Re: x15: kernel crash: LR is at sysc_enable_opt_clocks

[Arnd Bergmann]

On Wed, Sep 14, 2022, at 2:46 PM, Naresh Kamboju wrote:
> Following kernel crash noticed on arm TI x15 device while booting the
> mainline kernel 6.0.0-rc4.
>
> Reported-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx>
> [    3.059600] random: crng init done
> [    3.059631] 8<--- cut here ---
> [    3.066101] Unable to handle kernel paging request at virtual
> address adacafae
> [    3.073394] [adacafae] *pgd=00000000
> [    3.076995] Internal error: Oops: 5 [#1] SMP ARM


> [    3.781494]  clk_enable from sysc_enable_opt_clocks+0x4c/0xa0
> [    3.787261]  sysc_enable_opt_clocks from sysc_probe+0xda0/0x1598
> [    3.793304]  sysc_probe from platform_probe+0x64/0xc0

So it's getting an invalid clk pointer in sysc_enable_opt_clocks.

> [    3.798400]  platform_probe from really_probe+0xe8/0x41c
> [    3.803741]  really_probe from __driver_probe_device+0xa8/0x20c
> [    3.809692]  __driver_probe_device from driver_probe_device+0x38/0xc8
> [    3.816192]  driver_probe_device from __device_attach_driver+0xb4/0x130
> [    3.822845]  __device_attach_driver from bus_for_each_drv+0x84/0xc8
> [    3.829162]  bus_for_each_drv from __device_attach+0xb0/0x210
> [    3.834930]  __device_attach from bus_probe_device+0x8c/0x94
> [    3.840637]  bus_probe_device from device_add+0x3ec/0x924
> [    3.846069]  device_add from of_platform_device_create_pdata+0x98/0xc8
> [    3.852630]  of_platform_device_create_pdata from
> of_platform_bus_create+0x200/0x4d8
> [    3.860412]  of_platform_bus_create from of_platform_populate+0x9c/0x138
> [    3.867156]  of_platform_populate from simple_pm_bus_probe+0xac/0xd0
> [    3.873565]  simple_pm_bus_probe from platform_probe+0x64/0xc0
> [    3.879425]  platform_probe from really_probe+0xe8/0x41c

It looks strange to recursively go through simple_pm_bus_probe()
here four times before getting to the ti,sysc driver.

>
> Full test log link,
>   - https://lkft.validation.linaro.org/scheduler/job/5508159
>

What is the easiest way to find out how long this job
has been failing, and what the last successful build
was?

       Arnd