Re: [PATCH] hw-detect: Do not attempt to load ide-generic if paravirt is active.

[Bjørn Mork <bjorn@mork.no>]

Ian Campbell <ijc@hellion.org.uk> writes:

> The lguest and Xen paravirt_ops implementations both reserve all IO
> ports in the 0-65535 range which prevents ide-generic from loading 

Excuse me for being daft, but wasn't that the whole point of
paravirt-ioport?


/*
 * Reserve the whole legacy IO space to prevent any legacy drivers
 * from wasting time probing for their hardware.  This is a fairly
 * brute-force approach to disabling all non-virtual drivers.
 *
 * Note that this must be called very early to have any effect.
 */
int paravirt_disable_iospace(void)
{
        return request_resource(&ioport_resource, &reserve_ioports);
}


> and leads to an error dialog being presented to the user.

so, add a hack to work around the problems introduced by another hack? 
Does this scale?  How many drivers need to be prevented from attempting
to load just to avoid the (correct) error messages caused by the initial
brute force hack?

And there are already kernel patches like this in the queue, disabling
paravirt-ioport for Xen: http://patchwork.kernel.org/patch/22432/

Which I believe means that you just as well could remove the hack from
lguest too.  It has always been wrong.  The prevent "wasting time
probing for their hardware" argument is just as valid for real
hardware.  But the result is replacing probing by guessing...



Bjørn