From: Petr Mladek <pmladek@suse.com>
To: John Ogness <john.ogness@linutronix.de>
Cc: linux-kernel@vger.kernel.org,
Andrea Parri <andrea.parri@amarulasolutions.com>,
Sergey Senozhatsky <sergey.senozhatsky@gmail.com>,
Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>,
Steven Rostedt <rostedt@goodmis.org>,
Brendan Higgins <brendanhiggins@google.com>,
Peter Zijlstra <peterz@infradead.org>,
Thomas Gleixner <tglx@linutronix.de>,
Linus Torvalds <torvalds@linux-foundation.org>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: dataring_push() barriers Re: [RFC PATCH v4 1/9] printk-rb: add a new printk ringbuffer implementation
Date: Tue, 20 Aug 2019 15:50:04 +0200 [thread overview]
Message-ID: <20190820135004.7vatbrzphfsgsnw2@pathway.suse.cz> (raw)
In-Reply-To: <20190807222634.1723-2-john.ogness@linutronix.de>
On Thu 2019-08-08 00:32:26, John Ogness wrote:
> +/**
> + * dataring_push() - Reserve a data block in the data array.
> + *
> + * @dr: The data ringbuffer to reserve data in.
> + *
> + * @size: The size to reserve.
> + *
> + * @desc: A pointer to a descriptor to store the data block information.
> + *
> + * @id: The ID of the descriptor to be associated.
> + * The data block will not be set with @id, but rather initialized with
> + * a value that is explicitly different than @id. This is to handle the
> + * case when newly available garbage by chance matches the descriptor
> + * ID.
> + *
> + * This function expects to move the head pointer forward. If this would
> + * result in overtaking the data array index of the tail, the tail data block
> + * will be invalidated.
> + *
> + * Return: A pointer to the reserved writer data, otherwise NULL.
> + *
> + * This will only fail if it was not possible to invalidate the tail data
> + * block.
> + */
> +char *dataring_push(struct dataring *dr, unsigned int size,
> + struct dr_desc *desc, unsigned long id)
> +{
> + unsigned long begin_lpos;
> + unsigned long next_lpos;
> + struct dr_datablock *db;
> + bool ret;
> +
> + to_db_size(&size);
> +
> + do {
> + /* fA: */
> + ret = get_new_lpos(dr, size, &begin_lpos, &next_lpos);
> +
> + /*
> + * fB:
> + *
> + * The data ringbuffer tail may have been pushed (by this or
> + * any other task). The updated @tail_lpos must be visible to
> + * all observers before changes to @begin_lpos, @next_lpos, or
> + * @head_lpos by this task are visible in order to allow other
> + * tasks to recognize the invalidation of the data
> + * blocks.
This sounds strange. The write barrier should be done only on CPU
that really modified tail_lpos. I.e. it should be in _dataring_pop()
after successful dr->tail_lpos modification.
> + * This pairs with the smp_rmb() in _dataring_pop() as well as
> + * any reader task using smp_rmb() to post-validate data that
> + * has been read from a data block.
> +
> + * Memory barrier involvement:
> + *
> + * If dE reads from fE, then dI reads from fA->eA.
> + * If dC reads from fG, then dI reads from fA->eA.
> + * If dD reads from fH, then dI reads from fA->eA.
> + * If mC reads from fH, then mF reads from fA->eA.
> + *
> + * Relies on:
> + *
> + * FULL MB between fA->eA and fE
> + * matching
> + * RMB between dE and dI
> + *
> + * FULL MB between fA->eA and fG
> + * matching
> + * RMB between dC and dI
> + *
> + * FULL MB between fA->eA and fH
> + * matching
> + * RMB between dD and dI
> + *
> + * FULL MB between fA->eA and fH
> + * matching
> + * RMB between mC and mF
> + */
> + smp_mb();
All these comments talk about sychronization against read barriers.
It means that we would need a write barrier here. But it does
not make much sense to do write barrier before actually
writing dr->head_lpos.
After all I think that we do not need any barrier here.
The write barrier for dr->tail_lpos should be in
_dataring_pop(). The read barrier is not needed because
we are not reading anything here.
Instead we should put a barrier after modyfying dr->head_lpos,
see below.
> + if (!ret) {
> + /*
> + * Force @desc permanently invalid to minimize risk
> + * of the descriptor later unexpectedly being
> + * determined as valid due to overflowing/wrapping of
> + * @head_lpos. An unaligned @begin_lpos can never
> + * point to a data block and having the same value
> + * for @begin_lpos and @next_lpos is also invalid.
> + */
> +
> + /* fC: */
> + WRITE_ONCE(desc->begin_lpos, 1);
> +
> + /* fD: */
> + WRITE_ONCE(desc->next_lpos, 1);
> +
> + return NULL;
> + }
> + /* fE: */
> + } while (atomic_long_cmpxchg_relaxed(&dr->head_lpos, begin_lpos,
> + next_lpos) != begin_lpos);
> +
We need a write barrier here to make sure that dr->head_lpos
is updated before we start updating other values, e.g.
db->id below.
