] task: ffffffc1ec637080 ti: ffffffc1a2834000 task.ti: ffffffc1a2834000
[20742.368205] PC is at osq_lock+0x54/0x18c
[20742.372124] LR is at mutex_optimistic_spin+0x94/0x1c8
[20742.377166] pc : [<ffffffc0000ea578>] lr : [<ffffffc0000e9c28>] pstate: 60000045
[20742.384547] sp : ffffffc1a2837c00
[20742.387855] x29: ffffffc1a2837c00 x28: 0000000000000000 
[20742.393183] x27: 0000000000000000 x26: ffffffc1aebcf038 
[20742.398507] x25: ffffffc1ec637080 x24: 0000000000000000 
[20742.403830] x23: 0000000000000000 x22: 0000000000000000 
[20742.409154] x21: 0000000000000000 x20: 00000000c82f5f60 
[20742.414477] x19: ffffffc1aebcf018 x18: ffffffc000bc7e38 
[20742.419800] x17: ffffffc000b54a60 x16: ffffffc000b54a60 
[20742.425123] x15: ffffffc000b54a60 x14: 0000000000000001 
[20742.430446] x13: 00000026e42decd9 x12: 000000000000bb4a 
[20742.435770] x11: 000000000000c36a x10: 0000000000000400 
[20742.441093] x9 : 0000000000000000 x8 : ffffffc1f5daf570 
[20742.446416] x7 : 0000000000000018 x6 : ffffffc001443510 
[20742.451739] x5 : ffffffc001214000 x4 : 0000000000000000 
[20742.457062] x3 : 00000000c82f5fbf x2 : ffffffc1f5dc1000 
[20742.462386] x1 : ffffffc001241148 x0 : ffffffc1aebcf038 
[20742.467708] 
[20742.469196] Process kworker/5:1 (pid: 3161, stack limit = 0xffffffc1a2834020)
[20742.476318] Call trace:
[20742.478762] [<ffffffc0000ea578>] osq_lock+0x54/0x18c
[20742.483722] [<ffffffc000b46694>] __mutex_lock_interruptible_slowpath+0x3c/0x1d8
[20742.491020] [<ffffffc000b46878>] mutex_lock_interruptible+0x48/0x4c
[20742.497280] [<ffffffc00090a104>] vi_notify_channel_reset+0x38/0x88
[20742.503451] [<ffffffc0007a06a0>] tegra_channel_handle_error+0x5c/0xa0
[20742.509883] [<ffffffc0007a06f4>] tegra_channel_status_worker+0x10/0x18
[20742.516403] [<ffffffc0000bbca0>] process_one_work+0x150/0x448
[20742.522140] [<ffffffc0000bc0cc>] worker_thread+0x134/0x40c
[20742.527619] [<ffffffc0000c1968>] kthread+0xe0/0xf4
[20742.532403] [<ffffffc000084f90>] ret_from_fork+0x10/0x40
[20742.537707] ---[ end trace 2f108655eea162ba ]---
[20742.544975] note: kworker/5:1[3161] exited with preempt_count 1
[20742.550978] Unable to handle kernel paging request at virtual address ffffffffffffffd8
[20742.558882] pgd = ffffffc19efab000
[20742.562276] [ffffffffffffffd8] *pgd=0000000000000000, *pud=0000000000000000
[20742.569250] Internal error: Oops: 96000005 [#2] PREEMPT SMP
[20742.574811] Modules linked in:
[20742.577871] CPU: 5 PID: 3161 Comm: kworker/5:1 Tainted: G      D         4.4.38 #135
[20742.585600] Hardware name: quill (DT)
[20742.589260] task: ffffffc1ec637080 ti: ffffffc1a2834000 task.ti: ffffffc1a2834000
[20742.596732] PC is at kthread_data+0x4/0xc
[20742.600734] LR is at wq_worker_sleeping+0x10/0xc4
[20742.605428] pc : [<ffffffc0000c2038>] lr : [<ffffffc0000bcb78>] pstate: 600000c5
[20742.612807] sp : ffffffc1a2837860
[20742.616113] x29: ffffffc1a2837860 x28: ffffffc1a2834000 
[20742.621435] x27: 0000000000000000 x26: ffffffc001241000 
[20742.626756] x25: 0000000000000005 x24: ffffffc000b458d0 
[20742.632077] x23: 0000000000000000 x22: ffffffc1ec637610 
[20742.637398] x21: ffffffc001213000 x20: ffffffc1ec637080 
[20742.642719] x19: ffffffc1f5dc0500 x18: ffffffc000bc7e38 
[20742.648041] x17: 000000000000000e x16: 0000000000000007 
[20742.653364] x15: 0000000000000001 x14: 0000000000000007 
[20742.658686] x13: 000000000000000e x12: 0000000000000013 
[20742.664008] x11: 000000000000001a x10: 000000000023b282 
[20742.669330] x9 : 000000000023b282 x8 : 0000000000000230 
[20742.674651] x7 : 0000000000000000 x6 : 00000000018e0ec4 
[20742.679974] x5 : ffffffc1f5dc0500 x4 : ffffffc1f5dc0ee0 
[20742.685298] x3 : 000000000001af3b x2 : ffffffc1ecc03000 
