Re: [PATCH v3 1/4] fs: kiocb: introduce IOCB_DONT_DIRTY_PAGE flag for direct IO

From: Ming Lei <ming.lei@canonical.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Dave Kleikamp <dave.kleikamp@oracle.com>,
	Jens Axboe <axboe@kernel.dk>, Zach Brown <zab@zabbo.net>,
	Maxim Patlasov <mpatlasov@parallels.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Tejun Heo <tj@kernel.org>
Subject: Re: [PATCH v3 1/4] fs: kiocb: introduce IOCB_DONT_DIRTY_PAGE flag for direct IO
Date: Thu, 7 May 2015 20:12:31 +0800	[thread overview]
Message-ID: <CACVXFVOJH-7eKvpFe+9vN_0kyo8Thik+Mn75PTtmNhMdyjxoVw@mail.gmail.com> (raw)
In-Reply-To: <20150507071618.GA7595@infradead.org>

On Thu, May 7, 2015 at 3:16 PM, Christoph Hellwig <hch@infradead.org> wrote:
> On Thu, May 07, 2015 at 01:08:23AM +0800, Ming Lei wrote:
>> When direct IO is submitted from kernel, it is often unnecessary
>> to dirty pages, for example of loop, dirtying pages have been
>> considered in the upper filesystem(over loop) side already, and
>> they don't need to be dirtied again.
>>
>> So this patch introduces IOCB_DONT_DIRTY_PAGE flag for direct IO,
>> and other ITER_BVEC and ITER_KVEC cases might benefit from it too.
>>
>> The patch is based on previous Dave's patch.
>
> in-kernel I/O should never dirty the pages written to or read from,
> so just checking the iov_iter type should be enough.

In case of loop, it is quite specific about dirtying READ pages in
direct IO because fs over loop has considered dirtying these pages
already.

For other cases of ITER_BVEC or ITER_KVEC, if the page is
anonymous or mapped, dirtying is still needed, otherwise
the new written data may be dropped like direct-io from user space.

So I suggest to use the flag of IOCB_DONT_DIRTY_PAGE for
the purpose.

Thanks,
Ming Lei