Create: add support for RAID0 layouts.

Since Linux 5.4 a layout is needed for RAID0 arrays with
varying device sizes.
This patch makes the layout of an array visible (via --examine)
and sets the layout on newly created arrays.
--layout=dangerous
can be used to avoid setting a layout so that they array
can be used on older kernels.

Tested-by: dann frazier <dann.frazier@canonical.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>

Create.c

@@ -51,6 +51,9 @@ static int default_layout(struct supertype *st, int level, int verbose)
 		default: /* no layout */
 			layout = 0;
 			break;
+		case 0:
+			layout = RAID0_ORIG_LAYOUT;
+			break;
 		case 10:
 			layout = 0x102; /* near=2, far=1 */
 			if (verbose > 0)
@@ -950,6 +953,11 @@ int Create(struct supertype *st, char *mddev,
 				if (rv) {
 					pr_err("ADD_NEW_DISK for %s failed: %s\n",
 					       dv->devname, strerror(errno));
+					if (errno == EINVAL &&
+					    info.array.level == 0) {
+						pr_err("Possibly your kernel doesn't support RAID0 layouts.\n");
+						pr_err("Either upgrade, or use --layout=dangerous\n");
+					}
 					goto abort_locked;
 				}
 				break;
@@ -1046,6 +1054,9 @@ int Create(struct supertype *st, char *mddev,
 			if (ioctl(mdfd, RUN_ARRAY, &param)) {
 				pr_err("RUN_ARRAY failed: %s\n",
 				       strerror(errno));
+				if (errno == 524 /* ENOTSUP */ &&
+				    info.array.level == 0)
+					cont_err("Please use --layout=original or --layout=alternate\n");
 				if (info.array.chunk_size & (info.array.chunk_size-1)) {
 					cont_err("Problem may be that chunk size is not a power of 2\n");
 				}

Detail.c

@@ -525,6 +525,11 @@ int Detail(char *dev, struct context *c)
 			printf("            Layout : %s\n",
 			       str ? str : "-unknown-");
 		}
+		if (array.level == 0 && array.layout) {
+			str = map_num(r0layout, array.layout);
+			printf("            Layout : %s\n",
+			       str ? str : "-unknown-");
+		}
 		if (array.level == 6) {
 			str = map_num(r6layout, array.layout);
 			printf("            Layout : %s\n",

maps.c

@@ -73,6 +73,18 @@ mapping_t r6layout[] = {
 	{ NULL, UnSet }
 };
 
+/* raid0 layout is only needed because of a bug in 3.14 which changed
+ * the effective layout of raid0 arrays with varying device sizes.
+ */
+mapping_t r0layout[] = {
+	{ "original", RAID0_ORIG_LAYOUT},
+	{ "alternate", RAID0_ALT_MULTIZONE_LAYOUT},
+	{ "1", 1}, /* aka ORIG */
+	{ "2", 2}, /* aka ALT */
+	{ "dangerous", 0},
+	{ NULL, UnSet},
+};
+
 mapping_t pers[] = {
 	{ "linear", LEVEL_LINEAR},
 	{ "raid0", 0},

md.4

@@ -193,6 +193,20 @@ smallest device has been exhausted, the RAID0 driver starts
 collecting chunks into smaller stripes that only span the drives which
 still have remaining space.
 
+A bug was introduced in linux 3.14 which changed the layout of blocks in
+a RAID0 beyond the region that is striped over all devices.  This bug
+does not affect an array with all devices the same size, but can affect
+other RAID0 arrays.
+
+Linux 5.4 (and some stable kernels to which the change was backported)
+will not normally assemble such an array as it cannot know which layout
+to use.  There is a module parameter "raid0.default_layout" which can be
+set to "1" to force the kernel to use the pre-3.14 layout or to "2" to
+force it to use the 3.14-and-later layout.  when creating a new RAID0
+array,
+.I mdadm
+will record the chosen layout in the metadata in a way that allows newer
+kernels to assemble the array without needing a module parameter.
 
 .SS RAID1
 

mdadm.8.in

@@ -593,6 +593,8 @@ to change the RAID level in some cases.  See LEVEL CHANGES below.
 This option configures the fine details of data layout for RAID5, RAID6,
 and RAID10 arrays, and controls the failure modes for
 .IR faulty .
+It can also be used for working around a kernel bug with RAID0, but generally
+doesn't need to be used explicitly.
 
