*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
+ * Copyright (c) 2011, 2015 by Delphix. All rights reserved.
* Copyright (c) 2014, Joyent, Inc. All rights reserved.
- * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
+ * Copyright 2014 HybridCluster. All rights reserved.
+ * Copyright 2016 RackTop Systems.
* Copyright (c) 2016 Actifio, Inc. All rights reserved.
*/
/* Set this tunable to TRUE to replace corrupt data with 0x2f5baddb10c */
int zfs_send_corrupt_data = B_FALSE;
-int zfs_send_queue_length = 16 * 1024 * 1024;
-int zfs_recv_queue_length = 16 * 1024 * 1024;
+int zfs_send_queue_length = SPA_MAXBLOCKSIZE;
+int zfs_recv_queue_length = SPA_MAXBLOCKSIZE;
+/* Set this tunable to FALSE to disable setting of DRR_FLAG_FREERECORDS */
+int zfs_send_set_freerecords_bit = B_TRUE;
static char *dmu_recv_tag = "dmu_recv_tag";
-static const char *recv_clone_name = "%recv";
+const char *recv_clone_name = "%recv";
#define BP_SPAN(datablkszsec, indblkshift, level) \
(((uint64_t)datablkszsec) << (SPA_MINBLOCKSHIFT + \
(level) * (indblkshift - SPA_BLKPTRSHIFT)))
+static void byteswap_record(dmu_replay_record_t *drr);
+
struct send_thread_arg {
bqueue_t q;
dsl_dataset_t *ds; /* Dataset to traverse */
int flags; /* flags to pass to traverse_dataset */
int error_code;
boolean_t cancel;
+ zbookmark_phys_t resume;
};
struct send_block_record {
{
dump_bytes_io_t *dbi = (dump_bytes_io_t *)arg;
dmu_sendarg_t *dsp = dbi->dbi_dsp;
- dsl_dataset_t *ds = dsp->dsa_os->os_dsl_dataset;
+ dsl_dataset_t *ds = dmu_objset_ds(dsp->dsa_os);
ssize_t resid; /* have to get resid to get detailed errno */
- ASSERT0(dbi->dbi_len % 8);
+
+ /*
+ * The code does not rely on len being a multiple of 8. We keep
+ * this assertion because of the corresponding assertion in
+ * receive_read(). Keeping this assertion ensures that we do not
+ * inadvertently break backwards compatibility (causing the assertion
+ * in receive_read() to trigger on old software). Newer feature flags
+ * (such as raw send) may break this assertion since they were
+ * introduced after the requirement was made obsolete.
+ */
+
+ ASSERT(dbi->dbi_len % 8 == 0 ||
+ (dsp->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) != 0);
dsp->dsa_err = vn_rdwr(UIO_WRITE, dsp->dsa_vp,
(caddr_t)dbi->dbi_buf, dbi->dbi_len,
{
ASSERT3U(offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum),
==, sizeof (dmu_replay_record_t) - sizeof (zio_cksum_t));
- fletcher_4_incremental_native(dsp->dsa_drr,
+ (void) fletcher_4_incremental_native(dsp->dsa_drr,
offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum),
&dsp->dsa_zc);
- if (dsp->dsa_drr->drr_type != DRR_BEGIN) {
+ if (dsp->dsa_drr->drr_type == DRR_BEGIN) {
+ dsp->dsa_sent_begin = B_TRUE;
+ } else {
ASSERT(ZIO_CHECKSUM_IS_ZERO(&dsp->dsa_drr->drr_u.
drr_checksum.drr_checksum));
dsp->dsa_drr->drr_u.drr_checksum.drr_checksum = dsp->dsa_zc;
}
- fletcher_4_incremental_native(&dsp->dsa_drr->
+ if (dsp->dsa_drr->drr_type == DRR_END) {
+ dsp->dsa_sent_end = B_TRUE;
+ }
+ (void) fletcher_4_incremental_native(&dsp->dsa_drr->
drr_u.drr_checksum.drr_checksum,
sizeof (zio_cksum_t), &dsp->dsa_zc);
if (dump_bytes(dsp, dsp->dsa_drr, sizeof (dmu_replay_record_t)) != 0)
return (SET_ERROR(EINTR));
if (payload_len != 0) {
- fletcher_4_incremental_native(payload, payload_len,
+ (void) fletcher_4_incremental_native(payload, payload_len,
&dsp->dsa_zc);
if (dump_bytes(dsp, payload, payload_len) != 0)
return (SET_ERROR(EINTR));
return (0);
}
+/*
+ * Fill in the drr_free struct, or perform aggregation if the previous record is
+ * also a free record, and the two are adjacent.
+ *
+ * Note that we send free records even for a full send, because we want to be
+ * able to receive a full send as a clone, which requires a list of all the free
+ * and freeobject records that were generated on the source.
+ */
static int
dump_free(dmu_sendarg_t *dsp, uint64_t object, uint64_t offset,
uint64_t length)
* that the receiving system doesn't have any dbufs in the range
* being freed. This is always true because there is a one-record
* constraint: we only send one WRITE record for any given
- * object+offset. We know that the one-record constraint is
+ * object,offset. We know that the one-record constraint is
* true because we always send data in increasing order by
* object,offset.
*
(object == dsp->dsa_last_data_object &&
offset > dsp->dsa_last_data_offset));
- /*
- * If we are doing a non-incremental send, then there can't
- * be any data in the dataset we're receiving into. Therefore
- * a free record would simply be a no-op. Save space by not
- * sending it to begin with.
- */
- if (!dsp->dsa_incremental)
- return (0);
-
- if (length != -1ULL && offset + length < offset)
- length = -1ULL;
-
/*
* If there is a pending op, but it's not PENDING_FREE, push it out,
* since free block aggregation can only be done for blocks of the
if (dsp->dsa_pending_op == PENDING_FREE) {
/*
- * There should never be a PENDING_FREE if length is -1
- * (because dump_dnode is the only place where this
- * function is called with a -1, and only after flushing
- * any pending record).
+ * There should never be a PENDING_FREE if length is
+ * DMU_OBJECT_END (because dump_dnode is the only place where
+ * this function is called with a DMU_OBJECT_END, and only after
+ * flushing any pending record).
*/
- ASSERT(length != -1ULL);
+ ASSERT(length != DMU_OBJECT_END);
/*
* Check to see whether this free block can be aggregated
* with pending one.
*/
if (drrf->drr_object == object && drrf->drr_offset +
drrf->drr_length == offset) {
- drrf->drr_length += length;
+ if (offset + length < offset)
+ drrf->drr_length = DMU_OBJECT_END;
+ else
+ drrf->drr_length += length;
return (0);
} else {
/* not a continuation. Push out pending record */
dsp->dsa_drr->drr_type = DRR_FREE;
drrf->drr_object = object;
drrf->drr_offset = offset;
- drrf->drr_length = length;
+ if (offset + length < offset)
+ drrf->drr_length = DMU_OBJECT_END;
+ else
+ drrf->drr_length = length;
drrf->drr_toguid = dsp->dsa_toguid;
- if (length == -1ULL) {
+ if (length == DMU_OBJECT_END) {
if (dump_record(dsp, NULL, 0) != 0)
return (SET_ERROR(EINTR));
} else {
}
static int
-dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type,
- uint64_t object, uint64_t offset, int blksz, const blkptr_t *bp, void *data)
+dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type, uint64_t object,
+ uint64_t offset, int lsize, int psize, const blkptr_t *bp, void *data)
{
+ uint64_t payload_size;
+ boolean_t raw = (dsp->dsa_featureflags & DMU_BACKUP_FEATURE_RAW);
struct drr_write *drrw = &(dsp->dsa_drr->drr_u.drr_write);
/*
(object == dsp->dsa_last_data_object &&
offset > dsp->dsa_last_data_offset));
dsp->dsa_last_data_object = object;
- dsp->dsa_last_data_offset = offset + blksz - 1;
+ dsp->dsa_last_data_offset = offset + lsize - 1;
/*
* If there is any kind of pending aggregation (currently either
drrw->drr_object = object;
drrw->drr_type = type;
drrw->drr_offset = offset;
- drrw->drr_length = blksz;
drrw->drr_toguid = dsp->dsa_toguid;
- if (bp == NULL || BP_IS_EMBEDDED(bp)) {
+ drrw->drr_logical_size = lsize;
+
+ /* only set the compression fields if the buf is compressed or raw */
+ if (raw || lsize != psize) {
+ ASSERT(!BP_IS_EMBEDDED(bp));
+ ASSERT3S(psize, >, 0);
+
+ if (raw) {
+ ASSERT(BP_IS_PROTECTED(bp));
+
+ /*
+ * This is a raw protected block so we need to pass
+ * along everything the receiving side will need to
+ * interpret this block, including the byteswap, salt,
+ * IV, and MAC.
+ */
+ if (BP_SHOULD_BYTESWAP(bp))
+ drrw->drr_flags |= DRR_RAW_BYTESWAP;
+ zio_crypt_decode_params_bp(bp, drrw->drr_salt,
+ drrw->drr_iv);
+ zio_crypt_decode_mac_bp(bp, drrw->drr_mac);
+ } else {
+ /* this is a compressed block */
+ ASSERT(dsp->dsa_featureflags &
+ DMU_BACKUP_FEATURE_COMPRESSED);
+ ASSERT(!BP_SHOULD_BYTESWAP(bp));
+ ASSERT(!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)));
+ ASSERT3U(BP_GET_COMPRESS(bp), !=, ZIO_COMPRESS_OFF);
+ ASSERT3S(lsize, >=, psize);
+ }
+
+ /* set fields common to compressed and raw sends */
+ drrw->drr_compressiontype = BP_GET_COMPRESS(bp);
+ drrw->drr_compressed_size = psize;
+ payload_size = drrw->drr_compressed_size;
+ } else {
+ payload_size = drrw->drr_logical_size;
+ }
+
+ if (bp == NULL || BP_IS_EMBEDDED(bp) || (BP_IS_PROTECTED(bp) && !raw)) {
/*
- * There's no pre-computed checksum for partial-block
- * writes or embedded BP's, so (like
- * fletcher4-checkummed blocks) userland will have to
- * compute a dedup-capable checksum itself.
+ * There's no pre-computed checksum for partial-block writes,
+ * embedded BP's, or encrypted BP's that are being sent as
+ * plaintext, so (like fletcher4-checkummed blocks) userland
+ * will have to compute a dedup-capable checksum itself.
