cvs commit: src/sys/dev/mpt mpt.c mpt.h mpt_freebsd.cmpt_freebsd.hmpi_raid.h

[Matthew Jacob]

Thanks. When I'm ready for a checkin I'll contact uyou.


On Wed, 4 Jun 2003, Ruslan Ermilov wrote:

> On Tue, Jun 03, 2003 at 02:42:10PM -0700, Matthew Jacob wrote:
> >
> > > This doesn't have any relation to the problem under HTT+SMP
> > > with mpt(4) I was having, does it?
> >
> > Not directly.
> >
> > I'm sorting through some of the Domain Validation code I got from
> > someboy at LSI-Logic. This current checkin is in support of that.
> >
> > There's a possibility that some of the issues that have been seen (not
> > necessarily yours) are related to weaknesses in MPT firmware when Domain
> > Validation is *not* performed.
> >
> > For example, if I connect an Ultra2 disk to an Ultra320 and have things
> > all negotiate correctly, and then power cycle the Ultra2 disk, the ARQ
> > data (auto request sense) for the subsequent check condition appears to
> > be mangled (I can't decide whether it was shifted by a byte or endian
> > fouled up). Nobody else has ever reported this, so I'm assuming it's due
> > to me not doing DV. Oh well.
> >
> Thanks.  If you would like to ask me to try any patches for RELENG_4,
> you are welcome to.  I have a 100% reproducible case here.
>
>
> Cheers,
> --
> Ruslan Ermilov		Sysadmin and DBA,
> ru at sunbay.com		Sunbay Software Ltd,
> ru at FreeBSD.org		FreeBSD committer
>