Tarantool development patches archive
 help / color / mirror / Atom feed
From: Vladislav Shpilevoy via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: Serge Petrenko <sergepetrenko@tarantool.org>, gorcunov@gmail.com
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH v3] wal: introduce limits on simultaneous writes
Date: Mon, 1 Mar 2021 23:05:26 +0100	[thread overview]
Message-ID: <a2a7d894-eb5d-d2db-66fc-dc5cd3f33a7a@tarantool.org> (raw)
In-Reply-To: <17046c4d-b239-f3d8-f150-941026096b1e@tarantool.org>

Hi! Thanks for the fixes!

>>> @@ -159,6 +276,13 @@ journal_write(struct journal_entry *entry)
>>>   static inline int
>>>   journal_write_async(struct journal_entry *entry)
>>>   {
>>> +    /*
>>> +     * It's the job of the caller to check whether the queue is full prior
>>> +     * to submitting the request.
>>> +     */
>>> +    assert(!journal_queue_is_full() || journal_queue.is_ready);
>>> +    journal_queue_on_append(entry);
>> 8. Probably must assert that waiters list is empty. Otherwise you
>> could go out of order.
> 
> It's not empty by the time the first entry gets to 'journal_write_async'.
> Everyone else is waken up, but not yet removed from the queue.
> 
> Looks like we cannot determine whether a write is called after waiting in queue
> or not.

It bothers me a lot, the rest looks good. We don't have any protection against
a reorder, even not an assertion. How about this? (I didn't test):

====================
--- a/src/box/journal.c
+++ b/src/box/journal.c
@@ -69,8 +69,11 @@ void
 journal_queue_wakeup(void)
 {
 	struct rlist *list = &journal_queue.waiters;
-	if (!rlist_empty(list) && !journal_queue_is_full())
+	if (!rlist_empty(list) && !journal_queue_is_full()) {
 		fiber_wakeup(rlist_first_entry(list, struct fiber, state));
+		--journal_queue.waiter_count;
+		assert(journal_queue.waiter_count >= 0);
+	}
 }
 
 void
@@ -84,7 +87,6 @@ journal_queue_wait(void)
 	 * Will be waken up by either queue emptying or a synchronous write.
 	 */
 	fiber_yield();
-	--journal_queue.waiter_count;
 	journal_queue_wakeup();
 }
 
@@ -96,5 +98,6 @@ journal_queue_flush(void)
 	struct rlist *list = &journal_queue.waiters;
 	while (!rlist_empty(list))
 		fiber_wakeup(rlist_first_entry(list, struct fiber, state));
+	journal_queue.waiter_count = 0;
 	journal_queue_wait();
 }
diff --git a/src/box/journal.h b/src/box/journal.h
index 5f0e0accd..ea56043e2 100644
--- a/src/box/journal.h
+++ b/src/box/journal.h
@@ -260,6 +260,7 @@ journal_write_async(struct journal_entry *entry)
 	 * to submitting the request.
 	 */
 	journal_queue_on_append(entry);
+	assert(journal_queue.waiter_count == 0);
 
 	return current_journal->write_async(current_journal, entry);
 }

  reply	other threads:[~2021-03-01 22:05 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-24 19:35 Serge Petrenko via Tarantool-patches
2021-02-24 19:40 ` Serge Petrenko via Tarantool-patches
2021-02-25 13:05 ` Konstantin Osipov via Tarantool-patches
2021-02-26  0:57   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-26  7:18     ` Konstantin Osipov via Tarantool-patches
2021-02-26 20:23       ` Vladislav Shpilevoy via Tarantool-patches
2021-02-26 21:20         ` Konstantin Osipov via Tarantool-patches
2021-02-26 22:44           ` Vladislav Shpilevoy via Tarantool-patches
2021-02-27 13:27             ` Konstantin Osipov via Tarantool-patches
2021-03-01 19:15   ` Serge Petrenko via Tarantool-patches
2021-03-01 21:46     ` Konstantin Osipov via Tarantool-patches
2021-02-26  0:56 ` Vladislav Shpilevoy via Tarantool-patches
2021-03-01 19:08   ` Serge Petrenko via Tarantool-patches
2021-03-01 22:05     ` Vladislav Shpilevoy via Tarantool-patches [this message]
2021-03-02 17:51       ` Serge Petrenko via Tarantool-patches
2021-03-03 20:59         ` Vladislav Shpilevoy via Tarantool-patches
2021-03-09 15:10           ` Serge Petrenko via Tarantool-patches
2021-03-09 19:49 ` Vladislav Shpilevoy via Tarantool-patches
2021-03-10  8:18   ` Konstantin Osipov via Tarantool-patches
2021-03-12 17:10     ` Serge Petrenko via Tarantool-patches
2021-03-13 19:14       ` Konstantin Osipov via Tarantool-patches
2021-03-15 23:42       ` Vladislav Shpilevoy via Tarantool-patches
2021-03-16  6:45         ` Konstantin Osipov via Tarantool-patches
2021-03-16 20:27           ` Vladislav Shpilevoy via Tarantool-patches
2021-03-16 10:19         ` Serge Petrenko via Tarantool-patches
2021-03-16 20:48           ` Vladislav Shpilevoy via Tarantool-patches
2021-03-17 12:14             ` Serge Petrenko via Tarantool-patches
2021-03-17 21:02           ` Vladislav Shpilevoy via Tarantool-patches
2021-03-19 11:32             ` Serge Petrenko via Tarantool-patches
2021-03-19 15:36 ` Kirill Yukhin via Tarantool-patches

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=a2a7d894-eb5d-d2db-66fc-dc5cd3f33a7a@tarantool.org \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=gorcunov@gmail.com \
    --cc=sergepetrenko@tarantool.org \
    --cc=v.shpilevoy@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH v3] wal: introduce limits on simultaneous writes' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox