[patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[John W. Linville]

Patch to add IOI Media Bay 4-in-1 media reader to the SCSI quirk list...

"It works for me!"  Pretty simple patch, really...

John

diff -urNp linux-2.6.5-1.358/drivers/scsi/scsi_devinfo.c linux/drivers/scsi/scsi_devinfo.c
--- linux-2.6.5-1.358/drivers/scsi/scsi_devinfo.c	2004-05-08 08:56:41.000000000 -0400
+++ linux/drivers/scsi/scsi_devinfo.c	2004-08-11 06:08:00.000000000 -0400
@@ -155,6 +155,7 @@ static struct {
 	{"HP", "C1557A", NULL, BLIST_FORCELUN},
 	{"IBM", "AuSaV1S2", NULL, BLIST_FORCELUN},
 	{"IBM", "ProFibre 4000R", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
+	{"IOI", "Media Bay", NULL, BLIST_FORCELUN},
 	{"iomega", "jaz 1GB", "J.86", BLIST_NOTQ | BLIST_NOLUN},
 	{"IOMEGA", "Io20S         *F", NULL, BLIST_KEY},
 	{"INSITE", "Floptical   F*8I", NULL, BLIST_KEY},

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[Patrick Mansfield]

We seem to be getting quite a few of these. In theory we could add a line
like this for every multi-lun SCSI device.

Can you instead try booting with scsi_mod.max_luns=8 (or such) or build
with SCSI_MULTI_LUN enabled?

You can also use the scsi devinfo stuff, but the above is easier and
should work fine, unless you have some device that hangs when IO is sent
to a non-zero LUN.

The same for 2.4, but AFAIR the option name is max_scsi_luns (but there is
no scsi devinfo imethod).

-- Patrick Mansfield

On Thu, Aug 12, 2004 at 05:37:16PM -0400, John W. Linville wrote:
> Patch to add IOI Media Bay 4-in-1 media reader to the SCSI quirk list...
> 
> "It works for me!"  Pretty simple patch, really...
> 
> John
> 
> diff -urNp linux-2.6.5-1.358/drivers/scsi/scsi_devinfo.c linux/drivers/scsi/scsi_devinfo.c
> --- linux-2.6.5-1.358/drivers/scsi/scsi_devinfo.c	2004-05-08 08:56:41.000000000 -0400
> +++ linux/drivers/scsi/scsi_devinfo.c	2004-08-11 06:08:00.000000000 -0400
> @@ -155,6 +155,7 @@ static struct {
>  	{"HP", "C1557A", NULL, BLIST_FORCELUN},
>  	{"IBM", "AuSaV1S2", NULL, BLIST_FORCELUN},
>  	{"IBM", "ProFibre 4000R", "*", BLIST_SPARSELUN | BLIST_LARGELUN},
> +	{"IOI", "Media Bay", NULL, BLIST_FORCELUN},
>  	{"iomega", "jaz 1GB", "J.86", BLIST_NOTQ | BLIST_NOLUN},
>  	{"IOMEGA", "Io20S         *F", NULL, BLIST_KEY},
>  	{"INSITE", "Floptical   F*8I", NULL, BLIST_KEY},
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[John W. Linville]

Patrick Mansfield wrote:

>We seem to be getting quite a few of these. In theory we could add a line
>like this for every multi-lun SCSI device.
>  
>

Isn't that what the quirk list is for?

>Can you instead try booting with scsi_mod.max_luns=8 (or such) or build
>with SCSI_MULTI_LUN enabled?
>  
>

That works for my box, but what about for others?  Like those who may 
have both a multi-lun device and a single-lun device that hangs on a 
non-zero lun?  What about the average luser who can't be bothered to 
hack-up his startup scripts or *gasp* rebuild his kernel?

It seems like the quirk list is there for a reason.  If we start 
rejecting certain devices, then what is the criteria for a device to 
actually make it on the list?

