Re: [PATCH v14 14/21] xfs: add flags to xfs_free_eofblocks() to pass down to block processing

["Darrick J. Wong" <djwong@kernel.org>]

On Mon, Aug 03, 2026 at 10:08:04PM +0200, Andrey Albershteyn wrote:
> Add a flags parameter to xfs_free_eofblocks() to support selective
> extent unmapping. Add two flags for unmapping all extents (unwritten and
> normal) and fsverity leftover extents (only unwritten ones, leaving
> normal in place).

Er, why do we need this?  Is this to clear out unwritten merkle tree
blocks after a failed fsverity enrollment or something?  The commit
message should say a little bit more about why anyone needs this.

> Signed-off-by: Andrey Albershteyn <aalbersh@kernel.org>
> ---
>  fs/xfs/libxfs/xfs_bmap.c | 56 +++++++++++++++++++++++++++++-----------
>  fs/xfs/libxfs/xfs_bmap.h |  6 ++++-
>  fs/xfs/xfs_bmap_util.c   | 20 ++++++++++----
>  fs/xfs/xfs_bmap_util.h   | 13 +++++++++-
>  fs/xfs/xfs_file.c        |  2 +-
>  fs/xfs/xfs_icache.c      |  2 +-
>  fs/xfs/xfs_inode.c       |  2 +-
>  7 files changed, 76 insertions(+), 25 deletions(-)
> 
> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> index cc48f6e20e80..1d8d157a9dfa 100644
> --- a/fs/xfs/libxfs/xfs_bmap.c
> +++ b/fs/xfs/libxfs/xfs_bmap.c
> @@ -6144,15 +6144,12 @@ xfs_bmap_validate_extent(
>  			XFS_IS_REALTIME_INODE(ip), whichfork, irec);
>  }
>  
> -/*
> - * Used in xfs_itruncate_extents().  This is the maximum number of extents
> - * freed from a file in a single transaction.
> - */
> -#define	XFS_ITRUNC_MAX_EXTENTS	2
> -
>  /*
>   * Unmap every extent in part of an inode's fork.  We don't do any higher level
>   * invalidation work at all.
> + *
> + * The XFS_BMAPI_UNWRITTEN could be passed to remove only unwritten extents,
> + * leaving out normal extents in place.

What flags do we support here?  ATTRFORK, NODISCARD, and UNWRITTEN?  Can
that be documented in the comment or turned into asserts?

>   */
>  int
>  xfs_bunmapi_range(
> @@ -6162,23 +6159,52 @@ xfs_bunmapi_range(
>  	xfs_fileoff_t		startoff,
>  	xfs_fileoff_t		endoff)
>  {
> -	xfs_filblks_t		unmap_len = endoff - startoff + 1;
>  	int			error = 0;
> +	int			nimaps = 1;
> +	int			done = 0;
> +	struct xfs_bmbt_irec	imap;
> +	int			read_flags =
> +			flags & (XFS_BMAPI_ATTRFORK | XFS_BMAPI_ENTIRE);

Why BMAPI_ENTIRE?

> +	xfs_exntst_t		exntst = XFS_EXT_NORM;
>  
>  	xfs_assert_ilocked(ip, XFS_ILOCK_EXCL);
>  
> -	while (unmap_len > 0) {
> -		ASSERT((*tpp)->t_highest_agno == NULLAGNUMBER);
> -		error = __xfs_bunmapi(*tpp, ip, startoff, &unmap_len, flags,
> -				XFS_ITRUNC_MAX_EXTENTS);
> +	if (flags & XFS_BMAPI_UNWRITTEN)
> +		exntst = XFS_EXT_UNWRITTEN;
> +
> +	while (startoff < endoff) {
> +		nimaps = 1;
> +
> +		error = xfs_bmapi_read(ip, startoff, endoff - startoff + 1,
> +				&imap, &nimaps, read_flags);
>  		if (error)
>  			goto out;
>  
> -		/* free the just unmapped extents */
> -		error = xfs_defer_finish(tpp);
> -		if (error)
> +		if (nimaps == 0)
>  			goto out;
> -		cond_resched();
> +
> +		if ((exntst == XFS_EXT_UNWRITTEN) &&
> +				(imap.br_state != exntst)) {

No need for parentheses around these condition checks.