 The layout of the RAID5 parity block can be one of
 .BR left\-asymmetric ,
@@ -652,7 +654,7 @@ option to set subsequent failure modes.
 "clear" or "none" will remove any pending or periodic failure modes,
 and "flush" will clear any persistent faults.
 
-Finally, the layout options for RAID10 are one of 'n', 'o' or 'f' followed
+The layout options for RAID10 are one of 'n', 'o' or 'f' followed
 by a small number.  The default is 'n2'.  The supported options are:
 
 .I 'n'
@@ -677,6 +679,32 @@ devices in the array.  It does not need to divide evenly into that
 number (e.g. it is perfectly legal to have an 'n2' layout for an array
 with an odd number of devices).
 
+A bug introduced in Linux 3.14 means that RAID0 arrays
+.B "with devices of differing sizes"
+started using a different layout.  This could lead to
+data corruption.  Since Linux 5.4 (and various stable releases that received
+backports), the kernel will not accept such an array unless
+a layout is explictly set.  It can be set to
+.RB ' original '
+or
+.RB ' alternate '.
+When creating a new array,
+.I mdadm
+will select
+.RB ' original '
+by default, so the layout does not normally need to be set.
+An array created for either
+.RB ' original '
+or
+.RB ' alternate '
+will not be recognized by an (unpatched) kernel prior to 5.4.  To create
+a RAID0 array with devices of differing sizes that can be used on an
+older kernel, you can set the layout to
+.RB ' dangerous '.
+This will use whichever layout the running kernel supports, so the data
+on the array may become corrupt when changing kernel from pre-3.14 to a
+later kernel.
+
 When an array is converted between RAID5 and RAID6 an intermediate
 RAID6 layout is used in which the second parity block (Q) is always on
 the last device.  To convert a RAID5 to RAID6 and leave it in this new

mdadm.c

@@ -550,6 +550,14 @@ int main(int argc, char *argv[])
 				pr_err("raid level must be given before layout.\n");
 				exit(2);
 
+			case 0:
+				s.layout = map_name(r0layout, optarg);
+				if (s.layout == UnSet) {
+					pr_err("layout %s not understood for raid0.\n",
+						optarg);
+					exit(2);
+				}
+				break;
 			case 5:
 				s.layout = map_name(r5layout, optarg);
 				if (s.layout == UnSet) {

mdadm.h

@@ -763,7 +763,8 @@ extern int restore_stripes(int *dest, unsigned long long *offsets,
 
 extern char *map_num(mapping_t *map, int num);
 extern int map_name(mapping_t *map, char *name);
-extern mapping_t r5layout[], r6layout[], pers[], modes[], faultylayout[];
+extern mapping_t r0layout[], r5layout[], r6layout[],
+	pers[], modes[], faultylayout[];
 extern mapping_t consistency_policies[], sysfs_array_states[];
 
 extern char *map_dev_preferred(int major, int minor, int create,
@@ -1758,6 +1759,11 @@ char *xstrdup(const char *str);
 #define makedev(M,m) (((M)<<8) | (m))
 #endif
 
+enum r0layout {
+	RAID0_ORIG_LAYOUT = 1,
+	RAID0_ALT_MULTIZONE_LAYOUT = 2,
+};
+
 /* for raid4/5/6 */
 #define ALGORITHM_LEFT_ASYMMETRIC	0
 #define ALGORITHM_RIGHT_ASYMMETRIC	1

super0.c

@@ -1291,6 +1291,12 @@ static int validate_geometry0(struct supertype *st, int level,
 	if (*chunk == UnSet)
 		*chunk = DEFAULT_CHUNK;
 
+	if (level == 0 && layout != UnSet) {
+		if (verbose)
+			pr_err("0.90 metadata does not support layouts for RAID0\n");
+		return 0;
+	}
+
 	if (!subdev)
 		return 1;
 

super1.c

@@ -43,7 +43,7 @@ struct mdp_superblock_1 {
 
 	__u64	ctime;		/* lo 40 bits are seconds, top 24 are microseconds or 0*/
 	__u32	level;		/* -4 (multipath), -1 (linear), 0,1,4,5 */
-	__u32	layout;		/* only for raid5 currently */
+	__u32	layout;		/* used for raid5, raid6, raid10, and raid0 */
 	__u64	size;		/* used size of component devices, in 512byte sectors */
 