John

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[Dave Jones]

On Thu, Aug 12, 2004 at 07:00:53PM -0400, John W. Linville wrote:
 > Patrick Mansfield wrote:
 > 
 > >We seem to be getting quite a few of these. In theory we could add a line
 > >like this for every multi-lun SCSI device.
 > Isn't that what the quirk list is for?
 > 
 > >Can you instead try booting with scsi_mod.max_luns=8 (or such) or build
 > >with SCSI_MULTI_LUN enabled?
 > 
 > That works for my box, but what about for others?  Like those who may 
 > have both a multi-lun device and a single-lun device that hangs on a 
 > non-zero lun?  What about the average luser who can't be bothered to 
 > hack-up his startup scripts or *gasp* rebuild his kernel?
 > 
 > It seems like the quirk list is there for a reason.  If we start 
 > rejecting certain devices, then what is the criteria for a device to 
 > actually make it on the list?

I hit the same problem when I tried to get James to merge around a half dozen
or so entries (thats since grown to almost double that).  One possibility
is to fix this in the USB storage layer. (Assume all USB SCSI devices are
capable of handling MULTI_LUN without problem).

I believe Pete Zaitcev was writing an alternative usb-storage driver
that did something along these lines, but I'm not sure of the details.

	Dave

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[Patrick Mansfield]

On Thu, Aug 12, 2004 at 07:00:53PM -0400, John W. Linville wrote:
> Patrick Mansfield wrote:
> 
> >We seem to be getting quite a few of these. In theory we could add a line
> >like this for every multi-lun SCSI device.
> > 
> >
> 
> Isn't that what the quirk list is for?

We should not add to the list for devices that behave as expected. I would
hope the number of borken BLIST_NOLUN devices is much smaller than the number
of good BLIST_FORCELUN devices.

The list is backwards compatible, so we have flags in there that are not
required, like BLIST_FORCELUN.

> >Can you instead try booting with scsi_mod.max_luns=8 (or such) or build
> >with SCSI_MULTI_LUN enabled?
> > 
> >
> 
> That works for my box, but what about for others?  Like those who may 
> have both a multi-lun device and a single-lun device that hangs on a 
> non-zero lun?  What about the average luser who can't be bothered to 
> hack-up his startup scripts or *gasp* rebuild his kernel?

Users should first set max_luns (or use a kernel built with
SCSI_MULTI_LUN), and then if they find a device that breaks have it added
to the quirks as BLIST_NOLUN. In the meantime they can boot using the
devinfo flag with BLIST_NOLUN.

I do not recall any reports (on linux-scsi) of devices that can't handle
LUN 0 requests, there is no change in the number of SCSI_NOLUN devices
from 2.4.21 to 2.6 (but I didn't check older kernels).

I had thought it was best to build without SCSI_MULTI_LUN, but given
the lack of SCSI_NOLUN additions, most users are better off with it on.

> It seems like the quirk list is there for a reason.  If we start 
> rejecting certain devices, then what is the criteria for a device to 
> actually make it on the list?

Only add borken devices to the list, so it is a list of bad devices rather
than a list of good ones.

-- Patrick Mansfield

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[John W. Linville]

Patrick Mansfield wrote:

>Only add borken devices to the list, so it is a list of bad devices rather
>than a list of good ones.
>  
>

I now see the comments about the list being deprecated.  I withdraw the 
2.6 patch.  I still think the 2.4 patch is wortwhile, unless there is an 
overhaul going-on there as well.

John

Re: [patch] 2.6 -- add IOI Media Bay to SCSI quirk list

[Marcelo Tosatti]

On Thu, Aug 12, 2004 at 09:21:56PM -0400, John W. Linville wrote:
> Patrick Mansfield wrote:
> 
> >Only add borken devices to the list, so it is a list of bad devices rather
> >than a list of good ones.
> > 
> >
> 
> I now see the comments about the list being deprecated.  I withdraw the 
> 2.6 patch.  I still think the 2.4 patch is wortwhile, unless there is an 
> overhaul going-on there as well.

Not on 2.4.x, no. Care to write & test a patch?