Would it be clearer if this was:

		/* caller only wants to unmap unwritten extents */
		if ((flags & XFS_BMAPI_UNWRITTEN) &&
		    imap.br_state != XFS_EXT_UNWRITTEN) {
			...
		}

> +			startoff = imap.br_startoff + imap.br_blockcount;
> +			continue;
> +		}
> +
> +		done = 0;
> +		while (!done) {
> +			ASSERT((*tpp)->t_highest_agno == NULLAGNUMBER);
> +			error = xfs_bunmapi(*tpp, ip, imap.br_startoff,
> +					imap.br_blockcount, flags, 0, &done);
> +			if (error)
> +				goto out;
> +
> +			/* free the just unmapped extent */
> +			error = xfs_defer_finish(tpp);
> +			if (error)
> +				goto out;
> +			cond_resched();
> +		}
> +
> +		startoff = imap.br_startoff + imap.br_blockcount;
>  	}
>  out:
>  	return error;
> diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
> index d5f2729305fa..0f36431d9936 100644
> --- a/fs/xfs/libxfs/xfs_bmap.h
> +++ b/fs/xfs/libxfs/xfs_bmap.h
> @@ -90,6 +90,9 @@ struct xfs_bmalloca {
>  /* Try to align allocations to the extent size hint */
>  #define XFS_BMAPI_EXTSZALIGN	(1u << 11)
>  
> +/* Process unwritten extents only. Used for unmapping */
> +#define XFS_BMAPI_UNWRITTEN	(1u << 12)
> +
>  #define XFS_BMAPI_FLAGS \
>  	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
>  	{ XFS_BMAPI_METADATA,	"METADATA" }, \
> @@ -102,7 +105,8 @@ struct xfs_bmalloca {
>  	{ XFS_BMAPI_COWFORK,	"COWFORK" }, \
>  	{ XFS_BMAPI_NODISCARD,	"NODISCARD" }, \
>  	{ XFS_BMAPI_NORMAP,	"NORMAP" },\
> -	{ XFS_BMAPI_EXTSZALIGN,	"EXTSZALIGN" }
> +	{ XFS_BMAPI_EXTSZALIGN,	"EXTSZALIGN" }, \
> +	{ XFS_BMAPI_UNWRITTEN,	"UNWRITTEN" }
>  
>  
>  static inline int xfs_bmapi_aflag(int w)
> diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
> index c88b9ade7389..6323eac48fc8 100644
> --- a/fs/xfs/xfs_bmap_util.c
> +++ b/fs/xfs/xfs_bmap_util.c
> @@ -574,11 +574,13 @@ xfs_can_free_eofblocks(
>   */
>  int
>  xfs_free_eofblocks(
> -	struct xfs_inode	*ip)
> +	struct xfs_inode	*ip,
> +	int			flags)
>  {
>  	struct xfs_trans	*tp;
>  	struct xfs_mount	*mp = ip->i_mount;
>  	int			error;
> +	int			bmapi_flags = XFS_BMAPI_NODISCARD;
>  
>  	/* Attach the dquots to the inode up front. */
>  	error = xfs_qm_dqattach(ip);
> @@ -593,15 +595,20 @@ xfs_free_eofblocks(
>  	 *
>  	 * Note that this means we also leave speculative preallocations in
>  	 * place for preallocated files.
> +	 *
> +	 * Clean up delalloc reservations for fsverity too as those won't be
> +	 * used
>  	 */
> -	if (ip->i_diflags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)) {
> +	if (ip->i_diflags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND) ||
> +			(flags & XFS_FREE_FSVERITY)) {

Oh, this patch keeps going after adding new flags to xfs_bunmapi.

Uh, this part should be a separate patch then.

