On Thu, 2003-03-13 at 15:17, Doug Ledford wrote:
> Mark Haverkamp wrote:
>
> > Does the cmd_per_lun element of the Scsi_Host_Template structure serve
> > more than one purpose?
>
> Only when inappropriately abused by LLDD authors. The cmd_per_lun value
> is suppossed to be for untagged devices only! If you have a tape drive
> that doesn't support tagged commands but you want to be able to
> internally have the next command queued up and ready to go when the
> current command completes (in order to keep it streaming better), then
> you can set cmd_per_lun to 2 and you will get two outstanding commands
> for this device at a time. I used that so that in my interrupt handler
> I could send the next command to the device before I passed the
> completed command up to the SCSI layer.
>
> > In scsi_alloc_sdev it is passed into
> > scsi_adjust_queue_depth. In the aacraid case this is 512. Later the
> > aacraid driver (in aac_slave_configure) sets the queue depth to either
> > 128 for tagged or 1 if not.
>
> That's where you are suppossed to set the queue depth on tagged devices.
Then it sounds like the aacraid driver could set cmd_per_lun to a small
number like one since the real queue depth will be set later in
aac_slave_configure.
Mark.
-- Mark Haverkamp <markh@osdl.org>- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Mar 15 2003 - 22:00:37 EST