From mboxrd@z Thu Jan 1 00:00:00 1970 From: Yong Zhang Subject: Re: Kernel rwlock design, Multicore and IGMP Date: Fri, 12 Nov 2010 21:00:17 +0800 Message-ID: <20101112130017.GA9752@zhy> References: <1289489007.17691.1310.camel@edumazet-laptop> <20101112071323.GB5660@cr0.nay.redhat.com> <1289546874.17691.1774.camel@edumazet-laptop> <20101112081945.GA5949@cr0.nay.redhat.com> <20101112091818.GB5949@cr0.nay.redhat.com> Reply-To: Yong Zhang Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Eric Dumazet , Cypher Wu , linux-kernel@vger.kernel.org, netdev To: =?utf-8?Q?Am=C3=A9rico?= Wang Return-path: Content-Disposition: inline In-Reply-To: <20101112091818.GB5949@cr0.nay.redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Fri, Nov 12, 2010 at 05:18:18PM +0800, Am=C3=A9rico Wang wrote: > On Fri, Nov 12, 2010 at 05:09:45PM +0800, Yong Zhang wrote: > >On Fri, Nov 12, 2010 at 4:19 PM, Am=C3=A9rico Wang wrote: > >> On Fri, Nov 12, 2010 at 08:27:54AM +0100, Eric Dumazet wrote: > >>>Le vendredi 12 novembre 2010 =C3=A0 15:13 +0800, Am=C3=A9rico Wang= a =C3=A9crit : > >>>> On Fri, Nov 12, 2010 at 11:32:59AM +0800, Cypher Wu wrote: > >>>> >On Thu, Nov 11, 2010 at 11:23 PM, Eric Dumazet wrote: > >>>> >> Le jeudi 11 novembre 2010 =C3=A0 21:49 +0800, Cypher Wu a =C3= =A9crit : > >>>> >> > >>>> >> Hi > >>>> >> > >>>> >> CC netdev, since you ask questions about network stuff _and_ = rwlock > >>>> >> > >>>> >> > >>>> >>> I'm using TILEPro and its rwlock in kernel is a liitle diffe= rent than > >>>> >>> other platforms. It have a priority for write lock that when= tried it > >>>> >>> will block the following read lock even if read lock is hold= by > >>>> >>> others. Its code can be read in Linux Kernel 2.6.36 in > >>>> >>> arch/tile/lib/spinlock_32.c. > >>>> >> > >>>> >> This seems a bug to me. > >>>> >> > >>>> >> read_lock() can be nested. We used such a schem in the past i= n iptables > >>>> >> (it can re-enter itself), > >>>> >> and we used instead a spinlock(), but with many discussions w= ith lkml > >>>> >> and Linus himself if I remember well. > >>>> >> > >>>> >It seems not a problem that read_lock() can be nested or not si= nce > >>>> >rwlock doesn't have 'owner', it's just that should we give > >>>> >write_lock() a priority than read_lock() since if there have a = lot > >>>> >read_lock()s then they'll starve write_lock(). > >>>> >We should work out a well defined behavior so all the > >>>> >platform-dependent raw_rwlock has to design under that principl= e. > >>>> > >>> > >>>AFAIK, Lockdep allows read_lock() to be nested. > >>> > >>>> It is a known weakness of rwlock, it is designed like that. :) > >>>> > >>> > >>>Agreed. > >>> > >> > >> Just for record, both Tile and X86 implement rwlock with a write-b= ias, > >> this somewhat reduces the write-starvation problem. > > > >Are you sure(on x86)? > > > >It seems that we never realize writer-bias rwlock. > > >=20 > Try >=20 > % grep RW_LOCK_BIAS -nr arch/x86 >=20 > *And* read the code to see how it works. :) If read_lock()/write_lock() fails, the subtracted value(1 for read_lock() and RW_LOCK_BIAS for write_lock()) is added back. So reader and writer will contend on the same lock fairly. And RW_LOCK_BIAS based rwlock is a variant of sighed-test rwlock, so it works in the same way to highest-bit-set mode rwlock. Seem you're cheated by it's name(RW_LOCK_BIAS). :) Or am I missing something? Thanks, Yong >=20 > Note, on Tile, it uses a little different algorithm.