svn commit: r253550 - head/sys/dev/mps

[Slawa Olhovchenkov]

On Mon, Aug 26, 2013 at 04:43:35PM -0600, Kenneth D. Merry wrote:

> On Tue, Aug 27, 2013 at 02:25:26 +0400, Slawa Olhovchenkov wrote:
> > On Mon, Aug 26, 2013 at 03:40:27PM -0600, Kenneth D. Merry wrote:
> > 
> > > On Tue, Aug 13, 2013 at 19:30:29 +0400, Slawa Olhovchenkov wrote:
> > > > On Mon, Aug 12, 2013 at 01:02:29PM -0600, Kenneth D. Merry wrote:
> > > > 
> > > > > If you really want one now, I've attached a patch from stable/9 on June
> > > > > 27th.  It may or may not apply now.
> > > > 
> > > > Its apply witch litle edit.
> > > > 
> > > > > > > Now that it's done, I hope to merge that change to stable/9 this week.
> > > > > > 
> > > > > > Also, I see strange behaviour of LSI 9211-8i -- after some activity
> > > > > > got timeout and HBA put in looped reset.
> > > > > 
> > > > > That's not good.  Try with the newer driver and see whether if affects the
> > > > > behavior.
> > > > 
> > > > Witch this driver I got same behavior: after intesive i/o got timeout,
> > > > driver try re-init controller without success. On all HDD activity LED
> > > > ON w/o blinking at this moment.
> > > 
> > > Do you have any logs or dmesg output showing the problem?
> > 
> > Currenly not awailable, but simmilar as here
> > http://lists.freebsd.org/pipermail/freebsd-current/2013-May/041943.html
> 
> That one is relatively recent, but all it shows is timeouts.  There are
> lots of potential causes.

In my case only timeouts showed.

> Without dmesg output, we don't know what firmware version he's running, and
> that could be an issue.  As can SATA drives behind an expander.

firmware try bundled (v14 IR), v16 IR from LSI site,
v14 IT from supermicro site. No expander, but SATA drive.

> > or here http://forums.freebsd.org/showthread.php?t=28252.
> 
> This one shows lots of IOC terminated errors, but it looks like his problem
> was solved by upgrading his SSD firmware.

It's only as sample of dmesg output (in my dmesg no other questionably
information).
Firmware of SATA disk marked as compatible on LSI site (ST91000640NS SN02).

> > At the time of problem -- no any messages in the log or concole, just
> > stopping all operations. After some time (aprox 30-60 seconds) --
> > circular messages about timeout, reset and massive 'SCSI command
> > timeout'.
> 
> What was your topology?  i.e. mps(4) controller connected directly to the
> drives, or via an expander?  SAS or SATA drives?

directly to 8 SATA drives. drives ST91000640NS SN02

> > > I have merged the latest driver changes to stable/9, including a bug fix
> > > for out of chain frame handling.
> > 
> > Now I can't test anymore -- controller replaced to asr.
> 
> An asr(4) controller?

Yes.