From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Wed, 11 Sep 2013 09:54:44 +1000 From: Paul Mackerras To: Alexander Graf Subject: Re: [PATCH 2/6] powerpc: Provide for giveup_fpu/altivec to save state in alternate location Message-ID: <20130910235444.GD4455@iris.ozlabs.ibm.com> References: <20130910102000.GA28145@iris.ozlabs.ibm.com> <20130910102110.GC28145@iris.ozlabs.ibm.com> <675235E3-6991-4DB1-90A4-935E826F6413@suse.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <675235E3-6991-4DB1-90A4-935E826F6413@suse.de> Cc: kvm-ppc@vger.kernel.org, kvm@vger.kernel.org, linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, Sep 10, 2013 at 12:12:47PM -0500, Alexander Graf wrote: > > On 10.09.2013, at 05:21, Paul Mackerras wrote: > > > @@ -212,6 +212,7 @@ struct thread_struct { > > #endif > > #endif > > struct thread_fp_state fp_state; > > + struct thread_fp_state *fp_save_area; > > Why do you need these pointers? Couldn't you handle everything you need through preempt notifiers? As you note in your review of a later patch, no, I need the pointer so that if in-kernel code wants to use FP or VSX, potentially in the context of this same process, it knows where to save the FP/VSX state away to. Paul.