[20742.690620] x1 : 0000000000000005 x0 : 0000000000000000 
[20742.695939] 
[20742.697426] Process kworker/5:1 (pid: 3161, stack limit = 0xffffffc1a2834020)
[20742.704546] Call trace:
[20742.706988] [<ffffffc0000c2038>] kthread_data+0x4/0xc
[20742.712032] [<ffffffc000b454f8>] __schedule+0x348/0x6dc
[20742.717246] [<ffffffc000b458d0>] schedule+0x44/0xa8
[20742.722115] [<ffffffc0000a6c60>] do_exit+0x5b0/0x9a0
[20742.727073] [<ffffffc000089c08>] bug_handler.part.3+0x0/0x7c
[20742.732725] [<ffffffc00009c7c0>] __do_kernel_fault.part.7+0x74/0x84
[20742.738981] [<ffffffc00009ab10>] cpu_enable_pan+0x0/0x48
[20742.744282] [<ffffffc0000813a0>] do_mem_abort+0x40/0xa0
[20742.749496] [<ffffffc000084540>] el1_da+0x18/0x78
[20742.754191] [<ffffffc000b46694>] __mutex_lock_interruptible_slowpath+0x3c/0x1d8
[20742.761486] [<ffffffc000b46878>] mutex_lock_interruptible+0x48/0x4c
[20742.767743] [<ffffffc00090a104>] vi_notify_channel_reset+0x38/0x88
[20742.773913] [<ffffffc0007a06a0>] tegra_channel_handle_error+0x5c/0xa0
[20742.780342] [<ffffffc0007a06f4>] tegra_channel_status_worker+0x10/0x18
[20742.786859] [<ffffffc0000bbca0>] process_one_work+0x150/0x448
[20742.792594] [<ffffffc0000bc0cc>] worker_thread+0x134/0x40c
[20742.798070] [<ffffffc0000c1968>] kthread+0xe0/0xf4
[20742.802852] [<ffffffc000084f90>] ret_from_fork+0x10/0x40
[20742.808154] ---[ end trace 2f108655eea162bb ]---
[20742.814923] Fixing recursive fault but reboot is needed!
  1. I also tried to run samples 12_camera_v4l2_cuda when I terminated application by “ctrl +c” errors would flush on terminal as below
[ 2851.779072] [VuNguyen] ov10635_power_on 1881
[ 2851.865184] [vunguyen] ov10635_s_stream [+] 1
[ 2851.869550] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[ 2854.740915] [vunguyen] ov10635_s_stream [+] 0
[ 2854.747412] [VuNguyen] ov10635_power_off 1918
[ 2854.765714] camera_v4l2_cud[4549]: unhandled level 2 translation fault (11) at 0x00000000, esr 0x92000006
[ 2854.775475] pgd = ffffffc1944bc000
[ 2854.778926] [00000000] *pgd=00000002144bb003, *pud=00000002144bb003, *pmd=0000000000000000
[ 2854.787476] 
[ 2854.788982] CPU: 3 PID: 4549 Comm: camera_v4l2_cud Not tainted 4.4.38 #135
[ 2854.796037] Hardware name: quill (DT)
[ 2854.799745] task: ffffffc06c95be80 ti: ffffffc194498000 task.ti: ffffffc194498000
[ 2854.807339] PC is at 0x7f7a9b94e0
[ 2854.810678] LR is at 0x7f7a9b9490
[ 2854.814043] pc : [<0000007f7a9b94e0>] lr : [<0000007f7a9b9490>] pstate: 60000000
[ 2854.821563] sp : 0000007fd57dddf0
[ 2854.824898] x29: 0000007fd57dde40 x28: 0000007fd57de348 
[ 2854.830361] x27: 00000000c0145608 x26: 0000000000000006 
[ 2854.835792] x25: 000000007d000000 x24: 0000000000000004 
[ 2854.841171] x23: 0000000000478540 x22: 0000000000000000 
[ 2854.846597] x21: 0000000000000000 x20: 0000000000000000 
[ 2854.852053] x19: 0000000000000000 x18: 0000000000000000 
[ 2854.857439] x17: 0000007f7a988fb0 x16: 0000007f7926b7d8 
[ 2854.862911] x15: 001b27029a80f646 x14: 0000000000000000 
[ 2854.868302] x13: 00000001f4000000 x12: 0000000000000017 
[ 2854.873731] x11: 00000000000de6ea x10: 000000005a5c56c0 
[ 2854.879154] x9 : 00000000000af270 x8 : 00000000000af270 
[ 2854.884531] x7 : 0000000000000000 x6 : 0000000000000001 
[ 2854.890008] x5 : 0000000000000000 x4 : 0000000000478428 
[ 2854.895427] x3 : 0000000000000000 x2 : 0000000000000000 
[ 2854.900848] x1 : 0000000000000000 x0 : 0000000000478540 
[ 2854.906235] 
[ 2854.907763] Library at 0x7f7a9b94e0: 0x7f7a9b5000 /usr/lib/aarch64-linux-gnu/libv4l/plugins/libv4l2_nvvidconv.so
[ 2854.918014] Library at 0x7f7a9b9490: 0x7f7a9b5000 /usr/lib/aarch64-linux-gnu/libv4l/plugins/libv4l2_nvvidconv.so
[ 2854.928223] vdso base = 0x7f7a9f7000