 	__u32	chunksize;	/* in 512byte sectors */
@@ -144,6 +144,7 @@ struct misc_dev_info {
 #define	MD_FEATURE_JOURNAL		512 /* support write journal */
 #define	MD_FEATURE_PPL			1024 /* support PPL */
 #define	MD_FEATURE_MUTLIPLE_PPLS	2048 /* support for multiple PPLs */
+#define	MD_FEATURE_RAID0_LAYOUT		4096 /* layout is meaningful in RAID0 */
 #define	MD_FEATURE_ALL			(MD_FEATURE_BITMAP_OFFSET	\
 					|MD_FEATURE_RECOVERY_OFFSET	\
 					|MD_FEATURE_RESHAPE_ACTIVE	\
@@ -155,6 +156,7 @@ struct misc_dev_info {
 					|MD_FEATURE_JOURNAL		\
 					|MD_FEATURE_PPL			\
 					|MD_FEATURE_MULTIPLE_PPLS	\
+					|MD_FEATURE_RAID0_LAYOUT	\
 					)
 
 static int role_from_sb(struct mdp_superblock_1 *sb)
@@ -498,6 +500,11 @@ static void examine_super1(struct supertype *st, char *homehost)
 	printf("         Events : %llu\n",
 	       (unsigned long long)__le64_to_cpu(sb->events));
 	printf("\n");
+	if (__le32_to_cpu(sb->level) == 0 &&
+	    (sb->feature_map & __cpu_to_le32(MD_FEATURE_RAID0_LAYOUT))) {
+		c = map_num(r0layout, __le32_to_cpu(sb->layout));
+		printf("         Layout : %s\n", c?c:"-unknown-");
+	}
 	if (__le32_to_cpu(sb->level) == 5) {
 		c = map_num(r5layout, __le32_to_cpu(sb->layout));
 		printf("         Layout : %s\n", c?c:"-unknown-");
@@ -1646,6 +1653,7 @@ struct devinfo {
 	int fd;
 	char *devname;
 	long long data_offset;
+	unsigned long long dev_size;
 	mdu_disk_info_t disk;
 	struct devinfo *next;
 };
@@ -1687,6 +1695,7 @@ static int add_to_super1(struct supertype *st, mdu_disk_info_t *dk,
 	di->devname = devname;
 	di->disk = *dk;
 	di->data_offset = data_offset;
+	get_dev_size(fd, NULL, &di->dev_size);
 	di->next = NULL;
 	*dip = di;
 
@@ -1888,10 +1897,25 @@ static int write_init_super1(struct supertype *st)
 	unsigned long long sb_offset;
 	unsigned long long data_offset;
 	long bm_offset;
+	int raid0_need_layout = 0;
 
 	for (di = st->info; di; di = di->next) {
 		if (di->disk.state & (1 << MD_DISK_JOURNAL))
 			sb->feature_map |= __cpu_to_le32(MD_FEATURE_JOURNAL);
+		if (sb->level == 0 && sb->layout != 0) {
+			struct devinfo *di2 = st->info;
+			unsigned long long s1, s2;
+			s1 = di->dev_size;
+			if (di->data_offset != INVALID_SECTORS)
+				s1 -= di->data_offset;
+			s1 /= __le32_to_cpu(sb->chunksize);
+			s2 = di2->dev_size;
+			if (di2->data_offset != INVALID_SECTORS)
+				s2 -= di2->data_offset;
+			s2 /= __le32_to_cpu(sb->chunksize);
+			if (s1 != s2)
+				raid0_need_layout = 1;
+		}
 	}
 
 	for (di = st->info; di; di = di->next) {
@@ -2039,6 +2063,10 @@ static int write_init_super1(struct supertype *st)
 			sb->bblog_offset = 0;
 		}
 
+		/* RAID0 needs a layout if devices aren't all the same size */
+		if (raid0_need_layout)
+			sb->feature_map |= __cpu_to_le32(MD_FEATURE_RAID0_LAYOUT);
+
 		sb->sb_csum = calc_sb_1_csum(sb);
 		rv = store_super1(st, di->fd);