From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Campbell Subject: Re: [PATCH XEN v3 14/22] tools: foreignmemory: provide xenforeignmemory_unmap. Date: Wed, 7 Oct 2015 16:30:50 +0100 Message-ID: <1444231850.1410.77.camel@citrix.com> References: <1444226543.1410.53.camel@citrix.com> <1444227341-875-14-git-send-email-ian.campbell@citrix.com> <56153432.1070002@citrix.com> <1444231005.1410.71.camel@citrix.com> <56153830.6010601@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <56153830.6010601@citrix.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: Andrew Cooper , ian.jackson@eu.citrix.com, wei.liu2@citrix.com, xen-devel@lists.xen.org List-Id: xen-devel@lists.xenproject.org On Wed, 2015-10-07 at 16:20 +0100, Andrew Cooper wrote: > On 07/10/15 16:16, Ian Campbell wrote: > > On Wed, 2015-10-07 at 16:03 +0100, Andrew Cooper wrote: > > > On 07/10/15 15:15, Ian Campbell wrote: > > > > > > What is the plan WRT fixing the existing APIs? > > Which ones do you mean? The xc_map_foreign_* which aren't moved here? > > > > My intention is to deprecate them and switch everything over to the API > > provided by this library. IOW xenforeignmemory_map() will be the sole > > (at > > least ABI stable way) way to map foreign memory. > > > > I don't want to go replicate the "there's ten ways to do it" aspect of > > the > > current interfaces in this new library. > > > > I think the selected interface (which matches the old _bulk one) > > provides > > sufficient flexibility to be usable going forward. Evidence of this is > > that > > all the old libxc interfaces are now implemented in terms of it. > > Sorry about not being clear. I meant "At what point do we fix the > (eventual) API in the new libraries", with reference to the suggested > s/unsigned int/size_t/ change. Ah, I think we should aim to release 4.7 with this being declared stable. I don't foresee the s/unsigned int/size_t/ in libxenforeignmemory being especially problematic wrt that. There are only a handful of users in tree, to update and the rest are still using the legacy xc_map_foreign wrappers, which would be the appropriate place for any impedance matching until those users get ported over. Ian.