Re: [PATCH 2/5] libxfs: add function to free all buffers in bcache

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

On Tue, Mar 06, 2018 at 03:54:51PM -0600, Eric Sandeen wrote:
> libxfs_bcache_purge simply moves all "free" buffers
> onto the xfs_buf_freelist mru list; add a new function to
> actually free them when we tear everything down, so leak
> checkers don't go nuts about lots of unfreed xfs_bufs
> at exit.
> 
> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
> ---
>  libxfs/init.c      |  5 ++++-
>  libxfs/libxfs_io.h |  1 +
>  libxfs/rdwr.c      | 18 ++++++++++++++++++
>  3 files changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/libxfs/init.c b/libxfs/init.c
> index 7bde8b7..c7d73b6 100644
> --- a/libxfs/init.c
> +++ b/libxfs/init.c
> @@ -888,8 +888,11 @@ libxfs_umount(xfs_mount_t *mp)
>  void
>  libxfs_destroy(void)
>  {
> -	manage_zones(1);
> +	/* Free everything from the buffer cache before freeing buffer zone */
> +	libxfs_bcache_purge();
> +	libxfs_bcache_free();

/me wonders if we should warn about unflushed buffers here, but afaict
all the existing programs take care of this, so...

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>

--D

>  	cache_destroy(libxfs_bcache);
> +	manage_zones(1);
>  }
>  
>  int
> diff --git a/libxfs/libxfs_io.h b/libxfs/libxfs_io.h
> index 78b6780..6de6fcb 100644
> --- a/libxfs/libxfs_io.h
> +++ b/libxfs/libxfs_io.h
> @@ -188,6 +188,7 @@ extern void	libxfs_readbuf_verify(struct xfs_buf *bp,
>  			const struct xfs_buf_ops *ops);
>  extern xfs_buf_t *libxfs_getsb(struct xfs_mount *, int);
>  extern void	libxfs_bcache_purge(void);
> +extern void	libxfs_bcache_free(void);
>  extern void	libxfs_bcache_flush(void);
>  extern void	libxfs_purgebuf(xfs_buf_t *);
>  extern int	libxfs_bcache_overflowed(void);
> diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
> index 3c5def2..81701b7 100644
> --- a/libxfs/rdwr.c
> +++ b/libxfs/rdwr.c
> @@ -1275,6 +1275,24 @@ libxfs_bulkrelse(
>  }
>  
>  /*
> + * Free everything from the xfs_buf_freelist MRU, used at final teardown
> + */
> +void
> +libxfs_bcache_free(void)
> +{
> +	struct list_head	*cm_list;
> +	xfs_buf_t		*bp, *next;
> +
> +	cm_list = &xfs_buf_freelist.cm_list;
> +	list_for_each_entry_safe(bp, next, cm_list, b_node.cn_mru) {
> +		free(bp->b_addr);
> +		if (bp->b_maps != &bp->__b_map)
> +			free(bp->b_maps);
> +		kmem_zone_free(xfs_buf_zone, bp);
> +	}
> +}
> +
> +/*
>   * When a buffer is marked dirty, the error is cleared. Hence if we are trying
>   * to flush a buffer prior to cache reclaim that has an error on it it means
>   * we've already tried to flush it and it failed. Prevent repeated corruption
> -- 
> 1.8.3.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html