On 5/2/2012 1:39 PM, Trent Nelson wrote:
[Resending from non-broken MTA.]

Hi Matt,

     isp(4) mentions the following sysctl options:

        dev.isp.N.loop_down_limit
            This value says how long to wait in seconds after loop has gone
            down before giving up and expiring all of the devices that were
            visible.  The default is 300 seconds (5 minutes).  A separate
            (nonadjustable) timeout is used when booting to not stop booting
            on lack of FC connectivity.

        dev.isp.N.gone_device_time
            This value says how long to wait for devices to reappear if they
            (temporarily) disappear due to loop or fabric events.  While this
            timeout is running, I/O to those devices will simply be held.

        dev.isp.N.wwnn
            This is the readonly World Wide Node Name value for this port.

        dev.isp.N.wwpn
            This is the readonly World Wide Port Name value for this port.

     I'm not seeing any of these in 9.0-stable.  There doesn't seem to be any
     occurrence of SYSCTL_ADD_* in /usr/src/sys/dev/isp.

------------------------------------------------------------------------
r227126 | mjacob | 2011-11-05 17:44:40 -0700 (Sat, 05 Nov 2011) | 4 lines

Implement the sysctl's for fibre channel that are listed in the man page.


MFC after:      3 days

------------------------------------------------------------------------
Guess it didn't make it into RELENG_9.


     Thoughts?

     Out of interest, what's the current state of isp on FreeBSD?  I had a peek
     at NetBSD and OpenBSD's isp drivers and it looks like you've been involved
     in all of them at some point or another.

     Are they all in sync or is one leading the pack?  (I'm keen to hear if
     there are isp changes on other platforms I can merge back into FreeBSD;
     all the x86/amd64 boxes on the Snakebite network run FreeBSD and have
     QLogic PCI-X 23xx (single&  dual port) cards hooking them up to the SAN,
     so the isp driver is a pretty critical part of my infrastructure.)


Yes, I wrote them all. This driver is part of a multiplatform driver that I started at NASA/Ames in 1997 that has run on FreeBSD, OpenBSD, NetBSD, BSDi, Solaris, Linux and few other random systems. At the time I started it, I was of the opinion that it didn't cost much extra effort and performance to write drivers in OS-independent and OS-dependent pieces. This all goes back to my days at Sun doing the Solaris DDI/DKI and involvement in the OBIOS and other "common framework" toads. Nowadays it hardly matters.

FreeBSD is the most up to date, but certainly has bugs in target mode. I no longer am a committer for OpenBSD and haven't refreshed the NetBSD port in quite some time.

I've just gotten some paid work that will allow me time to fix a number of the items that are currently broken, but because I believe that the following two things are true, there isn't a lot of plans for major work on it at this time:

 * Fibre Channel is not very interesting at this point. IMO it's now in
   legacy mode while CNA/100Gb and/or faster and faster Infiniband will
   become the long haul data transport. SAS and/or PCIe GEN3 will
   handle the within rack transport.
 * I'm /really/ tired of this driver.


23XX cards are pretty old at this point. In fact, the 25XX cards are pretty old.

Did you have specific bugs you were concerned with?

Note, btw, that freebsd-scsi is a better list to bring up FC or SCSI related topics.

_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[email protected]"

Reply via email to