I was wondering is there any differences between V4L2_MEMORY_MMAP and V4L2_MEMORY_DMABUF could cause that problem. Or because of the other reason.

Thanks
Vu Nguyen

2

Hi,
Please share a sample to reproduce the issue. Thanks.

3

Hi DaneLLL,

https://drive.google.com/open?id=1sEXDcliDWsh02M9DSxuBQvub4gunnd8L

Plsz download and extract to “/home/nvidia/tegra_multimedia_api/samples” and just run “make” command. You can change NUM_CAM_SUPPORT value in “.h” file.

I found one strange thing that if I enable render everything worked well, but if render was disabled then easily to see below log as well as easily stuck.

[   78.792273] tegra-vi4 15700000.vi: Status:  5 channel:00 frame:0000
[   78.798654] tegra-vi4 15700000.vi:          timestamp sof 88469448320 eof 88469628512 data 0x00000001
[   78.808015] tegra-vi4 15700000.vi:          capture_id 210 stream  0 vchan  0

Thanks and Best Regards,
Vu Nguyen

4

Hi DaneLLL,

Did you try my application.
FYI, if I only capture frames from 3 cameras and display with OpenCV it would never happen this issue.

Thanks,
Vu Nguyen

5

Hi Vu,
We don’t have board with multiple YUV sensors so we try multiple USB cameras. The issue is not seen with USB cameras.

