2002-09-19 21:03:46

by Reg Clemens

[permalink] [raw]
Subject: Dont understand hdc=ide-scsi behaviour.


I dont understand the behaviour of kernel 2.4.18 (and probably all others) when
I put the line
hdc=ide-scsi
on the load line.

I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
get it for BOTH hdc and hdd, the cdwriter and the zip drive.

After starting this way (with hdc=ide-scsi), I find that
/dev/cdrom2 -> /dev/scd0
and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)

I would EXPECT to get to them via /dev/hdd1 or /dev/hdd4.

Did I miss something or is this a bug????

Reg.Clemens
[email protected]


2002-09-19 23:58:35

by Allan Duncan

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.

Reg Clemens wrote:
>
> I dont understand the behaviour of kernel 2.4.18 (and probably all others) when
> I put the line
> hdc=ide-scsi
> on the load line.
>
> I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
> get it for BOTH hdc and hdd, the cdwriter and the zip drive.
>
> After starting this way (with hdc=ide-scsi), I find that
> /dev/cdrom2 -> /dev/scd0
> and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)
>
> I would EXPECT to get to them via /dev/hdd1 or /dev/hdd4.
>
> Did I miss something or is this a bug????

I presume you are putting the "hdc=ide-scsi" as a kernel param.
I do similarly, EXCEPT the CDwriter is on its own IDE bus.
I suspect that the hdc=ide-scsi thing will apply to both master and slave
on an IDE channel - whether that is intended/mandated I can't say.
If you move the zip to hdb all should be well, or buy an IDE PiC and get
more channels if you have run out.

2002-09-20 05:49:50

by Andre Hedrick

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.


Because only the lastest code base under -ac and maybe 2.5 is beginning to
to be device selectable.

hdc=scsi
insmod ide-cd
insmod ide-scsi

Be done

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Thu, 19 Sep 2002, Reg Clemens wrote:

>
> I dont understand the behaviour of kernel 2.4.18 (and probably all others) when
> I put the line
> hdc=ide-scsi
> on the load line.
>
> I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
> get it for BOTH hdc and hdd, the cdwriter and the zip drive.
>
> After starting this way (with hdc=ide-scsi), I find that
> /dev/cdrom2 -> /dev/scd0
> and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)
>
> I would EXPECT to get to them via /dev/hdd1 or /dev/hdd4.
>
> Did I miss something or is this a bug????
>
> Reg.Clemens
> [email protected]
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>

2002-09-20 06:47:19

by Brad Hards

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Fri, 20 Sep 2002 07:08, Reg Clemens wrote:
> I dont understand the behaviour of kernel 2.4.18 (and probably all others)
> when I put the line
> hdc=ide-scsi
> on the load line.
>
> I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
> get it for BOTH hdc and hdd, the cdwriter and the zip drive.
>
> After starting this way (with hdc=ide-scsi), I find that
> /dev/cdrom2 -> /dev/scd0
> and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)
There are two slightly different things happening, I think.

1. When you say hdc=ide-scsi, you are telling the IDE system that you don't
want to use the normal IDE interfaces to userland (such as ide-cdrom), but
instead want all access to this device to be accessed through the SCSI
midlayer (and associated SCSI interfaces, like the sg and scd drivers). So
ide-scsi becomes the driver, instead of ide-cdrom. You should be able to see
this in /proc/ide/hdc/driver

2. ide-scsi is greedy, and will grab any IDE device without a driver. ATAPI
floppy devices (hopefully) like your zip drive need the IDE floppy device
driver, which is probably not loaded. What does CONFIG_BLK_DEV_IDEFLOPPY
equal in your kernel config?

> I would EXPECT to get to them via /dev/hdd1 or /dev/hdd4.
And you will, with the right driver loaded :-)

> Did I miss something or is this a bug????
If you load ide-floppy before ide-scsi, and it still doesn't work, then there
is a bug.

Brad
- --
http://conf.linux.org.au. 22-25Jan2003. Perth, Australia. Birds in Black.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE9isQoW6pHgIdAuOMRAtfvAJ9QxzwAyyaLFRIHisEiZ9oEzGm9ngCfZMHB
X/OxH0BykeRSrQKAKj22u2Y=
=E/AW
-----END PGP SIGNATURE-----

2002-09-20 07:41:22

by Miles Lane

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.

