From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Paul E. McKenney" Subject: Re: [PATCH 1/2] rcu: add rcu_access_pointer and rcu_dereference_protect Date: Wed, 7 Apr 2010 08:59:29 -0700 Message-ID: <20100407155929.GB2481@linux.vnet.ibm.com> References: <20100407135732.12414.16416.stgit@warthog.procyon.org.uk> <1270652210.8141.9.camel@edumazet-laptop> Reply-To: paulmck@linux.vnet.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Cc: David Howells , Trond.Myklebust@netapp.com, linux-nfs@vger.kernel.org, linux-kernel@vger.kernel.org To: Eric Dumazet Return-path: Received: from e4.ny.us.ibm.com ([32.97.182.144]:42947 "EHLO e4.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932581Ab0DGP7f (ORCPT ); Wed, 7 Apr 2010 11:59:35 -0400 In-Reply-To: <1270652210.8141.9.camel@edumazet-laptop> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Wed, Apr 07, 2010 at 04:56:50PM +0200, Eric Dumazet wrote: > Le mercredi 07 avril 2010 =E0 14:57 +0100, David Howells a =E9crit : > > From: Paul E. McKenney > >=20 > > This patch adds variants of rcu_dereference() that handle situation= s > > where the RCU-protected data structure cannot change, perhaps due t= o > > our holding the update-side lock, or where the RCU-protected pointe= r is > > only to be fetched, not dereferenced. > >=20 > > The new rcu_access_pointer() primitive is for the case where the po= inter > > is be fetch and not dereferenced. This primitive may be used witho= ut > > protection, RCU or otherwise, due to the fact that it uses ACCESS_O= NCE(). > >=20 > > The new rcu_dereference_protect() primitive is for the case where u= pdates > > are prevented, for example, due to holding the update-side lock. T= his > > primitive does neither ACCESS_ONCE() nor smp_read_barrier_depends()= , so > > can only be used when updates are somehow prevented. > >=20 > > Suggested-by: David Howells > > Signed-off-by: Paul E. McKenney > > Signed-off-by: David Howells > > --- > >=20 > > include/linux/rcupdate.h | 34 ++++++++++++++++++++++++++++++++++ > > 1 files changed, 34 insertions(+), 0 deletions(-) > >=20 > > diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h > > index 872a98e..a1b14b6 100644 > > --- a/include/linux/rcupdate.h > > +++ b/include/linux/rcupdate.h > > @@ -209,9 +209,43 @@ static inline int rcu_read_lock_sched_held(voi= d) > > rcu_dereference_raw(p); \ > > }) > > =20 > > +/** > > + * rcu_access_pointer - fetch RCU pointer with no dereferencing > > + * > > + * Return the value of the specified RCU-protected pointer, but om= it the > > + * smp_read_barrier_depends() and keep the ACCESS_ONCE(). This is= useful > > + * when the value of this pointer is accessed, but the pointer is = not > > + * dereferenced, for example, when testing an RCU-protected pointe= r against > > + * NULL. This may also be used in cases where update-side locks p= revent > > + * the value of the pointer from changing, but rcu_dereference_pro= tect() > > + * is a lighter-weight primitive for this use case. > > + */ > > +#define rcu_access_pointer(p) \ > > + ({ \ > > + ACCESS_ONCE(p); \ > > + }) > > + > > +/** > > + * rcu_dereference_protected - fetch RCU pointer when updates prev= ented > > + * > > + * Return the value of the specified RCU-protected pointer, but om= it > > + * both the smp_read_barrier_depends() and the ACCESS_ONCE(). Thi= s > > + * is useful in cases where update-side locks prevent the value of= the > > + * pointer from changing. Please note that this primitive does -n= ot- > > + * prevent the compiler from repeating this reference or combining= it > > + * with other references, so it should not be used without protect= ion > > + * of appropriate locks. > > + */ > > +#define rcu_dereference_protected(p) \ > > + ({ \ > > + (p); \ > > + }) > > + > > #else /* #ifdef CONFIG_PROVE_RCU */ > > =20 > > #define rcu_dereference_check(p, c) rcu_dereference_raw(p) > > +#define rcu_access_pointer(p) ACCESS_ONCE(p) > > +#define rcu_dereference_protect(p) (p) > > =20 > > #endif /* #else #ifdef CONFIG_PROVE_RCU */ > > =20 > >=20 > > -- >=20 > This is not the version Paul posted.=20 I blew the name -- rcu_dereference_protected() is in fact a better name= =2E > Removing checks just to shutup warnings ? >=20 > All the point is to get lockdep assistance, and you throw it away. >=20 > We want to explicit the condition, so that RCU users can explicitly > state what protects their data. What Eric said!!! ;-) Thanx, Paul