*/
drrw->drr_checksumtype = ZIO_CHECKSUM_OFF;
} else {
drrw->drr_checksumtype = BP_GET_CHECKSUM(bp);
- if (zio_checksum_table[drrw->drr_checksumtype].ci_dedup)
- drrw->drr_checksumflags |= DRR_CHECKSUM_DEDUP;
+ if (zio_checksum_table[drrw->drr_checksumtype].ci_flags &
+ ZCHECKSUM_FLAG_DEDUP)
+ drrw->drr_flags |= DRR_CHECKSUM_DEDUP;
DDK_SET_LSIZE(&drrw->drr_key, BP_GET_LSIZE(bp));
DDK_SET_PSIZE(&drrw->drr_key, BP_GET_PSIZE(bp));
DDK_SET_COMPRESS(&drrw->drr_key, BP_GET_COMPRESS(bp));
+ DDK_SET_CRYPT(&drrw->drr_key, BP_IS_PROTECTED(bp));
drrw->drr_key.ddk_cksum = bp->blk_cksum;
}
- if (dump_record(dsp, data, blksz) != 0)
+ if (dump_record(dsp, data, payload_size) != 0)
return (SET_ERROR(EINTR));
return (0);
}
if (dsp->dsa_pending_op != PENDING_NONE) {
if (dump_record(dsp, NULL, 0) != 0)
- return (EINTR);
+ return (SET_ERROR(EINTR));
dsp->dsa_pending_op = PENDING_NONE;
}
decode_embedded_bp_compressed(bp, buf);
if (dump_record(dsp, buf, P2ROUNDUP(drrw->drr_psize, 8)) != 0)
- return (EINTR);
+ return (SET_ERROR(EINTR));
return (0);
}
static int
-dump_spill(dmu_sendarg_t *dsp, uint64_t object, int blksz, void *data)
+dump_spill(dmu_sendarg_t *dsp, const blkptr_t *bp, uint64_t object, void *data)
{
struct drr_spill *drrs = &(dsp->dsa_drr->drr_u.drr_spill);
+ uint64_t blksz = BP_GET_LSIZE(bp);
if (dsp->dsa_pending_op != PENDING_NONE) {
if (dump_record(dsp, NULL, 0) != 0)
drrs->drr_length = blksz;
drrs->drr_toguid = dsp->dsa_toguid;
+ /* handle raw send fields */
+ if (dsp->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) {
+ ASSERT(BP_IS_PROTECTED(bp));
+
+ if (BP_SHOULD_BYTESWAP(bp))
+ drrs->drr_flags |= DRR_RAW_BYTESWAP;
+ drrs->drr_compressiontype = BP_GET_COMPRESS(bp);
+ drrs->drr_compressed_size = BP_GET_PSIZE(bp);
+ zio_crypt_decode_params_bp(bp, drrs->drr_salt, drrs->drr_iv);
+ zio_crypt_decode_mac_bp(bp, drrs->drr_mac);
+ }
+
if (dump_record(dsp, data, blksz) != 0)
return (SET_ERROR(EINTR));
return (0);
dump_freeobjects(dmu_sendarg_t *dsp, uint64_t firstobj, uint64_t numobjs)
{
struct drr_freeobjects *drrfo = &(dsp->dsa_drr->drr_u.drr_freeobjects);
+ uint64_t maxobj = DNODES_PER_BLOCK *
+ (DMU_META_DNODE(dsp->dsa_os)->dn_maxblkid + 1);
- /* See comment in dump_free(). */
- if (!dsp->dsa_incremental)
- return (0);
+ /*
+ * ZoL < 0.7 does not handle large FREEOBJECTS records correctly,
+ * leading to zfs recv never completing. to avoid this issue, don't
+ * send FREEOBJECTS records for object IDs which cannot exist on the
+ * receiving side.
+ */
+ if (maxobj > 0) {
+ if (maxobj < firstobj)
+ return (0);
+
+ if (maxobj < firstobj + numobjs)
+ numobjs = maxobj - firstobj;
+ }
/*
* If there is a pending op, but it's not PENDING_FREEOBJECTS,
}
static int
-dump_dnode(dmu_sendarg_t *dsp, uint64_t object, dnode_phys_t *dnp)
+dump_dnode(dmu_sendarg_t *dsp, const blkptr_t *bp, uint64_t object,
+ dnode_phys_t *dnp)
{
struct drr_object *drro = &(dsp->dsa_drr->drr_u.drr_object);
+ int bonuslen;
+
+ if (object < dsp->dsa_resume_object) {
+ /*
+ * Note: when resuming, we will visit all the dnodes in
+ * the block of dnodes that we are resuming from. In
+ * this case it's unnecessary to send the dnodes prior to
+ * the one we are resuming from. We should be at most one
+ * block's worth of dnodes behind the resume point.
+ */
+ ASSERT3U(dsp->dsa_resume_object - object, <,
+ 1 << (DNODE_BLOCK_SHIFT - DNODE_SHIFT));
+ return (0);
+ }
if (dnp == NULL || dnp->dn_type == DMU_OT_NONE)
return (dump_freeobjects(dsp, object, 1));
drro->drr_bonustype = dnp->dn_bonustype;
drro->drr_blksz = dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT;
drro->drr_bonuslen = dnp->dn_bonuslen;
+ drro->drr_dn_slots = dnp->dn_extra_slots + 1;
drro->drr_checksumtype = dnp->dn_checksum;
drro->drr_compress = dnp->dn_compress;
drro->drr_toguid = dsp->dsa_toguid;
drro->drr_blksz > SPA_OLD_MAXBLOCKSIZE)
drro->drr_blksz = SPA_OLD_MAXBLOCKSIZE;
- if (dump_record(dsp, DN_BONUS(dnp),
- P2ROUNDUP(dnp->dn_bonuslen, 8)) != 0) {
- return (SET_ERROR(EINTR));
+ bonuslen = P2ROUNDUP(dnp->dn_bonuslen, 8);
+
+ if ((dsp->dsa_featureflags & DMU_BACKUP_FEATURE_RAW)) {
+ ASSERT(BP_IS_ENCRYPTED(bp));
+
+ if (BP_SHOULD_BYTESWAP(bp))
+ drro->drr_flags |= DRR_RAW_BYTESWAP;
+
+ /* needed for reconstructing dnp on recv side */
+ drro->drr_maxblkid = dnp->dn_maxblkid;
+ drro->drr_indblkshift = dnp->dn_indblkshift;
+ drro->drr_nlevels = dnp->dn_nlevels;
+ drro->drr_nblkptr = dnp->dn_nblkptr;
+
+ /*
+ * Since we encrypt the entire bonus area, the (raw) part
+ * beyond the bonuslen is actually nonzero, so we need
+ * to send it.
+ */
+ if (bonuslen != 0) {
+ drro->drr_raw_bonuslen = DN_MAX_BONUS_LEN(dnp);
+ bonuslen = drro->drr_raw_bonuslen;
+ }
}
+ if (dump_record(dsp, DN_BONUS(dnp), bonuslen) != 0)
+ return (SET_ERROR(EINTR));
+
/* Free anything past the end of the file. */
if (dump_free(dsp, object, (dnp->dn_maxblkid + 1) *
- (dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT), -1ULL) != 0)
+ (dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT), DMU_OBJECT_END) != 0)
return (SET_ERROR(EINTR));
if (dsp->dsa_err != 0)
return (SET_ERROR(EINTR));
return (0);
}
+static int
+dump_object_range(dmu_sendarg_t *dsp, const blkptr_t *bp, uint64_t firstobj,
+ uint64_t numslots)
+{
+ struct drr_object_range *drror =
+ &(dsp->dsa_drr->drr_u.drr_object_range);
+
+ /* we only use this record type for raw sends */
+ ASSERT(BP_IS_PROTECTED(bp));
+ ASSERT(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_RAW);
+ ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF);
+ ASSERT3U(BP_GET_TYPE(bp), ==, DMU_OT_DNODE);
+ ASSERT0(BP_GET_LEVEL(bp));
+
+ if (dsp->dsa_pending_op != PENDING_NONE) {
+ if (dump_record(dsp, NULL, 0) != 0)
+ return (SET_ERROR(EINTR));
+ dsp->dsa_pending_op = PENDING_NONE;
+ }
+
+ bzero(dsp->dsa_drr, sizeof (dmu_replay_record_t));
+ dsp->dsa_drr->drr_type = DRR_OBJECT_RANGE;
+ drror->drr_firstobj = firstobj;
+ drror->drr_numslots = numslots;
+ drror->drr_toguid = dsp->dsa_toguid;
+ if (BP_SHOULD_BYTESWAP(bp))
+ drror->drr_flags |= DRR_RAW_BYTESWAP;
+ zio_crypt_decode_params_bp(bp, drror->drr_salt, drror->drr_iv);
+ zio_crypt_decode_mac_bp(bp, drror->drr_mac);
+
+ if (dump_record(dsp, NULL, 0) != 0)
+ return (SET_ERROR(EINTR));
+ return (0);
+}
+
static boolean_t
backup_do_embed(dmu_sendarg_t *dsp, const blkptr_t *bp)
{
* Compression function must be legacy, or explicitly enabled.
*/
if ((BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_LEGACY_FUNCTIONS &&
- !(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_EMBED_DATA_LZ4)))
+ !(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_LZ4)))
return (B_FALSE);
/*
uint64_t record_size;
int err = 0;
+ ASSERT(zb->zb_object == DMU_META_DNODE_OBJECT ||
+ zb->zb_object >= sta->resume.zb_object);
+ ASSERT3P(sta->ds, !=, NULL);
+
if (sta->cancel)
return (SET_ERROR(EINTR));
struct send_thread_arg *st_arg = arg;
int err;
struct send_block_record *data;
+ fstrans_cookie_t cookie = spl_fstrans_mark();
if (st_arg->ds != NULL) {
- err = traverse_dataset(st_arg->ds, st_arg->fromtxg,
- st_arg->flags, send_cb, arg);
+ err = traverse_dataset_resume(st_arg->ds,
+ st_arg->fromtxg, &st_arg->resume,
+ st_arg->flags, send_cb, st_arg);
+
if (err != EINTR)
st_arg->error_code = err;
}
data = kmem_zalloc(sizeof (*data), KM_SLEEP);
data->eos_marker = B_TRUE;
bqueue_enqueue(&st_arg->q, data, 1);
+ spl_fstrans_unmark(cookie);
+ thread_exit();
}
/*
spa_t *spa = ds->ds_dir->dd_pool->dp_spa;
dmu_object_type_t type = bp ? BP_GET_TYPE(bp) : DMU_OT_NONE;
int err = 0;
- dnode_phys_t *blk;
- uint64_t dnobj;
ASSERT3U(zb->zb_level, >=, 0);
+ ASSERT(zb->zb_object == DMU_META_DNODE_OBJECT ||
+ zb->zb_object >= dsa->dsa_resume_object);
+
+ /*
+ * All bps of an encrypted os should have the encryption bit set.
