Re: [PATCH v4.2 5/5] xfs: xfs_defer_capture should absorb remaining transaction reservation

[Christoph Hellwig <hch@lst.de>]

On Thu, Oct 01, 2020 at 09:21:03PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> When xfs_defer_capture extracts the deferred ops and transaction state
> from a transaction, it should record the transaction reservation type
> from the old transaction so that when we continue the dfops chain, we
> still use the same reservation parameters.
> 
> Doing this means that the log item recovery functions get to determine
> the transaction reservation instead of abusing tr_itruncate in yet
> another part of xfs.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
> v4.2: save only the log reservation, and hardcode logcount and flags.
> ---
>  fs/xfs/libxfs/xfs_defer.c |    3 +++
>  fs/xfs/libxfs/xfs_defer.h |    3 +++
>  fs/xfs/xfs_log_recover.c  |   17 ++++++++++++++---
>  3 files changed, 20 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/xfs/libxfs/xfs_defer.c b/fs/xfs/libxfs/xfs_defer.c
> index 10aeae7353ab..e19dc1ced7e6 100644
> --- a/fs/xfs/libxfs/xfs_defer.c
> +++ b/fs/xfs/libxfs/xfs_defer.c
> @@ -579,6 +579,9 @@ xfs_defer_ops_capture(
>  	dfc->dfc_blkres = tp->t_blk_res - tp->t_blk_res_used;
>  	dfc->dfc_rtxres = tp->t_rtx_res - tp->t_rtx_res_used;
>  
> +	/* Preserve the log reservation size. */
> +	dfc->dfc_logres = tp->t_log_res;
> +
>  	return dfc;
>  }
>  
> diff --git a/fs/xfs/libxfs/xfs_defer.h b/fs/xfs/libxfs/xfs_defer.h
> index 5c0e59b69ffa..6cde6f0713f7 100644
> --- a/fs/xfs/libxfs/xfs_defer.h
> +++ b/fs/xfs/libxfs/xfs_defer.h
> @@ -79,6 +79,9 @@ struct xfs_defer_capture {
>  	/* Block reservations for the data and rt devices. */
>  	unsigned int		dfc_blkres;
>  	unsigned int		dfc_rtxres;
> +
> +	/* Log reservation saved from the transaction. */
> +	unsigned int		dfc_logres;

> +		struct xfs_trans_res	resv;
> +
> +		/*
> +		 * Create a new transaction reservation from the captured
> +		 * information.  Set logcount to 1 to force the new transaction
> +		 * to regrant every roll so that we can make forward progress
> +		 * in recovery no matter how full the log might be.
> +		 */
> +		resv.tr_logres = dfc->dfc_logres;
> +		resv.tr_logcount = 1;
> +		resv.tr_logflags = XFS_TRANS_PERM_LOG_RES;
> +
> +		error = xfs_trans_alloc(mp, &resv, dfc->dfc_blkres,
> +				dfc->dfc_rtxres, XFS_TRANS_RESERVE, &tp);

What about just embedding the struct xfs_trans_res into
struct xfs_defer_capture directly?  That probably also means merging
this and the previous patch.