linux-kernel - Re: [PATCH v6 5/7] xfs: Support atomic write for statx

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <753babd8-de34-422f-9958-a2a06b503ca6@oracle.com>
Date: Tue, 1 Oct 2024 09:29:35 +0100
From: John Garry <john.g.garry@...cle.com>
To: "Darrick J. Wong" <djwong@...nel.org>
Cc: axboe@...nel.dk, brauner@...nel.org, viro@...iv.linux.org.uk, jack@...e.cz,
        dchinner@...hat.com, hch@....de, cem@...nel.org,
        linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-xfs@...r.kernel.org, linux-fsdevel@...r.kernel.org, hare@...e.de,
        martin.petersen@...cle.com, catherine.hoang@...cle.com,
        mcgrof@...nel.org, ritesh.list@...il.com, ojaswin@...ux.ibm.com
Subject: Re: [PATCH v6 5/7] xfs: Support atomic write for statx

On 30/09/2024 17:37, Darrick J. Wong wrote:
> On Mon, Sep 30, 2024 at 12:54:36PM +0000, John Garry wrote:
>> Support providing info on atomic write unit min and max for an inode.
>>
>> For simplicity, currently we limit the min at the FS block size. As for
>> max, we limit also at FS block size, as there is no current method to
>> guarantee extent alignment or granularity for regular files.
>>
>> Signed-off-by: John Garry <john.g.garry@...cle.com>
>> ---
>>   fs/xfs/xfs_inode.h | 17 +++++++++++++++++
>>   fs/xfs/xfs_iops.c  | 24 ++++++++++++++++++++++++
>>   2 files changed, 41 insertions(+)
>>
>> diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
>> index 1c62ee294a5a..1ea73402d592 100644
>> --- a/fs/xfs/xfs_inode.h
>> +++ b/fs/xfs/xfs_inode.h
>> @@ -332,6 +332,23 @@ static inline bool xfs_inode_has_atomicwrites(struct xfs_inode *ip)
>>   	return ip->i_diflags2 & XFS_DIFLAG2_ATOMICWRITES;
>>   }
>>   
>> +static inline bool
>> +xfs_inode_can_atomicwrite(
>> +	struct xfs_inode	*ip)
>> +{
>> +	struct xfs_mount	*mp = ip->i_mount;
>> +	struct xfs_buftarg	*target = xfs_inode_buftarg(ip);
>> +
>> +	if (!xfs_inode_has_atomicwrites(ip))
>> +		return false;
>> +	if (mp->m_sb.sb_blocksize < target->bt_bdev_awu_min)
>> +		return false;
>> +	if (mp->m_sb.sb_blocksize > target->bt_bdev_awu_max)
>> +		return false;
>> +
>> +	return true;
>> +}
>> +
>>   /*
>>    * In-core inode flags.
>>    */
>> diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
>> index ee79cf161312..915d057db9bb 100644
>> --- a/fs/xfs/xfs_iops.c
>> +++ b/fs/xfs/xfs_iops.c
>> @@ -570,6 +570,23 @@ xfs_stat_blksize(
>>   	return max_t(uint32_t, PAGE_SIZE, mp->m_sb.sb_blocksize);
>>   }
>>   
>> +static void
>> +xfs_get_atomic_write_attr(
>> +	struct xfs_inode	*ip,
>> +	unsigned int		*unit_min,
>> +	unsigned int		*unit_max)
>> +{
>> +	struct xfs_mount	*mp = ip->i_mount;
>> +	struct xfs_sb		*sbp = &mp->m_sb;
>> +
>> +	if (!xfs_inode_can_atomicwrite(ip)) {
>> +		*unit_min = *unit_max = 0;
>> +		return;
>> +	}
>> +
>> +	*unit_min = *unit_max = sbp->sb_blocksize;
> 
> Ok, so we're only supporting untorn writes if they're exactly the fs
> blocksize, and 1 fsblock is between awu_min/max.  That simplifies a lot
> of things. :)
> 
> Not supporting sub-fsblock atomic writes means that we'll never hit the
> directio COW fallback code, which uses the pagecache.

My original idea (with forcealign) was to support 1FSB and larger.

I suppose that with a larger FS block size we might want to support 
sub-fsblock atomic writes. However, for the moment, I don't see a need 
to support this.

> 
> Not supporting multi-fsblock atomic writes means that you don't have to
> figure out how to ensure that we always do cow on forcealign
> granularity.  Though as I pointed out elsewhere in this thread, that's a
> forcealign problem.

Sure

> 
> Yay! ;)
> 
>> +}
>> +
>>   STATIC int
>>   xfs_vn_getattr(
>>   	struct mnt_idmap	*idmap,
>> @@ -643,6 +660,13 @@ xfs_vn_getattr(
>>   			stat->dio_mem_align = bdev_dma_alignment(bdev) + 1;
>>   			stat->dio_offset_align = bdev_logical_block_size(bdev);
>>   		}
>> +		if (request_mask & STATX_WRITE_ATOMIC) {
>> +			unsigned int unit_min, unit_max;
>> +
>> +			xfs_get_atomic_write_attr(ip, &unit_min, &unit_max);
>> +			generic_fill_statx_atomic_writes(stat,
>> +				unit_min, unit_max);
> 
> Consistent indenting and wrapping, please:

ok

> 
> 			xfs_get_atomic_write_attr(ip, &unit_min,
> 					&unit_max);
> 			generic_fill_statx_atomic_writes(stat,
> 					unit_min, unit_max);
> 
> 
> With that fixed,
> Reviewed-by: Darrick J. Wong <djwong@...nel.org>

ok, thanks!