Re: Reliability of RPC services

[Michal Suchanek]

On 4/26/06, Jonathan S. Shapiro <address@hidden> wrote:
> On Wed, 2006-04-26 at 10:00 +0200, Bas Wijnen wrote:
>
> > Obviously user-provided drivers cannot be allowed to do that directly (but
> > they may do it through calls to the framework).  However, for user-provided
> > drivers I was thinking about devices that the users brings with him.  There 
> > it
> > would be the bus driver (usb, firewire) which does the DMA calls, and the
> > actual driver just talks to the bus driver.
>
> This isn't how things work. The DMA chip isn't generally associated with
> the bus. It's generally on the card. USB may be something that could be
> handled as a special case.

No, PCI (and PCI mods like AGP or CardBus) is a special case. Drivers
of devices connected to this bus have to be trusted because they could
do all kinds of things given access to the bus.

Devices connected to [PS]ATA, USB, FireWire, SCSI, parallel, etc.
ports do not need trusted drivers. In some cases the device could lock
up the bus. But even if more than one device is connected to the same
bus the problematic device can be simply unplugged.

Thanks

Michal