On Wed, Sep 24, 2014 at 5:12 PM, Ian Campbell <Ian.Campbell@citrix.com>
wrote:

> On Tue, 2014-09-23 at 15:14 +0200, Tamas K Lengyel wrote:
> > Define the ARM specific test_and_set_bit functions and switch
> > to use maximum gpfn as the limit to setting permissions. Also,
> > move HAS_MEM_ACCESS definition into config.
>
> Please could you mention the removal of the x86 comment about atomicity
> here, since the reasoning for its correctness will be of interest to
> future code archaeologists.
>

Ack. I'm still hoping we can figure out if its the code or the comments
that's wrong. I tend to think its the comment as it smells like a
copy-paste.


>
> > -    rc = xc_set_mem_access(xch, domain_id, default_access, 0,
> > -                           xenaccess->domain_info->max_pages);
> > +    rc = xc_set_mem_access(xch, domain_id, default_access, START_PFN,
> > +                           (xenaccess->max_gpfn - START_PFN) );
>
> So the reason why info->max_pages is no use any more is because ARM
> starts at non-zero and has multiple banks?
>
> I suppose that's fair enough.
>
> Ian.
>

Yeap. It's not a problem to start at 0, this is just a bit of optimization
from the user's side that I would consider best-practice. The user could
also include some logic to skip the hole in-between the banks (if there are
multiple banks), but that I felt is too much complexity to include here.

Tamas