Skip to contentSkip to conversation

improve the discard / reset zones code in mkfs

PATCH In review Christoph Hellwig 2 participants 9 messages Oct 29, 2025

lore
Share & tools
lore
#0 lore raw

Hi all,

this series refactors the discard and reset zones code in mkfs a bit, and then allows discarding for the data device even when the RT devices sits on a zones where discard doesn't make much sense.

Note: this sits on top of the "improve a few messages in mkfs and xfs_copy" series sent a while ago.

#1 lore raw

PATCH 1/4 mkfs: move clearing LIBXFS_DIRECT into check_device_type

Keep it close to the block device vs regular file logic and remove the checks for each device in the caller.

---
 mkfs/xfs_mkfs.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
mkfs/xfs_mkfs.c +14 -13
index cb7c20e3aa18..3ccd37920321 100644
@@ -1330,6 +1330,7 @@ nr_cpus(void)
 
 static void
 check_device_type(
+	struct cli_params	*cli,
 	struct libxfs_dev	*dev,
 	bool			no_size,
 	bool			dry_run,
@@ -1375,6 +1376,13 @@ check_device_type(
 			dev->isfile = 1;
 		else if (!dry_run)
 			dev->create = 1;
+
+		/*
+		 * Explicitly disable direct IO for image files so we don't
+		 * error out on sector size mismatches between the new
+		 * filesystem and the underlying host filesystem.
+		 */
+		cli->xi->flags &= ~LIBXFS_DIRECT;
 		return;
 	}
 
@@ -2378,21 +2386,14 @@ validate_sectorsize(
 	 * Before anything else, verify that we are correctly operating on
 	 * files or block devices and set the control parameters correctly.
 	 */
-	check_device_type(&cli->xi->data, !cli->dsize, dry_run, "data", "d");
+	check_device_type(cli, &cli->xi->data, !cli->dsize, dry_run,
+			"data", "d");
 	if (!cli->loginternal)
-		check_device_type(&cli->xi->log, !cli->logsize, dry_run, "log",
-				"l");
+		check_device_type(cli, &cli->xi->log, !cli->logsize, dry_run,
+				"log", "l");
 	if (cli->xi->rt.name)
-		check_device_type(&cli->xi->rt, !cli->rtsize, dry_run, "RT",
-				"r");
-
-	/*
-	 * Explicitly disable direct IO for image files so we don't error out on
-	 * sector size mismatches between the new filesystem and the underlying
-	 * host filesystem.
-	 */
-	if (cli->xi->data.isfile || cli->xi->log.isfile || cli->xi->rt.isfile)
-		cli->xi->flags &= ~LIBXFS_DIRECT;
+		check_device_type(cli, &cli->xi->rt, !cli->rtsize, dry_run,
+				"RT", "r");
 
 	memset(ft, 0, sizeof(*ft));
 	get_topology(cli->xi, ft, force_overwrite);
Signature · 1 line
2.47.3
Reviewed-by on v1 1/4
#2 lore raw

Re: [PATCH 1/4] mkfs: move clearing LIBXFS_DIRECT into check_device_type

Quoted 4 lines · Christoph Hellwig

Keep it close to the block device vs regular file logic and remove the checks for each device in the caller.

I guess it makes sense to move the clearing of LIBXFS_DIRECT to where we determine that one of the device paths pointed to a regular file...

--D

Quoted 62 lines
---
 mkfs/xfs_mkfs.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
mkfs/xfs_mkfs.c +14 -13
index cb7c20e3aa18..3ccd37920321 100644
@@ -1330,6 +1330,7 @@ nr_cpus(void)
 
 static void
 check_device_type(
+	struct cli_params	*cli,
 	struct libxfs_dev	*dev,
 	bool			no_size,
 	bool			dry_run,
@@ -1375,6 +1376,13 @@ check_device_type(
 			dev->isfile = 1;
 		else if (!dry_run)
 			dev->create = 1;
+
+		/*
+		 * Explicitly disable direct IO for image files so we don't
+		 * error out on sector size mismatches between the new
+		 * filesystem and the underlying host filesystem.
+		 */
+		cli->xi->flags &= ~LIBXFS_DIRECT;
 		return;
 	}
 
