Date: 31 Aug 2015 11:21:25 -0400
Message-ID: <20150831152125.2143.qmail@ns.horizon.com>
From: "George Spelvin" <linux@horizon.com>
To: linux@horizon.com, oleg@redhat.com
Subject: Re: [PATCH] task_work: remove fifo ordering guarantee
Cc: eric.dumazet@gmail.com, linux-kernel@vger.kernel.org, mingo@kernel.org
In-Reply-To: <20150831132210.GD31015@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1214
Lines: 30

Oleg Nesterov wrote:
> Actually you need a single tail pointer, See 158e1645e07f3e9f7e49.
> But this doesn't matter.

(This uses a circularly linked list, keeping a pointer to the
tail, and tail->next pointing to the head.)

Yes, if you're not trying to be lockless, that works quite well.

> And this also means that the next writer which does task_work_add() +
> task_work_cancel() will be suprised. Worse, this means that work->func()
> doesn't own its callback_head/container_of. The previous tail is visible
> to task_work_run().

I forgot about task_work_cancel()!  Yes, supporting that as well would
be a problem, and make a messy algorithm even messier.

> Perhaps I missed something. But to me this all looks too clever ;)
> Personally I'd prefer to just add another spinlock_t.
>
> But so far I hope we can keep this stupid but simple "reverse the list"
> loop.

It *is* too complicated.  That's why I described it: to discourage people
from implementing it.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/