On 4/28/2016 8:39 PM, Jan Beulich wrote:
On 28.04.16 at 14:12, <yu.c.zh...@linux.intel.com> wrote:
I'm still confused why do we need this, especially at such critical
moment. IIUC HVMMEM type is used to get/set mem type, why would someone
define a HVMMEM type but not use it here?
Who knows. And as said - the patch can go in as is, I just inquired
because I like to avoid future code churn whenever possible, i.e.
when a certain way of coding makes it less likely for the code
needing touching again compared to some other variant, I'd
generally like that to be used (as long as it's not meaningfully worse
in other respects).
Thanks Jan.
So my understanding is that this patch does not need any change any
more.
As to your concern, I still do not have any better thought.
And this hole is a problem because of the old mistake I have made in
previous version. Could we be careful in the future review to avoid
another hole(besides the HVMMEM_unused one which is unavoidable with
HVMMEM_ioreq_server), and if this can not be avoided, we try to find a
more graceful solution by then? :)
Jan
Yu
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
