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:   Mon, 25 Nov 2019 20:03:20 +0100
From:   Javier =?utf-8?B?R29uesOhbGV6?= <javier@javigon.com>
To:     Damien Le Moal <Damien.LeMoal@wdc.com>
Cc:     "jaegeuk@kernel.org" <jaegeuk@kernel.org>,
        "yuchao0@huawei.com" <yuchao0@huawei.com>,
        "linux-f2fs-devel@lists.sourceforge.net" 
        <linux-f2fs-devel@lists.sourceforge.net>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        Javier =?utf-8?B?R29uesOhbGV6?= <javier.gonz@samsung.com>
Subject: Re: [PATCH] f2fs: disble physical prealloc in LSF mount
Message-ID: <20191125190320.g7beal27nc5ubju7@mpHalley>
References: <20191122085952.12754-1-javier@javigon.com>
 <BYAPR04MB58166AE029D919C6610D8404E74A0@BYAPR04MB5816.namprd04.prod.outlook.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <BYAPR04MB58166AE029D919C6610D8404E74A0@BYAPR04MB5816.namprd04.prod.outlook.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 25.11.2019 00:48, Damien Le Moal wrote:
>On 2019/11/22 18:00, Javier González wrote:
>> From: Javier González <javier.gonz@samsung.com>
>>
>> Fix file system corruption when using LFS mount (e.g., in zoned
>> devices). Seems like the fallback into buffered I/O creates an
>> inconsistency if the application is assuming both read and write DIO. I
>> can easily reproduce a corruption with a simple RocksDB test.
>>
>> Might be that the f2fs_forced_buffered_io path brings some problems too,
>> but I have not seen other failures besides this one.
>>
>> Problem reproducible without a zoned block device, simply by forcing
>> LFS mount:
>>
>>   $ sudo mkfs.f2fs -f -m /dev/nvme0n1
>>   $ sudo mount /dev/nvme0n1 /mnt/f2fs
>>   $ sudo  /opt/rocksdb/db_bench  --benchmarks=fillseq --use_existing_db=0
>>   --use_direct_reads=true --use_direct_io_for_flush_and_compaction=true
>>   --db=/mnt/f2fs --num=5000 --value_size=1048576 --verify_checksum=1
>>   --block_size=65536
>>
>> Note that the options that cause the problem are:
>>   --use_direct_reads=true --use_direct_io_for_flush_and_compaction=true
>>
>> Fixes: f9d6d0597698 ("f2fs: fix out-place-update DIO write")
>>
>> Signed-off-by: Javier González <javier.gonz@samsung.com>
>> ---
>>  fs/f2fs/data.c | 3 ---
>>  1 file changed, 3 deletions(-)
>>
>> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
>> index 5755e897a5f0..b045dd6ab632 100644
>> --- a/fs/f2fs/data.c
>> +++ b/fs/f2fs/data.c
>> @@ -1081,9 +1081,6 @@ int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from)
>>  			return err;
>>  	}
>>
>> -	if (direct_io && allow_outplace_dio(inode, iocb, from))
>> -		return 0;
>
>Since for LFS mode, all DIOs can end up out of place, I think that it
>may be better to change allow_outplace_dio() to always return true in
>the case of LFS mode. So may be something like:
>
>static inline int allow_outplace_dio(struct inode *inode,
>			struct kiocb *iocb, struct iov_iter *iter)
>{
>	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
>	int rw = iov_iter_rw(iter);
>
>	return test_opt(sbi, LFS) ||
>	 	(rw == WRITE && !block_unaligned_IO(inode, iocb, iter));
>}
>
>instead of the original:
>
>static inline int allow_outplace_dio(struct inode *inode,
>			struct kiocb *iocb, struct iov_iter *iter)
>{
>	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
>	int rw = iov_iter_rw(iter);
>
>	return (test_opt(sbi, LFS) && (rw == WRITE) &&
>				!block_unaligned_IO(inode, iocb, iter));
>}
>
>Thoughts ?
>

I see what you mean and it makes sense. However, the problem I am seeing
occurs when allow_outplace_dio() returns true, as this is what creates
the inconsistency between the write being buffered and the read being
DIO.

I did test the code you propose and, as expected, it still triggered the
corruption.

>> -
>>  	if (is_inode_flag_set(inode, FI_NO_PREALLOC))
>>  		return 0;
>>
>>
>
>
>-- 
>Damien Le Moal
>Western Digital Research

Thanks,
Javier