[PATCH] fix computation related to FC_EX_CLOSED in FC local port

[PATCH] fix computation related to FC_EX_CLOSED in FC local port

[Hillf Danton]

In fc_lport_error(), FC_EX_CLOSED could be understood to be no more
rettries, so resetting local port is reasonable and necessary.

When processing response to flogi request, FC_EX_CLOSED should be
treated as other errors, since the state machine of local port could
reach reset.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
---

--- a/drivers/scsi/libfc/fc_lport.c	2010-11-01 19:54:12.000000000 +0800
+++ b/drivers/scsi/libfc/fc_lport.c	2010-12-16 23:46:06.000000000 +0800
@@ -1013,7 +1013,7 @@ static void fc_lport_error(struct fc_lpo
 		     lport->retry_count);

 	if (PTR_ERR(fp) == -FC_EX_CLOSED)
-		return;
+		goto reset;

 	/*
 	 * Memory allocation failure, or the exchange timed out
@@ -1029,6 +1029,7 @@ static void fc_lport_error(struct fc_lpo

 		schedule_delayed_work(&lport->retry_work, delay);
 	} else
+ reset:
 		fc_lport_enter_reset(lport);
 }

@@ -1428,9 +1429,6 @@ void fc_lport_flogi_resp(struct fc_seq *

 	FC_LPORT_DBG(lport, "Received a FLOGI %s\n", fc_els_resp_type(fp));

-	if (fp == ERR_PTR(-FC_EX_CLOSED))
-		return;
-
 	mutex_lock(&lport->lp_mutex);

 	if (lport->state != LPORT_ST_FLOGI) {

Re: [Open-FCoE] [PATCH] fix computation related to FC_EX_CLOSED in FC local port

[Vasu Dev]

On Thu, 2010-12-16 at 23:54 +0800, Hillf Danton wrote:
> In fc_lport_error(), FC_EX_CLOSED could be understood to be no more
> rettries, so resetting local port is reasonable and necessary.
> 

It is not reasonable/necessary in this case as FC_EX_CLOSED is issued on
lport reset thru exch block reset called there, so doing lport reset
again in same call flow is not needed unless I'm missing something to do
reset again here. Are hitting any bug which got fixed by this patch ?

  
> When processing response to flogi request, FC_EX_CLOSED should be
> treated as other errors, since the state machine of local port could
> reach reset.
> 
> Signed-off-by: Hillf Danton <dhillf@gmail.com>
> ---
> 
> --- a/drivers/scsi/libfc/fc_lport.c	2010-11-01 19:54:12.000000000 +0800
> +++ b/drivers/scsi/libfc/fc_lport.c	2010-12-16 23:46:06.000000000 +0800
> @@ -1013,7 +1013,7 @@ static void fc_lport_error(struct fc_lpo
>  		     lport->retry_count);
> 
>  	if (PTR_ERR(fp) == -FC_EX_CLOSED)
> -		return;
> +		goto reset;
> 
>  	/*
>  	 * Memory allocation failure, or the exchange timed out
> @@ -1029,6 +1029,7 @@ static void fc_lport_error(struct fc_lpo
> 
>  		schedule_delayed_work(&lport->retry_work, delay);
>  	} else
> + reset:
>  		fc_lport_enter_reset(lport);
>  }
> 
> @@ -1428,9 +1429,6 @@ void fc_lport_flogi_resp(struct fc_seq *
> 
>  	FC_LPORT_DBG(lport, "Received a FLOGI %s\n", fc_els_resp_type(fp));
> 
> -	if (fp == ERR_PTR(-FC_EX_CLOSED))
> -		return;
> -

Like other resp handlers in lport.c, above code is required instead just
removing for  fc_lport_flogi_resp, this is to allow exch clean up on
exch block reset.

	Vasu

>  	mutex_lock(&lport->lp_mutex);
> 
>  	if (lport->state != LPORT_ST_FLOGI) {
> _______________________________________________
> devel mailing list
> devel@lists.open-fcoe.org
> https://lists.open-fcoe.org/mailman/listinfo/devel

Re: [Open-FCoE] [PATCH] fix computation related to FC_EX_CLOSED in FC local port

[Hillf Danton]

On Fri, Dec 17, 2010 at 4:05 AM, Vasu Dev <vasu.dev@linux.intel.com> wrote:
> On Thu, 2010-12-16 at 23:54 +0800, Hillf Danton wrote:
>> In fc_lport_error(), FC_EX_CLOSED could be understood to be no more
>> rettries, so resetting local port is reasonable and necessary.
>>
>
> It is not reasonable/necessary in this case as FC_EX_CLOSED is issued on
> lport reset thru exch block reset called there, so doing lport reset
> again in same call flow is not needed unless I'm missing something to do
> reset again here. Are hitting any bug which got fixed by this patch ?
>
Umm..why could lport reset then when no more retries, first?
And how about the state machine of lport then if nothing done, second?

It looks not buggy, I think.
>
>> When processing response to flogi request, FC_EX_CLOSED should be
>> treated as other errors, since the state machine of local port could
>> reach reset.
>>
>> Signed-off-by: Hillf Danton <dhillf@gmail.com>
>> ---
>>
>> --- a/drivers/scsi/libfc/fc_lport.c   2010-11-01 19:54:12.000000000 +0800
>> +++ b/drivers/scsi/libfc/fc_lport.c   2010-12-16 23:46:06.000000000 +0800
>> @@ -1013,7 +1013,7 @@ static void fc_lport_error(struct fc_lpo
>>                    lport->retry_count);
>>
>>       if (PTR_ERR(fp) == -FC_EX_CLOSED)
>> -             return;
>> +             goto reset;
>>
>>       /*
>>        * Memory allocation failure, or the exchange timed out
>> @@ -1029,6 +1029,7 @@ static void fc_lport_error(struct fc_lpo
>>
>>               schedule_delayed_work(&lport->retry_work, delay);
>>       } else
>> + reset:
>>               fc_lport_enter_reset(lport);
>>  }
>>
>> @@ -1428,9 +1429,6 @@ void fc_lport_flogi_resp(struct fc_seq *
>>
>>       FC_LPORT_DBG(lport, "Received a FLOGI %s\n", fc_els_resp_type(fp));
>>
>> -     if (fp == ERR_PTR(-FC_EX_CLOSED))
>> -             return;
>> -
>
> Like other resp handlers in lport.c, above code is required instead just
> removing for  fc_lport_flogi_resp, this is to allow exch clean up on
> exch block reset.
First, this error could be processed correctly after lock.
And it could be also processed with more cares not only in the above
error handler.

And the state machine of lport is the major concern of the patch,
simply returning
helps little.

Cheers
Hillf

>
>        Vasu
>
>>       mutex_lock(&lport->lp_mutex);
>>
>>       if (lport->state != LPORT_ST_FLOGI) {
>> _______________________________________________
>> devel mailing list
>> devel@lists.open-fcoe.org
>> https://lists.open-fcoe.org/mailman/listinfo/devel
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html