Re: [PATCH 1/5] fc class: add fc host default default dev loss setting

[Andrew Vasquez <andrew.vasquez@qlogic.com>]

On Mon, 09 Aug 2010, Mike Christie wrote:

> On 08/09/2010 10:00 AM, Mike Christie wrote:
> > On 08/09/2010 09:53 AM, James Smart wrote:
> >> Mike,
> >>
> >> why the motivation to get the dev_loss_tmo from the LLDD ? I would have
> >> assumed this would have become a base fc_host attribute which is rd/wr,
> >> the default coming from the module parameter, with the "set" propagating
> >> to the LLDD to change all rports to the host value (either new direct
> >> callback, or loop which calls rport set routine).
> >
> > qla2xxx and fnic set the default/initial dev_loss_tmo based on some
> > value they get from firmware.
> >
> 
> The qla2xxx mention is not right.
> 
> Andrew,
> 
> It looks like when the port_down_retry_count is set the 
> login_retry_count could be adjusted.

Actualy it's the other way around.  login_retry_count is initially
seeded with NVRAM values, then based on value, reassigned to
port_down_retry_count.

> When the dev_loss_tmo is set by 
> sysfs does qla2xxx, possibly want to adjust the login_retry_count?

I'd prefer to continue to have port_down_retry_count stay as a seeded
value to the transport (fc_host).

-- av