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

Re: [Xen-devel] [PATCH] linux/i386: allow CONFIG_HIGHPTE on i386 (take 2)


  • To: Jan Beulich <jbeulich@xxxxxxxxxx>, <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • From: Keir Fraser <keir@xxxxxxxxxxxxx>
  • Date: Fri, 12 Jan 2007 10:26:49 +0000
  • Delivery-date: Fri, 12 Jan 2007 02:26:32 -0800
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: Acc2NCrxaZu2oqInEdu5iwAX8io7RQ==
  • Thread-topic: [Xen-devel] [PATCH] linux/i386: allow CONFIG_HIGHPTE on i386 (take 2)



On 10/1/07 15:53, "Jan Beulich" <jbeulich@xxxxxxxxxx> wrote:

> While, as discussed, the performance impact of this option is certainly higher
> than on native Linux, the option should not be entirely disallowed if people
> want to sacrifice performance for less lowmem pressure.
> 
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxxxx>

Much of the suckiness will come from the fact that highptes are being pinned
for their lifetime. This means that you don't avoid emulation traps for
updates during process fork() and exit(), which are big wins. Obviously  the
problem is that kmap_atomic_pte() doesn't know whether to map the page
writeable or read-only (is it part of a pinned pagetable yet?). We could
steal a page flag, or use page->private, or something like that do give this
info... This patch is fine for now though.

 -- 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®.