From: Kirill Tkhai Date: Fri, 23 Aug 2013 15:43:56 +0000 Subject: Re: [RFC] sparc64: Duplicate kernel text on every NUMA node Message-Id: <12781377272636@web6h.yandex.ru> List-Id: References: <8311377178790@web14g.yandex.ru> In-Reply-To: <8311377178790@web14g.yandex.ru> MIME-Version: 1.0 Content-Type: text/plain; charset="windows-1252" Content-Transfer-Encoding: quoted-printable To: sparclinux@vger.kernel.org 23.08.2013, 09:23, "David Miller" : > From: Kirill Tkhai > Date: Thu, 22 Aug 2013 17:39:50 +0400 > >> =9AThere is an idea to duplicate kernel text on every NUMA node. Node >> =9Aaccess is the fastest, so this should help to execute kernel text a >> =9Alittle faster. > > Interesting change. > > What other major architectures do this? It looks like nobody supports this. I saw an old patch for x86 (https://lkm= l.org/lkml/2003/6/16/226) and several others, but they were not integrated. So, sparc64 can be the fi= rst. Other architectures have no as cheap replication as sparc64 has. They must modify page table top every switch_to(), while sparc64 is free from this th= ing. And it looks like they have to flush all instruction cache every context sw= itch. I'll work on the patch a little bit more and test it in different boundary = cases, before submit it as a ready patch. Thanks, Kirill