From: Eric Biggers <firstname.lastname@example.org> To: Alan Stern <email@example.com> Cc: "Darrick J. Wong" <firstname.lastname@example.org>, email@example.com, firstname.lastname@example.org, "Paul E . McKenney" <email@example.com>, firstname.lastname@example.org, Akira Yokosawa <email@example.com>, Andrea Parri <firstname.lastname@example.org>, Boqun Feng <email@example.com>, Daniel Lustig <firstname.lastname@example.org>, Dave Chinner <email@example.com>, David Howells <firstname.lastname@example.org>, Jade Alglave <email@example.com>, Luc Maranget <firstname.lastname@example.org>, Nicholas Piggin <email@example.com>, Peter Zijlstra <firstname.lastname@example.org>, Will Deacon <email@example.com> Subject: Re: [PATCH] tools/memory-model: document the "one-time init" pattern Date: Fri, 17 Jul 2020 19:00:09 -0700 Message-ID: <20200718020009.GE2183@sol.localdomain> (raw) In-Reply-To: <20200718012555.GA1168834@rowland.harvard.edu> On Fri, Jul 17, 2020 at 09:25:55PM -0400, Alan Stern wrote: > On Fri, Jul 17, 2020 at 05:58:57PM -0700, Eric Biggers wrote: > > On Fri, Jul 17, 2020 at 01:53:40PM -0700, Darrick J. Wong wrote: > > > > +There are also cases in which the smp_load_acquire() can be replaced by > > > > +the more lightweight READ_ONCE(). (smp_store_release() is still > > > > +required.) Specifically, if all initialized memory is transitively > > > > +reachable from the pointer itself, then there is no control dependency > > > > > > I don't quite understand what "transitively reachable from the pointer > > > itself" means? Does that describe the situation where all the objects > > > reachable through the object that the global struct foo pointer points > > > at are /only/ reachable via that global pointer? > > > > > > > The intent is that "transitively reachable" means that all initialized memory > > can be reached by dereferencing the pointer in some way, e.g. p->a->b->c. > > > > It could also be the case that allocating the object initializes some global or > > static data, which isn't reachable in that way. Access to that data would then > > be a control dependency, which a data dependency barrier wouldn't work for. > > > > It's possible I misunderstood something. (Note the next paragraph does say that > > using READ_ONCE() is discouraged, exactly for this reason -- it can be hard to > > tell whether it's correct.) Suggestions of what to write here are appreciated. > > Perhaps something like this: > > Specifically, if the only way to reach the initialized memory > involves dereferencing the pointer itself then READ_ONCE() is > sufficient. This is because there will be an address dependency > between reading the pointer and accessing the memory, which will > ensure proper ordering. But if some of the initialized memory > is reachable some other way (for example, if it is global or > static data) then there need not be an address dependency, > merely a control dependency (checking whether the pointer is > non-NULL). Control dependencies do not always ensure ordering > -- certainly not for reads, and depending on the compiler, > possibly not for some writes -- and therefore a load-acquire is > necessary. > > Perhaps this is more wordy than you want, but it does get the important > ideas across. > How about: There are also cases in which the smp_load_acquire() can be replaced by the more lightweight READ_ONCE(). (smp_store_release() is still required.) Specifically, if the only way to reach the initialized memory involves dereferencing the pointer itself, then the data dependency barrier provided by READ_ONCE() is sufficient. However, if some of the initialized memory is reachable some other way (for example, if it is global or static data) then there need not be an address dependency, merely a control dependency (checking whether the pointer is non-NULL). READ_ONCE() is *not* sufficient in that case. The optimization of replacing smp_load_acquire() with READ_ONCE() is discouraged for nontrivial data structures, since it can be difficult to determine if it is correct. In particular, for complex data structures the correctness of the READ_ONCE() optimization may depend on internal implementation details of other kernel subsystems.
next prev parent reply index Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-07-17 4:44 Eric Biggers 2020-07-17 5:49 ` Sedat Dilek 2020-07-17 12:35 ` Matthew Wilcox 2020-07-17 14:26 ` Alan Stern 2020-07-17 17:47 ` Matthew Wilcox 2020-07-17 17:51 ` Alan Stern 2020-07-18 1:02 ` Eric Biggers 2020-07-27 12:51 ` Matthew Wilcox 2020-07-17 21:05 ` Darrick J. Wong 2020-07-18 0:44 ` Darrick J. Wong 2020-07-18 1:38 ` Eric Biggers 2020-07-18 2:13 ` Matthew Wilcox 2020-07-18 5:28 ` Eric Biggers 2020-07-18 14:35 ` Alan Stern 2020-07-20 2:07 ` Dave Chinner 2020-07-20 9:00 ` Peter Zijlstra 2020-07-27 15:17 ` Alan Stern 2020-07-27 15:28 ` Matthew Wilcox 2020-07-27 16:01 ` Paul E. McKenney 2020-07-27 16:31 ` Alan Stern 2020-07-27 16:59 ` Matthew Wilcox 2020-07-27 19:13 ` Alan Stern 2020-07-17 20:53 ` Darrick J. Wong 2020-07-18 0:58 ` Eric Biggers 2020-07-18 1:25 ` Alan Stern 2020-07-18 1:40 ` Matthew Wilcox 2020-07-18 2:00 ` Dave Chinner 2020-07-18 14:21 ` Alan Stern 2020-07-18 2:00 ` Eric Biggers [this message] 2020-07-18 1:42 ` Dave Chinner 2020-07-18 14:08 ` Alan Stern 2020-07-20 1:33 ` Dave Chinner 2020-07-20 14:52 ` Alan Stern 2020-07-20 15:37 ` Darrick J. Wong 2020-07-20 15:39 ` Matthew Wilcox 2020-07-20 16:04 ` Paul E. McKenney 2020-07-20 16:48 ` peterz 2020-07-20 22:06 ` Paul E. McKenney 2020-07-20 16:12 ` Alan Stern
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20200718020009.GE2183@sol.localdomain \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ --firstname.lastname@example.org \ --email@example.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
Linux-Fsdevel Archive on lore.kernel.org Archives are clonable: git clone --mirror https://lkml.kernel.org/linux-fsdevel/0 linux-fsdevel/git/0.git # If you have public-inbox 1.1+ installed, you may # initialize and index your mirror using the following commands: public-inbox-init -V2 linux-fsdevel linux-fsdevel/ https://lkml.kernel.org/linux-fsdevel \ firstname.lastname@example.org public-inbox-index linux-fsdevel Example config snippet for mirrors Newsgroup available over NNTP: nntp://nntp.lore.kernel.org/org.kernel.vger.linux-fsdevel AGPL code for this site: git clone https://public-inbox.org/public-inbox.git