Messages in this thread | | | Date | Tue, 25 Jul 2000 15:34:03 +0100 (BST) | From | James Sutherland <> | Subject | Re: Direct access to hardware |
| |
On Tue, 25 Jul 2000, Stephen Frost wrote:
> On Tue, 25 Jul 2000, James Sutherland wrote: > > > On Mon, 24 Jul 2000, Horst von Brand wrote: > > > To get what you are asking for would give a kernel source of a gigabyte or > > > so (just add up all the funny things you might want to send to a random > > > IDE, SCSI, FireWire, USB, ... device, consider that downloadable firmware > > > is becomming the norm, and devices are proliferating like never before), > > > and _that_ doesn't scale at all. > > > > A gigabyte? Hardly. It's an if statement we need, that's all. OK, a couple > > of if statements per subsystem will add up to quite a few Kb - but that'll > > hardly break the bank. > > That depends, are you talking about just filtering the known 'bad' > things, or filtering down to where only valid commands (with valid data > mind you, which is a cute thing to try and determine in kernel space) work? > > Not that it really matter much, both lead to the dark side and loads > of cruft that will never be cleaned out, even once the hardware is fixed and > such filtering of 'invalid' commands with 'invalid' data, or valid commands > with 'invalid' data is useless.
Simply block all the non-ATA commands. These should never be issued by anything other than a vendor diagnostics program - and if you're running one of those, WTF are you doing in a multi-user desktop/server OS at the time? The commands in question are very specialised and rare. The manufacturer could just supply a boot image on their WWW site, containing a simple OS (FreeDOS, cut down Linux, whatever) plus their utility.
James.
- 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/
| |