Explanation could be the following:
-----------------------------------
This drive does not appreciate initiators that are very unfair at
arbitrating for the SCSI BUS and tells such initiators about by using
unexpected bus free conditions. I can reproduce at will the problem
using a SYM53C896 controller.
An SCSI initiator that is configured as ID 7 has the highest priority with
regards to BUS arbitration protocol. If this initiator starts arbitration
immediately after the BUS FREE condition, then it will for sure win
arbitration every time. Doing so, it does not allow other devices to
reselect and continue or complete tasks.
Result is that devices are just force-feeded with SCSI commands while
their arse is just blocked at the same time. :-)
We should not be surprised if some seem to suffer a bit of such a
situation. :)
By the way, the UNEXPECTED BUS FREE condition is used by devices to report
serious errors to initiator and should not make problems. The behaviour of
the driver seems not normal to me.
Gérard.
On Wed, 26 May 1999, Nick Vermeer* wrote:
> To: [EMAIL PROTECTED]
> Subject: Raid problems
>
> I have linux-2.2.6 + raid-0145-2.2.6 patches running on an SMP Dual PIII-500 w/
> 512MB ram
>
> 2 Adaptec 2940U2W's w/ IBM 9.1 Gig 10,000RPM LVD drives
>
> Current config
>
> SCSI ID Controler Card
> 0 scsi1
> 0 scsi0
> 1 scsi1
> 2 scsi0
> 3 scsi1
> 4 scsi0
>
>
> Scsi ID's 1+2+3+4 are raid 5'ed and the two 0's are RAID 1
>
> Here is the problem.. Every three days (give or take a few hours) the machine
> crashes with the following error
>
> May 22 00:45:10 shamu kernel: (scsi1:0:1:-1) Unexpected busfree, LASTPHASE=0x40,
>SEQADDR = 0x15d
> May 22 00:45:10 shamu kernel: (scsi1:0:1:0) No active SCB for reconnecting target -
>Issuing BUS DEVICE RESET..
> May 22 00:45:10 shamu kernel: (scsi1:0:1:0) SAVED_TCL=0x10, ARG_1=0xff,
>SEQADDR=0x10b
>
> I believed this to be a hardware error, but I have replaced the
> drive,case,cable and changed the configuration of the chain and the problem
> continues. Also whenever the machine is halted, the scsi id 1 is kicked out of
> the raid array as not in sync. Any ideas? The other chain (which is identical
> physically) has no problems
>
>
>
> ---
> Nick Vermeer Internet Express
> Network Administator 1518 5th ave.
> (309)736-8000 Moline IL, 61265
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to [EMAIL PROTECTED]
>
>