@@ -2378,21 +2386,14 @@ validate_sectorsize(
 	 * Before anything else, verify that we are correctly operating on
 	 * files or block devices and set the control parameters correctly.
 	 */
-	check_device_type(&cli->xi->data, !cli->dsize, dry_run, "data", "d");
+	check_device_type(cli, &cli->xi->data, !cli->dsize, dry_run,
+			"data", "d");
 	if (!cli->loginternal)
-		check_device_type(&cli->xi->log, !cli->logsize, dry_run, "log",
-				"l");
+		check_device_type(cli, &cli->xi->log, !cli->logsize, dry_run,
+				"log", "l");
 	if (cli->xi->rt.name)
-		check_device_type(&cli->xi->rt, !cli->rtsize, dry_run, "RT",
-				"r");
-
-	/*
-	 * Explicitly disable direct IO for image files so we don't error out on
-	 * sector size mismatches between the new filesystem and the underlying
-	 * host filesystem.
-	 */
-	if (cli->xi->data.isfile || cli->xi->log.isfile || cli->xi->rt.isfile)
-		cli->xi->flags &= ~LIBXFS_DIRECT;
+		check_device_type(cli, &cli->xi->rt, !cli->rtsize, dry_run,
+				"RT", "r");
 
 	memset(ft, 0, sizeof(*ft));
 	get_topology(cli->xi, ft, force_overwrite);
Signature · 1 line
2.47.3
#3 lore raw

PATCH 2/4 libxfs: cleanup get_topology

In reply to Christoph Hellwig

Add a libxfs_ prefix to the name, clear the structure in the helper instead of in the callers, and use a bool to pass a boolean argument.

---
 libxfs/topology.c | 9 +++++----
 libxfs/topology.h | 7 ++-----
 mkfs/xfs_mkfs.c   | 3 +--
 repair/sb.c       | 3 +--
 4 files changed, 9 insertions(+), 13 deletions(-)
libxfs/topology.c +5 -4
index 7764687beac0..366165719c84 100644
@@ -224,7 +224,7 @@ static void
 blkid_get_topology(
 	const char		*device,
 	struct device_topology	*dt,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
 	blkid_topology tp;
 	blkid_probe pr;
@@ -317,7 +317,7 @@ static void
 get_device_topology(
 	struct libxfs_dev	*dev,
 	struct device_topology	*dt,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
 	struct stat		st;
 
@@ -364,11 +364,12 @@ get_device_topology(
 }
 
 void
-get_topology(
+libxfs_get_topology(
 	struct libxfs_init	*xi,
 	struct fs_topology	*ft,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
+	memset(ft, 0, sizeof(*ft));
 	get_device_topology(&xi->data, &ft->data, force_overwrite);
 	get_device_topology(&xi->rt, &ft->rt, force_overwrite);
 	get_device_topology(&xi->log, &ft->log, force_overwrite);
libxfs/topology.h +2 -5
index f0ca65f3576e..3688d56b542f 100644
@@ -25,11 +25,8 @@ struct fs_topology {
 	struct device_topology	log;
 };
 
-void
-get_topology(
-	struct libxfs_init	*xi,
-	struct fs_topology	*ft,
-	int			force_overwrite);
+void libxfs_get_topology(struct libxfs_init *xi, struct fs_topology *ft,
+		bool force_overwrite);
 
 extern void
 calc_default_ag_geometry(
mkfs/xfs_mkfs.c +1 -2
index 3ccd37920321..0ba7798eccf6 100644
@@ -2395,8 +2395,7 @@ validate_sectorsize(
 		check_device_type(cli, &cli->xi->rt, !cli->rtsize, dry_run,
 				"RT", "r");
 
-	memset(ft, 0, sizeof(*ft));
-	get_topology(cli->xi, ft, force_overwrite);
+	libxfs_get_topology(cli->xi, ft, force_overwrite);
 
 	/* set configured sector sizes in preparation for checks */
 	if (!cli->sectorsize) {
repair/sb.c +1 -2
index 0e4827e04678..ee1cc63fae64 100644
@@ -184,8 +184,7 @@ guess_default_geometry(
 	uint64_t		dblocks;
 	int			multidisk;
 
-	memset(&ft, 0, sizeof(ft));
-	get_topology(x, &ft, 1);
+	libxfs_get_topology(x, &ft, true);
 
 	/*
 	 * get geometry from get_topology result.
Signature · 1 line
2.47.3
Reviewed-by on v1 2/4
#4 lore raw

Re: [PATCH 2/4] libxfs: cleanup get_topology

Quoted 4 lines · Christoph Hellwig

Add a libxfs_ prefix to the name, clear the structure in the helper instead of in the callers, and use a bool to pass a boolean argument.

