Re: [PATCH] switch scsi upper driver probing to the driver model

[Christoph Hellwig <hch@lst.de>]

On Sat, May 17, 2003 at 01:59:52PM +1000, Douglas Gilbert wrote:
> Is that "_and_" symmetric? That is, can one have the sg
> device in preference to the sd device? If so, how is
> that selected?

You can. In fact there's nothing in this patch that preferes sd over sg
except the existing link order.  A better way to control the driver
matching would be nice but that nothing that should be done in the
scsi code but rather a generic driver model thing.

> Christoph,
> This patch would have some interesting side effects:
>    - no generic interface (apart from SCSI_IOCTL_SEND_COMMAND)
>      for tape devices (st and osst). While the maintainers of
>      st and osst might like this, not all the user apps would :-)

I have some WIP code to make parts of drivers/block/scsi_ioctl.c
working for scsi char drivers.

>    - the extra ioctls that sg defines are either fudged
>      or missing. This might upset programs other than
>      cdrecord; cdparanoia comes to mind

Hmm, what does cdparanoia complain about?  And yes, I think we
need to kill all those return 0 ioctls from the block layer -
they might have helped to bring up the SG_IO code with cdrecord
but they are a really stupid idea long term.

> It was my understanding that the device model has been
> recently modified by Greg KH to cope with the dual nature
> of the interface that sg causes. Evidently similar situations
> occur elsewhere in the kernel.

Greg, any comments on this?  I haven't found any code like that
in drivers/base/ and I can't understand how this is supposed to work..`