+ * If this is not true it indicates tampering and we report an error.
+ */
+ if (dsa->dsa_os->os_encrypted &&
+ !BP_IS_HOLE(bp) && !BP_USES_CRYPT(bp)) {
+ spa_log_error(spa, zb);
+ zfs_panic_recover("unencrypted block in encrypted "
+ "object set %llu", ds->ds_object);
+ return (SET_ERROR(EIO));
+ }
+
if (zb->zb_object != DMU_META_DNODE_OBJECT &&
DMU_OBJECT_IS_SPECIAL(zb->zb_object)) {
return (0);
} else if (BP_IS_HOLE(bp)) {
uint64_t span = BP_SPAN(dblkszsec, indblkshift, zb->zb_level);
uint64_t offset = zb->zb_blkid * span;
- err = dump_free(dsa, zb->zb_object, offset, span);
+ /* Don't dump free records for offsets > DMU_OBJECT_END */
+ if (zb->zb_blkid == 0 || span <= DMU_OBJECT_END / zb->zb_blkid)
+ err = dump_free(dsa, zb->zb_object, offset, span);
} else if (zb->zb_level > 0 || type == DMU_OT_OBJSET) {
return (0);
} else if (type == DMU_OT_DNODE) {
- int blksz = BP_GET_LSIZE(bp);
+ int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT;
arc_flags_t aflags = ARC_FLAG_WAIT;
arc_buf_t *abuf;
- int i;
+ enum zio_flag zioflags = ZIO_FLAG_CANFAIL;
+
+ if (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) {
+ ASSERT(BP_IS_ENCRYPTED(bp));
+ ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF);
+ zioflags |= ZIO_FLAG_RAW;
+ }
ASSERT0(zb->zb_level);
if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf,
- ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL,
- &aflags, zb) != 0)
+ ZIO_PRIORITY_ASYNC_READ, zioflags, &aflags, zb) != 0)
return (SET_ERROR(EIO));
- blk = abuf->b_data;
- dnobj = zb->zb_blkid * (blksz >> DNODE_SHIFT);
- for (i = 0; i < blksz >> DNODE_SHIFT; i++) {
- err = dump_dnode(dsa, dnobj + i, blk + i);
- if (err != 0)
- break;
+ dnode_phys_t *blk = abuf->b_data;
+ uint64_t dnobj = zb->zb_blkid * epb;
+
+ /*
+ * Raw sends require sending encryption parameters for the
+ * block of dnodes. Regular sends do not need to send this
+ * info.
+ */
+ if (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) {
+ ASSERT(arc_is_encrypted(abuf));
+ err = dump_object_range(dsa, bp, dnobj, epb);
}
- (void) arc_buf_remove_ref(abuf, &abuf);
+
+ if (err == 0) {
+ for (int i = 0; i < epb;
+ i += blk[i].dn_extra_slots + 1) {
+ err = dump_dnode(dsa, bp, dnobj + i, blk + i);
+ if (err != 0)
+ break;
+ }
+ }
+ arc_buf_destroy(abuf, &abuf);
} else if (type == DMU_OT_SA) {
arc_flags_t aflags = ARC_FLAG_WAIT;
arc_buf_t *abuf;
- int blksz = BP_GET_LSIZE(bp);
+ enum zio_flag zioflags = ZIO_FLAG_CANFAIL;
+
+ if (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) {
+ ASSERT(BP_IS_PROTECTED(bp));
+ zioflags |= ZIO_FLAG_RAW;
+ }
if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf,
- ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL,
- &aflags, zb) != 0)
+ ZIO_PRIORITY_ASYNC_READ, zioflags, &aflags, zb) != 0)
return (SET_ERROR(EIO));
- err = dump_spill(dsa, zb->zb_object, blksz, abuf->b_data);
- (void) arc_buf_remove_ref(abuf, &abuf);
+ err = dump_spill(dsa, bp, zb->zb_object, abuf->b_data);
+ arc_buf_destroy(abuf, &abuf);
} else if (backup_do_embed(dsa, bp)) {
/* it's an embedded level-0 block of a regular object */
int blksz = dblkszsec << SPA_MINBLOCKSHIFT;
int blksz = dblkszsec << SPA_MINBLOCKSHIFT;
uint64_t offset;
+ /*
+ * If we have large blocks stored on disk but the send flags
+ * don't allow us to send large blocks, we split the data from
+ * the arc buf into chunks.
+ */
+ boolean_t split_large_blocks = blksz > SPA_OLD_MAXBLOCKSIZE &&
+ !(dsa->dsa_featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS);
+
+ /*
+ * Raw sends require that we always get raw data as it exists
+ * on disk, so we assert that we are not splitting blocks here.
+ */
+ boolean_t request_raw =
+ (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_RAW) != 0;
+
+ /*
+ * We should only request compressed data from the ARC if all
+ * the following are true:
+ * - stream compression was requested
+ * - we aren't splitting large blocks into smaller chunks
+ * - the data won't need to be byteswapped before sending
+ * - this isn't an embedded block
+ * - this isn't metadata (if receiving on a different endian
+ * system it can be byteswapped more easily)
+ */
+ boolean_t request_compressed =
+ (dsa->dsa_featureflags & DMU_BACKUP_FEATURE_COMPRESSED) &&
+ !split_large_blocks && !BP_SHOULD_BYTESWAP(bp) &&
+ !BP_IS_EMBEDDED(bp) && !DMU_OT_IS_METADATA(BP_GET_TYPE(bp));
+
+ IMPLY(request_raw, !split_large_blocks);
+ IMPLY(request_raw, BP_IS_PROTECTED(bp));
ASSERT0(zb->zb_level);
+ ASSERT(zb->zb_object > dsa->dsa_resume_object ||
+ (zb->zb_object == dsa->dsa_resume_object &&
+ zb->zb_blkid * blksz >= dsa->dsa_resume_offset));
+
+ ASSERT3U(blksz, ==, BP_GET_LSIZE(bp));
+
+ enum zio_flag zioflags = ZIO_FLAG_CANFAIL;
+ if (request_raw)
+ zioflags |= ZIO_FLAG_RAW;
+ else if (request_compressed)
+ zioflags |= ZIO_FLAG_RAW_COMPRESS;
+
if (arc_read(NULL, spa, bp, arc_getbuf_func, &abuf,
- ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL,
- &aflags, zb) != 0) {
+ ZIO_PRIORITY_ASYNC_READ, zioflags, &aflags, zb) != 0) {
if (zfs_send_corrupt_data) {
- uint64_t *ptr;
/* Send a block filled with 0x"zfs badd bloc" */
- abuf = arc_buf_alloc(spa, blksz, &abuf,
- ARC_BUFC_DATA);
+ abuf = arc_alloc_buf(spa, &abuf, ARC_BUFC_DATA,
+ blksz);
+ uint64_t *ptr;
for (ptr = abuf->b_data;
(char *)ptr < (char *)abuf->b_data + blksz;
ptr++)
offset = zb->zb_blkid * blksz;
- if (!(dsa->dsa_featureflags &
- DMU_BACKUP_FEATURE_LARGE_BLOCKS) &&
- blksz > SPA_OLD_MAXBLOCKSIZE) {
+ if (split_large_blocks) {
+ ASSERT0(arc_is_encrypted(abuf));
+ ASSERT3U(arc_get_compression(abuf), ==,
+ ZIO_COMPRESS_OFF);
char *buf = abuf->b_data;
while (blksz > 0 && err == 0) {
int n = MIN(blksz, SPA_OLD_MAXBLOCKSIZE);
err = dump_write(dsa, type, zb->zb_object,
- offset, n, NULL, buf);
+ offset, n, n, NULL, buf);
offset += n;
buf += n;
blksz -= n;
}
} else {
- err = dump_write(dsa, type, zb->zb_object,
- offset, blksz, bp, abuf->b_data);
+ err = dump_write(dsa, type, zb->zb_object, offset,
+ blksz, arc_buf_size(abuf), bp, abuf->b_data);
}
- (void) arc_buf_remove_ref(abuf, &abuf);
+ arc_buf_destroy(abuf, &abuf);
}
ASSERT(err == 0 || err == EINTR);
*/
static int
dmu_send_impl(void *tag, dsl_pool_t *dp, dsl_dataset_t *to_ds,
- zfs_bookmark_phys_t *ancestor_zb, boolean_t is_clone, boolean_t embedok,
- boolean_t large_block_ok, int outfd, vnode_t *vp, offset_t *off)
+ zfs_bookmark_phys_t *ancestor_zb, boolean_t is_clone,
+ boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
+ boolean_t rawok, int outfd, uint64_t resumeobj, uint64_t resumeoff,
+ vnode_t *vp, offset_t *off)
{
objset_t *os;
dmu_replay_record_t *drr;
uint64_t fromtxg = 0;
uint64_t featureflags = 0;
struct send_thread_arg to_arg;
+ void *payload = NULL;
+ size_t payload_len = 0;
struct send_block_record *to_data;
err = dmu_objset_from_ds(to_ds, &os);
return (err);
}
+ /*
+ * If this is a non-raw send of an encrypted ds, we can ensure that
+ * the objset_phys_t is authenticated. This is safe because this is
+ * either a snapshot or we have owned the dataset, ensuring that
+ * it can't be modified.