Looks good,

--D

Quoted 94 lines
---
 libxfs/topology.c | 9 +++++----
 libxfs/topology.h | 7 ++-----
 mkfs/xfs_mkfs.c   | 3 +--
 repair/sb.c       | 3 +--
 4 files changed, 9 insertions(+), 13 deletions(-)
libxfs/topology.c +5 -4
index 7764687beac0..366165719c84 100644
@@ -224,7 +224,7 @@ static void
 blkid_get_topology(
 	const char		*device,
 	struct device_topology	*dt,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
 	blkid_topology tp;
 	blkid_probe pr;
@@ -317,7 +317,7 @@ static void
 get_device_topology(
 	struct libxfs_dev	*dev,
 	struct device_topology	*dt,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
 	struct stat		st;
 
@@ -364,11 +364,12 @@ get_device_topology(
 }
 
 void
-get_topology(
+libxfs_get_topology(
 	struct libxfs_init	*xi,
 	struct fs_topology	*ft,
-	int			force_overwrite)
+	bool			force_overwrite)
 {
+	memset(ft, 0, sizeof(*ft));
 	get_device_topology(&xi->data, &ft->data, force_overwrite);
 	get_device_topology(&xi->rt, &ft->rt, force_overwrite);
 	get_device_topology(&xi->log, &ft->log, force_overwrite);
libxfs/topology.h +2 -5
index f0ca65f3576e..3688d56b542f 100644
@@ -25,11 +25,8 @@ struct fs_topology {
 	struct device_topology	log;
 };
 
-void
-get_topology(
-	struct libxfs_init	*xi,
-	struct fs_topology	*ft,
-	int			force_overwrite);
+void libxfs_get_topology(struct libxfs_init *xi, struct fs_topology *ft,
+		bool force_overwrite);
 
 extern void
 calc_default_ag_geometry(
mkfs/xfs_mkfs.c +1 -2
index 3ccd37920321..0ba7798eccf6 100644
@@ -2395,8 +2395,7 @@ validate_sectorsize(
 		check_device_type(cli, &cli->xi->rt, !cli->rtsize, dry_run,
 				"RT", "r");
 
-	memset(ft, 0, sizeof(*ft));
-	get_topology(cli->xi, ft, force_overwrite);
+	libxfs_get_topology(cli->xi, ft, force_overwrite);
 
 	/* set configured sector sizes in preparation for checks */
 	if (!cli->sectorsize) {
repair/sb.c +1 -2
index 0e4827e04678..ee1cc63fae64 100644
@@ -184,8 +184,7 @@ guess_default_geometry(
 	uint64_t		dblocks;
 	int			multidisk;
 
-	memset(&ft, 0, sizeof(ft));
-	get_topology(x, &ft, 1);
+	libxfs_get_topology(x, &ft, true);
 
 	/*
 	 * get geometry from get_topology result.
Signature · 1 line
2.47.3
#5 lore raw

PATCH 3/4 mkfs: remove duplicate struct libxfs_init arguments

In reply to Christoph Hellwig

The libxfs_init structure instance is pointed to by cli_params, so use that were it already exists instead of passing an additional argument.

---
 mkfs/xfs_mkfs.c | 49 ++++++++++++++++++++++---------------------------
 1 file changed, 22 insertions(+), 27 deletions(-)
mkfs/xfs_mkfs.c +22 -27
index 0ba7798eccf6..09a69af31be5 100644
@@ -4005,8 +4005,7 @@ ddev_is_solidstate(
 static void
 calc_concurrency_ag_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	uint64_t		try_agsize;
 	uint64_t		def_agsize;
@@ -4074,11 +4073,10 @@ out:
 static void
 calculate_initial_ag_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	if (cli->data_concurrency > 0) {
-		calc_concurrency_ag_geometry(cfg, cli, xi);
+		calc_concurrency_ag_geometry(cfg, cli);
 	} else if (cli->agsize) {	/* User-specified AG size */
 		cfg->agsize = getnum(cli->agsize, &dopts, D_AGSIZE);
 
@@ -4099,8 +4097,9 @@ _("agsize (%s) not a multiple of fs blk size (%d)\n"),
 		cfg->agcount = cli->agcount;
 		cfg->agsize = cfg->dblocks / cfg->agcount +
 				(cfg->dblocks % cfg->agcount != 0);
-	} else if (cli->data_concurrency == -1 && ddev_is_solidstate(xi)) {
-		calc_concurrency_ag_geometry(cfg, cli, xi);
+	} else if (cli->data_concurrency == -1 &&
+		   ddev_is_solidstate(cli->xi)) {
+		calc_concurrency_ag_geometry(cfg, cli);
 	} else {
 		calc_default_ag_geometry(cfg->blocklog, cfg->dblocks,
 					 cfg->dsunit, &cfg->agsize,
@@ -4360,8 +4359,7 @@ rtdev_is_solidstate(
 static void
 calc_concurrency_rtgroup_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	uint64_t		try_rgsize;
 	uint64_t		def_rgsize;
@@ -4468,8 +4466,7 @@ _("realtime group count (%llu) must be less than the maximum (%u)\n"),
 static void
 calculate_rtgroup_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	if (!cli->sb_feat.metadir) {
 		cfg->rgcount = 0;
@@ -4510,8 +4507,9 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
 		cfg->rgsize = cfg->rtblocks;
 		cfg->rgcount = 0;
 	} else if (cli->rtvol_concurrency > 0 ||
-		   (cli->rtvol_concurrency == -1 && rtdev_is_solidstate(xi))) {
-		calc_concurrency_rtgroup_geometry(cfg, cli, xi);
+		   (cli->rtvol_concurrency == -1 &&
+		    rtdev_is_solidstate(cli->xi))) {
+		calc_concurrency_rtgroup_geometry(cfg, cli);
 	} else if (is_power_of_2(cfg->rtextblocks)) {
 		cfg->rgsize = calc_rgsize_extsize_power(cfg);
 		cfg->rgcount = cfg->rtblocks / cfg->rgsize +
@@ -4538,7 +4536,6 @@ 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;
@@ -4547,7 +4544,8 @@ adjust_nr_zones(
 	if (zt->rt.nr_zones)
 		max_zones = zt->rt.nr_zones;
 	else
-		max_zones = DTOBT(xi->rt.size, cfg->blocklog) / cfg->rgsize;
+		max_zones = DTOBT(cli->xi->rt.size, cfg->blocklog) /
+				cfg->rgsize;
 
