Re: [PATCH] net: usb: lan78xx: Fix lost EEPROM read timeout error(-ETIMEDOUT) in lan78xx_read_raw_eeprom

[Bhanu Seshu Kumar Valluri]

On 01/10/25 13:54, Oleksij Rempel wrote:
> On Wed, Oct 01, 2025 at 01:40:56PM +0530, Bhanu Seshu Kumar Valluri wrote:
>> On 01/10/25 13:12, Oleksij Rempel wrote:
>>> Hi,
>>>
>>> On Wed, Oct 01, 2025 at 10:07:21AM +0530, Bhanu Seshu Kumar Valluri wrote:
>>>> On 01/10/25 06:09, Jakub Kicinski wrote:
>>>>> On Tue, 30 Sep 2025 14:19:02 +0530 Bhanu Seshu Kumar Valluri wrote:
>>>>>> +	if (dev->chipid == ID_REV_CHIP_ID_7800_) {
>>>>>> +		int rc = lan78xx_write_reg(dev, HW_CFG, saved);
>>>>>> +		/* If USB fails, there is nothing to do */
>>>>>> +		if (rc < 0)
>>>>>> +			return rc;
>>>>>> +	}
>>>>>> +	return ret;
>>>>>
>>>>> I don't think you need to add and handle rc here separately?
>>>>> rc can only be <= so save the answer to ret and "fall thru"?
>>>>
>>>> The fall thru path might have been reached with ret holding EEPROM read timeout
>>>> error status. So if ret is used instead of rc it might over write the ret with 0 when 
>>>> lan78xx_write_reg returns success and timeout error status would be lost.
>>>
>>> Ack, I see. It may happen if communication with EEPROM will fail. The same
>>> would happen on write path too. Is it happened with real HW or it is
>>> some USB emulation test? For me it is interesting why EEPROM is timed
>>> out.
>>
>> The sysbot's log with message "EEPROM read operation timeout" confirms that EEPROM read
>> timeout occurring. I tested the same condition on EVB-LAN7800LC by simulating 
>> timeout during probe.
> 
> Do you simulating timeout during probe by modifying the code, or it is
> real HW issue?
> 

On my real hardware timeout didn't occur. So I simulated it once by modifying the code
to confirm the BUG. The BUG has occurred confirming syzbot finding.

Thanks.