Re: Regression caused by "eeprom: at24: Probe for DDR3 thermal sensor in the SPD case" - "sysfs: cannot create duplicate filename"

["Krzysztof Olędzki" <ole@ans.pl>]

On 03.08.2024 at 10:19, Heiner Kallweit wrote:
> On 23.07.2024 16:12, Krzysztof Olędzki wrote:
>> On 06.07.2024 at 18:42, Krzysztof Olędzki wrote:
>>> On 02.07.2024 at 13:25, Heiner Kallweit wrote:
>>>> On 23.06.2024 20:47, Krzysztof Olędzki wrote:
>>>>> Hi,
>>>>>
>>>>> After upgrading kernel to Linux 6.6.34 on one of my systems, I noticed "sysfs: cannot create duplicate filename" and i2c registration errors in dmesg, please see below.
>>>>>
>>>>> This seems to be related to https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-6.6.y&id=4d5ace787273cb159bfdcf1c523df957938b3e42 - reverting the change fixes the problem.
>>>
>>> <CUT>
>>>
>>>>
>>>> Could you please test whether the attached two experimental patches fix the issue for you?
>>>> They serialize client device instantiation per I2C adapter, and include the client device
>>>> name in the check whether a bus address is busy.
>>>
>>> Sadly, they crash the kernel.
>>>
>>> I will get serial console attached there next week, so will be able to capture the full crash.
>>> For now, I was able to obtain a photo. I'm very sorry for the quality, just wanted to provide
>>> something for now.
>>
>> Sorry it took me so long - my attempts to coordinate setting up serial console
>> were not successful, so it had to wait for me to go there in person...
>>
>> I have attached complete dmesg, summary:
>>
>> [   10.905953] rtmutex deadlock detected
>> [   10.909959] WARNING: CPU: 5 PID: 83 at kernel/locking/rtmutex.c:1642 __rt_mutex_slowlock_locked.constprop.0+0x10f/0x1a5
>> [   10.920961] CPU: 5 PID: 83 Comm: kworker/u16:3 Not tainted 6.6.34-o5 #1
>> [   10.929970] Hardware name: Dell Inc. PowerEdge T110 II/0PM2CW, BIOS 2.10.0 05/24/2018
>> [   10.938954] Workqueue: events_unbound async_run_entry_fn
>>
>>
>> [   11.336954] BUG: scheduling while atomic: kworker/u16:3/83/0x00000002
>> [   11.342953] Preemption disabled at:
>> [   11.342953] [<0000000000000000>] 0x0
>> [   11.350953] CPU: 5 PID: 83 Comm: kworker/u16:3 Tainted: G        W          6.6.34-o5 #1
>> [   11.361954] Hardware name: Dell Inc. PowerEdge T110 II/0PM2CW, BIOS 2.10.0 05/24/2018
>> [   11.369953] Workqueue: events_unbound async_run_entry_fn
>>
> Thanks a lot for the comprehensive info. Reason for the deadlock is that calls to
> i2c_new_client_device() can be nested. So another solution approach is needed.
> I'd appreciate if you could test also the new version below.

The patch did not apply cleanly for Linux-6.6, so I had to tweak it a little
bit for the include/linux/i2c.h part, but it does seem to work. Everything
gets detected and there are no warning / errors:


[    8.311414] i2c i2c-12: 4/4 memory slots populated (from DMI)
[    8.314112] at24 12-0050: 256 byte spd EEPROM, read-only
[    8.314856] i2c i2c-12: Successfully instantiated SPD at 0x50
[    8.317513] at24 12-0051: 256 byte spd EEPROM, read-only
[    8.318252] i2c i2c-12: Successfully instantiated SPD at 0x51
[    8.320909] at24 12-0052: 256 byte spd EEPROM, read-only
[    8.322126] i2c i2c-12: Successfully instantiated SPD at 0x52
[    8.325538] at24 12-0053: 256 byte spd EEPROM, read-only
[    8.326789] i2c i2c-12: Successfully instantiated SPD at 0x53

# sensors|grep -A2 jc42
jc42-i2c-12-19
Adapter: SMBus I801 adapter at 3000
temp1:        +36.5°C  (low  =  +0.0°C)
--
jc42-i2c-12-1b
Adapter: SMBus I801 adapter at 3000
temp1:        +35.0°C  (low  =  +0.0°C)
--
jc42-i2c-12-1a
Adapter: SMBus I801 adapter at 3000
temp1:        +36.0°C  (low  =  +0.0°C)
--
jc42-i2c-12-18
Adapter: SMBus I801 adapter at 3000
temp1:        +36.5°C  (low  =  +0.0°C)

Feel free to add:
Tested-by: Krzysztof Piotr Oledzki <ole@ans.pl>

Thanks,
 Krzysztof