Best Regards,
Petr
> + db = to_datablock(dr, begin_lpos);
> +
> + /*
> + * fF:
> + *
> + * @db->id is a garbage value and could possibly match the @id. This
> + * would be a problem because the data block would be considered
> + * valid before the writer has finished with it (i.e. before the
> + * writer has set @id). Force some other ID value.
> + */
> + WRITE_ONCE(db->id, id - 1);
>
> + /*
> + * fG:
> + *
> + * Ensure that @db->id is initialized to a wrong ID value before
> + * setting @begin_lpos so that there is no risk of accidentally
> + * matching a data block to a descriptor before the writer is finished
> + * with it (i.e. before the writer has set the correct @id). This
> + * pairs with the _acquire() in _dataring_pop().
> + *
> + * Memory barrier involvement:
> + *
> + * If dC reads from fG, then dF reads from fF.
> + *
> + * Relies on:
> + *
> + * RELEASE from fF to fG
> + * matching
> + * ACQUIRE from dC to dF
> + */
> + smp_store_release(&desc->begin_lpos, begin_lpos);
> +
> + /* fH: */
> + WRITE_ONCE(desc->next_lpos, next_lpos);
> +
> + /* If this data block wraps, use @data from the content data block. */
> + if (DATA_WRAPS(dr, begin_lpos) != DATA_WRAPS(dr, next_lpos))
> + db = to_datablock(dr, 0);
> +
> + return &db->data[0];
> +}
next prev parent reply other threads:[~2019-08-20 13:50 UTC|newest]
Thread overview: 131+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-07 22:26 [RFC PATCH v4 0/9] printk: new ringbuffer implementation John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 1/9] printk-rb: add a new printk " John Ogness
2019-08-20 8:15 ` numlist_pop(): " Petr Mladek
2019-08-21 5:41 ` John Ogness
2019-09-04 12:19 ` Peter Zijlstra
2019-08-20 8:22 ` assign_desc() barriers: " Petr Mladek
2019-08-20 14:14 ` Petr Mladek
2019-08-21 5:52 ` John Ogness
2019-08-22 11:53 ` Petr Mladek
2019-08-25 2:06 ` John Ogness
2019-08-26 8:21 ` John Ogness
2019-08-20 8:55 ` comments style: " Petr Mladek
2019-08-20 9:27 ` Sergey Senozhatsky
2019-08-21 5:46 ` John Ogness
2019-08-22 13:50 ` Petr Mladek
2019-08-22 17:38 ` Andrea Parri
2019-08-23 10:47 ` Petr Mladek
2019-08-23 14:27 ` Andrea Parri
2019-08-23 9:49 ` Sergey Senozhatsky
2019-08-23 5:54 ` Sergey Senozhatsky
2019-08-23 10:29 ` Petr Mladek
2019-08-21 5:42 ` John Ogness
2019-08-22 12:44 ` Petr Mladek
2019-08-20 13:50 ` Petr Mladek [this message]
2019-08-25 2:42 ` dataring_push() barriers " John Ogness
2019-08-27 14:36 ` Petr Mladek
2019-08-28 13:43 ` John Ogness
2019-08-20 15:12 ` datablock reuse races " Petr Mladek
2019-08-23 9:21 ` numlist_push() barriers " Petr Mladek
2019-08-26 8:34 ` Andrea Parri
2019-08-26 8:43 ` Andrea Parri
2019-08-26 14:10 ` Petr Mladek
2019-08-26 16:01 ` Andrea Parri
2019-08-26 22:36 ` John Ogness
2019-08-27 7:40 ` Petr Mladek
2019-08-27 14:28 ` John Ogness
2019-08-27 15:07 ` Petr Mladek
2019-08-28 10:24 ` John Ogness
2019-08-23 17:18 ` numlist API " Petr Mladek
2019-08-26 23:57 ` John Ogness
2019-08-27 13:03 ` Petr Mladek
2019-08-28 7:13 ` John Ogness
2019-08-28 8:58 ` Petr Mladek
2019-08-28 14:03 ` John Ogness
2019-08-29 11:28 ` Petr Mladek
2019-09-03 7:58 ` Sergey Senozhatsky
2019-08-30 14:48 ` dataring " Petr Mladek
2019-08-07 22:26 ` [RFC PATCH v4 2/9] printk-rb: add test module John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 3/9] printk-rb: fix missing includes/exports John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 4/9] printk-rb: initialize new descriptors as invalid John Ogness
2019-08-20 9:23 ` Petr Mladek
2019-08-20 10:16 ` Sergey Senozhatsky
