Re: [PATCH] Paging and memory sharing for HVM guests

[Patrick Colp <pjcolp@cs.ubc.ca>]

>> Introducing another bogus failure indicator for the mmap_batch
>> privcmd operations seems rather undesirable - we'll already need to
>> find a backwards-compatible solution to the current (broken) or-ing
>> in of 0xf0000000 (broken because MFNs can now be more than
>> 28 bits wide).
> 
> Surly you mean > 30 bits wide?
> Anyway, I'll let Patrick comment on that, since he is the author of
> this bit of the code.

I realise that this isn't the best approach. However, as you point out,
there already is 0xf0000000 in the code. I thought it easiest just to latch
on to that for now, although you're quite right that a better solution
needs to be found, preferably something that will solve both these problems.


>> Removing support for IOCTL_PRIVCMD_MMAP from the pv-ops
>> implementation seems pretty unrelated, so should probably be a
>> separate patch.
> 
> Forwarding this Q to Patrick again.

Yes, you're probably right. A while back I submitted a patch which changed
libxc to use only mmap_batch, which is now in the xen-unstable tree. This
was done with the idea of unifying the mmap interface (although ideally it
would be a nicer one that what is there currently, but that's an entirely
different discussion). This all came about while working on this project,
which is why it's been included in the patches here. It could easily be
resubmitted as an independent patch, if that is preferred.


Patrick