>-----Original Message-----
>From: Mike Christie [mailto:[EMAIL PROTECTED]
>Sent: Thursday, August 28, 2008 7:23 PM
>To: Dev, Vasu
>Cc: [email protected]
>Subject: Re: [Open-FCoE] [PATCH] libfc: fixed ESB_ST_ABNORMAL handling
>(ver-2)
>
>Mike Christie wrote:
>> Vasu Dev wrote:
>>> @@ -410,16 +449,19 @@ static void fc_exch_timeout(unsigned long
ep_arg)
>>>             if (e_stat & ESB_ST_REC_QUAL)
>>>                     fc_exch_rrq(ep);
>>>             goto done;
>>> -   } else if (!(e_stat & ESB_ST_ABNORMAL)) {
>>> -           resp = ep->resp;
>>> -           arg = ep->resp_arg;
>>> +   } else if (ep->fh_type != FC_TYPE_FCP) {
>>
>> I do not think this works because srr is type FCP isn't it?. If that
>> times out we probably want to want to handle it. Currently we
possibly
>> retry the rec and then if that fails send an abort for the scsi
command.
>
>I mean retry srr.
>
>> Not sure how handy the retries are, but at the very lest we need to
call
>> exch_done on the srr ep, and we need to release the hold on the scsi
>> command's fsp.

This issue was already there before this patch, since prior to this
patch "} else if (!(e_stat & ESB_ST_ABNORMAL))" check would have
prevented any -FC_EX_TIMEOUT event to any upper layer of fc_exch.c since
ESB_ST_ABNORMAL is set on sending abort and it would have remained set
on exch timeout for no response to abort request.

However you are correct that this is an issue for FCP and need to be
fixed when -FC_EX_TIMEOUT events are passed up to FCP and that will also
require adding -FC_EX_TIMEOUT handling in all exch resp handlers of FCP.

So I'm updating comment work and description of this patch that this
patch is only fixing -FC_EX_TIMEOUT for only non FCP cases and FCP need
to be fixed later in another patch. Is that okay?

_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel

Reply via email to