Jens Axboe wrote:
>
> On Sat, Jan 22 2000, H. Peter Anvin wrote:
> > > I invite you to read drivers/cdrom.c and include/linux/cdrom.h. The
> > > CDROM_SEND_PACKET ioctl has been added to the interface during the 2.3.x
> > > series, and allows you to send arbitrary packets to a low-level cdrom
> > > driver that supports CDC_GENERIC_PACKET (i.e. ide and scsi).
> > >
> > > /me thinks the uniform cdrom driver is very cool, but then I'm biased.
> > >
> >
> > This seems like the wrong way of doing this. This *definitely*
> > shouldn't be CDROM-specific.
>
> It was put in the CD-ROM subsystem, since it was designed to be
> independent of the specific CD-ROM type (allow queueing of commands
> to ATAPI devices and not through ide-scsi). The CDROM_SEND_PACKET
> was just a natural extension of the generic_packet() hook added
> for ATAPI and SCSI drives, which was originally added to provide
> uniform support for DVD CSS on both types. Later it was extended
> to cover audio commands as well.
>
> I have been thinking about the earlier comment you made, the one
> about providing ATAPI support through SCSI emulation purely like
> Windows NT does (and other OS' too). Provide a nice packet layer
> that doesn't differentiate between ATAPI and SCSI, commands
> could be queued through SG or similar. This exact discussion
> was raised on linux-scsi about a year ago, if I remember correctly.
>
Well, that's exactly what ide-scsi does, so doing that is definitely an
option (and is currently the *only* way to do this.) If so we should
ditch the IDE drivers completely and do everything through ide-scsi.
The other option is to make a generic /dev/hg* interface for *all*
IDE/ATAPI devices.
-hpa
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Jan 23 2000 - 21:00:28 EST