From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ian Campbell Subject: Re: [Xen-devel] [PATCH v4 2/7] xen/arm: introduce HYPERVISOR_platform_op on arm and arm64 Date: Fri, 20 Nov 2015 12:18:34 +0000 Message-ID: <1448021914.5647.139.camel@citrix.com> References: <1447349448-22610-2-git-send-email-stefano.stabellini@eu.citrix.com> <5645E868.30706@citrix.com> <1447666977.15629.9.camel@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Stefano Stabellini Cc: Julien Grall , xen-devel@lists.xensource.com, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org List-Id: xen-devel@lists.xenproject.org On Fri, 2015-11-20 at 11:58 +0000, Stefano Stabellini wrote: > On Mon, 16 Nov 2015, Ian Campbell wrote: > > On Fri, 2015-11-13 at 18:10 +0000, Stefano Stabellini wrote: > > > > > > I agree with your point (I thought about it myself) but the current > > > assembly scheme for hypercalls doesn't work well with that. I would > > > have > > > to introduce, and maintain going forward, two special hypercall > > > implementations in assembly, one for arm and another for arm64, just > > > to > > > set interface_version. I don't think it is worth it; I prefer to have > > > to > > > maintain the explicit interface_version setting at the call sites > > > (that > > > today is just one). > > > > You could give the bare assembly stub a different name (append _core or > > _raw or something) and make HYPERVISOR_platform_op a C wrapper for it > > which > > DTRT. > > I had an idea. I just need to > > #define HYPERVISOR_platform_op_raw HYPERVISOR_platform_op The need for this #define is a bit unfortunate, but the alternatives (e.g. a suffix argument to the HYPERCALL*() macros or a RAWHYPERCALL variant) would seem to suck more, so I say go for it. Ian.