Re: [PATCH RFC 10/10] xfs: Allow block allocator to take an alignment hint

[John Garry <john.g.garry@oracle.com>]

On 05/02/2025 19:20, Darrick J. Wong wrote:
> On Tue, Feb 04, 2025 at 12:01:27PM +0000, John Garry wrote:
>> When issuing an atomic write by the CoW method, give the block allocator a
>> hint to naturally align the data blocks.
>>
>> This means that we have a better chance to issuing the atomic write via
>> HW offload next time.
>>
>> Signed-off-by: John Garry <john.g.garry@oracle.com>
>> ---
>>   fs/xfs/libxfs/xfs_bmap.c | 7 ++++++-
>>   fs/xfs/libxfs/xfs_bmap.h | 6 +++++-
>>   fs/xfs/xfs_reflink.c     | 8 ++++++--
>>   3 files changed, 17 insertions(+), 4 deletions(-)
>>
>> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
>> index 40ad22fb808b..7a3910018dee 100644
>> --- a/fs/xfs/libxfs/xfs_bmap.c
>> +++ b/fs/xfs/libxfs/xfs_bmap.c
>> @@ -3454,6 +3454,12 @@ xfs_bmap_compute_alignments(
>>   		align = xfs_get_cowextsz_hint(ap->ip);
>>   	else if (ap->datatype & XFS_ALLOC_USERDATA)
>>   		align = xfs_get_extsz_hint(ap->ip);
>> +
>> +	if (align > 1 && ap->flags & XFS_BMAPI_NALIGN)
>> +		args->alignment = align;
>> +	else
>> +		args->alignment = 1;
>> +
>>   	if (align) {
>>   		if (xfs_bmap_extsize_align(mp, &ap->got, &ap->prev, align, 0,
>>   					ap->eof, 0, ap->conv, &ap->offset,
>> @@ -3781,7 +3787,6 @@ xfs_bmap_btalloc(
>>   		.wasdel		= ap->wasdel,
>>   		.resv		= XFS_AG_RESV_NONE,
>>   		.datatype	= ap->datatype,
>> -		.alignment	= 1,
>>   		.minalignslop	= 0,
>>   	};
>>   	xfs_fileoff_t		orig_offset;
>> diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
>> index 4b721d935994..d68b594c3fa2 100644
>> --- a/fs/xfs/libxfs/xfs_bmap.h
>> +++ b/fs/xfs/libxfs/xfs_bmap.h
>> @@ -87,6 +87,9 @@ struct xfs_bmalloca {
>>   /* Do not update the rmap btree.  Used for reconstructing bmbt from rmapbt. */
>>   #define XFS_BMAPI_NORMAP	(1u << 10)
>>   
>> +/* Try to naturally align allocations */
>> +#define XFS_BMAPI_NALIGN	(1u << 11)
>> +
>>   #define XFS_BMAPI_FLAGS \
>>   	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
>>   	{ XFS_BMAPI_METADATA,	"METADATA" }, \
>> @@ -98,7 +101,8 @@ struct xfs_bmalloca {
>>   	{ XFS_BMAPI_REMAP,	"REMAP" }, \
>>   	{ XFS_BMAPI_COWFORK,	"COWFORK" }, \
>>   	{ XFS_BMAPI_NODISCARD,	"NODISCARD" }, \
>> -	{ XFS_BMAPI_NORMAP,	"NORMAP" }
>> +	{ XFS_BMAPI_NORMAP,	"NORMAP" },\
>> +	{ XFS_BMAPI_NALIGN,	"NALIGN" }
> 
> Tihs isn't really "naturally" aligned, is it?  It really means "try to
> align allocations to the extent size hint", which isn't required to be a
> power of two.

Sure, so I would expect that the user will set extsize/cowextsize 
according to the size what we want to do atomics for, and we can align 
to that. I don't think that it makes a difference that either extsize 
isn't mandated to be a power-of-2.

So then I should rename to XFS_BMAPI_EXTSZALIGN or something like that - ok?

Thanks,
John


> 
> --D
> 
>>   
>>   
>>   static inline int xfs_bmapi_aflag(int w)
>> diff --git a/fs/xfs/xfs_reflink.c b/fs/xfs/xfs_reflink.c
>> index 60c986300faa..198fb5372f10 100644
>> --- a/fs/xfs/xfs_reflink.c
>> +++ b/fs/xfs/xfs_reflink.c
>> @@ -445,6 +445,11 @@ xfs_reflink_fill_cow_hole(
>>   	int			nimaps;
>>   	int			error;
>>   	bool			found;
>> +	uint32_t		bmapi_flags = XFS_BMAPI_COWFORK |
>> +					XFS_BMAPI_PREALLOC;
>> +
>> +	if (atomic)
>> +		bmapi_flags |= XFS_BMAPI_NALIGN;
>>   
>>   	resaligned = xfs_aligned_fsb_count(imap->br_startoff,
>>   		imap->br_blockcount, xfs_get_cowextsz_hint(ip));
>> @@ -478,8 +483,7 @@ xfs_reflink_fill_cow_hole(
>>   	/* Allocate the entire reservation as unwritten blocks. */
>>   	nimaps = 1;
>>   	error = xfs_bmapi_write(tp, ip, imap->br_startoff, imap->br_blockcount,
>> -			XFS_BMAPI_COWFORK | XFS_BMAPI_PREALLOC, 0, cmap,
>> -			&nimaps);
>> +			bmapi_flags, 0, cmap, &nimaps);
>>   	if (error)
>>   		goto out_trans_cancel;
>>   
>> -- 
>> 2.31.1
>>
>>