Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <24c88c4f-aea5-1fb7-0ead-95c88629d72b@marcan.st>
Date:   Tue, 16 Aug 2022 15:28:50 +0900
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:91.0)
 Gecko/20100101 Thunderbird/91.12.0
Subject: Re: [PATCH] workqueue: Fix memory ordering race in queue_work*()
Content-Language: en-US
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Herbert Xu <herbert@gondor.apana.org.au>,
        Will Deacon <will@kernel.org>, Tejun Heo <tj@kernel.org>,
        peterz@infradead.org, jirislaby@kernel.org, maz@kernel.org,
        mark.rutland@arm.com, boqun.feng@gmail.com,
        catalin.marinas@arm.com, oneukum@suse.com,
        roman.penyaev@profitbricks.com, asahi@lists.linux.dev,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        stable@vger.kernel.org
References: <YvqaK3hxix9AaQBO@slm.duckdns.org>
 <YvsZ6vObgLaDeSZk@gondor.apana.org.au>
 <CAHk-=wgSNiT5qJX53RHtWECsUiFq6d6VWYNAvu71ViOEan07yw@mail.gmail.com>
 <cd51b422-89f3-1856-5d3b-d6e5b0029085@marcan.st>
 <CAHk-=wjfLT7nL8pV8RWATpjgm0zDtUwT8UMtroqnGcXRjN8tgw@mail.gmail.com>
From:   Hector Martin <marcan@marcan.st>
In-Reply-To: <CAHk-=wjfLT7nL8pV8RWATpjgm0zDtUwT8UMtroqnGcXRjN8tgw@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2022/08/16 14:52, Linus Torvalds wrote:
> I think I understand *why* it's broken - it looks like a "harmless"
> optimization. After all, if the bitop doesn't do anything, there's
> nothing to order it with.
> 
> It makes a certain amount of sense - as long as you don't think about
> it too hard.
> 
> The reason it is completely and utterly broken is that it's not
> actually just "the bitop doesn't do anything". Even when it doesn't
> change the bit value, just the ordering of the read of the old bit
> value can be meaningful, exactly for that case of "I added more work
> to the queue, I need to set the bit to tell the consumers, and if I'm
> the first person to set the bit I may need to wake the consumer up".

This is the same reason I argued queue_work() itself needs to have a
similar guarantee, even when it doesn't queue work (and I updated the
doc to match). If test_and_set_bit() is used in this kind of context
often in the kernel, clearly the current implementation/doc clashes with
that.

As I said, I don't have any particular beef in this fight, but this is
horribly broken on M1/2 right now, so I'll send a patch to change the
bitops instead and you all can fight it out over which way is correct :)

-- 
Hector Martin (marcan@marcan.st)
Public Key: https://mrcn.st/pub