 	if (!cli->rgcount)
 		cfg->rgcount += XFS_RESERVED_ZONES;
@@ -4576,7 +4574,6 @@ static void
 calculate_zone_geometry(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	struct zone_topology	*zt)
 {
 	if (cfg->rtblocks == 0) {
@@ -4645,7 +4642,7 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
 	}
 
 	if (cli->rtsize || cli->rgcount)
-		adjust_nr_zones(cfg, cli, xi, zt);
+		adjust_nr_zones(cfg, cli, zt);
 
 	if (cfg->rgcount < XFS_MIN_ZONES)  {
 		fprintf(stderr,
@@ -4984,7 +4981,6 @@ static uint64_t
 calc_concurrency_logblocks(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	unsigned int		max_tx_bytes)
 {
 	uint64_t		log_bytes;
@@ -4992,7 +4988,7 @@ calc_concurrency_logblocks(
 	unsigned int		new_logblocks;
 
 	if (cli->log_concurrency < 0) {
-		if (!ddev_is_solidstate(xi))
+		if (!ddev_is_solidstate(cli->xi))
 			goto out;
 
 		cli->log_concurrency = nr_cpus();
@@ -5160,7 +5156,6 @@ static void
 calculate_log_size(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	struct xfs_mount	*mp)
 {
 	struct xfs_sb		*sbp = &mp->m_sb;
@@ -5225,8 +5220,8 @@ _("external log device size %lld blocks too small, must be at least %lld blocks\
 		if (cfg->lsunit) {
 			uint64_t	max_logblocks;
 
-			max_logblocks = min(DTOBT(xi->log.size, cfg->blocklog),
-					    XFS_MAX_LOG_BLOCKS);
+			max_logblocks = min(XFS_MAX_LOG_BLOCKS,
+				DTOBT(cli->xi->log.size, cfg->blocklog));
 			align_log_size(cfg, cfg->lsunit, max_logblocks);
 		}
 
@@ -5261,7 +5256,7 @@ _("max log size %d smaller than min log size %d, filesystem is too small\n"),
 
 		if (cli->log_concurrency != 0)
 			cfg->logblocks = calc_concurrency_logblocks(cfg, cli,
-							xi, max_tx_bytes);
+							max_tx_bytes);
 
 		/* But don't go below a reasonable size */
 		cfg->logblocks = max(cfg->logblocks,
@@ -6135,12 +6130,12 @@ main(
 	 * dependent on device sizes. Once calculated, make sure everything
 	 * aligns to device geometry correctly.
 	 */
-	calculate_initial_ag_geometry(&cfg, &cli, &xi);
+	calculate_initial_ag_geometry(&cfg, &cli);
 	align_ag_geometry(&cfg, &ft);
 	if (cfg.sb_feat.zoned)
-		calculate_zone_geometry(&cfg, &cli, &xi, &zt);
+		calculate_zone_geometry(&cfg, &cli, &zt);
 	else
-		calculate_rtgroup_geometry(&cfg, &cli, &xi);
+		calculate_rtgroup_geometry(&cfg, &cli);
 
 	calculate_imaxpct(&cfg, &cli);
 
@@ -6164,7 +6159,7 @@ main(
 	 * With the mount set up, we can finally calculate the log size
 	 * constraints and do default size calculations and final validation
 	 */
-	calculate_log_size(&cfg, &cli, &xi, mp);
+	calculate_log_size(&cfg, &cli, mp);
 
 	finish_superblock_setup(&cfg, mp, sbp);
 
Signature · 1 line
2.47.3
Reviewed-by on v1 3/4
#6 lore raw

Re: [PATCH 3/4] mkfs: remove duplicate struct libxfs_init arguments

Quoted 4 lines · Christoph Hellwig

The libxfs_init structure instance is pointed to by cli_params, so use that were it already exists instead of passing an additional argument.