+ */
+ if (!rawok && os->os_encrypted &&
+ arc_is_unauthenticated(os->os_phys_buf)) {
+ zbookmark_phys_t zb;
+
+ SET_BOOKMARK(&zb, to_ds->ds_object, ZB_ROOT_OBJECT,
+ ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
+ err = arc_untransform(os->os_phys_buf, os->os_spa,
+ &zb, B_FALSE);
+ if (err != 0) {
+ dsl_pool_rele(dp, tag);
+ return (err);
+ }
+
+ ASSERT0(arc_is_unauthenticated(os->os_phys_buf));
+ }
+
drr = kmem_zalloc(sizeof (dmu_replay_record_t), KM_SLEEP);
drr->drr_type = DRR_BEGIN;
drr->drr_u.drr_begin.drr_magic = DMU_BACKUP_MAGIC;
DMU_SET_STREAM_HDRTYPE(drr->drr_u.drr_begin.drr_versioninfo,
DMU_SUBSTREAM);
+ bzero(&to_arg, sizeof (to_arg));
+
#ifdef _KERNEL
if (dmu_objset_type(os) == DMU_OST_ZFS) {
uint64_t version;
}
#endif
- if (large_block_ok && to_ds->ds_feature_inuse[SPA_FEATURE_LARGE_BLOCKS])
+ /* raw sends imply large_block_ok */
+ if ((large_block_ok || rawok) &&
+ to_ds->ds_feature_inuse[SPA_FEATURE_LARGE_BLOCKS])
featureflags |= DMU_BACKUP_FEATURE_LARGE_BLOCKS;
- if (embedok &&
+ if (to_ds->ds_feature_inuse[SPA_FEATURE_LARGE_DNODE])
+ featureflags |= DMU_BACKUP_FEATURE_LARGE_DNODE;
+
+ /* encrypted datasets will not have embedded blocks */
+ if ((embedok || rawok) && !os->os_encrypted &&
spa_feature_is_active(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA)) {
featureflags |= DMU_BACKUP_FEATURE_EMBED_DATA;
- if (spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS))
- featureflags |= DMU_BACKUP_FEATURE_EMBED_DATA_LZ4;
+ }
+
+ /* raw send implies compressok */
+ if (compressok || rawok)
+ featureflags |= DMU_BACKUP_FEATURE_COMPRESSED;
+ if (rawok && os->os_encrypted)
+ featureflags |= DMU_BACKUP_FEATURE_RAW;
+
+ if ((featureflags &
+ (DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_COMPRESSED |
+ DMU_BACKUP_FEATURE_RAW)) != 0 &&
+ spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) {
+ featureflags |= DMU_BACKUP_FEATURE_LZ4;
+ }
+
+ if (resumeobj != 0 || resumeoff != 0) {
+ featureflags |= DMU_BACKUP_FEATURE_RESUMING;
}
DMU_SET_FEATUREFLAGS(drr->drr_u.drr_begin.drr_versioninfo,
drr->drr_u.drr_begin.drr_toguid = dsl_dataset_phys(to_ds)->ds_guid;
if (dsl_dataset_phys(to_ds)->ds_flags & DS_FLAG_CI_DATASET)
drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_CI_DATA;
+ if (zfs_send_set_freerecords_bit)
+ drr->drr_u.drr_begin.drr_flags |= DRR_FLAG_FREERECORDS;
if (ancestor_zb != NULL) {
drr->drr_u.drr_begin.drr_fromguid =
dsp->dsa_off = off;
dsp->dsa_toguid = dsl_dataset_phys(to_ds)->ds_guid;
dsp->dsa_pending_op = PENDING_NONE;
- dsp->dsa_incremental = (ancestor_zb != NULL);
dsp->dsa_featureflags = featureflags;
+ dsp->dsa_resume_object = resumeobj;
+ dsp->dsa_resume_offset = resumeoff;
mutex_enter(&to_ds->ds_sendstream_lock);
list_insert_head(&to_ds->ds_sendstreams, dsp);
dsl_dataset_long_hold(to_ds, FTAG);
dsl_pool_rele(dp, tag);
- if (dump_record(dsp, NULL, 0) != 0) {
+ /* handle features that require a DRR_BEGIN payload */
+ if (featureflags &
+ (DMU_BACKUP_FEATURE_RESUMING | DMU_BACKUP_FEATURE_RAW)) {
+ nvlist_t *keynvl = NULL;
+ nvlist_t *nvl = fnvlist_alloc();
+
+ if (featureflags & DMU_BACKUP_FEATURE_RESUMING) {
+ dmu_object_info_t to_doi;
+ err = dmu_object_info(os, resumeobj, &to_doi);
+ if (err != 0) {
+ fnvlist_free(nvl);
+ goto out;
+ }
+
+ SET_BOOKMARK(&to_arg.resume, to_ds->ds_object,
+ resumeobj, 0,
+ resumeoff / to_doi.doi_data_block_size);
+
+ fnvlist_add_uint64(nvl, "resume_object", resumeobj);
+ fnvlist_add_uint64(nvl, "resume_offset", resumeoff);
+ }
+
+ if (featureflags & DMU_BACKUP_FEATURE_RAW) {
+ ASSERT(os->os_encrypted);
+
+ err = dsl_crypto_populate_key_nvlist(to_ds, &keynvl);
+ if (err != 0) {
+ fnvlist_free(nvl);
+ goto out;
+ }
+
+ fnvlist_add_nvlist(nvl, "crypt_keydata", keynvl);
+ }
+
+ payload = fnvlist_pack(nvl, &payload_len);
+ drr->drr_payloadlen = payload_len;
+ fnvlist_free(keynvl);
+ fnvlist_free(nvl);
+ }
+
+ err = dump_record(dsp, payload, payload_len);
+ fnvlist_pack_free(payload, payload_len);
+ if (err != 0) {
err = dsp->dsa_err;
goto out;
}
- err = bqueue_init(&to_arg.q, zfs_send_queue_length,
+ err = bqueue_init(&to_arg.q,
+ MAX(zfs_send_queue_length, 2 * zfs_max_recordsize),
offsetof(struct send_block_record, ln));
to_arg.error_code = 0;
to_arg.cancel = B_FALSE;
to_arg.ds = to_ds;
to_arg.fromtxg = fromtxg;
to_arg.flags = TRAVERSE_PRE | TRAVERSE_PREFETCH;
+ if (rawok)
+ to_arg.flags |= TRAVERSE_NO_DECRYPT;
(void) thread_create(NULL, 0, send_traverse_thread, &to_arg, 0, curproc,
TS_RUN, minclsyspri);
if (dump_record(dsp, NULL, 0) != 0)
err = dsp->dsa_err;
-
out:
mutex_enter(&to_ds->ds_sendstream_lock);
list_remove(&to_ds->ds_sendstreams, dsp);
mutex_exit(&to_ds->ds_sendstream_lock);
+ VERIFY(err != 0 || (dsp->dsa_sent_begin && dsp->dsa_sent_end));
+
kmem_free(drr, sizeof (dmu_replay_record_t));
kmem_free(dsp, sizeof (dmu_sendarg_t));
int
dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
- boolean_t embedok, boolean_t large_block_ok,
- int outfd, vnode_t *vp, offset_t *off)
+ boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
+ boolean_t rawok, int outfd, vnode_t *vp, offset_t *off)
{
dsl_pool_t *dp;
dsl_dataset_t *ds;
dsl_dataset_t *fromds = NULL;
+ ds_hold_flags_t dsflags = (rawok) ? 0 : DS_HOLD_FLAG_DECRYPT;
int err;
err = dsl_pool_hold(pool, FTAG, &dp);
if (err != 0)
return (err);
- err = dsl_dataset_hold_obj(dp, tosnap, FTAG, &ds);
+ err = dsl_dataset_hold_obj_flags(dp, tosnap, dsflags, FTAG, &ds);
if (err != 0) {
dsl_pool_rele(dp, FTAG);
return (err);
err = dsl_dataset_hold_obj(dp, fromsnap, FTAG, &fromds);
if (err != 0) {
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
dsl_pool_rele(dp, FTAG);
return (err);
}
is_clone = (fromds->ds_dir != ds->ds_dir);
dsl_dataset_rele(fromds, FTAG);
err = dmu_send_impl(FTAG, dp, ds, &zb, is_clone,
- embedok, large_block_ok, outfd, vp, off);
+ embedok, large_block_ok, compressok, rawok, outfd,
+ 0, 0, vp, off);
} else {
err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE,
- embedok, large_block_ok, outfd, vp, off);
+ embedok, large_block_ok, compressok, rawok, outfd,
+ 0, 0, vp, off);
}
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (err);
}
int
-dmu_send(const char *tosnap, const char *fromsnap,
- boolean_t embedok, boolean_t large_block_ok,
- int outfd, vnode_t *vp, offset_t *off)
+dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok,
+ boolean_t large_block_ok, boolean_t compressok, boolean_t rawok,
+ int outfd, uint64_t resumeobj, uint64_t resumeoff, vnode_t *vp,
+ offset_t *off)
{
dsl_pool_t *dp;
dsl_dataset_t *ds;
int err;
+ ds_hold_flags_t dsflags = (rawok) ? 0 : DS_HOLD_FLAG_DECRYPT;
boolean_t owned = B_FALSE;
if (fromsnap != NULL && strpbrk(fromsnap, "@#") == NULL)
* We are sending a filesystem or volume. Ensure
* that it doesn't change by owning the dataset.
*/
- err = dsl_dataset_own(dp, tosnap, FTAG, &ds);
+ err = dsl_dataset_own(dp, tosnap, dsflags, FTAG, &ds);
owned = B_TRUE;
} else {
- err = dsl_dataset_hold(dp, tosnap, FTAG, &ds);
+ err = dsl_dataset_hold_flags(dp, tosnap, dsflags, FTAG, &ds);
}
if (err != 0) {
dsl_pool_rele(dp, FTAG);
err = dsl_bookmark_lookup(dp, fromsnap, ds, &zb);
}
if (err != 0) {
- dsl_dataset_rele(ds, FTAG);
+ if (owned)
+ dsl_dataset_disown(ds, dsflags, FTAG);
+ else
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+
dsl_pool_rele(dp, FTAG);
return (err);
}
err = dmu_send_impl(FTAG, dp, ds, &zb, is_clone,
- embedok, large_block_ok, outfd, vp, off);
+ embedok, large_block_ok, compressok, rawok,
+ outfd, resumeobj, resumeoff, vp, off);
} else {
err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE,
- embedok, large_block_ok, outfd, vp, off);
+ embedok, large_block_ok, compressok, rawok,
+ outfd, resumeobj, resumeoff, vp, off);
}
if (owned)
- dsl_dataset_disown(ds, FTAG);
+ dsl_dataset_disown(ds, dsflags, FTAG);
else
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+
return (err);
}
static int
-dmu_adjust_send_estimate_for_indirects(dsl_dataset_t *ds, uint64_t size,
- uint64_t *sizep)
+dmu_adjust_send_estimate_for_indirects(dsl_dataset_t *ds, uint64_t uncompressed,
+ uint64_t compressed, boolean_t stream_compressed, uint64_t *sizep)
{
int err;
+ uint64_t size;
/*
* Assume that space (both on-disk and in-stream) is dominated by
* data. We will adjust for indirect blocks and the copies property,
* but ignore per-object space used (eg, dnodes and DRR_OBJECT records).
*/
+ uint64_t recordsize;
+ uint64_t record_count;
+ objset_t *os;
+ VERIFY0(dmu_objset_from_ds(ds, &os));
+
+ /* Assume all (uncompressed) blocks are recordsize. */
+ if (os->os_phys->os_type == DMU_OST_ZVOL) {
+ err = dsl_prop_get_int_ds(ds,
+ zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &recordsize);
+ } else {
+ err = dsl_prop_get_int_ds(ds,
+ zfs_prop_to_name(ZFS_PROP_RECORDSIZE), &recordsize);
+ }
+ if (err != 0)
+ return (err);
+ record_count = uncompressed / recordsize;
+
+ /*
+ * If we're estimating a send size for a compressed stream, use the
+ * compressed data size to estimate the stream size. Otherwise, use the
+ * uncompressed data size.
