From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeremy Fitzhardinge Subject: Re: [PATCH 3/6] xen-gntdev: Remove unneeded structures from grant_map tracking data Date: Tue, 14 Dec 2010 13:56:09 -0800 Message-ID: <4D07E7F9.8080006@goop.org> References: <1292338553-20575-1-git-send-email-dgdegra@tycho.nsa.gov> <1292338553-20575-4-git-send-email-dgdegra@tycho.nsa.gov> <4D07DE69.3030607@goop.org> <4D07E71B.3020606@tycho.nsa.gov> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4D07E71B.3020606@tycho.nsa.gov> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Daniel De Graaf Cc: xen-devel@lists.xensource.com, Ian.Campbell@citrix.com List-Id: xen-devel@lists.xenproject.org On 12/14/2010 01:52 PM, Daniel De Graaf wrote: > On 12/14/2010 04:15 PM, Jeremy Fitzhardinge wrote: >> On 12/14/2010 06:55 AM, Daniel De Graaf wrote: >>> The entire hypercall argument list isn't required; only selected >>> fields from the hypercall need to be tracked between the ioctl, map, >>> and unmap operations. >> Is the rationale of this patch to save memory? If so, how much does it >> save. >> >> (This patch seems sensible in principle, but it doesn't seem to save >> much complexity.) >> >> J > This will also allow easier testing of what pages need to be unmapped > (more obvious in the HVM version). I also find it less confusing to > populate the hypercall arguments immediately before the hypercall, but > that's likely a matter of opinion. It only saves 46 bytes per page, so > if it seems more complex it could be dropped. I like it in general. See the other mail I just sent - you can use the multicall API to remove all the allocations for the arguments, and that should help a lot. J