Yay fewer args,

--D

Quoted 185 lines
---
 mkfs/xfs_mkfs.c | 49 ++++++++++++++++++++++---------------------------
 1 file changed, 22 insertions(+), 27 deletions(-)
mkfs/xfs_mkfs.c +22 -27
index 0ba7798eccf6..09a69af31be5 100644
@@ -4005,8 +4005,7 @@ ddev_is_solidstate(
 static void
 calc_concurrency_ag_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	uint64_t		try_agsize;
 	uint64_t		def_agsize;
@@ -4074,11 +4073,10 @@ out:
 static void
 calculate_initial_ag_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	if (cli->data_concurrency > 0) {
-		calc_concurrency_ag_geometry(cfg, cli, xi);
+		calc_concurrency_ag_geometry(cfg, cli);
 	} else if (cli->agsize) {	/* User-specified AG size */
 		cfg->agsize = getnum(cli->agsize, &dopts, D_AGSIZE);
 
@@ -4099,8 +4097,9 @@ _("agsize (%s) not a multiple of fs blk size (%d)\n"),
 		cfg->agcount = cli->agcount;
 		cfg->agsize = cfg->dblocks / cfg->agcount +
 				(cfg->dblocks % cfg->agcount != 0);
-	} else if (cli->data_concurrency == -1 && ddev_is_solidstate(xi)) {
-		calc_concurrency_ag_geometry(cfg, cli, xi);
+	} else if (cli->data_concurrency == -1 &&
+		   ddev_is_solidstate(cli->xi)) {
+		calc_concurrency_ag_geometry(cfg, cli);
 	} else {
 		calc_default_ag_geometry(cfg->blocklog, cfg->dblocks,
 					 cfg->dsunit, &cfg->agsize,
@@ -4360,8 +4359,7 @@ rtdev_is_solidstate(
 static void
 calc_concurrency_rtgroup_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	uint64_t		try_rgsize;
 	uint64_t		def_rgsize;
@@ -4468,8 +4466,7 @@ _("realtime group count (%llu) must be less than the maximum (%u)\n"),
 static void
 calculate_rtgroup_geometry(
 	struct mkfs_params	*cfg,
-	struct cli_params	*cli,
-	struct libxfs_init	*xi)
+	struct cli_params	*cli)
 {
 	if (!cli->sb_feat.metadir) {
 		cfg->rgcount = 0;
@@ -4510,8 +4507,9 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
 		cfg->rgsize = cfg->rtblocks;
 		cfg->rgcount = 0;
 	} else if (cli->rtvol_concurrency > 0 ||
-		   (cli->rtvol_concurrency == -1 && rtdev_is_solidstate(xi))) {
-		calc_concurrency_rtgroup_geometry(cfg, cli, xi);
+		   (cli->rtvol_concurrency == -1 &&
+		    rtdev_is_solidstate(cli->xi))) {
+		calc_concurrency_rtgroup_geometry(cfg, cli);
 	} else if (is_power_of_2(cfg->rtextblocks)) {
 		cfg->rgsize = calc_rgsize_extsize_power(cfg);
 		cfg->rgcount = cfg->rtblocks / cfg->rgsize +
@@ -4538,7 +4536,6 @@ 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;
@@ -4547,7 +4544,8 @@ adjust_nr_zones(
 	if (zt->rt.nr_zones)
 		max_zones = zt->rt.nr_zones;
 	else
-		max_zones = DTOBT(xi->rt.size, cfg->blocklog) / cfg->rgsize;
+		max_zones = DTOBT(cli->xi->rt.size, cfg->blocklog) /
+				cfg->rgsize;
 
 	if (!cli->rgcount)
 		cfg->rgcount += XFS_RESERVED_ZONES;
@@ -4576,7 +4574,6 @@ static void
 calculate_zone_geometry(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	struct zone_topology	*zt)
 {
 	if (cfg->rtblocks == 0) {
@@ -4645,7 +4642,7 @@ _("rgsize (%s) not a multiple of fs blk size (%d)\n"),
 	}
 
 	if (cli->rtsize || cli->rgcount)
-		adjust_nr_zones(cfg, cli, xi, zt);
+		adjust_nr_zones(cfg, cli, zt);
 
 	if (cfg->rgcount < XFS_MIN_ZONES)  {
 		fprintf(stderr,
@@ -4984,7 +4981,6 @@ static uint64_t
 calc_concurrency_logblocks(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	unsigned int		max_tx_bytes)
 {
 	uint64_t		log_bytes;
@@ -4992,7 +4988,7 @@ calc_concurrency_logblocks(
 	unsigned int		new_logblocks;
 