+ */
+ size = stream_compressed ? compressed : uncompressed;
+
/*
* Subtract out approximate space used by indirect blocks.
* Assume most space is used by data blocks (non-indirect, non-dnode).
- * Assume all blocks are recordsize. Assume ditto blocks and
- * internal fragmentation counter out compression.
+ * Assume no ditto blocks or internal fragmentation.
*
* Therefore, space used by indirect blocks is sizeof(blkptr_t) per
- * block, which we observe in practice.
+ * block.
*/
- uint64_t recordsize;
- err = dsl_prop_get_int_ds(ds, "recordsize", &recordsize);
- if (err != 0)
- return (err);
- size -= size / recordsize * sizeof (blkptr_t);
+ size -= record_count * sizeof (blkptr_t);
/* Add in the space for the record associated with each block. */
- size += size / recordsize * sizeof (dmu_replay_record_t);
+ size += record_count * sizeof (dmu_replay_record_t);
*sizep = size;
}
int
-dmu_send_estimate(dsl_dataset_t *ds, dsl_dataset_t *fromds, uint64_t *sizep)
+dmu_send_estimate(dsl_dataset_t *ds, dsl_dataset_t *fromds,
+ boolean_t stream_compressed, uint64_t *sizep)
{
int err;
- uint64_t size;
+ uint64_t uncomp, comp;
ASSERT(dsl_pool_config_held(ds->ds_dir->dd_pool));
if (fromds != NULL && !dsl_dataset_is_before(ds, fromds, 0))
return (SET_ERROR(EXDEV));
- /* Get uncompressed size estimate of changed data. */
+ /* Get compressed and uncompressed size estimates of changed data. */
if (fromds == NULL) {
- size = dsl_dataset_phys(ds)->ds_uncompressed_bytes;
+ uncomp = dsl_dataset_phys(ds)->ds_uncompressed_bytes;
+ comp = dsl_dataset_phys(ds)->ds_compressed_bytes;
} else {
- uint64_t used, comp;
+ uint64_t used;
err = dsl_dataset_space_written(fromds, ds,
- &used, &comp, &size);
+ &used, &comp, &uncomp);
if (err != 0)
return (err);
}
- err = dmu_adjust_send_estimate_for_indirects(ds, size, sizep);
+ err = dmu_adjust_send_estimate_for_indirects(ds, uncomp, comp,
+ stream_compressed, sizep);
+ /*
+ * Add the size of the BEGIN and END records to the estimate.
+ */
+ *sizep += 2 * sizeof (dmu_replay_record_t);
return (err);
}
+struct calculate_send_arg {
+ uint64_t uncompressed;
+ uint64_t compressed;
+};
+
/*
* Simple callback used to traverse the blocks of a snapshot and sum their
- * uncompressed size
+ * uncompressed and compressed sizes.
*/
/* ARGSUSED */
static int
dmu_calculate_send_traversal(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
const zbookmark_phys_t *zb, const dnode_phys_t *dnp, void *arg)
{
- uint64_t *spaceptr = arg;
+ struct calculate_send_arg *space = arg;
if (bp != NULL && !BP_IS_HOLE(bp)) {
- *spaceptr += BP_GET_UCSIZE(bp);
+ space->uncompressed += BP_GET_UCSIZE(bp);
+ space->compressed += BP_GET_PSIZE(bp);
}
return (0);
}
*/
int
dmu_send_estimate_from_txg(dsl_dataset_t *ds, uint64_t from_txg,
- uint64_t *sizep)
+ boolean_t stream_compressed, uint64_t *sizep)
{
int err;
- uint64_t size = 0;
+ struct calculate_send_arg size = { 0 };
ASSERT(dsl_pool_config_held(ds->ds_dir->dd_pool));
* traverse the blocks of the snapshot with birth times after
* from_txg, summing their uncompressed size
*/
- err = traverse_dataset(ds, from_txg, TRAVERSE_POST,
+ err = traverse_dataset(ds, from_txg,
+ TRAVERSE_POST | TRAVERSE_NO_DECRYPT,
dmu_calculate_send_traversal, &size);
+
if (err)
return (err);
- err = dmu_adjust_send_estimate_for_indirects(ds, size, sizep);
+ err = dmu_adjust_send_estimate_for_indirects(ds, size.uncompressed,
+ size.compressed, stream_compressed, sizep);
return (err);
}
uint64_t val;
int error;
dsl_pool_t *dp = ds->ds_dir->dd_pool;
+ struct drr_begin *drrb = drba->drba_cookie->drc_drrb;
+ uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo);
+ boolean_t encrypted = ds->ds_dir->dd_crypto_obj != 0;
+ boolean_t raw = (featureflags & DMU_BACKUP_FEATURE_RAW) != 0;
/* temporary clone name must not exist */
error = zap_lookup(dp->dp_meta_objset,
dsl_dataset_t *snap;
uint64_t obj = dsl_dataset_phys(ds)->ds_prev_snap_obj;
+ /* Can't perform a raw receive on top of a non-raw receive */
+ if (!encrypted && raw)
+ return (SET_ERROR(EINVAL));
+
/* Find snapshot in this dir that matches fromguid. */
while (obj != 0) {
error = dsl_dataset_hold_obj(dp, obj, FTAG,
/* if full, then must be forced */
if (!drba->drba_cookie->drc_force)
return (SET_ERROR(EEXIST));
- /* start from $ORIGIN@$ORIGIN, if supported */
- drba->drba_snapobj = dp->dp_origin_snap != NULL ?
- dp->dp_origin_snap->ds_object : 0;
+
+ /*
+ * We don't support using zfs recv -F to blow away
+ * encrypted filesystems. This would require the
+ * dsl dir to point to the old encryption key and
+ * the new one at the same time during the receive.
+ */
+ if ((!encrypted && raw) || encrypted)
+ return (SET_ERROR(EINVAL));
+
+ drba->drba_snapobj = 0;
}
return (0);
struct drr_begin *drrb = drba->drba_cookie->drc_drrb;
uint64_t fromguid = drrb->drr_fromguid;
int flags = drrb->drr_flags;
+ ds_hold_flags_t dsflags = 0;
int error;
uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo);
dsl_dataset_t *ds;
/* already checked */
ASSERT3U(drrb->drr_magic, ==, DMU_BACKUP_MAGIC);
+ ASSERT(!(featureflags & DMU_BACKUP_FEATURE_RESUMING));
if (DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo) ==
DMU_COMPOUNDSTREAM ||
spa_version(dp->dp_spa) < SPA_VERSION_SA)
return (SET_ERROR(ENOTSUP));
+ if (drba->drba_cookie->drc_resumable &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EXTENSIBLE_DATASET))
+ return (SET_ERROR(ENOTSUP));
+
/*
* The receiving code doesn't know how to translate a WRITE_EMBEDDED
- * record to a plan WRITE record, so the pool must have the
+ * record to a plain WRITE record, so the pool must have the
* EMBEDDED_DATA feature enabled if the stream has WRITE_EMBEDDED
* records. Same with WRITE_EMBEDDED records that use LZ4 compression.
*/
if ((featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) &&
!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA))
return (SET_ERROR(ENOTSUP));
- if ((featureflags & DMU_BACKUP_FEATURE_EMBED_DATA_LZ4) &&
+ if ((featureflags & DMU_BACKUP_FEATURE_LZ4) &&
!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS))
return (SET_ERROR(ENOTSUP));
/*
* The receiving code doesn't know how to translate large blocks
* to smaller ones, so the pool must have the LARGE_BLOCKS
- * feature enabled if the stream has LARGE_BLOCKS.
+ * feature enabled if the stream has LARGE_BLOCKS. Same with
+ * large dnodes.
*/
if ((featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) &&
!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_BLOCKS))
return (SET_ERROR(ENOTSUP));
+ if ((featureflags & DMU_BACKUP_FEATURE_LARGE_DNODE) &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_DNODE))
+ return (SET_ERROR(ENOTSUP));
+
+ if ((featureflags & DMU_BACKUP_FEATURE_RAW)) {
+ /* raw receives require the encryption feature */
+ if (!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_ENCRYPTION))
+ return (SET_ERROR(ENOTSUP));
+ } else {
+ dsflags |= DS_HOLD_FLAG_DECRYPT;
+ }
- error = dsl_dataset_hold(dp, tofs, FTAG, &ds);
+ error = dsl_dataset_hold_flags(dp, tofs, dsflags, FTAG, &ds);
if (error == 0) {
/* target fs already exists; recv into temp clone */
/* Can't recv a clone into an existing fs */
- if (flags & DRR_FLAG_CLONE) {
- dsl_dataset_rele(ds, FTAG);
+ if (flags & DRR_FLAG_CLONE || drba->drba_origin) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (SET_ERROR(EINVAL));
}
error = recv_begin_check_existing_impl(drba, ds, fromguid);
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
} else if (error == ENOENT) {
/* target fs does not exist; must be a full backup or clone */
- char buf[MAXNAMELEN];
+ char buf[ZFS_MAX_DATASET_NAME_LEN];
/*
* If it's a non-clone incremental, we are missing the
drba->drba_origin))
return (SET_ERROR(ENOENT));
+ /*
+ * If we're receiving a full send as a clone, and it doesn't
+ * contain all the necessary free records and freeobject
+ * records, reject it.
