> Not only that but MULTI_LUN jacks up the ide-scsi driver. At least I am
> quite certain I don't have 7 CDRW's, 7 DVDROM's, and 7 CDROM's connected
> to my machine.


From: Gadi Oxman <[EMAIL PROTECTED]>
To: Dan Hollis <[EMAIL PROTECTED]>
Cc: Wakko Warner <[EMAIL PROTECTED]>,
        Tim Walberg <[EMAIL PROTECTED]>, [EMAIL PROTECTED]
Subject: Re: Oops with 2.2.13pre17 and ide-scsi
On Mon, 18 Oct 1999, Dan Hollis wrote:

> On Mon, 18 Oct 1999, Wakko Warner wrote:
> > > This happens when you compile in "probe all luns" with ide-scsi enabled.
> > > Disable "probe all luns" and the 8 targed ids will go away. Guaranteed.
> > > It should probably be counted as an ide-scsi bug? (Do any ide devices
> > > legitimately have more than one ide-scsi lun?)
> > Isn't there a way to fix this so that probing all luns works correctly with
> > ide-scsi?
> 
> Ask the ide-scsi maintainer I guess.

multi-lun IDE devices do exist, so in that sense the driver is correct
in allowing multi-lun probes. It is just that most IDE devices ignore
the LUN field and respond to each LUN. This is usually just a minor
inconvinience, as the device is detected eight times.

Over the times, there have been reports of actual kernel oopes which seemed
related to multi-lun support on the surface, and I repeatedly got patches
which disabled multi-lun probing, even though the multi-lun support iself
can't be the cause of the problem; it just allows the SCSI subsystem
to send packet commands with a lun field different than zero. Any kernel
crash has its reasons in another (real) bug.

That said, there might be a way to support multi-lun devices and still
detect normal devices just once. multi-lun devices are supposed to have
a "max lun" field in response to the ATAPI IDENTIFY command, and perhaps
multi-lun devices return x > 0 in those fields, while other return 0
(or perhaps 7?).

If anybody is interested in testing this, add something like:

        if (drive->id)
                printk("%s: last-lun: %d\n", drive->name,
                        drive->id->word126 & 0x7);

to idescsi_setup in drivers/scsi/ide-scsi.c. To know if we can use this
field, we need test results both from users of multi-lun IDE devices
(such as PD-CD drives), and from users of single-lun IDE devices.

> > what if I have a cdchanger on a scsi card and I use ide-scsi with another
> > cdrom?
> 
> Youre screwed for now 8)
> 
> -Dan

No, having a cd-changer on another SCSI card should not be related to
ide-scsi. Enabling multi-lun should add support for both.

Gadi


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
Please read the FAQ at http://www.tux.org/lkml/


-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]

Reply via email to