Re: [RFC PATCH] libxfs: add missing agfl free deferred op type

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Thu, Oct 25, 2018 at 05:38:10PM -0700, Darrick J. Wong wrote:
> Soo... I tried to run xfs_repair in gdb on arekm's big filesystem, but
> we forgot to port deferred agfl free defer ops type to xfsprogs so it
> blew up earlier this afternoon.  This is a patch to add it in, but since
> it takes a while to get far enough to start rebuilding directories, this
> is basically an untested patch.
> 
> Which I'm sending out in case sandeen/dchinner are doing the same thing
> I am. :P

FWIW this has now passed the usual testing, so strike the 'RFC' from the
title.

--D

> --D
> 
> ---
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> When we added a new defer op type for agfl block freeing to the kernel,
> we forgot to add that type to the userspace side of things.  This will
> cause spontaneous combustion in xfs_repair if we try to rebuild
> directories.
> 
> Fixes: d5c1b462232 ("xfs: defer agfl block frees when dfops is available")
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  libxfs/defer_item.c |   44 ++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 44 insertions(+)
> 
> diff --git a/libxfs/defer_item.c b/libxfs/defer_item.c
> index b7185c1a..82aa32a8 100644
> --- a/libxfs/defer_item.c
> +++ b/libxfs/defer_item.c
> @@ -118,11 +118,55 @@ static const struct xfs_defer_op_type xfs_extent_free_defer_type = {
>  	.cancel_item	= xfs_extent_free_cancel_item,
>  };
>  
> +/*
> + * AGFL blocks are accounted differently in the reserve pools and are not
> + * inserted into the busy extent list.
> + */
> +STATIC int
> +xfs_agfl_free_finish_item(
> +	struct xfs_trans		*tp,
> +	struct list_head		*item,
> +	void				*done_item,
> +	void				**state)
> +{
> +	struct xfs_mount		*mp = tp->t_mountp;
> +	struct xfs_extent_free_item	*free;
> +	struct xfs_buf			*agbp;
> +	int				error;
> +	xfs_agnumber_t			agno;
> +	xfs_agblock_t			agbno;
> +
> +	free = container_of(item, struct xfs_extent_free_item, xefi_list);
> +	ASSERT(free->xefi_blockcount == 1);
> +	agno = XFS_FSB_TO_AGNO(mp, free->xefi_startblock);
> +	agbno = XFS_FSB_TO_AGBNO(mp, free->xefi_startblock);
> +
> +	error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
> +	if (!error)
> +		error = xfs_free_agfl_block(tp, agno, agbno, agbp,
> +					    &free->xefi_oinfo);
> +	kmem_free(free);
> +	return error;
> +}
> +
> +/* sub-type with special handling for AGFL deferred frees */
> +static const struct xfs_defer_op_type xfs_agfl_free_defer_type = {
> +	.type		= XFS_DEFER_OPS_TYPE_AGFL_FREE,
> +	.diff_items	= xfs_extent_free_diff_items,
> +	.create_intent	= xfs_extent_free_create_intent,
> +	.abort_intent	= xfs_extent_free_abort_intent,
> +	.log_item	= xfs_extent_free_log_item,
> +	.create_done	= xfs_extent_free_create_done,
> +	.finish_item	= xfs_agfl_free_finish_item,
> +	.cancel_item	= xfs_extent_free_cancel_item,
> +};
> +
>  /* Register the deferred op type. */
>  void
>  xfs_extent_free_init_defer_op(void)
>  {
>  	xfs_defer_init_op_type(&xfs_extent_free_defer_type);
> +	xfs_defer_init_op_type(&xfs_agfl_free_defer_type);
>  }
>  
>  /* Reverse Mapping */