On Thu, 2002-09-19 at 23:46, Brad Hards wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Fri, 20 Sep 2002 07:08, Reg Clemens wrote:
> > I dont understand the behaviour of kernel 2.4.18 (and probably all others)
> > when I put the line
> > hdc=ide-scsi
> > on the load line.
> >
> > I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
> > get it for BOTH hdc and hdd, the cdwriter and the zip drive.
> >
> > After starting this way (with hdc=ide-scsi), I find that
> > /dev/cdrom2 -> /dev/scd0
> > and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)
> There are two slightly different things happening, I think.
>
> 1. When you say hdc=ide-scsi, you are telling the IDE system that you don't
> want to use the normal IDE interfaces to userland (such as ide-cdrom), but
> instead want all access to this device to be accessed through the SCSI
> midlayer (and associated SCSI interfaces, like the sg and scd drivers). So
> ide-scsi becomes the driver, instead of ide-cdrom. You should be able to see
> this in /proc/ide/hdc/driver
>
> 2. ide-scsi is greedy, and will grab any IDE device without a driver. ATAPI
> floppy devices (hopefully) like your zip drive need the IDE floppy device
> driver, which is probably not loaded. What does CONFIG_BLK_DEV_IDEFLOPPY
> equal in your kernel config?
>
> > I would EXPECT to get to them via /dev/hdd1 or /dev/hdd4.
> And you will, with the right driver loaded :-)
>
> > Did I miss something or is this a bug????
> If you load ide-floppy before ide-scsi, and it still doesn't work, then there
> is a bug.

Well, aren't things going to get even more confusing when we
try to support devices like the Lacie DVD/CD Rewritable
combo drive? Are we going to do a better job of simply making
all usable interfaces available, so we no longer need to switch
between drivers or twiddle driver load order?

Miles

2002-09-20 08:00:23

by Brad Hards

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

WARNING: this email is even more clueless than my normal posting standard.

On Sat, 21 Sep 2002 00:46, Miles Lane wrote:
> Well, aren't things going to get even more confusing when we
> try to support devices like the Lacie DVD/CD Rewritable
> combo drive? Are we going to do a better job of simply making
> all usable interfaces available, so we no longer need to switch
> between drivers or twiddle driver load order?


Don't know the product, but I take it that this is a single IDE drive that can
read and write CDs, and read (and perhaps write) DVDs? If you want to write,
use ide-scsi+sg+scd, else just use ide-cdrom.

Until we have another userspace and kernel combination for writing, you are
stuck with cdrecord and sg. It'd be an vaguelly interesting exercise to have
a CD burning app that worked natively with ide-cdrom, but I have about a
million more important things to do before I'd start on that :)

If anyone is interested in coding this, what I'd really like to see is a CD
burning library that works with sg and ide-cdrom. You'd probably need to
modify ide-cdrom - I have no real idea about ATAPI or any other block device.

Then we could move away from GUIs that are really all just running the same
command line application, and get some more interesting handling going.

Brad

- --
http://conf.linux.org.au. 22-25Jan2003. Perth, Australia. Birds in Black.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE9itVHW6pHgIdAuOMRAumwAKCUzrDf7t4iRAHBWzSWyxHLLzOYQgCgvH73
URyJjN3asd8ZfKUa003K8sc=
=1Aws
-----END PGP SIGNATURE-----

2002-09-20 18:41:31

by Paul Bristow

[permalink] [raw]
Subject: Re: Dont understand hdc=ide-scsi behaviour.

Use the line
hdc=ide-scsi hdd=ide-floppy

and all should be well, even if using ide-floppy as a module

Miles Lane wrote:

>On Thu, 2002-09-19 at 23:46, Brad Hards wrote:
>
>
>>-----BEGIN PGP SIGNED MESSAGE-----
>>Hash: SHA1
>>
>>On Fri, 20 Sep 2002 07:08, Reg Clemens wrote:
>>
>>
>>>I dont understand the behaviour of kernel 2.4.18 (and probably all others)
>>>when I put the line
>>> hdc=ide-scsi
>>>on the load line.
>>>
>>>I would EXPECT to get the ide-scsi driver for hdc (my cdwriter) but instead
>>>get it for BOTH hdc and hdd, the cdwriter and the zip drive.
>>>
>>>After starting this way (with hdc=ide-scsi), I find that
>>> /dev/cdrom2 -> /dev/scd0
>>>and that to access the zip drive I have to use /dev/sda1 (or /dev/sda4)
>>>
>>>
>>There are two slightly different things happening, I think.
>>
>>1. When you say hdc=ide-scsi, you are telling the IDE system that you don't
>>want to use the normal IDE interfaces to userland (such as ide-cdrom), but
>>instead want all access to this device to be accessed through the SCSI
>>midlayer (and associated SCSI interfaces, like the sg and scd drivers). So
>>ide-scsi becomes the driver, instead of ide-cdrom. You should be able to see
>>this in /proc/ide/hdc/driver
>>
>>2. ide-scsi is greedy, and will grab any IDE device without a driver. ATAPI
>>floppy devices (hopefully) like your zip drive need the IDE floppy device
>>driver, which is probably not loaded. What does CONFIG_BLK_DEV_IDEFLOPPY
>>equal in your kernel config?
>>
>>
--

Paul

Email: [email protected]
Web: http://paulbristow.net
ICQ: 11965223