[dm-devel] Question on dm-mp sense handling

[Mike Anderson]

Is someone still working on the sense buffer handling code for bios
to be used by dm-mp?

The ifdef'd out code in dm-emc.c and dm_scsi_err_handler appears to only
allow the interpretation of sense data if available and does not provide
any detail on non-sense based failures (DRIVER_TIMEOUT vs DID_* status).

An old patch that Mike Christie submitted was a start at mapping errors to
BLK error flags but did not receive much feedback. Is there some reason to
not decode down at the lower levels with a possible interface for vendor
specific mapping?

Old post can be found here.
http://marc.theaimsgroup.com/?l=linux-kernel&m=107961883914541&w=2

The reason I am asking the question is I am looking into issues of a
single device mapper hardware handler supporting block devices exported by
a DASD driver or through the SCSI subsystem.

-andmike
--
Michael Anderson
andmike at us.ibm.com