fs/zfs/vdev_label.c

31  * The vdev label serves several distinct purposes:
49  * label, it only consumes the information in the first three cases.  The
57  * Before describing the contents of the label, it's important to understand how
74  * the same transaction group (10).  Each label is mirrored and checksummed, so
75  * that we can detect when we fail partway through writing the label.
80  * 	1. For each vdev, update 'L1' to the new label
82  * 	3. For each vdev, update 'L2' to the new label
84  * Given arbitrary failure, we can determine the correct label to use based on
91  * Another added complexity is that not every label is updated when the config
93  * every label for every device in the pool.  This means that both L1 and L2 may
101  * The vdev label consists of two distinct parts, and is wrapped within the
102  * vdev_label_t structure.  The label includes 8k of padding to permit legacy
105  * The first half of the label is a packed nvlist which contains pool wide
109  * The latter half of the label consists of a redundant array of uberblocks.
123  * 	txg		Transaction group in which this label was written
129  * Each leaf device label also contains the following:
157  * Basic routines to read and write from a vdev label.
161 vdev_label_offset(uint64_t psize, int l, uint64_t offset)
163 	ASSERT(offset < sizeof (vdev_label_t));
166 	return (offset + l * sizeof (vdev_label_t) + (l < VDEV_LABELS / 2 ?
171  * Returns back the vdev label associated with the passed in offset.
174 vdev_label_number(uint64_t psize, uint64_t offset)
178 	if (offset >= psize - VDEV_LABEL_END_SIZE) {
179 		offset -= psize - VDEV_LABEL_END_SIZE;
180 		offset += (VDEV_LABELS / 2) * sizeof (vdev_label_t);
182 	l = offset / sizeof (vdev_label_t);
187 vdev_label_read(zio_t *zio, vdev_t *vd, int l, void *buf, uint64_t offset,
195 	    vdev_label_offset(vd->vdev_psize, l, offset),
201 vdev_label_write(zio_t *zio, vdev_t *vd, int l, void *buf, uint64_t offset,
211 	    vdev_label_offset(vd->vdev_psize, l, offset),
440  * Returns the configuration from the label of the given vdev. For vdevs
441  * which don't have a txg value stored on their label (i.e. spares/cache)
443  * the configuration from the first valid label we find. Otherwise,
444  * find the most up-to-date label that does not exceed the specified
468 		nvlist_t *label = NULL;
478 		    &label, 0) == 0) {
485 			 * configuration from the first valid label we
488 			error = nvlist_lookup_uint64(label,
491 				config = label;
496 				config = fnvlist_dup(label);
500 		if (label != NULL) {
501 			nvlist_free(label);
502 			label = NULL;
527 	nvlist_t *label;
535 	 * Read the label, if any, and perform some basic sanity checks.
537 	if ((label = vdev_label_read_config(vd, -1ULL)) == NULL)
540 	(void) nvlist_lookup_uint64(label, ZPOOL_CONFIG_CREATE_TXG,
543 	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE,
545 	    nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID,
547 		nvlist_free(label);
552 	    (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID,
554 	    nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_TXG,
556 		nvlist_free(label);
560 	nvlist_free(label);
575 	 * unused) label.  This is only an error if the create transaction
632  * Initialize a vdev label.  We check to make sure each leaf device is not in
633  * use, and writable.  We put down an initial label which we will later
634  * overwrite with a complete label.  Note that it's important to do this
643 	nvlist_t *label;
737 	 * Initialize its label.
743 	 * Generate a label describing the pool and our top-level vdev.
751 		 * For inactive hot spares, we generate a special label that
753 		 * label if we are adding a hot spare, or if we are removing an
757 		VERIFY(nvlist_alloc(&label, NV_UNIQUE_NAME, KM_SLEEP) == 0);
759 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_VERSION,
761 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_POOL_STATE,
763 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_GUID,
768 		 * For level 2 ARC devices, add a special label.
770 		VERIFY(nvlist_alloc(&label, NV_UNIQUE_NAME, KM_SLEEP) == 0);
772 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_VERSION,
774 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_POOL_STATE,
776 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_GUID,
783 		label = spa_config_generate(spa, vd, txg, B_FALSE);
790 		VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_CREATE_TXG,
797 	error = nvlist_pack(label, &buf, &buflen, NV_ENCODE_XDR, KM_SLEEP);
799 		nvlist_free(label);
850 	nvlist_free(label);
1001  * configuration. First, we read the uberblock array of each label of each
1029 	 * It's possible that the best uberblock was discovered on a label
1156 	nvlist_t *label;
1171 	 * Generate a label describing the top-level config to which we belong.
1173 	label = spa_config_generate(vd->vdev_spa, vd, txg, B_FALSE);
1181 	if (nvlist_pack(label, &buf, &buflen, NV_ENCODE_XDR, KM_SLEEP) == 0) {
1192 	nvlist_free(label);
1260 	 * Normally, we don't want to try too hard to write every label and
1263 	 * single label out, we should retry with ZIO_FLAG_TRYHARD before
1310 	 * the new labels to disk to ensure that all even-label updates
1341 	 * to disk to ensure that all odd-label updates are committed to