Part Number: AM62A7
Hello,
I am currently working on upgrading a custom AM62A-based board to the SDK 11.1 release. During boot testing, I experienced a rare warning error that I believe is related to the wave5 driver. A snippet of the issue can be seen below:
[ OK ] Found device /dev/mmcblk0p1.
Starting File System Check on /dev/mmcblk0p1...
[ OK ] Started Simple Network Management Protocol (SNMP) Daemon..
[ OK ] Found device /dev/fb0.
Starting Start psplash boot splash screen...
[ OK ] Stopped Virtual Console Setup.
[ OK ] Started Start psplash boot splash screen.
[ OK ] Started Start psplash-systemd progress communication helper.
Starting Virtual Console Setup...
[ 9.710311] ------------[ cut here ]------------
[ 9.717473] WARNING: CPU: 2 PID: 748 at /kernel/kthread.c:1430 kthread_destroy_worker+0x84/0x98
[ 9.726214] Modules linked in: rpmsg_ctrl rpmsg_char snd_soc_tlv320aic26 onboard_usb_dev crct10dif_ce snd_soc_simple_card snd_soc_simple_card_utils e5010_jpeg_enc v4l2_jpeg pwm_omap_dmtimer display_connector wave5(-) videobuf2_dma_contig v4l2_mem2mem videobuf2_v4l2 rtc_ti_k3 k3_j72xx_bandgap rti_wdt videobuf2_memops videobuf2_common ti_k3_dsp_remoteproc tidss overlay videodev ti_k3_common ti_k3_r5_remoteproc snd_soc_davinci_mcasp drm_dma_helper snd_soc_ti_udma mcrc64 snd_soc_ti_edma sa2ul mc snd_soc_ti_sdma drm_display_helper ti_tfp410 drm_kms_helper omap_hwspinlock ltc2945 spi_omap2_mcspi omap_mailbox pwm_tiehrpwm at24 cfg80211 bluetooth ecdh_generic ecc rfkill cryptodev(O) fuse drm drm_panel_orientation_quirks backlight ipv6
[ 9.790760] CPU: 2 UID: 0 PID: 748 Comm: rmmod Tainted: G O 6.12.43-g-g897da7ab5fc0 #1
[ 9.790792] Tainted: [O]=OOT_MODULE
[ 9.790795] Hardware name: Critical Link MitySOM-AM62A (DT)
[ 9.790800] pstate: 80000005 (Nzcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 9.790808] pc : kthread_destroy_worker+0x84/0x98
[ 9.790831] lr : kthread_destroy_worker+0x30/0x98
[ 9.790838] sp : ffff8000833abc20
[ 9.790842] x29: ffff8000833abc20 x28: ffff00000538a180 x27: 0000000000000000
[ 9.790857] x26: 0000000000000000 x25: 0000000000000000 x24: 0000000000000000
[ 9.790868] x23: ffff000000662490 x22: ffff000001b40c90 x21: ffff80007a1bc5e0
[ 9.790879] x20: ffff000008538000 x19: ffff00000b275500 x18: 0000000000000000
[ 9.790890] x17: ffff7ffffc606000 x16: ffff800081cd8000 x15: 0000000000000000
[ 9.790901] x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
[ 9.790912] x11: 00000000000000c0 x10: 0000000000000a10 x9 : ffff800080ed85a8
[ 9.790923] x8 : ffff00000538abf0 x7 : 00000001f650a7e2 x6 : 0000000000000000
[ 9.790934] x5 : 0000000000000003 x4 : ffff000008538028 x3 : 0000000000000000
[ 9.790944] x2 : 0000000000000001 x1 : ffff000001fdc250 x0 : ffff00000b275508
[ 9.790956] Call trace:
[ 9.790960] kthread_destroy_worker+0x84/0x98
[ 9.790970] wave5_vpu_remove+0xd0/0xe8 [wave5]
[ 9.791003] platform_remove+0x30/0x58
[ 9.791016] device_remove+0x54/0x90
[ 9.791024] device_release_driver_internal+0x1d4/0x238
[ 9.791032] driver_detach+0x54/0xa8
[ 9.791040] bus_remove_driver+0x74/0xd0
[ 9.791047] driver_unregister+0x38/0x70
[ 9.791055] platform_driver_unregister+0x1c/0x30
[ 9.791064] wave5_vpu_driver_exit+0x18/0x1358 [wave5]
[ 9.791084] __arm64_sys_delete_module+0x1c8/0x2b8
[ 9.791094] invoke_syscall+0x50/0x120
[ 9.791105] el0_svc_common.constprop.0+0xc8/0xf0
[ 9.791115] do_el0_svc+0x24/0x38
[ 9.791123] el0_svc+0x28/0x98
[ 9.791134] el0t_64_sync_handler+0x120/0x130
[ 9.791143] el0t_64_sync+0x190/0x198
[ 9.791153] ---[ end trace 0000000000000000 ]---
[ 9.820194] vdec 30210000.video-codec: Runtime PM usage count underflow!
[ 9.908506] vdec 30210000.video-codec: error -ENXIO: IRQ index 0 not found
[ 10.024628] vdec 30210000.video-codec: failed to get irq resource, falling back to polling
[ OK ] Finished File System Check on /dev/[ 10.034012] vdec 30210000.video-codec: OPP table not found in device tree
mmcblk0p1.
[ 10.055246] vdec 30210000.video-codec: Added wave5 driver with caps: 'ENCODE' 'DECODE'
[ 10.063438] vdec 30210000.video-codec: Product Code: 0x521c
[ 10.069674] vdec 30210000.video-codec: Firmware Revision: 363254
Mounting /run/media/mmcblk0p1...
Starting Save/Restore Sound Card State...
Starting File System Check on /dev/mmcblk1p1...I believe this is an issue with the wave5 driver because of the given context of this dump occuring near where we modprobe the wave5 driver and because the wave5_vpu_remove() function in drivers/media/platform/chips-media/wave5/wave5-vpu.c calls kthread_destroy_worker(). The issue could be related to a race condition and this happening too soon; which it then tries to destroy something that was never fully created/initialized causing the warning dump to occur. I am also aware that the two messages outputted directly after the trace are not an error but a warning that hr timer polling is being used as AM62A has no interrupt routed (via this E2E).
Is there any insight into why this issue occurs and a possible solution?