+ */
+ if (fromguid == 0 && drba->drba_origin &&
+ !(flags & DRR_FLAG_FREERECORDS))
+ return (SET_ERROR(EINVAL));
+
/* Open the parent of tofs */
- ASSERT3U(strlen(tofs), <, MAXNAMELEN);
+ ASSERT3U(strlen(tofs), <, sizeof (buf));
(void) strlcpy(buf, tofs, strrchr(tofs, '/') - tofs + 1);
- error = dsl_dataset_hold(dp, buf, FTAG, &ds);
+ error = dsl_dataset_hold_flags(dp, buf, dsflags, FTAG, &ds);
if (error != 0)
return (error);
error = dsl_fs_ss_limit_check(ds->ds_dir, 1,
ZFS_PROP_FILESYSTEM_LIMIT, NULL, drba->drba_cred);
if (error != 0) {
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (error);
}
error = dsl_fs_ss_limit_check(ds->ds_dir, 1,
ZFS_PROP_SNAPSHOT_LIMIT, NULL, drba->drba_cred);
if (error != 0) {
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (error);
}
if (drba->drba_origin != NULL) {
dsl_dataset_t *origin;
- error = dsl_dataset_hold(dp, drba->drba_origin,
- FTAG, &origin);
+
+ error = dsl_dataset_hold_flags(dp, drba->drba_origin,
+ dsflags, FTAG, &origin);
if (error != 0) {
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (error);
}
if (!origin->ds_is_snapshot) {
- dsl_dataset_rele(origin, FTAG);
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(origin, dsflags, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (SET_ERROR(EINVAL));
}
- if (dsl_dataset_phys(origin)->ds_guid != fromguid) {
- dsl_dataset_rele(origin, FTAG);
- dsl_dataset_rele(ds, FTAG);
+ if (dsl_dataset_phys(origin)->ds_guid != fromguid &&
+ fromguid != 0) {
+ dsl_dataset_rele_flags(origin, dsflags, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
return (SET_ERROR(ENODEV));
}
- dsl_dataset_rele(origin, FTAG);
+ dsl_dataset_rele_flags(origin,
+ dsflags, FTAG);
}
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
error = 0;
}
return (error);
{
dmu_recv_begin_arg_t *drba = arg;
dsl_pool_t *dp = dmu_tx_pool(tx);
+ objset_t *mos = dp->dp_meta_objset;
struct drr_begin *drrb = drba->drba_cookie->drc_drrb;
const char *tofs = drba->drba_cookie->drc_tofs;
+ uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo);
dsl_dataset_t *ds, *newds;
+ objset_t *os;
uint64_t dsobj;
+ ds_hold_flags_t dsflags = 0;
int error;
- uint64_t crflags;
-
- crflags = (drrb->drr_flags & DRR_FLAG_CI_DATA) ?
- DS_FLAG_CI_DATASET : 0;
+ uint64_t crflags = 0;
+ dsl_crypto_params_t *dcpp = NULL;
+ dsl_crypto_params_t dcp = { 0 };
+
+ if (drrb->drr_flags & DRR_FLAG_CI_DATA)
+ crflags |= DS_FLAG_CI_DATASET;
+ if ((featureflags & DMU_BACKUP_FEATURE_RAW) == 0) {
+ dsflags |= DS_HOLD_FLAG_DECRYPT;
+ } else {
+ dcp.cp_cmd = DCP_CMD_RAW_RECV;
+ }
- error = dsl_dataset_hold(dp, tofs, FTAG, &ds);
+ error = dsl_dataset_hold_flags(dp, tofs, dsflags, FTAG, &ds);
if (error == 0) {
/* create temporary clone */
dsl_dataset_t *snap = NULL;
+
if (drba->drba_snapobj != 0) {
VERIFY0(dsl_dataset_hold_obj(dp,
drba->drba_snapobj, FTAG, &snap));
+ } else {
+ /* we use the dcp whenever we are not making a clone */
+ dcpp = &dcp;
}
+
dsobj = dsl_dataset_create_sync(ds->ds_dir, recv_clone_name,
- snap, crflags, drba->drba_cred, tx);
+ snap, crflags, drba->drba_cred, dcpp, tx);
if (drba->drba_snapobj != 0)
dsl_dataset_rele(snap, FTAG);
- dsl_dataset_rele(ds, FTAG);
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
} else {
dsl_dir_t *dd;
const char *tail;
if (drba->drba_origin != NULL) {
VERIFY0(dsl_dataset_hold(dp, drba->drba_origin,
FTAG, &origin));
+ } else {
+ /* we use the dcp whenever we are not making a clone */
+ dcpp = &dcp;
}
/* Create new dataset. */
- dsobj = dsl_dataset_create_sync(dd,
- strrchr(tofs, '/') + 1,
- origin, crflags, drba->drba_cred, tx);
+ dsobj = dsl_dataset_create_sync(dd, strrchr(tofs, '/') + 1,
+ origin, crflags, drba->drba_cred, dcpp, tx);
if (origin != NULL)
dsl_dataset_rele(origin, FTAG);
dsl_dir_rele(dd, FTAG);
drba->drba_cookie->drc_newfs = B_TRUE;
}
- VERIFY0(dsl_dataset_own_obj(dp, dsobj, dmu_recv_tag, &newds));
+ VERIFY0(dsl_dataset_own_obj(dp, dsobj, dsflags, dmu_recv_tag, &newds));
+ VERIFY0(dmu_objset_from_ds(newds, &os));
+
+ if (drba->drba_cookie->drc_resumable) {
+ dsl_dataset_zapify(newds, tx);
+ if (drrb->drr_fromguid != 0) {
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_FROMGUID,
+ 8, 1, &drrb->drr_fromguid, tx));
+ }
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_TOGUID,
+ 8, 1, &drrb->drr_toguid, tx));
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_TONAME,
+ 1, strlen(drrb->drr_toname) + 1, drrb->drr_toname, tx));
+ uint64_t one = 1;
+ uint64_t zero = 0;
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_OBJECT,
+ 8, 1, &one, tx));
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_OFFSET,
+ 8, 1, &zero, tx));
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_BYTES,
+ 8, 1, &zero, tx));
+ if (featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) {
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_LARGEBLOCK,
+ 8, 1, &one, tx));
+ }
+ if (featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) {
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_EMBEDOK,
+ 8, 1, &one, tx));
+ }
+ if (featureflags & DMU_BACKUP_FEATURE_COMPRESSED) {
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_COMPRESSOK,
+ 8, 1, &one, tx));
+ }
+ if (featureflags & DMU_BACKUP_FEATURE_RAW) {
+ VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_RAWOK,
+ 8, 1, &one, tx));
+ }
+ }
+
+ /*
+ * Usually the os->os_encrypted value is tied to the presence of a
+ * DSL Crypto Key object in the dd. However, that will not be received
+ * until dmu_recv_stream(), so we set the value manually for now.
+ */
+ if (featureflags & DMU_BACKUP_FEATURE_RAW) {
+ os->os_encrypted = B_TRUE;
+ drba->drba_cookie->drc_raw = B_TRUE;
+ }
dmu_buf_will_dirty(newds->ds_dbuf, tx);
dsl_dataset_phys(newds)->ds_flags |= DS_FLAG_INCONSISTENT;
/*
- * If we actually created a non-clone, we need to create the
- * objset in our new dataset.
+ * If we actually created a non-clone, we need to create the objset
+ * in our new dataset. If this is a raw send we postpone this until
+ * dmu_recv_stream() so that we can allocate the metadnode with the
+ * properties from the DRR_BEGIN payload.
*/
- if (BP_IS_HOLE(dsl_dataset_get_blkptr(newds))) {
+ rrw_enter(&newds->ds_bp_rwlock, RW_READER, FTAG);
+ if (BP_IS_HOLE(dsl_dataset_get_blkptr(newds)) &&
+ (featureflags & DMU_BACKUP_FEATURE_RAW) == 0) {
(void) dmu_objset_create_impl(dp->dp_spa,
newds, dsl_dataset_get_blkptr(newds), drrb->drr_type, tx);
}
+ rrw_exit(&newds->ds_bp_rwlock, FTAG);
drba->drba_cookie->drc_ds = newds;
spa_history_log_internal_ds(newds, "receive", tx, "");
}
-/*
- * NB: callers *MUST* call dmu_recv_stream() if dmu_recv_begin()
- * succeeds; otherwise we will leak the holds on the datasets.
- */
-int
-dmu_recv_begin(char *tofs, char *tosnap, struct drr_begin *drrb,
- boolean_t force, char *origin, dmu_recv_cookie_t *drc)
+static int
+dmu_recv_resume_begin_check(void *arg, dmu_tx_t *tx)
{
- dmu_recv_begin_arg_t drba = { 0 };
- dmu_replay_record_t *drr;
+ dmu_recv_begin_arg_t *drba = arg;
+ dsl_pool_t *dp = dmu_tx_pool(tx);
+ struct drr_begin *drrb = drba->drba_cookie->drc_drrb;
+ int error;
+ ds_hold_flags_t dsflags = 0;
+ uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo);
+ dsl_dataset_t *ds;
+ const char *tofs = drba->drba_cookie->drc_tofs;
- bzero(drc, sizeof (dmu_recv_cookie_t));
- drc->drc_drrb = drrb;
- drc->drc_tosnap = tosnap;
- drc->drc_tofs = tofs;
- drc->drc_force = force;
- drc->drc_cred = CRED();
+ /* already checked */
+ ASSERT3U(drrb->drr_magic, ==, DMU_BACKUP_MAGIC);
+ ASSERT(featureflags & DMU_BACKUP_FEATURE_RESUMING);
- if (drrb->drr_magic == BSWAP_64(DMU_BACKUP_MAGIC))
- drc->drc_byteswap = B_TRUE;
- else if (drrb->drr_magic != DMU_BACKUP_MAGIC)
+ if (DMU_GET_STREAM_HDRTYPE(drrb->drr_versioninfo) ==
+ DMU_COMPOUNDSTREAM ||
+ drrb->drr_type >= DMU_OST_NUMTYPES)
return (SET_ERROR(EINVAL));
- drr = kmem_zalloc(sizeof (dmu_replay_record_t), KM_SLEEP);
- drr->drr_type = DRR_BEGIN;
- drr->drr_u.drr_begin = *drc->drc_drrb;
- if (drc->drc_byteswap) {
- fletcher_4_incremental_byteswap(drr,
- sizeof (dmu_replay_record_t), &drc->drc_cksum);
- } else {
- fletcher_4_incremental_native(drr,
- sizeof (dmu_replay_record_t), &drc->drc_cksum);
- }
- kmem_free(drr, sizeof (dmu_replay_record_t));
-
- if (drc->drc_byteswap) {
- drrb->drr_magic = BSWAP_64(drrb->drr_magic);
- drrb->drr_versioninfo = BSWAP_64(drrb->drr_versioninfo);
- drrb->drr_creation_time = BSWAP_64(drrb->drr_creation_time);
- drrb->drr_type = BSWAP_32(drrb->drr_type);
- drrb->drr_toguid = BSWAP_64(drrb->drr_toguid);
- drrb->drr_fromguid = BSWAP_64(drrb->drr_fromguid);
- }
-
- drba.drba_origin = origin;
- drba.drba_cookie = drc;
- drba.drba_cred = CRED();
-
- return (dsl_sync_task(tofs, dmu_recv_begin_check, dmu_recv_begin_sync,
- &drba, 5, ZFS_SPACE_CHECK_NORMAL));
-}
+ /* Verify pool version supports SA if SA_SPILL feature set */
+ if ((featureflags & DMU_BACKUP_FEATURE_SA_SPILL) &&
+ spa_version(dp->dp_spa) < SPA_VERSION_SA)
+ return (SET_ERROR(ENOTSUP));
-struct receive_record_arg {
- dmu_replay_record_t header;
- void *payload; /* Pointer to a buffer containing the payload */
/*
- * If the record is a write, pointer to the arc_buf_t containing the
- * payload.
