Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sat, 18 Jan 2020 17:42:13 -0800
From:   Matthew Wilcox <willy@infradead.org>
To:     "yukuai (C)" <yukuai3@huawei.com>
Cc:     hch@infradead.org, darrick.wong@oracle.com,
        linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org, houtao1@huawei.com,
        zhengbin13@huawei.com, yi.zhang@huawei.com
Subject: Re: [RFC] iomap: fix race between readahead and direct write
Message-ID: <20200119014213.GA16943@bombadil.infradead.org>
References: <20200116063601.39201-1-yukuai3@huawei.com>
 <20200118230826.GA5583@bombadil.infradead.org>
 <f5328338-1a2d-38b4-283f-3fb97ad37133@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <f5328338-1a2d-38b4-283f-3fb97ad37133@huawei.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun, Jan 19, 2020 at 09:34:32AM +0800, yukuai (C) wrote:
> 
> 
> On 2020/1/19 7:08, Matthew Wilcox wrote:
> > It's worth noting that my patch series from earlier this week to
> > redesign the readahead API will fix this problem.  Direct write will block
> > on the locked pages in the page cache.
> 
> Thank you for your response!
> 
> In this case, direct write finish while page do not exist in the page
> cache. This is the fundamental condition of the race, because readahead
> won't allocate page if page exist in page cache.
> 
> By the way, in the current logic, if page exist in page cache, direct
> write need to hold lock for page in invalidate_inode_pages2_range().

Did you read my patch series?  The current code allocates pages,
but does not put them in the page cache until after iomap is called.
My patch series changes that to put the pages in the page cache as soon
as they're allocated, and before iomap is called.