From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Vrabel Subject: Re: [PATCHv13 1/2] gnttab: use per-VCPU maptrack free lists Date: Tue, 16 Jun 2015 10:49:17 +0100 Message-ID: <557FF11D.7020608@citrix.com> References: <1434379671-20327-1-git-send-email-david.vrabel@citrix.com> <1434379671-20327-2-git-send-email-david.vrabel@citrix.com> <557FDDD40200007800085368@mail.emea.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from mail6.bemta5.messagelabs.com ([195.245.231.135]) by lists.xen.org with esmtp (Exim 4.72) (envelope-from ) id 1Z4nUk-0002uI-D2 for xen-devel@lists.xenproject.org; Tue, 16 Jun 2015 09:49:26 +0000 In-Reply-To: <557FDDD40200007800085368@mail.emea.novell.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Jan Beulich , David Vrabel Cc: xen-devel@lists.xenproject.org, Malcolm Crossley , Keir Fraser , Ian Campbell , Tim Deegan List-Id: xen-devel@lists.xenproject.org On 16/06/15 07:27, Jan Beulich wrote: >>>> On 15.06.15 at 16:47, wrote: >> @@ -57,7 +58,7 @@ integer_param("gnttab_max_frames", max_grant_frames); >> * New options allow to set max_maptrack_frames and >> * map_grant_table_frames independently. >> */ >> -#define DEFAULT_MAX_MAPTRACK_FRAMES 256 >> +#define DEFAULT_MAX_MAPTRACK_FRAMES 1024 > > This should be undone by patch 2 now, too, or not be done here > in the first place. I anyway wonder - also because both patches > basically re-write __get_maptrack_handle() as well as the change > back to the first come first serve model here - whether folding the > second patch into this one wouldn't be better (albeit I realize that > this would cause an authorship conflict). We need to increase DEFAULT_MAX_MAPTRACK_FRAMES to at least 512 since the structure is now twice as big. We also want to avoid the slow steal path where possible so I think the extra doubling to account for unbalanced usage is still useful. Even with 1024 frames this is still only 4 MiB which seems fine to me. I would prefer to keep the steal patch separate since it's a self-contained extra bit. David