Kernel Oops in __blk_mq_all_tag_iter() using Raxda CM5

Kernel Oops in __blk_mq_all_tag_iter() using Raxda CM5

[Rick Koch]

Hello linux-block,

I am seeing a kernel Oops in early booting up of a Radxa CM5 board
running Armbian
on a 6.12.7 kernel.

I had reported a very similar Oops back on 2024-10-11 (Oct 11, 2024)
 https://lore.kernel.org/all/CANa58ee6EeT9V7Q=epoZdqYw3sLh1CZGNWqJ0UcKMp6eRfcd+Q@mail.gmail.com/T/

Which you had provided me with a patch that fixed the issue and has
not occurred again.

Here is the current Oops I am experiencing, maybe 1 out of every 10 boots.
It does lockup after this, requiring me to re-power/boot the board.

Starting kernel ...

[    1.582430] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
[    1.582984] Modules linked in:
[    1.583256] CPU: 6 UID: 0 PID: 46 Comm: cpuhp/6 Not tainted
6.12.7-edge-rockchip64 #7
[    1.583942] Hardware name: Saturn SDR, Radxa CM5, 8inch LCD/Touch panel (DT)
[    1.584557] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    1.585167] pc : __blk_mq_all_tag_iter+0x3c/0x274
[    1.585584] lr : blk_mq_all_tag_iter+0x14/0x20
[    1.585977] sp : ffff800082563c90
[    1.586268] x29: ffff800082563cc0 x28: ffff800081dd9c28 x27: 0000000000000000
[    1.586896] x26: 0000000000000092 x25: 0000000000000000 x24: ffff8000807ee3bc
[    1.587522] x23: ffff800082563d38 x22: 0000000000000004 x21: ffff800081dd9e70
[    1.588148] x20: ffff000101541400 x19: ffff8000807ee3bc x18: ffffffffffffffff
[    1.588775] x17: ffff80027d4f4000 x16: ffff8000821c8000 x15: 0000000000000000
[    1.589402] x14: ffff00010086e980 x13: 00000000000003e8 x12: 0000000000000001
[    1.590028] x11: 071c71c71c71c71c x10: 0000000000000ae0 x9 : ffff800082563a00
[    1.590655] x8 : ffff00010086f440 x7 : ffff0002feeecfc0 x6 : ffff800081df5558
[    1.591281] x5 : ffff800081ddd000 x4 : ffff00010086e900 x3 : 0000000000000004
[    1.591907] x2 : ffff800082563d38 x1 : ffff8000807ee3bc x0 : 0000000000000000
[    1.592534] Call trace:
[    1.592751]  __blk_mq_all_tag_iter+0x3c/0x274
[    1.593136]  blk_mq_all_tag_iter+0x14/0x20
[    1.593497]  blk_mq_hctx_notify_offline+0x164/0x230
[    1.593925]  cpuhp_invoke_callback+0x3ac/0x6a8
[    1.594320]  cpuhp_thread_fun+0xc4/0x1b0
[    1.594669]  smpboot_thread_fn+0x200/0x224
[    1.595032]  kthread+0x110/0x114
[    1.595320]  ret_from_fork+0x10/0x20
[    1.595639] Code: aa0203f7 f9427480 f90017e0 d2800000 (b9400724)
[    1.596172] ---[ end trace 0000000000000000 ]---

Re: Kernel Oops in __blk_mq_all_tag_iter() using Raxda CM5

[Ming Lei]

On Thu, Feb 06, 2025 at 10:29:34AM -0500, Rick Koch wrote:
> Hello linux-block,
> 
> I am seeing a kernel Oops in early booting up of a Radxa CM5 board running
> Armbian
> on a 6.12.7 kernel.
> 
> I had reported a very similar Oops back on 2024-10-11 (Oct 11, 2024)
> 
> https://lore.kernel.org/all/CANa58ee6EeT9V7Q=epoZdqYw3sLh1CZGNWqJ0UcKMp6eRfcd+Q@mail.gmail.com/T/
> 
> Which you had provided me with a patch that fixed the issue and has not
> occurred again.
> 
> Here is the current Oops I am experiencing, maybe 1 out of every 10 boots.
> It does lockup after this, requiring me to re-power/boot the board.
> 
> Starting kernel ...
> 
> [    1.582430] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
> [    1.582984] Modules linked in:
> [    1.583256] CPU: 6 UID: 0 PID: 46 Comm: cpuhp/6 Not tainted
> 6.12.7-edge-rockchip64 #7
> [    1.583942] Hardware name: Saturn SDR, Radxa CM5, 8inch LCD/Touch panel
> (DT)
> [    1.584557] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS
> BTYPE=--)
> [    1.585167] pc : __blk_mq_all_tag_iter+0x3c/0x274
> [    1.585584] lr : blk_mq_all_tag_iter+0x14/0x20

Can you share where '__blk_mq_all_tag_iter+0x3c' points to?

Also running kernel with KASAN enabled often provides more helpful hint.



Thanks,
Ming