So are you able to reprocude the issue via v4l2-ctl? Like ‘v4l2-ctl -d /dev/video0 & v4l2-ctl -d /dev/video1 & v4l2-ctl -d /dev/video2 &’

6

Hi DaneLLL,

If I use only v4l2 it would be OK, the problem only happen when I enabled Video Converter (Tegra APIs).
I think it might relate to shared memory between v4l2 and Video Converter.

Thanks

7

Hi Vu,
How about ‘./camera_v4l2_cuda -d /dev/video0 & ./camera_v4l2_cuda -d /dev/video1 &’ ?

8

Hi D

Hi DaneLLL,

It started well. However when I terminate the application Ubuntu has been stuck then I could control nothing except HW reboot. Debug log as below

[ 2592.623249] [vunguyen] ov10635_s_stream [+] 1
[ 2592.627702] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[ 2592.724977] [vunguyen] ov10635_s_stream [+] 1
[ 2592.729427] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[ 2592.735663] [vunguyen] ov10635_s_stream [+] 1
[ 2592.740061] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[ 2600.170786] [vunguyen] ov10635_s_stream [+] 0
[ 2600.177096] [VuNguyen] ov10635_power_off 1918
[ 2649.894820] tegra-vi4 15700000.vi: Status:  5 channel:00 frame:0000
[ 2649.894822] [vunguyen] ov10635_s_stream [+] 0
[ 2649.895784] [VuNguyen] ov10635_power_off 1918
[ 2649.909956] tegra-vi4 15700000.vi:          timestamp sof 2659566032512 eof 2659566150432 data 0x00000001
[ 2649.920825] tegra-vi4 15700000.vi:          capture_id 1719 stream  0 vchan  0
[ 2649.929344] tegra-vi4 15700000.vi: Status:  5 channel:01 frame:0000
[ 2649.935653] tegra-vi4 15700000.vi:          timestamp sof 2659575746144 eof 2659581550432 data 0x00000002
[ 2649.945228] tegra-vi4 15700000.vi:          capture_id 1716 stream  2 vchan  0
[ 2649.945251] Unable to handle kernel paging request at virtual address ffffffbe41c21b40

Thanks

9

Hi DaneLLL,

FYI,
If I opened 3 applications for 3 cameras on 3 terminal tabs and terminate one by one, everything would be OK.

10

Updated:

If I used thread as below it would be not stable.

int
main_thread(int dummy)
{
    context_t ctx;
    int error = 0;

    set_defaults(&ctx);

    /*CHECK_ERROR(parse_cmdline(&ctx, argc, argv), cleanup,
            "Invalid options specified");*/

    CHECK_ERROR(init_components(&ctx), cleanup,
            "Failed to initialize v4l2 components");

    CHECK_ERROR(prepare_buffers(&ctx), cleanup,
            "Failed to prepare v4l2 buffs");

    CHECK_ERROR(start_stream(&ctx), cleanup,
            "Failed to start streaming");

    CHECK_ERROR(start_capture(&ctx), cleanup,
            "Failed to start capturing")

    CHECK_ERROR(stop_stream(&ctx), cleanup,
            "Failed to stop streaming");

cleanup:
    if (ctx.cam_fd > 0)
        close(ctx.cam_fd);

    //if (ctx.renderer != NULL)
    //    delete ctx.renderer;

    if (ctx.egl_display && !eglTerminate(ctx.egl_display))
        printf("Failed to terminate EGL display connection\n");

    /*if (ctx.conv != NULL)
    {
        if (ctx.conv->isInError())
        {
            printf("Video converter is in error\n");
            error = 1;
        }
        delete ctx.conv;
    }*/

    if (ctx.g_buff != NULL)
    {
        for (unsigned i = 0; i < V4L2_BUFFERS_NUM; i++)
            if (ctx.g_buff[i].dmabuff_fd)
                NvBufferDestroy(ctx.g_buff[i].dmabuff_fd);
        free(ctx.g_buff);
    }

    while(!ctx.conv_output_plane_buf_queue->empty())
    {
        ctx.conv_output_plane_buf_queue->pop();
    }

    delete ctx.conv_output_plane_buf_queue;

    if (error)
        printf("App run failed\n");
    else
        printf("App run was successful\n");

    return -error;
}

int main (void) {
    std::thread encoderThread;
    int dummy = 0;

    encoderThread = std::thread (&main_thread, dummy);
    encoderThread.join();

    return 0;
}

Even though I run with only one camera but with thread it’s easy to see below debug log (rarely seen without thread using)