 	if (cli->log_concurrency < 0) {
-		if (!ddev_is_solidstate(xi))
+		if (!ddev_is_solidstate(cli->xi))
 			goto out;
 
 		cli->log_concurrency = nr_cpus();
@@ -5160,7 +5156,6 @@ static void
 calculate_log_size(
 	struct mkfs_params	*cfg,
 	struct cli_params	*cli,
-	struct libxfs_init	*xi,
 	struct xfs_mount	*mp)
 {
 	struct xfs_sb		*sbp = &mp->m_sb;
@@ -5225,8 +5220,8 @@ _("external log device size %lld blocks too small, must be at least %lld blocks\
 		if (cfg->lsunit) {
 			uint64_t	max_logblocks;
 
-			max_logblocks = min(DTOBT(xi->log.size, cfg->blocklog),
-					    XFS_MAX_LOG_BLOCKS);
+			max_logblocks = min(XFS_MAX_LOG_BLOCKS,
+				DTOBT(cli->xi->log.size, cfg->blocklog));
 			align_log_size(cfg, cfg->lsunit, max_logblocks);
 		}
 
@@ -5261,7 +5256,7 @@ _("max log size %d smaller than min log size %d, filesystem is too small\n"),
 
 		if (cli->log_concurrency != 0)
 			cfg->logblocks = calc_concurrency_logblocks(cfg, cli,
-							xi, max_tx_bytes);
+							max_tx_bytes);
 
 		/* But don't go below a reasonable size */
 		cfg->logblocks = max(cfg->logblocks,
@@ -6135,12 +6130,12 @@ main(
 	 * dependent on device sizes. Once calculated, make sure everything
 	 * aligns to device geometry correctly.
 	 */
-	calculate_initial_ag_geometry(&cfg, &cli, &xi);
+	calculate_initial_ag_geometry(&cfg, &cli);
 	align_ag_geometry(&cfg, &ft);
 	if (cfg.sb_feat.zoned)
-		calculate_zone_geometry(&cfg, &cli, &xi, &zt);
+		calculate_zone_geometry(&cfg, &cli, &zt);
 	else
-		calculate_rtgroup_geometry(&cfg, &cli, &xi);
+		calculate_rtgroup_geometry(&cfg, &cli);
 
 	calculate_imaxpct(&cfg, &cli);
 
@@ -6164,7 +6159,7 @@ main(
 	 * With the mount set up, we can finally calculate the log size
 	 * constraints and do default size calculations and final validation
 	 */
-	calculate_log_size(&cfg, &cli, &xi, mp);
+	calculate_log_size(&cfg, &cli, mp);
 
 	finish_superblock_setup(&cfg, mp, sbp);
 
Signature · 1 line
2.47.3
#7 lore raw

PATCH 4/4 mkfs: split zone reset from discard

In reply to Christoph Hellwig

Zone reset is a mandatory part of creating a file system on a zoned device, unlike discard, which can be skipped. It also is implemented a bit different, so just split the handling. This also means that we can now support the -K option to skip discard on the data section for zoned file systems.

---
 mkfs/xfs_mkfs.c | 112 +++++++++++++++++++++++-------------------------
 1 file changed, 53 insertions(+), 59 deletions(-)
mkfs/xfs_mkfs.c +53 -59
index 09a69af31be5..cd4f3ba4a549 100644
@@ -1607,34 +1607,6 @@ discard_blocks(int fd, uint64_t nsectors, int quiet)
 		printf("Done.\n");
 }
 
-static void
-reset_zones(
-	struct mkfs_params	*cfg,
-	int			fd,
-	uint64_t		start_sector,
-	uint64_t		nsectors,
-	int			quiet)
-{
-	struct blk_zone_range range = {
-		.sector		= start_sector,
-		.nr_sectors	= nsectors,
-	};
-
-	if (!quiet) {
-		printf("Resetting zones...");
-		fflush(stdout);
-	}
-
-	if (ioctl(fd, BLKRESETZONE, &range) < 0) {
-		if (!quiet)
-			printf(" FAILED (%d)\n", -errno);
-		exit(1);
-	}
-
-	if (!quiet)
-		printf("Done.\n");
-}
-
 static __attribute__((noreturn)) void
 illegal_option(
 	const char		*value,
@@ -3780,41 +3752,66 @@ discard_devices(
 	struct zone_topology	*zt,
 	int			quiet)
 {
-	/*
-	 * This function has to be called after libxfs has been initialized.
-	 */
-
 	if (!xi->data.isfile) {
 		uint64_t	nsectors = xi->data.size;
 
