cvs commit: src/sys/dev/disk/sbp sbp.c

[Joerg Sonnenberger]

On Sun, Jan 18, 2004 at 05:32:35PM +0900, Hidetoshi Shimokawa wrote:
> At Tue, 13 Jan 2004 09:32:12 -0800 (PST),
> Joerg Sonnenberger wrote:
> > 
> > joerg       2004/01/13 09:32:12 PST
> > 
> > DragonFly src repository
> > 
> >   Modified files:
> >     sys/dev/disk/sbp     sbp.c 
> >   Log:
> >   Add hw.firewire.sbp.tags to control tagging for SBP devices. The default
> >   behaviour was changed to disable tagged command queues, which is safer since
> >   not all firewire chips support TCQ.
> 
> As the nature of SBP-2, all commands are tagged and all target should
> support tagged queuing. I know some drives does't work well
> with command queue longer than 1, though. I digged the problem before and
> found that this is not because the target drive doesn't support the
> tagged queuing but it sometimes respones with bogus command chain
> status and this confuses the initiater(sbp.c).

Well, I read something about support for TCQ in the SBP specs, too.

> You can limit the queue length for such devices by:
> # camcontrol X:X:X tags -N 1

Last time I tried that didn't help. Hm. I will try to trace it down
(together with the if_wi timeout when using the firewire disk). But
first I have to get my internal HD back.

> I have installed DFB in one of my machine. I'd like to sync the
> firewire drivers in DFB with those in recent FreeBSD's tree.

This would be very useful.

> I'm interested in SSI(Single System Image) of DFB and
> I think the FireWire bus can be a efficient trasport layer for it.

Yeah, esp. since it is engineered by real intelligent people and
not forcefully put into market by MS ;-)

Do you know wether anyone has written a real IP-over-1394 driver?

Joerg

> 
> /\ Hidetoshi Shimokawa
> \/  simokawa at xxxxxxxxxxxxxxxxxxx
> PGP public key: http://www.sat.t.u-tokyo.ac.jp/~simokawa/pgp.html
>