[ 3373.109569] tegra-vi4 15700000.vi: Status:  5 channel:00 frame:0000
[ 3373.115980] tegra-vi4 15700000.vi:          timestamp sof 3382772358144 eof 3382772625440 data 0x00000001
[ 3373.125637] tegra-vi4 15700000.vi:          capture_id 10502 stream  0 vchan  0
11

Updated:

And when I started the application, sometime I faced this problem

[ 1454.668726] [vunguyen] ov10635_s_stream, i2c addr: 0x48
[ 1455.681792] tegra-vi4 15700000.vi: PXL_SOF syncpt timeout! err = -11
[ 1456.685754] tegra-vi4 15700000.vi: PXL_SOF syncpt timeout! err = -11
[ 1457.689822] tegra-vi4 15700000.vi: PXL_SOF syncpt timeout! err = -11
[ 1457.699388] [vunguyen] ov10635_s_stream [+] 0
[ 1458.693825] tegra-vi4 15700000.vi: PXL_SOF syncpt timeout! err = -11
[ 1458.700611] [vunguyen] ov10635_s_stream [+] 0
[ 1458.705130] nvcsi 150c0000.nvcsi: csi4_stream_check_status (2) ERROR_STATUS2VI_VC0 = 0x00000004
[ 1458.714026] nvcsi 150c0000.nvcsi: csi4_stream_check_status (2) INTR_STATUS 0x00010006
[ 1458.721967] nvcsi 150c0000.nvcsi: csi4_stream_check_status (2) ERR_INTR_STATUS 0x00010006
12

Hi Vu,
There is another sample for v4l2 source tegra_multimedia_api\samples\v4l2cuda

Could you also try it?

13

Hi DaneLLL,

Do you mean tegra_multimedia_api\samples\12_camera_v4l2_cuda ?
I’m working on it.

Thanks
Vu Nguyen

14

No. I mean v4l2cuda at

tegra_multimedia_api\samples\v4l2cuda
15

Hi DaneLLL,

It look like working but I want to use Nvidia Video Converter and Video Encoder as well also.

Thanks,
Vu Nguyen

16

Hi Vu,
Please try to put a delay at

// Push nv_buffer into conv output queue for conversion
            pthread_mutex_lock(&ctx->queue_lock);
            ctx->conv_output_plane_buf_queue->push(&ctx->g_buff[v4l2_buf.index]);
            pthread_cond_broadcast(&ctx->queue_cond);
            pthread_mutex_unlock(&ctx->queue_lock);

<b>usleep(100000);</b>

            // Enqueue camera buff
            // It might be more reasonable to wait for the completion of
            // VIC processing before enqueue current buff. But VIC processing
            // time is far less than camera frame interval, so we probably
            // don't need such synchonization.
            if (ioctl(ctx->cam_fd, VIDIOC_QBUF, &v4l2_buf))
                ERROR_RETURN("Failed to queue camera buffers: %s (%d)",
                        strerror(errno), errno);

It is possible to have race condition between VIC and camera capture here.

