On Mon, 24 Jul 2000, Andre Hedrick wrote:
> On Tue, 25 Jul 2000, Marc Lehmann wrote:
>
> > > because you have no business turning on "vender-unique"
> >
> > "you"?
>
> No, without loading vender-specific module to handle the calls.
> After loading such a tool, they have to filter for commands that they
> accept and reject all others.
>
> Basically an extension of the model presented in the beginning.
This is going to complicate matters, not simplyfy them.
The point is: by default setting, a drive MUST NOT take damage from
vendor-specific commands (except for the usual tear, that is). That is,
THE DRIVE filters illegal commands and returns error conditions
(possibly, another error condition needs to be defined for that).
IF the drive is to accept vendor-specific commands, e. g. for firmware
updates -- which could well be standardized -- this can be achieved
either by a jumper that needs to be changed, or by a command with a
digital signature that uses asymmetric cryptography, and the signature
shall comprise the command and its data.
There needs to be a flag in the inquiry data that indicates if the drive
is able to filter vendor-specific commands and (possibly another) if the
filtering is switched on. (Could be merged to a bit that indicates that
the drive is filtering commands.) Operating systems that feel the need
to may then choose to filter if this bis is missing.
Plus, vendors need to be told that they must not void people's drives'
warranties just because the drives get sent strange commands. This may
happen through defective memory in a machine even though the machine
actually complies to all parts of the ATA and ATAPI standards.
Heading to vendor-specific modules is the wrong way to go. The world
does not only consist of Winblows PCs, Suns and Linux boxes.
-- Matthias Andree- 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 : Mon Jul 31 2000 - 21:00:18 EST