+ * The receiving code doesn't know how to translate a WRITE_EMBEDDED
+ * record to a plain WRITE record, so the pool must have the
+ * EMBEDDED_DATA feature enabled if the stream has WRITE_EMBEDDED
+ * records. Same with WRITE_EMBEDDED records that use LZ4 compression.
*/
- arc_buf_t *write_buf;
- int payload_size;
+ if ((featureflags & DMU_BACKUP_FEATURE_EMBED_DATA) &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA))
+ return (SET_ERROR(ENOTSUP));
+ if ((featureflags & DMU_BACKUP_FEATURE_LZ4) &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS))
+ return (SET_ERROR(ENOTSUP));
+
+ /*
+ * The receiving code doesn't know how to translate large blocks
+ * to smaller ones, so the pool must have the LARGE_BLOCKS
+ * feature enabled if the stream has LARGE_BLOCKS. Same with
+ * large dnodes.
+ */
+ if ((featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_BLOCKS))
+ return (SET_ERROR(ENOTSUP));
+ if ((featureflags & DMU_BACKUP_FEATURE_LARGE_DNODE) &&
+ !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_LARGE_DNODE))
+ return (SET_ERROR(ENOTSUP));
+
+ /* 6 extra bytes for /%recv */
+ char recvname[ZFS_MAX_DATASET_NAME_LEN + 6];
+ (void) snprintf(recvname, sizeof (recvname), "%s/%s",
+ tofs, recv_clone_name);
+
+ if ((featureflags & DMU_BACKUP_FEATURE_RAW) == 0)
+ dsflags |= DS_HOLD_FLAG_DECRYPT;
+
+ if (dsl_dataset_hold_flags(dp, recvname, dsflags, FTAG, &ds) != 0) {
+ /* %recv does not exist; continue in tofs */
+ error = dsl_dataset_hold_flags(dp, tofs, dsflags, FTAG, &ds);
+ if (error != 0)
+ return (error);
+ }
+
+ /* check that ds is marked inconsistent */
+ if (!DS_IS_INCONSISTENT(ds)) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ /* check that there is resuming data, and that the toguid matches */
+ if (!dsl_dataset_is_zapified(ds)) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+ uint64_t val;
+ error = zap_lookup(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_TOGUID, sizeof (val), 1, &val);
+ if (error != 0 || drrb->drr_toguid != val) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ /*
+ * Check if the receive is still running. If so, it will be owned.
+ * Note that nothing else can own the dataset (e.g. after the receive
+ * fails) because it will be marked inconsistent.
+ */
+ if (dsl_dataset_has_owner(ds)) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EBUSY));
+ }
+
+ /* There should not be any snapshots of this fs yet. */
+ if (ds->ds_prev != NULL && ds->ds_prev->ds_dir == ds->ds_dir) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ /*
+ * Note: resume point will be checked when we process the first WRITE
+ * record.
+ */
+
+ /* check that the origin matches */
+ val = 0;
+ (void) zap_lookup(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_FROMGUID, sizeof (val), 1, &val);
+ if (drrb->drr_fromguid != val) {
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (SET_ERROR(EINVAL));
+ }
+
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+ return (0);
+}
+
+static void
+dmu_recv_resume_begin_sync(void *arg, dmu_tx_t *tx)
+{
+ dmu_recv_begin_arg_t *drba = arg;
+ dsl_pool_t *dp = dmu_tx_pool(tx);
+ const char *tofs = drba->drba_cookie->drc_tofs;
+ struct drr_begin *drrb = drba->drba_cookie->drc_drrb;
+ uint64_t featureflags = DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo);
+ dsl_dataset_t *ds;
+ objset_t *os;
+ ds_hold_flags_t dsflags = 0;
+ uint64_t dsobj;
+ /* 6 extra bytes for /%recv */
+ char recvname[ZFS_MAX_DATASET_NAME_LEN + 6];
+
+ (void) snprintf(recvname, sizeof (recvname), "%s/%s",
+ tofs, recv_clone_name);
+
+ if (featureflags & DMU_BACKUP_FEATURE_RAW) {
+ drba->drba_cookie->drc_raw = B_TRUE;
+ } else {
+ dsflags |= DS_HOLD_FLAG_DECRYPT;
+ }
+
+ if (dsl_dataset_hold_flags(dp, recvname, dsflags, FTAG, &ds) != 0) {
+ /* %recv does not exist; continue in tofs */
+ VERIFY0(dsl_dataset_hold_flags(dp, tofs, dsflags, FTAG, &ds));
+ drba->drba_cookie->drc_newfs = B_TRUE;
+ }
+
+ /* clear the inconsistent flag so that we can own it */
+ ASSERT(DS_IS_INCONSISTENT(ds));
+ dmu_buf_will_dirty(ds->ds_dbuf, tx);
+ dsl_dataset_phys(ds)->ds_flags &= ~DS_FLAG_INCONSISTENT;
+ dsobj = ds->ds_object;
+ dsl_dataset_rele_flags(ds, dsflags, FTAG);
+
+ VERIFY0(dsl_dataset_own_obj(dp, dsobj, dsflags, dmu_recv_tag, &ds));
+ VERIFY0(dmu_objset_from_ds(ds, &os));
+
+ dmu_buf_will_dirty(ds->ds_dbuf, tx);
+ dsl_dataset_phys(ds)->ds_flags |= DS_FLAG_INCONSISTENT;
+
+ rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
+ ASSERT(!BP_IS_HOLE(dsl_dataset_get_blkptr(ds)) ||
+ drba->drba_cookie->drc_raw);
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
+
+ drba->drba_cookie->drc_ds = ds;
+
+ spa_history_log_internal_ds(ds, "resume receive", tx, "");
+}
+
+/*
+ * NB: callers *MUST* call dmu_recv_stream() if dmu_recv_begin()
+ * succeeds; otherwise we will leak the holds on the datasets.
+ */
+int
+dmu_recv_begin(char *tofs, char *tosnap, dmu_replay_record_t *drr_begin,
+ boolean_t force, boolean_t resumable, char *origin, dmu_recv_cookie_t *drc)
+{
+ dmu_recv_begin_arg_t drba = { 0 };
+
+ bzero(drc, sizeof (dmu_recv_cookie_t));
+ drc->drc_drr_begin = drr_begin;
+ drc->drc_drrb = &drr_begin->drr_u.drr_begin;
+ drc->drc_tosnap = tosnap;
+ drc->drc_tofs = tofs;
+ drc->drc_force = force;
+ drc->drc_resumable = resumable;
+ drc->drc_cred = CRED();
+ drc->drc_clone = (origin != NULL);
+
+ if (drc->drc_drrb->drr_magic == BSWAP_64(DMU_BACKUP_MAGIC)) {
+ drc->drc_byteswap = B_TRUE;
+ (void) fletcher_4_incremental_byteswap(drr_begin,
+ sizeof (dmu_replay_record_t), &drc->drc_cksum);
+ byteswap_record(drr_begin);
+ } else if (drc->drc_drrb->drr_magic == DMU_BACKUP_MAGIC) {
+ (void) fletcher_4_incremental_native(drr_begin,
+ sizeof (dmu_replay_record_t), &drc->drc_cksum);
+ } else {
+ return (SET_ERROR(EINVAL));
+ }
+
+ drba.drba_origin = origin;
+ drba.drba_cookie = drc;
+ drba.drba_cred = CRED();
+
+ if (DMU_GET_FEATUREFLAGS(drc->drc_drrb->drr_versioninfo) &
+ DMU_BACKUP_FEATURE_RESUMING) {
+ return (dsl_sync_task(tofs,
+ dmu_recv_resume_begin_check, dmu_recv_resume_begin_sync,
+ &drba, 5, ZFS_SPACE_CHECK_NORMAL));
+ } else {
+ return (dsl_sync_task(tofs,
+ dmu_recv_begin_check, dmu_recv_begin_sync,
+ &drba, 5, ZFS_SPACE_CHECK_NORMAL));
+ }
+}
+
+struct receive_record_arg {
+ dmu_replay_record_t header;
+ void *payload; /* Pointer to a buffer containing the payload */
+ /*
+ * If the record is a write, pointer to the arc_buf_t containing the
+ * payload.
+ */
+ arc_buf_t *arc_buf;
+ int payload_size;
+ uint64_t bytes_read; /* bytes read from stream when record created */
boolean_t eos_marker; /* Marks the end of the stream */
bqueue_node_t node;
};
objset_t *os;
boolean_t byteswap;
bqueue_t q;
+
/*
* These three args are used to signal to the main thread that we're
* done.
kmutex_t mutex;
kcondvar_t cv;
boolean_t done;
+
int err;
/* A map from guid to dataset to help handle dedup'd streams. */
avl_tree_t *guid_to_ds_map;
+ boolean_t resumable;
+ boolean_t raw;
+ uint64_t last_object;
+ uint64_t last_offset;
+ uint64_t max_object; /* highest object ID referenced in stream */
+ uint64_t bytes_read; /* bytes read when current record created */
+
+ /* Encryption parameters for the last received DRR_OBJECT_RANGE */
+ uint64_t or_firstobj;
+ uint64_t or_numslots;
+ uint8_t or_salt[ZIO_DATA_SALT_LEN];
+ uint8_t or_iv[ZIO_DATA_IV_LEN];
+ uint8_t or_mac[ZIO_DATA_MAC_LEN];
+ boolean_t or_byteorder;
+};
+
+struct objlist {
+ list_t list; /* List of struct receive_objnode. */
+ /*
+ * Last object looked up. Used to assert that objects are being looked
+ * up in ascending order.
