Re: [PATCH RFC v2 1/2] drivers: pci: fix pci_mmap_fits() implementation for procfs mmap

[Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>]

On Tue, Nov 11, 2014 at 05:19:31PM +0000, Bjorn Helgaas wrote:

[...]

> > I read your comment as an agreement on the approach I took in my patch,
> > except for the current pci_resource_to_user() implementation(s), which I did
> > not touch.
> 
> Yes.  I have two things I'd like to clear up:
> 
>   1) Your patch changes behavior on platforms that implement their own
> pci_resource_to_user().  So I'd like to mention the details of that in
> the changelog, e.g., "procfs mmap on arches X, Y, Z has been broken
> since commit C, and this change fixes them."  ARM doesn't implement
> pci_resource_to_user(), so I don't think ARM is one of those arches.
> But I'd really like to include specifics on what those arches are, and
> what we think is currently broken, so their maintainers at least get a
> heads-up and can look for that  breakage.

I posted a v3, where I *tried* to bisect the commits that actually broke
the procfs interface and I added a commit log to explain why, it is very
hard to bisect a specific commit (given the dependency on the arch code)
and I do not have HW to test the fix on apart from ARM machines so there
is not much I can do on that side.

Please let me know what you think, thanks for having a look.

Lorenzo