17

Hi DaneLLL,

I tried but after several time running and stop, it still happen as below

[  400.295292] tegra-vi4 15700000.vi:          timestamp sof 409944985632 eof 409961203008 data 0x00000002
[  400.304754] tegra-vi4 15700000.vi:          capture_id 159 stream  2 vchan  0
[  400.311982] tegra-vi4 15700000.vi: Status:  5 channel:02 frame:0000
[  400.318358] tegra-vi4 15700000.vi:          timestamp sof 409965331040 eof 409965509248 data 0x00000004
[  400.327815] tegra-vi4 15700000.vi:          capture_id 194 stream  4 vchan  0
[  400.335047] Unable to handle kernel NULL pointer dereference at virtual address 00000000
[  400.335534] [vunguyen] ov10635_s_stream [+] 0
[  400.336895] [VuNguyen] ov10635_power_off 1918
[  400.350831] [VuNguyen] ov10635_power_off 1918
[  400.356348] pgd = ffffffc0014a5000
[  400.359770] [00000000] *pgd=000000026cdd6003, *pud=000000026cdd6003, *pmd=000000026cdd7003, *pte=00e8000003881707
[  400.370229] Internal error: Oops: 96000005 [#1] PREEMPT SMP
[  400.375792] Modules linked in:
[  400.378852] CPU: 4 PID: 30 Comm: kworker/4:0 Not tainted 4.4.38 #135
[  400.385191] Hardware name: quill (DT)
[  400.385900] [VuNguyen] ov10635_power_off 1918
[  400.393197] Workqueue: events tegra_channel_status_worker
[  400.398606] task: ffffffc1ec633200 ti: ffffffc1ec7c8000 task.ti: ffffffc1ec7c8000
[  400.406088] PC is at vi_notify_channel_reset+0x18/0x88
[  400.411221] LR is at tegra_channel_handle_error+0x5c/0xa0
[  400.416614] pc : [<ffffffc00090a0e4>] lr : [<ffffffc0007a06a0>] pstate: 60000045
[  400.423998] sp : ffffffc1ec7cbcf0
[  400.427309] x29: ffffffc1ec7cbcf0 x28: ffffffc1f5daec18 
[  400.432639] x27: ffffffc1f5daeee0 x26: ffffffc0013bb000 
[  400.437965] x25: 0000000000000000 x24: 0000000000000000 
[  400.443288] x23: ffffffc1f5db4400 x22: 0000000000000001 
[  400.448609] x21: 0000000000000000 x20: ffffffc1ec32d018 
[  400.453930] x19: 0000000000000000 x18: 0000000000000032 
[  400.459250] x17: 0000007f931711f8 x16: ffffffc000b54a60 
[  400.464572] x15: ffffffc000b54a60 x14: 0000000000000001 
[  400.469899] x13: 00000035f8bd1ee4 x12: 000000000006d908 
[  400.475222] x11: 000000000004b3fd x10: 000000000004b3fd 
[  400.480544] x9 : 0000000000000000 x8 : ffffffc1f5d9e570 
[  400.485867] x7 : 0000000000000018 x6 : ffffffc001443510 
[  400.491193] x5 : ffffffc1ec7cbd60 x4 : ffffffc1f5db0200 
[  400.496518] x3 : 0000000000000002 x2 : ffffffc1ec7cbd50 
[  400.501844] x1 : ffffffc1de097000 x0 : 0000000000000001

Sometime it looked more stable if I enable renderer

// Render the frame into display
    if (v4l2_buf->m.planes[0].bytesused)
        ctx->renderer->render(buffer->planes[0].fd);

Thanks
Vu Nguyen

18

Hi DaneLLL,

After I tried to increase num of v4l2 buffer from 4 to 10, issue gone. Thank for your support.

#define V4L2_BUFFERS_NUM    10                                                                         
#define MAX_QUEUE_BUFER     100

#define INFO(fmt, ...) \
    if (ctx->enable_verbose) \
        printf("INFO: %s(): (line:%d) " fmt "\n", __FUNCTION__, __LINE__, ##__VA_ARGS__);             

#define WARN(fmt, ...) \
        printf("WARN: %s(): (line:%d) " fmt "\n", __FUNCTION__, __LINE__, ##__VA_ARGS__);             

#define CHECK_ERROR(cond, label, fmt, ...) \                                                          
    if (!cond) { \
        error = 1; \
        printf("ERROR: %s(): (line:%d) " fmt "\n", __FUNCTION__, __LINE__, ##__VA_ARGS__); \          
        goto label; \                                                                                 
    }

VuNguyen