[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <48ECD9C8.4000700@redhat.com>
Date: Wed, 08 Oct 2008 12:03:20 -0400
From: Steven Rostedt <srostedt@...hat.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
CC: Steven Rostedt <rostedt@...dmis.org>,
Jeremy Fitzhardinge <jeremy@...p.org>,
Peter Zijlstra <a.p.zijlstra@...llo.nl>,
David Miller <davem@...emloft.net>, hugh@...itas.com,
mingo@...e.hu, akpm@...ux-foundation.org,
linux-kernel@...r.kernel.org, davej@...hat.com
Subject: Re: [RFC][PATCH 7/7] lockdep: spin_lock_nest_lock()
Linus Torvalds wrote:
> On Wed, 8 Oct 2008, Steven Rostedt wrote:
>
>>> And yes, if there is an outer lock, even the order of getting locks is
>>> irrelevant, as long as anybody who gets more than one inner lock always
>>> holds the outer one.
>>>
>> But I need to disagree on a programming practice style. Unlocking locks
>> in a non nested order is just bad programming practice.
>>
>
> No it is not.
>
> Unlocking locks in non-nested order can sometimes be very much the rigth
> thing to do, and thinking otherwise is (a) naive and (b) can generate
> totally unnecessary and pointless bugs.
>
> The thing is, sometimes you have to do it, and imposing totally made-up
> rules ("unlocks have to nest") just confuses everybody.
>
> The FACT is, that unlocks do not have to nest cleanly. That's a rock solid
> *FACT*. The locking order matters, and the unlocking order does not.
>
> And if you cannot accept that as a fact, and you then say "unlock order
> should matter just to keep things nice and clean", then you end up being
> screwed and/or confused when you can't hold to the unlock order.
>
> There are many perfectly valid reasons not to unlock in reverse order.
> Don't create make-believe rules that break those reasons for no gain.
>
Unfortunately, you cut out my comment that I stated "unless there is a
good reason not to",
which the below example is a good reason ;-)
> For example:
> - let's say that you have a singly-linked list of objects.
> - you need to lock all objects, do something, and then unlock all
> objects.
> - the *only* sane way to do that is to just traverse the list twice.
> - that means that you will unlock the objects in the same order you
> locked them, _not_ in reverse ("nested") order.
> - if you force a rule of "unlocks must be nested", then
>
> YOU ARE A F*CKING MORON.
>
> It's that simple. Don't do made-up rules that have no technical reason for
> them.
>
> Lock ordering matters. Unlock ordering does not. It really is that simple.
> Don't confuse the issue by claiming anything else.
>
Yes, I totally agree that there are good reasons not to unlock in
reverse order, and the example
you gave happens to be one of them.
I just find that seeing something like:
lock(A);
lock(B);
[do something]
unlock(A);
unlock(B);
just seems to be sloppy.
I wont harp on this, it only came up in conversation in which someone
pointed out your
post.
-- Steve
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists