Re: [PATCH v9 4/5] xfs: support shrinking unused space in the last AG

[Gao Xiang <hsiangkao@redhat.com>]

Hi Darrick,

On Wed, Mar 24, 2021 at 10:34:38AM -0700, Darrick J. Wong wrote:
> On Wed, Mar 24, 2021 at 09:06:20AM +0800, Gao Xiang wrote:
> > As the first step of shrinking, this attempts to enable shrinking
> > unused space in the last allocation group by fixing up freespace
> > btree, agi, agf and adjusting super block and use a helper
> > xfs_ag_shrink_space() to fixup the last AG.
> > 
> > This can be all done in one transaction for now, so I think no
> > additional protection is needed.
> > 
> > Reviewed-by: Darrick J. Wong <djwong@kernel.org>
> > Signed-off-by: Gao Xiang <hsiangkao@redhat.com>
> > ---
> >  fs/xfs/xfs_fsops.c | 84 +++++++++++++++++++++++++++-------------------
> >  fs/xfs/xfs_trans.c |  1 -
> >  2 files changed, 50 insertions(+), 35 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
> > index d1ba04124c28..9457b0691ece 100644
> > --- a/fs/xfs/xfs_fsops.c
> > +++ b/fs/xfs/xfs_fsops.c
> > @@ -91,23 +91,25 @@ xfs_growfs_data_private(
> >  	xfs_agnumber_t		nagcount;
> >  	xfs_agnumber_t		nagimax = 0;
> >  	xfs_rfsblock_t		nb, nb_div, nb_mod;
> > -	xfs_rfsblock_t		delta;
> > +	int64_t			delta;
> >  	bool			lastag_extended;
> >  	xfs_agnumber_t		oagcount;
> >  	struct xfs_trans	*tp;
> >  	struct aghdr_init_data	id = {};
> >  
> >  	nb = in->newblocks;
> > -	if (nb < mp->m_sb.sb_dblocks)
> > -		return -EINVAL;
> > -	if ((error = xfs_sb_validate_fsb_count(&mp->m_sb, nb)))
> > +	error = xfs_sb_validate_fsb_count(&mp->m_sb, nb);
> > +	if (error)
> >  		return error;
> > -	error = xfs_buf_read_uncached(mp->m_ddev_targp,
> > +
> > +	if (nb > mp->m_sb.sb_dblocks) {
> > +		error = xfs_buf_read_uncached(mp->m_ddev_targp,
> >  				XFS_FSB_TO_BB(mp, nb) - XFS_FSS_TO_BB(mp, 1),
> >  				XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);
> > -	if (error)
> > -		return error;
> > -	xfs_buf_relse(bp);
> > +		if (error)
> > +			return error;
> > +		xfs_buf_relse(bp);
> > +	}
> >  
> >  	nb_div = nb;
> >  	nb_mod = do_div(nb_div, mp->m_sb.sb_agblocks);
> > @@ -115,10 +117,16 @@ xfs_growfs_data_private(
> >  	if (nb_mod && nb_mod < XFS_MIN_AG_BLOCKS) {
> >  		nagcount--;
> >  		nb = (xfs_rfsblock_t)nagcount * mp->m_sb.sb_agblocks;
> > -		if (nb < mp->m_sb.sb_dblocks)
> > -			return -EINVAL;
> >  	}
> >  	delta = nb - mp->m_sb.sb_dblocks;
> > +	/*
> > +	 * Reject filesystems with a single AG because they are not
> > +	 * supported, and reject a shrink operation that would cause a
> > +	 * filesystem to become unsupported.
> > +	 */
> > +	if (delta < 0 && nagcount < 2)
> > +		return -EINVAL;
> > +
> >  	oagcount = mp->m_sb.sb_agcount;
> >  
> >  	/* allocate the new per-ag structures */
> > @@ -126,15 +134,22 @@ xfs_growfs_data_private(
> >  		error = xfs_initialize_perag(mp, nagcount, &nagimax);
> >  		if (error)
> >  			return error;
> > +	} else if (nagcount < oagcount) {
> > +		/* TODO: shrinking the entire AGs hasn't yet completed */
> > +		return -EINVAL;
> >  	}
> >  
> >  	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_growdata,
> > -			XFS_GROWFS_SPACE_RES(mp), 0, XFS_TRANS_RESERVE, &tp);
> > +			(delta > 0 ? XFS_GROWFS_SPACE_RES(mp) : -delta), 0,
> > +			XFS_TRANS_RESERVE, &tp);
> >  	if (error)
> >  		return error;
> >  
> > -	error = xfs_resizefs_init_new_ags(tp, &id, oagcount, nagcount,
> > -					  delta, &lastag_extended);
> > +	if (delta > 0)
> > +		error = xfs_resizefs_init_new_ags(tp, &id, oagcount, nagcount,
> > +						  delta, &lastag_extended);
> > +	else
> > +		error = xfs_ag_shrink_space(mp, &tp, nagcount - 1, -delta);
> 
> Assuming I don't hear anyone yelling NAK in the next day or so, I think
> I'll stage this for 5.13 with the following change to warn that the
> shrink feature is still EXPERIMENTAL:
> 
> By the way, are you going to send a patch to shrink the realtime device
> too?

My first priority will form the formal shrink-whole-AG patchset after this
is finalized.
I have little experience about rt device, maybe leave it later.

> 
> diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
> index 9457b0691ece..b33c894b6cf3 100644
> --- a/fs/xfs/xfs_fsops.c
> +++ b/fs/xfs/xfs_fsops.c
> @@ -145,11 +145,20 @@ xfs_growfs_data_private(
>  	if (error)
>  		return error;
>  
> -	if (delta > 0)
> +	if (delta > 0) {
>  		error = xfs_resizefs_init_new_ags(tp, &id, oagcount, nagcount,
>  						  delta, &lastag_extended);
> -	else
> +	} else {
> +		static struct ratelimit_state shrink_warning = \
> +			RATELIMIT_STATE_INIT("shrink_warning", 86400 * HZ, 1);
> +		ratelimit_set_flags(&shrink_warning, RATELIMIT_MSG_ON_RELEASE);
> +
> +		if (__ratelimit(&shrink_warning))
> +			xfs_alert(mp,
> +	"EXPERIMENTAL online shrink feature in use. Use at your own risk!");
> +
>  		error = xfs_ag_shrink_space(mp, &tp, nagcount - 1, -delta);
> +	}
>  	if (error)
>  		goto out_trans_cancel;

I'm fine with that, thanks for updating. :)

Thanks,
Gao Xiang

>  
> --D
> 
> >  	if (error)
> >  		goto out_trans_cancel;
> >  
> > @@ -169,28 +184,29 @@ xfs_growfs_data_private(
> >  	xfs_set_low_space_thresholds(mp);
> >  	mp->m_alloc_set_aside = xfs_alloc_set_aside(mp);
> >  
> > -	/*
> > -	 * If we expanded the last AG, free the per-AG reservation
> > -	 * so we can reinitialize it with the new size.
> > -	 */
> > -	if (lastag_extended) {
> > -		struct xfs_perag	*pag;
> > -
> > -		pag = xfs_perag_get(mp, id.agno);
> > -		error = xfs_ag_resv_free(pag);
> > -		xfs_perag_put(pag);
> > -		if (error)
> > -			return error;
> > +	if (delta > 0) {
> > +		/*
> > +		 * If we expanded the last AG, free the per-AG reservation
> > +		 * so we can reinitialize it with the new size.
> > +		 */
> > +		if (lastag_extended) {
> > +			struct xfs_perag	*pag;
> > +
> > +			pag = xfs_perag_get(mp, id.agno);
> > +			error = xfs_ag_resv_free(pag);
> > +			xfs_perag_put(pag);
> > +			if (error)
> > +				return error;
> > +		}
> > +		/*
> > +		 * Reserve AG metadata blocks. ENOSPC here does not mean there
> > +		 * was a growfs failure, just that there still isn't space for
> > +		 * new user data after the grow has been run.
> > +		 */
> > +		error = xfs_fs_reserve_ag_blocks(mp);
> > +		if (error == -ENOSPC)
> > +			error = 0;
> >  	}
> > -
> > -	/*
> > -	 * Reserve AG metadata blocks. ENOSPC here does not mean there was a
> > -	 * growfs failure, just that there still isn't space for new user data
> > -	 * after the grow has been run.
> > -	 */
> > -	error = xfs_fs_reserve_ag_blocks(mp);
> > -	if (error == -ENOSPC)
> > -		error = 0;
> >  	return error;
> >  
> >  out_trans_cancel:
> > diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
> > index b22a09e9daee..052274321993 100644
> > --- a/fs/xfs/xfs_trans.c
> > +++ b/fs/xfs/xfs_trans.c
> > @@ -436,7 +436,6 @@ xfs_trans_mod_sb(
> >  		tp->t_res_frextents_delta += delta;
> >  		break;
> >  	case XFS_TRANS_SB_DBLOCKS:
> > -		ASSERT(delta > 0);
> >  		tp->t_dblocks_delta += delta;
> >  		break;
> >  	case XFS_TRANS_SB_AGCOUNT:
> > -- 
> > 2.27.0
> > 
>