>  		if (ip->i_delayed_blks) {
>  			xfs_bmap_punch_delalloc_range(ip, XFS_DATA_FORK,
>  				round_up(XFS_ISIZE(ip), mp->m_sb.sb_blocksize),
>  				LLONG_MAX, NULL);
>  		}
>  		xfs_inode_clear_eofblocks_tag(ip);
> -		return 0;
> +		if (!(flags & XFS_FREE_FSVERITY))
> +			return 0;
>  	}
>  
>  	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate, 0, 0, 0, &tp);
> @@ -613,6 +620,9 @@ xfs_free_eofblocks(
>  	xfs_ilock(ip, XFS_ILOCK_EXCL);
>  	xfs_trans_ijoin(tp, ip, 0);
>  
> +	if (flags & XFS_FREE_FSVERITY)
> +		bmapi_flags |= XFS_BMAPI_UNWRITTEN;
> +
>  	/*
>  	 * Do not update the on-disk file size.  If we update the on-disk file
>  	 * size and then the system crashes before the contents of the file are
> @@ -620,7 +630,7 @@ xfs_free_eofblocks(
>  	 * bug).
>  	 */
>  	error = xfs_itruncate_extents_flags(&tp, ip, XFS_DATA_FORK,
> -				XFS_ISIZE(ip), XFS_BMAPI_NODISCARD);
> +				XFS_ISIZE(ip), bmapi_flags);
>  	if (error)
>  		goto err_cancel;
>  
> @@ -928,7 +938,7 @@ xfs_prepare_shift(
>  	 * into the accessible region of the file.
>  	 */
>  	if (xfs_can_free_eofblocks(ip)) {
> -		error = xfs_free_eofblocks(ip);
> +		error = xfs_free_eofblocks(ip, XFS_FREE_ALL);
>  		if (error)
>  			return error;
>  	}
> diff --git a/fs/xfs/xfs_bmap_util.h b/fs/xfs/xfs_bmap_util.h
> index eaaf094154b9..9ea3000466cc 100644
> --- a/fs/xfs/xfs_bmap_util.h
> +++ b/fs/xfs/xfs_bmap_util.h
> @@ -64,9 +64,20 @@ int	xfs_collapse_file_space(struct xfs_inode *, xfs_off_t offset,
>  int	xfs_insert_file_space(struct xfs_inode *, xfs_off_t offset,
>  		xfs_off_t len);
>  
> +/*
> + * Remove all extents and reservations beyond EOF
> + */
> +#define XFS_FREE_ALL		0
> +
> +/*
> + * Do the normal post EOF cleaning except don't remove normal extents, in other
> + * words, remove unwritten, delayed allocation and cow reservations

IOWs it preserves written blocks storing a merkle tree?

> + */
> +#define XFS_FREE_FSVERITY	1

/me wonders if this should be XFS_FREE_EOF_{ALL,PRESERVE_MERKLE} ?

Also, can we decide this from the XFS_DIFLAG2_VERITY state?

--D

> +
>  /* EOF block manipulation functions */
>  bool	xfs_can_free_eofblocks(struct xfs_inode *ip);
> -int	xfs_free_eofblocks(struct xfs_inode *ip);
> +int	xfs_free_eofblocks(struct xfs_inode *ip, int flags);
>  
>  int	xfs_swap_extents(struct xfs_inode *ip, struct xfs_inode *tip,
>  			 struct xfs_swapext *sx);
> diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
> index e9927688086d..43b8fd5a25c5 100644
> --- a/fs/xfs/xfs_file.c
> +++ b/fs/xfs/xfs_file.c
> @@ -1827,7 +1827,7 @@ xfs_file_release(
>  	    xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL)) {
>  		if (xfs_can_free_eofblocks(ip) &&
>  		    !xfs_iflags_test_and_set(ip, XFS_EOFBLOCKS_RELEASED))
> -			xfs_free_eofblocks(ip);
> +			xfs_free_eofblocks(ip, XFS_FREE_ALL);
>  		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
>  	}
>  
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index 9d8dd30bd927..2b3601bb28c9 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -1261,7 +1261,7 @@ xfs_inode_free_eofblocks(
>  	*lockflags |= XFS_IOLOCK_EXCL;
>  
>  	if (xfs_can_free_eofblocks(ip))
> -		return xfs_free_eofblocks(ip);
> +		return xfs_free_eofblocks(ip, XFS_FREE_ALL);
>  
>  	/* inode could be preallocated */
>  	trace_xfs_inode_free_eofblocks_invalid(ip);
> diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
> index 15279d22a894..63f346e2f1d5 100644
> --- a/fs/xfs/xfs_inode.c
> +++ b/fs/xfs/xfs_inode.c
> @@ -1436,7 +1436,7 @@ xfs_inactive(
>  		 * reference to the inode at this point anyways.
>  		 */
>  		if (xfs_can_free_eofblocks(ip))
> -			error = xfs_free_eofblocks(ip);
> +			error = xfs_free_eofblocks(ip, XFS_FREE_ALL);
>  
>  		goto out;
>  	}
> -- 
> 2.54.0
> 
>