On Mon, 13 Dec 1999, Chiaki Ishikawa wrote:
> X-PMC-CI-e-mail-id: 12062
>
> >FWIW, I solved the LUN-based changer problem in FreeBSD/CAM by adding a
> >sort-of scheduler to the CD driver. All CD devices that are part of a
> >changer are grouped together, and commands can only be issued to one LUN
> >at a time.
>
> It sounds to me a good solution. The choice of time quantum seems also
> reasonable.
> I am not a kernel hacker, but probably the
> checking of data structure to see if there are pending
> requests to different LUNs on the same target by traversing the
> linked list as mentioned in Eric Youngdale's post
> is functionally equivalent to what you describe here.
>
> Alan Cox has been mentioning partial overhaul of SCSI code and so I am
> not sure where the fix should go in (and also when).
The single-lun case is still just as broken as ever (although I
think I fixed it so that it iterates over the correct set of command
blocks, and hence it should detect when the other ones are busy).
The BSD solution sounds kind of like what I had in mind. I was
also thinking of a timer of sorts so that control would pass from one to
the next after an interval of inactivity of perhaps a second or so.
-Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
