On Fri, Jan 04, 2019 at 12:13:09AM -0800, Christopher Clark wrote: > 2) the p2m type of the guest-supplied memory for the ring. > > Roger raised a query about not pinning the p2m type of memory > used for the ring, and my response on 21st December is here: > > https://lists.xenproject.org/archives/html/xen-devel/2018-12/msg02204.html > > At the moment, I haven't changed that code. If the p2m type is changed > after ring registration, is it a problem? If not, then I think the code is > OK; but if so then a pointer to what makes it problematic would be > helpful to determine an appropriate next step.
My point was that you don't need to check every time that the gfn -> mfn translations for the ring are the same (as is done in argo_find_ring_mfns). AFAICT you take a reference to each page in the ring, so there's no need to check that the p2m mapping is still the same. I think the p2m type checks that you do are correct, and you should only allow p2m_ram_rw pages to be used for the ring. Roger. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
