AM62A7-Q1: wave5: H.265 encoder: unreferenced kernel paging request while encoding an RTP stream

Part Number: AM62A7-Q1

Hi,

I'm currently running a kernel 6.12.35. wave5 code on tag 11.01.04. I see the following kernel error, resulting in the system hanging.

[  609.727744] Unable to handle kernel paging request at virtual address dead000000000108
[  609.735693] Mem abort info:
[  609.738490]   ESR = 0x0000000096000044
[  609.742241]   EC = 0x25: DABT (current EL), IL = 32 bits
[  609.747553]   SET = 0, FnV = 0
[  609.750608]   EA = 0, S1PTW = 0
[  609.753744]   FSC = 0x04: level 0 translation fault
[  609.758611] Data abort info:
[  609.761484]   ISV = 0, ISS = 0x00000044, ISS2 = 0x00000000
[  609.766958]   CM = 0, WnR = 1, TnD = 0, TagAccess = 0
[  609.771998]   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
[  609.777298] [dead000000000108] address between user and kernel address ranges
[  609.784419] Internal error: Oops: 0000000096000044 [#1] PREEMPT SMP
[  609.790675] Modules linked in: rpmsg_ctrl rpmsg_char tpm_tis_spi inv_icm42600_spi tpm_tis_core panel_semtech_gs2962 inv_icm42600 inv_sensors_timestamp kfifo_buf
 snd_soc_simple_card crct10dif_ce snd_soc_simple_card_utils snd_soc_adau7002 e5010_jpeg_enc v4l2_jpeg rtc_ti_k3 k3_j72xx_bandgap ti_k3_dsp_remoteproc ti_k3_r5_remo
teproc ti_k3_common j721e_csi2rx wave5 snd_soc_davinci_mcasp cdns_csi2rx snd_soc_ti_udma videobuf2_dma_contig mcrc64 v4l2_mem2mem sa2ul videobuf2_v4l2 ar0823 snd_s
oc_ti_edma v4l2_fwnode videobuf2_memops omap_hwspinlock v4l2_async videobuf2_common snd_soc_ti_sdma videodev mc omap_mailbox at24 ina238 tps65219_pwrbutton spi_oma
p2_mcspi cfg80211 rfkill fuse ipv6
[  609.850492] CPU: 1 UID: 0 PID: 25 Comm: kworker/1:0 Not tainted 6.12.35-g16f5a07f-01131-g2a4b06c4e29a #1
[  609.859958] Hardware name: Benchmark Electronics 4K Camera Proto B (DT)
[  609.866558] Workqueue: events v4l2_m2m_device_run_work [v4l2_mem2mem]
[  609.873013] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[  609.879961] pc : v4l2_m2m_buf_remove_by_idx+0x84/0xe8 [v4l2_mem2mem]
[  609.886313] lr : v4l2_m2m_buf_remove_by_idx+0x28/0xe8 [v4l2_mem2mem]
[  609.892663] sp : ffff800081a4bbd0
[  609.895968] x29: ffff800081a4bbd0 x28: 0000000000000000 x27: 000000000007f800
[  609.903096] x26: 00000000aefd2800 x25: 0000000000000780 x24: ffff0000014efdc8
[  609.910224] x23: ffff0000014efc28 x22: ffff0000014efc00 x21: ffff0000014eff60
[  609.917351] x20: ffff0000014efdc8 x19: ffff000001daf800 x18: 0000000000000000
[  609.924478] x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000002
[  609.931605] x14: 0000000000000800 x13: 00000000000001e6 x12: 0000000000000000
[  609.938732] x11: 0000000000000000 x10: 00000000000009d0 x9 : ffff800081a4bcc0
[  609.945859] x8 : ffff800081a4bc88 x7 : 0000000000000000 x6 : ffff0000014eff50
[  609.952986] x5 : dead000000000100 x4 : dead000000000100 x3 : dead000000000122
[  609.960113] x2 : 0000000000000100 x1 : 0000000000000000 x0 : ffff000001dafc00
[  609.967242] Call trace:
[  609.969678]  v4l2_m2m_buf_remove_by_idx+0x84/0xe8 [v4l2_mem2mem]
[  609.975685]  start_encode+0x28c/0x554 [wave5]
[  609.980063]  wave5_vpu_enc_device_run+0x10c/0x230 [wave5]
[  609.985466]  v4l2_m2m_try_run+0x84/0x140 [v4l2_mem2mem]
[  609.990692]  v4l2_m2m_device_run_work+0x14/0x20 [v4l2_mem2mem]
[  609.996522]  process_one_work+0x148/0x28c
[  610.000532]  worker_thread+0x2d0/0x3d8
[  610.004277]  kthread+0x110/0x114
[  610.007500]  ret_from_fork+0x10/0x20
[  610.011074] Code: 91100060 aa0303f3 d2802002 f9424863 (f9000483)
[  610.017154] ---[ end trace 0000000000000000 ]---
[  610.021760] note: kworker/1:0[25] exited with irqs disabled
[  610.027441] note: kworker/1:0[25] exited with preempt_count 1
[  612.029229] ti-sci 44043000.system-controller: Mbox timedout in resp(caller: ti_sci_cmd_get_device_exclusive+0x18/0x24)
[  612.040102] ti-sci 44043000.system-controller: Mbox send fail -110

Any idea how to fix this?

Regards,

Bas Vermeulen

  • Hi Bas, 

    Can you provide us a way to reproduce the issue on our EVM? Also, does this package have the updated Gstreamer version that you are testing with(1.24.13)?

    Best Regards

    Suren

  • Hi Suren,

    Yes. this is with the updated gstreamer 1.24.13 running on there, and the modified VPAC/TIDL firmware. I currently don't have a way to reproduce it on the EVM, I just saw it in a longer test on our custom hardware. I'll see if I can find a way to reproduce on EVM, but am not sure when I'll have enough time to do so.

    We are encoding two streams, one 4K stream and a 1920x1088 stream, at 30 frames per second. The camera is pointed at the ceiling, so nothing much happening there.

    I'll try to find some time to reproduce on EVM once my SDK update is done.

    Regards,

    Bas Vermeulen