2019-08-21 5:56 ` John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 5/9] printk-rb: remove extra data buffer size allocation John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 6/9] printk-rb: adjust test module ringbuffer sizes John Ogness
2019-08-19 21:29 ` [PATCH] printk-rb: fix test module macro usage John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 7/9] printk-rb: increase size of seq and size variables John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 8/9] printk-rb: new functionality to support printk John Ogness
2019-08-20 9:59 ` Sergey Senozhatsky
2019-08-21 5:47 ` John Ogness
2019-08-07 22:26 ` [RFC PATCH v4 9/9] printk: use a new ringbuffer implementation John Ogness
2019-08-08 19:07 ` Linus Torvalds
2019-08-08 22:55 ` John Ogness
2019-08-08 23:33 ` Linus Torvalds
2019-08-08 23:45 ` Steven Rostedt
2019-08-09 0:21 ` Linus Torvalds
2019-08-09 0:48 ` Steven Rostedt
2019-08-09 1:15 ` Linus Torvalds
2019-08-09 11:15 ` Thomas Gleixner
2019-08-09 16:00 ` Linus Torvalds
2019-08-09 20:07 ` Thomas Gleixner
2019-08-09 20:20 ` Linus Torvalds
2019-08-09 6:14 ` Peter Zijlstra
2019-08-09 7:08 ` John Ogness
2019-08-09 15:57 ` Linus Torvalds
2019-08-10 5:53 ` Thomas Gleixner
2019-09-10 3:19 ` Sergey Senozhatsky
2019-08-12 9:54 ` Geert Uytterhoeven
2019-08-16 5:46 ` Dave Young
2019-08-16 5:54 ` Dave Young
2019-08-16 9:40 ` John Ogness
2019-09-04 12:35 ` [RFC PATCH v4 0/9] printk: " Peter Zijlstra
2019-09-05 13:05 ` Petr Mladek
2019-09-05 14:31 ` Peter Zijlstra
2019-09-05 15:38 ` Thomas Gleixner
2019-09-05 16:11 ` Steven Rostedt
2019-09-05 21:10 ` John Ogness
2019-09-06 9:39 ` Petr Mladek
2019-09-09 14:11 ` printk meeting at LPC Thomas Gleixner
2019-09-13 13:26 ` John Ogness
2019-09-13 14:48 ` Daniel Vetter
2019-09-15 13:47 ` John Ogness
2019-09-16 8:44 ` Daniel Vetter
2019-09-16 4:30 ` Tetsuo Handa
2019-09-16 10:46 ` Petr Mladek
2019-09-16 13:43 ` Steven Rostedt
2019-09-16 14:28 ` John Ogness
2019-09-17 8:11 ` Petr Mladek
2019-09-17 7:52 ` Petr Mladek
2019-09-17 13:02 ` Steven Rostedt
2019-09-17 13:12 ` Greg Kroah-Hartman
2019-09-17 13:37 ` Steven Rostedt
2019-09-17 14:08 ` Tetsuo Handa
2019-09-17 7:51 ` Sergey Senozhatsky
2019-09-18 1:25 ` Sergey Senozhatsky
2019-09-18 2:08 ` Steven Rostedt
2019-09-18 2:36 ` Sergey Senozhatsky
2019-09-18 5:19 ` Sergey Senozhatsky
2019-09-18 7:42 ` John Ogness
2019-09-18 8:10 ` Sergey Senozhatsky
2019-09-18 9:05 ` John Ogness
2019-09-18 9:11 ` Sergey Senozhatsky
2019-09-18 16:41 ` Petr Mladek
2019-09-18 16:48 ` Steven Rostedt
2019-09-24 14:24 ` Petr Mladek
2019-09-19 8:06 ` Daniel Vetter
2019-09-18 7:33 ` John Ogness
2019-09-18 8:08 ` Sergey Senozhatsky
2019-10-04 14:48 ` Tony Asleson
2019-10-07 12:01 ` Petr Mladek
2019-09-06 9:06 ` [RFC PATCH v4 0/9] printk: new ringbuffer implementation Peter Zijlstra
2019-09-06 10:09 ` Sergey Senozhatsky
2019-09-06 10:49 ` Peter Zijlstra
2019-09-06 13:44 ` Sergey Senozhatsky
2019-09-06 12:42 ` Petr Mladek
2019-09-06 14:01 ` Peter Zijlstra
2019-09-06 14:22 ` Peter Zijlstra
2019-09-06 19:53 ` Sergey Senozhatsky
2019-09-06 22:47 ` John Ogness
2019-09-08 22:18 ` Peter Zijlstra
2019-09-10 3:22 ` Sergey Senozhatsky
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=20190820135004.7vatbrzphfsgsnw2@pathway.suse.cz \
--to=pmladek@suse.com \
--cc=andrea.parri@amarulasolutions.com \
--cc=brendanhiggins@google.com \
--cc=gregkh@linuxfoundation.org \
--cc=john.ogness@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=sergey.senozhatsky.work@gmail.com \
--cc=sergey.senozhatsky@gmail.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).