On Fri, 11 Aug 2000, Eddie Williams wrote:
> > On Fri, 11 Aug 2000, Eddie Williams wrote:
> >
> > > Were there IO's going on to the disk when the power was dropped? Who is
> > > driving term power?
> >
> > Regardless. The kernel keeps dying even after the power and the failed
> > drive are back on and up in my aha2940 case.
>
> There are two ways to solve the problem: 1) is to avoid the problem in the
> first place and 2) to fix error handling so that the system recovers.
a proposal concerning 2):
The mid-layers allows HBA drivers to report unexpected bus resets by means
of scsi_report_bus_reset(). This should avoid reset wars. But I do not
understand why it is up to a particular HBA driver to make the system
stable in multiple initiator environments. There is no reason for
implementing this behavior in a specific driver but not in another one,
isn't it? It should not depend on the HBA driver. You can imagine any
combination of HBAs from different vendors driven by different drivers on
a single parallel bus, or whatever.
The mentioned behavior should depend on whether we have an environment
with several initiators or not. (Is it possible to detect this by means of
INQUIRY?) A more defensive error handling should be applied in such
environments. A compromise: let the user decide, i.e. during kernel
configuration by means of a SCSI midlayer option forcing defensive error
recovery.
Regards
Martin Peschke
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
