Received-SPF: pass (google.com: best guess record for domain of linux-ext4-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 18 Sep 2019 12:06:27 +0200
From:   Jan Kara <jack@suse.cz>
To:     "Theodore Y. Ts'o" <tytso@mit.edu>
Cc:     yangerkun <yangerkun@huawei.com>, jack@suse.cz,
        linux-ext4@vger.kernel.org, yi.zhang@huawei.com, houtao1@huawei.com
Subject: Re: [PATCH] ext4: fix a bug in ext4_wait_for_tail_page_commit
Message-ID: <20190918100627.GB25056@quack2.suse.cz>
References: <20190917084814.40370-1-yangerkun@huawei.com>
 <20190917153140.GF6762@mit.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190917153140.GF6762@mit.edu>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-ext4-owner@vger.kernel.org
Precedence: bulk

On Tue 17-09-19 11:31:40, Theodore Y. Ts'o wrote:
> On Tue, Sep 17, 2019 at 04:48:14PM +0800, yangerkun wrote:
> > No need to wait when offset equals to 0. And it will trigger a bug since
> > the latter __ext4_journalled_invalidatepage can free the buffers but leave
> > page still dirty.
> 
> That's only true if the block size == the page size, no?  If the
> offset is zero and the block size is 1k, we still need to wait.
> Shouldn't the better fix be:
> 
> > -	if (offset > PAGE_SIZE - i_blocksize(inode))
> > +	if (offset >= PAGE_SIZE - i_blocksize(inode))

No, what yangerkun wrote is correct. We don't have to wait for commit when
offset == 0 - truncate_inode_pages() should just happily process such page.
Also '>' in the above condition is correct. offset == PAGE_SIZE -
i_blocksize(inode) means one full block is getting truncated from the page
and we need to wait in that case to avoid jbd2_journal_invalidatepage()
failing with EBUSY when called from truncate_inode_pages().

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR