Re: [PATCH] Revert "scsi: libsas: Fix exp-attached device scan after probe failure scanned in again after probe failed"

[John Garry <john.g.garry@oracle.com>]

On 27/11/2025 00:59, yangxingui wrote:
> Kindly ping for upstream.
> 
> On 2025/10/21 15:34, Xingui Yang wrote:

Your reasons for revert is light on details.

>> This reverts commit ab2068a6fb84751836a84c26ca72b3beb349619d.
>>
>> As the disk may fall into an abnormal loop of probe when it fails to 
>> probe
>> due to physical reasons and cannot be repaired.

So for a faulty disk we get into a indefinite loop, right?

What about case where this was helping before?

>>
>> Signed-off-by: Xingui Yang <yangxingui@huawei.com>
>> ---
>>   drivers/scsi/libsas/sas_internal.h | 14 --------------
>>   1 file changed, 14 deletions(-)
>>
>> diff --git a/drivers/scsi/libsas/sas_internal.h b/drivers/scsi/libsas/ 
>> sas_internal.h
>> index 03d6ec1eb970..85948963fb97 100644
>> --- a/drivers/scsi/libsas/sas_internal.h
>> +++ b/drivers/scsi/libsas/sas_internal.h
>> @@ -145,20 +145,6 @@ static inline void sas_fail_probe(struct 
>> domain_device *dev, const char *func, i
>>           func, dev->parent ? "exp-attached" :
>>           "direct-attached",
>>           SAS_ADDR(dev->sas_addr), err);
>> -
>> -    /*
>> -     * If the device probe failed, the expander phy attached address
>> -     * needs to be reset so that the phy will not be treated as flutter
>> -     * in the next revalidation
>> -     */
>> -    if (dev->parent && !dev_is_expander(dev->dev_type)) {
>> -        struct sas_phy *phy = dev->phy;
>> -        struct domain_device *parent = dev->parent;
>> -        struct ex_phy *ex_phy = &parent->ex_dev.ex_phy[phy->number];
>> -
>> -        memset(ex_phy->attached_sas_addr, 0, SAS_ADDR_SIZE);
>> -    }
>> -
>>       sas_unregister_dev(dev->port, dev);
>>   }
>>
>