-		if (cfg->rtstart && zt->data.nr_zones) {
-			/*
-			 * Note that the zone reset here includes the LBA range
-			 * for the data device.
-			 *
-			 * This is because doing a single zone reset all on the
-			 * entire device (which the kernel automatically does
-			 * for us for a full device range) is a lot faster than
-			 * resetting each zone individually and resetting
-			 * the conventional zones used for the data device is a
-			 * no-op.
-			 */
-			reset_zones(cfg, xi->data.fd, 0,
-					cfg->rtstart + xi->rt.size, quiet);
+		if (cfg->rtstart && zt->data.nr_zones)
 			nsectors -= cfg->rtstart;
-		}
 		discard_blocks(xi->data.fd, nsectors, quiet);
 	}
-	if (xi->rt.dev && !xi->rt.isfile) {
-		if (zt->rt.nr_zones)
-			reset_zones(cfg, xi->rt.fd, 0, xi->rt.size, quiet);
-		else
-			discard_blocks(xi->rt.fd, xi->rt.size, quiet);
-	}
+	if (xi->rt.dev && !xi->rt.isfile && !zt->rt.nr_zones)
+		discard_blocks(xi->rt.fd, xi->rt.size, quiet);
 	if (xi->log.dev && xi->log.dev != xi->data.dev && !xi->log.isfile)
 		discard_blocks(xi->log.fd, xi->log.size, quiet);
 }
 
+static void
+reset_zones(
+	struct mkfs_params	*cfg,
+	struct libxfs_dev	*dev,
+	uint64_t		size,
+	bool			quiet)
+{
+	struct blk_zone_range range = {
+		.nr_sectors	= size,
+	};
+
+	if (!quiet) {
+		printf("Resetting zones...");
+		fflush(stdout);
+	}
+	if (ioctl(dev->fd, BLKRESETZONE, &range) < 0) {
+		if (!quiet)
+			printf(" FAILED (%d)\n", -errno);
+		exit(1);
+	}
+	if (!quiet)
+		printf("Done.\n");
+}
+
+static void
+reset_devices(
+	struct mkfs_params	*cfg,
+	struct libxfs_init	*xi,
+	struct zone_topology	*zt,
+	int			quiet)
+{
+	/*
+	 * Note that the zone reset here includes the conventional zones used
+	 * for the data device.
+	 *
+	 * It is done that way because doing a single zone reset all on the
+	 * entire device (which the kernel automatically does for us for a full
+	 * device range) is a lot faster than resetting each zone individually
+	 * and resetting the conventional zones used for the data device is a
+	 * no-op.
+	 */
+	if (!xi->data.isfile && zt->data.nr_zones && cfg->rtstart)
+		reset_zones(cfg, &xi->data, cfg->rtstart + xi->rt.size, quiet);
+	if (xi->rt.dev && !xi->rt.isfile && zt->rt.nr_zones)
+		reset_zones(cfg, &xi->rt, xi->rt.size, quiet);
+}
+
 static void
 validate_datadev(
 	struct mkfs_params	*cfg,
@@ -6196,13 +6193,10 @@ main(
 	/*
 	 * All values have been validated, discard the old device layout.
 	 */
-	if (cli.sb_feat.zoned && !discard) {
-		fprintf(stderr,
- _("-K not support for zoned file systems.\n"));
-		return 1;
-	}
 	if (discard && !dry_run)
-		discard_devices(&cfg, &xi, &zt, quiet);
+		discard_devices(&cfg, cli.xi, &zt, quiet);
+	if (cli.sb_feat.zoned && !dry_run)
+		reset_devices(&cfg, cli.xi, &zt, quiet);
 
 	/*
 	 * we need the libxfs buffer cache from here on in.
Signature · 1 line
2.47.3
Reviewed-by on v1 4/4
#8 lore raw

Re: [PATCH 4/4] mkfs: split zone reset from discard

Quoted 7 lines · Christoph Hellwig

Zone reset is a mandatory part of creating a file system on a zoned device, unlike discard, which can be skipped. It also is implemented a bit different, so just split the handling. This also means that we can now support the -K option to skip discard on the data section for zoned file systems.

