Re: [PATCH 31/45] xfs_mkfs: calculate zone overprovisioning when specifying size

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Apr 09, 2025 at 09:55:34AM +0200, Christoph Hellwig wrote:
> When size is specified for zoned file systems, calculate the required
> over provisioning to back the requested capacity.
> 
> Signed-off-by: Hans Holmberg <hans.holmberg@xxxxxxx>
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
>  mkfs/xfs_mkfs.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 46 insertions(+)
> 
> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> index 133ede8d8483..6b5eb9eb140a 100644
> --- a/mkfs/xfs_mkfs.c
> +++ b/mkfs/xfs_mkfs.c
> @@ -4413,6 +4413,49 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
>  			NBBY * (cfg->blocksize - sizeof(struct xfs_rtbuf_blkinfo)));
>  }
>  
> +/*
> + * If we're creating a zoned filesystem and the user specified a size, add
> + * enough over-provisioning to be able to back the requested amount of
> + * writable space.
> + */
> +static void
> +adjust_nr_zones(
> +	struct mkfs_params	*cfg,
> +	struct cli_params	*cli,
> +	struct libxfs_init	*xi,
> +	struct zone_topology	*zt)
> +{
> +	uint64_t		new_rtblocks, slack;
> +	unsigned int		max_zones;
> +
> +	if (zt->rt.nr_zones)
> +		max_zones = zt->rt.nr_zones;
> +	else
> +		max_zones = DTOBT(xi->rt.size, cfg->blocklog) / cfg->rgsize;
> +
> +	if (!cli->rgcount)
> +		cfg->rgcount += XFS_RESERVED_ZONES;
> +	if (cfg->rgcount > max_zones) {
> +		cfg->rgcount = max_zones;
> +		fprintf(stderr,
> +_("Warning: not enough zones for backing requested rt size due to\n"
> +  "over-provisioning needs, writeable size will be less than %s\n"),

Nit: "writable", not "writeable"

With that fixed,
Reviewed-by: "Darrick J. Wong" <djwong@xxxxxxxxxx>

--D

> +			cli->rtsize);
> +	}
> +	new_rtblocks = (cfg->rgcount * cfg->rgsize);
> +	slack = (new_rtblocks - cfg->rtblocks) % cfg->rgsize;
> +
> +	cfg->rtblocks = new_rtblocks;
> +	cfg->rtextents = cfg->rtblocks / cfg->rtextblocks;
> +
> +	/*
> +	 * Add the slack to the end of the last zone to the reserved blocks.
> +	 * This ensures the visible user capacity is exactly the one that the
> +	 * user asked for.
> +	 */
> +	cfg->rtreserved += (slack * cfg->blocksize);
> +}
> +
>  static void
>  calculate_zone_geometry(
>  	struct mkfs_params	*cfg,
> @@ -4485,6 +4528,9 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
>  		}
>  	}
>  
> +	if (cli->rtsize || cli->rgcount)
> +		adjust_nr_zones(cfg, cli, xi, zt);
> +
>  	if (cfg->rgcount < XFS_MIN_ZONES)  {
>  		fprintf(stderr,
>  _("realtime group count (%llu) must be greater than the minimum zone count (%u)\n"),
> -- 
> 2.47.2
> 
> 




[Index of Archives]     [XFS Filesystem Development (older mail)]     [Linux Filesystem Development]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux RAID]     [Linux SCSI]


  Powered by Linux