From owner-freebsd-arch@freebsd.org Wed Jul 8 07:36:09 2015 Return-Path: Delivered-To: freebsd-arch@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id CDFA49969D6 for ; Wed, 8 Jul 2015 07:36:09 +0000 (UTC) (envelope-from bright@mu.org) Received: from elvis.mu.org (elvis.mu.org [IPv6:2001:470:1f05:b76::196]) by mx1.freebsd.org (Postfix) with ESMTP id BDA261D76; Wed, 8 Jul 2015 07:36:09 +0000 (UTC) (envelope-from bright@mu.org) Received: from [IPv6:2601:645:8004:7515:1da0:325d:7d6:e304] (unknown [IPv6:2601:645:8004:7515:1da0:325d:7d6:e304]) by elvis.mu.org (Postfix) with ESMTPSA id AA062341F841; Wed, 8 Jul 2015 00:36:08 -0700 (PDT) Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 (1.0) Subject: Re: CFT/CFR: NUMA policy branch From: Alfred Perlstein X-Mailer: iPhone Mail (12H143) In-Reply-To: <1443707.QHq1OS6BQP@akita> Date: Wed, 8 Jul 2015 00:36:07 -0700 Cc: Alfred Perlstein , "freebsd-arch@freebsd.org" , Konstantin Belousov , Adrian Chadd Content-Transfer-Encoding: quoted-printable Message-Id: <295D42F8-6542-4665-B824-B61A7CD48282@mu.org> References: <2926903.YAk7qUEGf9@akita> <559CB61F.2070301@freebsd.org> <1443707.QHq1OS6BQP@akita> To: Rui Paulo X-BeenThere: freebsd-arch@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: Discussion related to FreeBSD architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 08 Jul 2015 07:36:09 -0000 > On Jul 7, 2015, at 11:44 PM, Rui Paulo wrote: >=20 >> On Tuesday 07 July 2015 22:33:19 Alfred Perlstein wrote: >>> On 7/7/15 7:43 PM, Rui Paulo wrote: >>> On Tuesday 07 July 2015 15:53:18 Adrian Chadd wrote: >>>>> I did not read further, the patch is half-done at best. >>>>=20 >>>> That's lovely. Meanwhile, people are actively using this thing. >>>=20 >>> It may not be perfect, but it's way more than half done. You might obje= ct >>> to introducing the syscalls, but procctl is still annoyingly limited. >> (not yelling at you Rui)... but really... Is that the problem?!!? Just >> write a userland library to abstract the kernel interface! >=20 > How can a library help? If you can't tell the kernel to apply a policy pe= r- > TID (procctl works by PID), it's useless for multi-threaded applications. >=20 The library would abstract away the kernel boundary concerns. So let's say w= e did what kib wanted and extended procctl to support tids, well at that poi= nt the syscalls made could be garbage collected and the library updated to c= all the correct kernel entry point.=20=