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

Re: [Xen-devel] [PATCH] Optimizing x86-64 xenlinux using global pages for user mode (take 2)


  • To: "Nakajima, Jun" <jun.nakajima@xxxxxxxxx>, <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • From: Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>
  • Date: Wed, 30 Aug 2006 22:18:18 +0100
  • Delivery-date: Wed, 30 Aug 2006 14:18:29 -0700
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: AcbL/5UqcHfKdRZfQTa9q9O4P2Kz2gAejrc/
  • Thread-topic: [Xen-devel] [PATCH] Optimizing x86-64 xenlinux using global pages for user mode (take 2)

On 30/8/06 7:43 am, "Nakajima, Jun" <jun.nakajima@xxxxxxxxx> wrote:

> The previous patch had a problem with the builder, which constructs the
> initial page tables with the USER bit on. And Xen couldn't distinguish
> kernel or use pages. It was partially checked avoiding the bug, and it's
> finished with this patch.

I'd like to avoid awareness of the extra PAGE_GUEST_KERNEL bit in guests and
tools as much as possible (in fact, entirely would be good).

There's no need for all the PAGE_KERNEL macros in Linux to include the new
flag. The absence of PAGE_USER will suffice to allow Xen to distinguish
kernel entries from user entries. All that is necessary is for Xen to add
the flag when it adds PAGE_USER to a kernel entry, so that kernel entries
remain disambiguated from user entries.

Also, you should use one of the high-order available bits (one of bits
52-60). Other OSes use the low-order available bits for their own purposes
-- we may as well avoid conflicting with them.

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

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