Nice feature improvement! ;)

--D

Quoted 156 lines
---
 mkfs/xfs_mkfs.c | 112 +++++++++++++++++++++++-------------------------
 1 file changed, 53 insertions(+), 59 deletions(-)
mkfs/xfs_mkfs.c +53 -59
index 09a69af31be5..cd4f3ba4a549 100644
@@ -1607,34 +1607,6 @@ discard_blocks(int fd, uint64_t nsectors, int quiet)
 		printf("Done.\n");
 }
 
-static void
-reset_zones(
-	struct mkfs_params	*cfg,
-	int			fd,
-	uint64_t		start_sector,
-	uint64_t		nsectors,
-	int			quiet)
-{
-	struct blk_zone_range range = {
-		.sector		= start_sector,
-		.nr_sectors	= nsectors,
-	};
-
-	if (!quiet) {
-		printf("Resetting zones...");
-		fflush(stdout);
-	}
-
-	if (ioctl(fd, BLKRESETZONE, &range) < 0) {
-		if (!quiet)
-			printf(" FAILED (%d)\n", -errno);
-		exit(1);
-	}
-
-	if (!quiet)
-		printf("Done.\n");
-}
-
 static __attribute__((noreturn)) void
 illegal_option(
 	const char		*value,
@@ -3780,41 +3752,66 @@ discard_devices(
 	struct zone_topology	*zt,
 	int			quiet)
 {
-	/*
-	 * This function has to be called after libxfs has been initialized.
-	 */
-
 	if (!xi->data.isfile) {
 		uint64_t	nsectors = xi->data.size;
 
-		if (cfg->rtstart && zt->data.nr_zones) {
-			/*
-			 * Note that the zone reset here includes the LBA range
-			 * for the data device.
-			 *
-			 * This is because doing a single zone reset all on the
-			 * entire device (which the kernel automatically does
-			 * for us for a full device range) is a lot faster than
-			 * resetting each zone individually and resetting
-			 * the conventional zones used for the data device is a
-			 * no-op.
-			 */
-			reset_zones(cfg, xi->data.fd, 0,
-					cfg->rtstart + xi->rt.size, quiet);
+		if (cfg->rtstart && zt->data.nr_zones)
 			nsectors -= cfg->rtstart;
-		}
 		discard_blocks(xi->data.fd, nsectors, quiet);
 	}
-	if (xi->rt.dev && !xi->rt.isfile) {
-		if (zt->rt.nr_zones)
-			reset_zones(cfg, xi->rt.fd, 0, xi->rt.size, quiet);
-		else
-			discard_blocks(xi->rt.fd, xi->rt.size, quiet);
-	}
+	if (xi->rt.dev && !xi->rt.isfile && !zt->rt.nr_zones)
+		discard_blocks(xi->rt.fd, xi->rt.size, quiet);
 	if (xi->log.dev && xi->log.dev != xi->data.dev && !xi->log.isfile)
 		discard_blocks(xi->log.fd, xi->log.size, quiet);
 }
 
+static void
+reset_zones(
+	struct mkfs_params	*cfg,
+	struct libxfs_dev	*dev,
+	uint64_t		size,
+	bool			quiet)
+{
+	struct blk_zone_range range = {
+		.nr_sectors	= size,
+	};
+
+	if (!quiet) {
+		printf("Resetting zones...");
+		fflush(stdout);
+	}
+	if (ioctl(dev->fd, BLKRESETZONE, &range) < 0) {
+		if (!quiet)
+			printf(" FAILED (%d)\n", -errno);
+		exit(1);
+	}
+	if (!quiet)
+		printf("Done.\n");
+}
+
+static void
+reset_devices(
+	struct mkfs_params	*cfg,
+	struct libxfs_init	*xi,
+	struct zone_topology	*zt,
+	int			quiet)
+{
+	/*
+	 * Note that the zone reset here includes the conventional zones used
+	 * for the data device.
+	 *
+	 * It is done that way because doing a single zone reset all on the
+	 * entire device (which the kernel automatically does for us for a full
+	 * device range) is a lot faster than resetting each zone individually
+	 * and resetting the conventional zones used for the data device is a
+	 * no-op.
+	 */
+	if (!xi->data.isfile && zt->data.nr_zones && cfg->rtstart)
+		reset_zones(cfg, &xi->data, cfg->rtstart + xi->rt.size, quiet);
+	if (xi->rt.dev && !xi->rt.isfile && zt->rt.nr_zones)
+		reset_zones(cfg, &xi->rt, xi->rt.size, quiet);
+}
+
 static void
 validate_datadev(
 	struct mkfs_params	*cfg,
@@ -6196,13 +6193,10 @@ main(
 	/*
 	 * All values have been validated, discard the old device layout.
 	 */
-	if (cli.sb_feat.zoned && !discard) {
-		fprintf(stderr,
- _("-K not support for zoned file systems.\n"));
-		return 1;
-	}
 	if (discard && !dry_run)
-		discard_devices(&cfg, &xi, &zt, quiet);
+		discard_devices(&cfg, cli.xi, &zt, quiet);
+	if (cli.sb_feat.zoned && !dry_run)
+		reset_devices(&cfg, cli.xi, &zt, quiet);
 
 	/*
 	 * we need the libxfs buffer cache from here on in.
Signature · 1 line
2.47.3