panic - sleeping thread on FreeBSD 8.0-stable / amd64

Jeremy Chadwick freebsd at jdc.parodius.com
Sat Mar 13 17:47:24 UTC 2010


On Sat, Mar 13, 2010 at 06:32:59PM +0100, Torfinn Ingolfsen wrote:
> On Sat, 13 Mar 2010 08:38:48 -0800
> Jeremy Chadwick <freebsd at jdc.parodius.com> wrote:
> 
> > That enclosure also doesn't state if it has a SAF-TE or SES-2 chip on
> > it.  It's impossible to tell from the photos since the metallic
> > enclosure cover up the backplane.
> 
> Unfortunately, the "manual" is very brief and lacks such technical details / specs.
> This is  a very simple enclosure; I don't think it even has  any sensors except for the fan and temperature sensors,
> which are available at the signal out connector on the backplane.

...which is probably wired to some sort of interface chip that would
toggle a buzzer/etc. if something goes awry.  SAF-TE and/or SES-2
provide this capability, but so can a very simple circuit.

How do you tune the thresholds for the temperature or fan?  If they're
DIP switches, then chances are SAF-TE or SES-2 aren't involved and it's
probably just some cheap/generic logic chip that does the work.

> > If it does, such chips can/will yank devices off the bus when the chip
> > considers the drive "faulty".  At least in the SCSI world with SES-2,
> 
> Well, the manual states that this is a host function:
> <quote>
> 2.2 HDD FAIL system
> 2.2.1 This product can not provide the signal, unless the HOST supports this function.
> 2.2.2 If the Host can provide HDD fail signal, connect the cable to the HOST and to the backplane HDD FAIL 1 - 2 - 3 - 4 - 5.
> 2.2.3 When the HDD is damaged, the light will show in red.
> </quote>
> There is a separate connector for HDD fail input signals on the backplane. There are also output signals for power, access for each drive.

Is it a 4-pin connector?  If so, what the manual is (horribly) trying to
document is probably an SGPIO connector.

SGPIO is a 4-pin connector which connects a SATA/SAS enclosure to a
SATA/SAS HBA (controller) so that the two have a direct way to signal
that a disk has physically failed -- otherwise, the controller has to
make guesses about the state of things, and timeouts can take a while.

SGPIO can also be used to provide other things; it's a generic
communication interface with an official specification.

-- 
| Jeremy Chadwick                                   jdc at parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                  Mountain View, CA, USA |
| Making life hard for others since 1977.              PGP: 4BD6C0CB |



More information about the freebsd-stable mailing list