Re: pciide detachment

["Jonathan A. Kollasch" <jakllsch%kollasch.net@localhost>]

On Sun, Sep 12, 2010 at 01:48:28PM -0500, David Young wrote:
> On Fri, Sep 10, 2010 at 08:45:38PM +0000, Jonathan A. Kollasch wrote:
> > Hi,
> > 
> > The attached patch lays the groundwork for detachment of
> > pciide(4)-family drivers.  Thoughts?
> 
> Thanks for working on this!  On two or three of my machines, pciide(4)
> is the only PCI device that I cannot detach.  To write the detachment
> routines was daunting.
> 
> How do your systems do under a pciide(4) detach & rescan cycle?

Seems to do well.  Over the course of a few hours of testing at a rate
of about 12 rescans per minute; no noticeable memory leaks or
mapping-already-mapped-bus_space problems arose in this code.

However if I pushed it too fast I'd sometimes get panics, these need
my attention.

The detach functions in the individual drivers are rather
simple, usually just a call to pciide_common_detach() and
pci_intr_disestablish().

        Jonathan Kollasch