From mboxrd@z Thu Jan 1 00:00:00 1970 From: Julien Grall Subject: Re: [PATCH v2 01/41] arm/acpi: Build numa for x86 only Date: Wed, 20 May 2015 16:49:48 +0100 Message-ID: <555CAD1C.1060203@citrix.com> References: <1431893048-5214-1-git-send-email-parth.dixit@linaro.org> <1431893048-5214-2-git-send-email-parth.dixit@linaro.org> <555CBF50020000780007C4C3@mail.emea.novell.com> <555CA675.1080600@citrix.com> <555CC752020000780007C540@mail.emea.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <555CC752020000780007C540@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 , Julien Grall , Parth Dixit Cc: keir@xen.org, ian.campbell@citrix.com, andrew.cooper3@citrix.com, tim@xen.org, xen-devel@lists.xen.org, stefano.stabellini@citrix.com, christoffer.dall@linaro.org List-Id: xen-devel@lists.xenproject.org On 20/05/15 16:41, Jan Beulich wrote: >>>> On 20.05.15 at 17:21, wrote: >> On 20/05/15 16:07, Jan Beulich wrote: >>>>>> On 17.05.15 at 22:03, wrote: >>>> From: Naresh Bhat >>>> >>>> Numa is currently not supported for arm in xen. >>> >>> This doesn't represent a valid reason for the code to be disabled, >>> the more that ARM will want to support NUMA sooner or later. A >>> valid reason would be if you'd have to introduce _a lot_ of code on >>> the ARM side in order to make this build - if that's the case, please >>> say so in the description, as from an architectural pov the absence >>> of the respective ACPI tables should be no issue at all. >> >> The acpi/numa.c is calling a lot of arch specific function implemented >> in (arch/x86/srat.c). The code in acpi/numa.c is very x86 related (such >> as x2apic). >> >> I think disable the compilation is preferable for now given that there >> is no board for testing the changes. > > Or, as said in reply to patch 2, stubbing out the missing pieces in > ARM code. As said there - let's not make more spaghetti of common > code than it already is. That would mean we will build a code that won't be use on ARM until we decide to implement NUMA. This is because the main function acpi_numa_init is not called and should not be called until someone figure out this is valid on ARM. What about defining a new define HAVE_NUMA? Regards, -- Julien Grall