Re: [PATCH] fakephp: Allocate PCI resources before adding the device

[Trent Piepho <xyzzy@speakeasy.org>]

On Tue, 25 Nov 2008, Darrick J. Wong wrote:
> On Tue, Nov 25, 2008 at 08:46:37PM -0800, Trent Piepho wrote:
> >
> > I've had a patch to fakephp that did something like this for a while, but I
> > called pci_bus_assign_resources() _after_ adding the devices with calls to
> > pci_bus_add_device().  It seems like that might be easier, to just add all
> > the devices and then call pci_bus_assign_resources() when done.  It appears
> > to work fine for me.  Is there a reason this is wrong?
>
> afaict, pci_bus_add_devices calls device_add to set up sysfs files and
> trigger a event that can (ultimately) cause a pci probe action to
> happen... but the probe will fail because the resources aren't ready.
> In any case, if a device shows up in sysfs I'd assume that to mean that
> the device is ready to go--all the BARs are reserved for the device,
> etc.  For sure, I woudn't expect to be racing
> pci_bus_assign_resources().

Ok, that makes sense.  The device I'm using fakephp for doesn't have a
kernel driver so I wouldn't have noticed that.

Have you tested this with a device that isn't present at boot?  I found
that I needed to a call to pci_enable_device() after assigning resources,
otherwise the BARs wouldn't be enabled.  This only happened if the device
wasn't present at boot time.

My hardware doesn't run on the latest kernel so I can't test it.  It looks
like there have been a bunch of pci hotplug changes so back porting this
might not be feasible.  It also looks a previous patch by Alex Chiang
completely changed the sysfs interface for fakephp.  I thought sysfs
interfaces were supposed to be stable?!  Also looks like it made fakephp
useless.  How are you supposed to figure out which "fake-n" directory is
the right one to disable the device you want?