[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH v2 for-4.9] x86/mm: Fix incorrect unmapping of 2MB and 1GB pages

>>> On 24.05.17 at 00:25, <boris.ostrovsky@xxxxxxxxxx> wrote:
> On 05/23/2017 10:32 AM, Boris Ostrovsky wrote:
>> On 05/23/2017 10:05 AM, Jan Beulich wrote:
>>> Further changes may then
>>> be needed to the splitting of large pages (in p2m_next_level())
>>> depending on whether INVALID_MFN entries can make it there.
>> Let me see what I can do here.
> TBH, I don't see what needs to be done in p2m_next_level(). mfn doesn't
> enter the calculation there.

The relevant variables are named "pfn" there, and so far I'm only
guessing there might be an issue (or actually I meanwhile thing it
should only be a cosmetic one): What's being read out of the old
PTE is being taken and incremented for each split PTE. That'll (in
the 2Mb page case) result in 4k PTEs referencing MFNs 0xffffffffff
and 0 ... 0x1fe. But generally this should be harmless, as these
are non-present PTEs, and the frame numbers read back out of
non-present PTEs should be of no interest to anyone.

I'm pondering to convert the code to use mfn_add() (perhaps we
should also have mfn_inc()), making the helper saturate.

In any event I hope to find time later today to look into the issue
myself. I'm kind of disappointed that there was no visible attempt
from Igor so far to help the situation.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.