Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] fat: Avoid oops when bdi->io_pages==0
To:     OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        fsdevel <linux-fsdevel@vger.kernel.org>
References: <87ft85osn6.fsf@mail.parknet.co.jp>
 <b4e1f741-989c-6c9d-b559-4c1ada88c499@kernel.dk>
 <87o8mq6aao.fsf@mail.parknet.co.jp>
 <4010690f-20ad-f7ba-b595-2e07b0fa2d94@kernel.dk>
 <87h7si68hn.fsf@mail.parknet.co.jp>
From:   Jens Axboe <axboe@kernel.dk>
Message-ID: <1f4d6aef-c75c-dd0f-02cb-a1246079c429@kernel.dk>
Date:   Mon, 31 Aug 2020 11:19:49 -0600
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <87h7si68hn.fsf@mail.parknet.co.jp>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 8/31/20 11:16 AM, OGAWA Hirofumi wrote:
> Jens Axboe <axboe@kernel.dk> writes:
> 
>> On 8/31/20 10:37 AM, OGAWA Hirofumi wrote:
>>> Jens Axboe <axboe@kernel.dk> writes:
>>>
>>>> I don't think we should work-around this here. What device is this on?
>>>> Something like the below may help.
>>>
>>> The reported bug is from nvme stack, and the below patch (I submitted
>>> same patch to you) fixed the reported case though. But I didn't verify
>>> all possible path, so I'd liked to use safer side.
>>>
>>> If block layer can guarantee io_pages!=0 instead, and can apply to
>>> stable branch (5.8+). It would work too.
>>
>> We really should ensure that ->io_pages is always set, imho, instead of
>> having to work-around it in other spots.
> 
> I think it is good too. However, the issue would be how to do it for
> stable branch.

Agree

> If you think that block layer patch is enough and submit to stable
> (5.8+) branch instead, I'm fine without fatfs patch. (Or removing
> workaround in fatfs with block layer patch later?)

Well, it should catch any block based case as we then set it when
allocating the queue. So should do the trick for all block based
cases at least.

-- 
Jens Axboe