Re: [PATCH] scsi: take module reference during async scan

[Tomas Henzl <thenzl@redhat.com>]

On 9/7/20 6:57 PM, Bart Van Assche wrote:
> On 2020-09-07 08:47, Tomas Henzl wrote:
>> During an async scan the driver shost->hostt structures are used,
>> that may cause issues when the driver is removed at that time.
>> As protection take the module reference.
>>
>> Signed-off-by: Tomas Henzl <thenzl@redhat.com>
>> ---
>>  drivers/scsi/scsi_scan.c | 8 ++++++++
>>  1 file changed, 8 insertions(+)
>>
>> diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
>> index f2437a757..c9cc0862c 100644
>> --- a/drivers/scsi/scsi_scan.c
>> +++ b/drivers/scsi/scsi_scan.c
>> @@ -1825,6 +1825,8 @@ static void do_scan_async(void *_data, async_cookie_t c)
>>  
>>  	do_scsi_scan_host(shost);
>>  	scsi_finish_async_scan(data);
>> +
>> +	module_put(shost->hostt->module);
>>  }
>>  
>>  /**
>> @@ -1848,6 +1850,12 @@ void scsi_scan_host(struct Scsi_Host *shost)
>>  		return;
>>  	}
>>  
>> +	/* protection against surprise driver removal
>> +	 * module_put is called from do_scan_async
>> +	 */
>> +	if (!try_module_get(shost->hostt->module))
>> +		return;
>> +
>>  	/* register with the async subsystem so wait_for_device_probe()
>>  	 * will flush this work
>>  	 */
> 
> Shouldn't scsi_autopm_put_host(shost) be called if try_module_get() fails?

Thanks. Yes it should, I'll post a V2 if James agrees to this patch in
general in a parallel thread.


> 
> Please also update the following comment in scsi_scan_host():
> 
> 	/* scsi_autopm_put_host(shost) is called in scsi_finish_async_scan() */
It's late here so I'm tired and I miss something but how should I update
it ?


Thanks,
Tomas
> 
> Thanks,
> 
> Bart.
> 
>