+ */
+ uint64_t last_lookup;
+};
+
+struct receive_objnode {
+ list_node_t node;
+ uint64_t object;
};
struct receive_arg {
objset_t *os;
vnode_t *vp; /* The vnode to read the stream from */
uint64_t voff; /* The current offset in the stream */
+ uint64_t bytes_read;
/*
* A record that has had its payload read in, but hasn't yet been handed
* off to the worker thread.
zio_cksum_t prev_cksum;
int err;
boolean_t byteswap;
+ boolean_t raw;
+ uint64_t featureflags;
/* Sorted list of objects not to issue prefetches for. */
- list_t ignore_obj_list;
-};
-
-struct receive_ign_obj_node {
- list_node_t node;
- uint64_t object;
+ struct objlist ignore_objlist;
};
typedef struct guid_map_entry {
uint64_t guid;
+ boolean_t raw;
dsl_dataset_t *gme_ds;
avl_node_t avlnode;
} guid_map_entry_t;
static int
guid_compare(const void *arg1, const void *arg2)
{
- const guid_map_entry_t *gmep1 = arg1;
- const guid_map_entry_t *gmep2 = arg2;
+ const guid_map_entry_t *gmep1 = (const guid_map_entry_t *)arg1;
+ const guid_map_entry_t *gmep2 = (const guid_map_entry_t *)arg2;
- if (gmep1->guid < gmep2->guid)
- return (-1);
- else if (gmep1->guid > gmep2->guid)
- return (1);
- return (0);
+ return (AVL_CMP(gmep1->guid, gmep2->guid));
}
static void
while ((gmep = avl_destroy_nodes(ca, &cookie)) != NULL) {
dsl_dataset_long_rele(gmep->gme_ds, gmep);
- dsl_dataset_rele(gmep->gme_ds, gmep);
+ dsl_dataset_rele_flags(gmep->gme_ds,
+ (gmep->raw) ? 0 : DS_HOLD_FLAG_DECRYPT, gmep);
kmem_free(gmep, sizeof (guid_map_entry_t));
}
avl_destroy(ca);
{
int done = 0;
- /* some things will require 8-byte alignment, so everything must */
- ASSERT0(len % 8);
+ /*
+ * The code doesn't rely on this (lengths being multiples of 8). See
+ * comment in dump_bytes.
+ */
+ ASSERT(len % 8 == 0 ||
+ (ra->featureflags & DMU_BACKUP_FEATURE_RAW) != 0);
while (done < len) {
ssize_t resid;
ra->voff, UIO_SYSSPACE, FAPPEND,
RLIM64_INFINITY, CRED(), &resid);
- if (resid == len - done)
- ra->err = SET_ERROR(EINVAL);
+ if (resid == len - done) {
+ /*
+ * Note: ECKSUM indicates that the receive
+ * was interrupted and can potentially be resumed.
+ */
+ ra->err = SET_ERROR(ECKSUM);
+ }
ra->voff += len - done - resid;
done = len - resid;
if (ra->err != 0)
return (ra->err);
}
+ ra->bytes_read += len;
+
ASSERT3U(done, ==, len);
return (0);
}
DO32(drr_object.drr_bonustype);
DO32(drr_object.drr_blksz);
DO32(drr_object.drr_bonuslen);
+ DO32(drr_object.drr_raw_bonuslen);
DO64(drr_object.drr_toguid);
+ DO64(drr_object.drr_maxblkid);
break;
case DRR_FREEOBJECTS:
DO64(drr_freeobjects.drr_firstobj);
DO64(drr_write.drr_object);
DO32(drr_write.drr_type);
DO64(drr_write.drr_offset);
- DO64(drr_write.drr_length);
+ DO64(drr_write.drr_logical_size);
DO64(drr_write.drr_toguid);
ZIO_CHECKSUM_BSWAP(&drr->drr_u.drr_write.drr_key.ddk_cksum);
DO64(drr_write.drr_key.ddk_prop);
+ DO64(drr_write.drr_compressed_size);
break;
case DRR_WRITE_BYREF:
DO64(drr_write_byref.drr_object);
DO64(drr_spill.drr_object);
DO64(drr_spill.drr_length);
DO64(drr_spill.drr_toguid);
+ DO64(drr_spill.drr_compressed_size);
+ DO32(drr_spill.drr_type);
+ break;
+ case DRR_OBJECT_RANGE:
+ DO64(drr_object_range.drr_firstobj);
+ DO64(drr_object_range.drr_numslots);
+ DO64(drr_object_range.drr_toguid);
break;
case DRR_END:
DO64(drr_end.drr_toguid);
return (1);
} else {
return (1 +
- ((DN_MAX_BONUSLEN - bonus_size) >> SPA_BLKPTRSHIFT));
+ ((DN_OLD_MAX_BONUSLEN -
+ MIN(DN_OLD_MAX_BONUSLEN, bonus_size)) >> SPA_BLKPTRSHIFT));
}
}
+static void
+save_resume_state(struct receive_writer_arg *rwa,
+ uint64_t object, uint64_t offset, dmu_tx_t *tx)
+{
+ int txgoff = dmu_tx_get_txg(tx) & TXG_MASK;
+
+ if (!rwa->resumable)
+ return;
+
+ /*
+ * We use ds_resume_bytes[] != 0 to indicate that we need to
+ * update this on disk, so it must not be 0.
+ */
+ ASSERT(rwa->bytes_read != 0);
+
+ /*
+ * We only resume from write records, which have a valid
+ * (non-meta-dnode) object number.
+ */
+ ASSERT(object != 0);
+
+ /*
+ * For resuming to work correctly, we must receive records in order,
+ * sorted by object,offset. This is checked by the callers, but
+ * assert it here for good measure.
+ */
+ ASSERT3U(object, >=, rwa->os->os_dsl_dataset->ds_resume_object[txgoff]);
+ ASSERT(object != rwa->os->os_dsl_dataset->ds_resume_object[txgoff] ||
+ offset >= rwa->os->os_dsl_dataset->ds_resume_offset[txgoff]);
+ ASSERT3U(rwa->bytes_read, >=,
+ rwa->os->os_dsl_dataset->ds_resume_bytes[txgoff]);
+
+ rwa->os->os_dsl_dataset->ds_resume_object[txgoff] = object;
+ rwa->os->os_dsl_dataset->ds_resume_offset[txgoff] = offset;
+ rwa->os->os_dsl_dataset->ds_resume_bytes[txgoff] = rwa->bytes_read;
+}
+
noinline static int
receive_object(struct receive_writer_arg *rwa, struct drr_object *drro,
- void *data)
+ void *data)
{
dmu_object_info_t doi;
dmu_tx_t *tx;
P2PHASE(drro->drr_blksz, SPA_MINBLOCKSIZE) ||
drro->drr_blksz < SPA_MINBLOCKSIZE ||
drro->drr_blksz > spa_maxblocksize(dmu_objset_spa(rwa->os)) ||
- drro->drr_bonuslen > DN_MAX_BONUSLEN) {
+ drro->drr_bonuslen >
+ DN_BONUS_SIZE(spa_maxdnodesize(dmu_objset_spa(rwa->os))) ||
+ drro->drr_dn_slots >
+ (spa_maxdnodesize(dmu_objset_spa(rwa->os)) >> DNODE_SHIFT)) {
return (SET_ERROR(EINVAL));
}
- err = dmu_object_info(rwa->os, drro->drr_object, &doi);
+ if (rwa->raw) {
+ /*
+ * We should have received a DRR_OBJECT_RANGE record
+ * containing this block and stored it in rwa.
+ */
+ if (drro->drr_object < rwa->or_firstobj ||
+ drro->drr_object >= rwa->or_firstobj + rwa->or_numslots ||
+ drro->drr_raw_bonuslen < drro->drr_bonuslen ||
+ drro->drr_indblkshift > SPA_MAXBLOCKSHIFT ||
+ drro->drr_nlevels > DN_MAX_LEVELS ||
+ drro->drr_nblkptr > DN_MAX_NBLKPTR ||
+ DN_SLOTS_TO_BONUSLEN(drro->drr_dn_slots) <
+ drro->drr_raw_bonuslen)
+ return (SET_ERROR(EINVAL));
+ } else {
+ if (drro->drr_flags != 0 || drro->drr_raw_bonuslen != 0 ||
+ drro->drr_indblkshift != 0 || drro->drr_nlevels != 0 ||
+ drro->drr_nblkptr != 0)
+ return (SET_ERROR(EINVAL));
+ }
- if (err != 0 && err != ENOENT)
+ err = dmu_object_info(rwa->os, drro->drr_object, &doi);
+ if (err != 0 && err != ENOENT && err != EEXIST)
return (SET_ERROR(EINVAL));
- object = err == 0 ? drro->drr_object : DMU_NEW_OBJECT;
+
+ if (drro->drr_object > rwa->max_object)
+ rwa->max_object = drro->drr_object;
/*
* If we are losing blkptrs or changing the block size this must
* be a new file instance. We must clear out the previous file
* contents before we can change this type of metadata in the dnode.
+ * Raw receives will also check that the indirect structure of the
+ * dnode hasn't changed.
*/
if (err == 0) {
- int nblkptr;
-
- nblkptr = deduce_nblkptr(drro->drr_bonustype,
+ uint32_t indblksz = drro->drr_indblkshift ?
+ 1ULL << drro->drr_indblkshift : 0;
+ int nblkptr = deduce_nblkptr(drro->drr_bonustype,
drro->drr_bonuslen);
- if (drro->drr_blksz != doi.doi_data_block_size ||
- nblkptr < doi.doi_nblkptr) {
+ object = drro->drr_object;
+
+ /* nblkptr will be bounded by the bonus size and type */
+ if (rwa->raw && nblkptr != drro->drr_nblkptr)
+ return (SET_ERROR(EINVAL));
+
+ if (rwa->raw &&
+ (drro->drr_blksz != doi.doi_data_block_size ||
+ nblkptr < doi.doi_nblkptr ||
+ indblksz != doi.doi_metadata_block_size ||
+ drro->drr_nlevels < doi.doi_indirection ||
+ drro->drr_dn_slots != doi.doi_dnodesize >> DNODE_SHIFT)) {
+ err = dmu_free_long_range_raw(rwa->os,
+ drro->drr_object, 0, DMU_OBJECT_END);
+ if (err != 0)
+ return (SET_ERROR(EINVAL));
+ } else if (drro->drr_blksz != doi.doi_data_block_size ||
+ nblkptr < doi.doi_nblkptr ||
+ drro->drr_dn_slots != doi.doi_dnodesize >> DNODE_SHIFT) {
err = dmu_free_long_range(rwa->os, drro->drr_object,
0, DMU_OBJECT_END);
if (err != 0)
return (SET_ERROR(EINVAL));
}
+
+ /*
+ * The dmu does not currently support decreasing nlevels
+ * on an object. For non-raw sends, this does not matter
+ * and the new object can just use the previous one's nlevels.
+ * For raw sends, however, the structure of the received dnode
+ * (including nlevels) must match that of the send side.
+ * Therefore, instead of using dmu_object_reclaim(), we must
+ * free the object completely and call dmu_object_claim_dnsize()
+ * instead.
+ */
+ if ((rwa->raw && drro->drr_nlevels < doi.doi_indirection) ||
+ drro->drr_dn_slots != doi.doi_dnodesize >> DNODE_SHIFT) {
+ if (rwa->raw) {
+ err = dmu_free_long_object_raw(rwa->os,
+ drro->drr_object);
+ } else {
+ err = dmu_free_long_object(rwa->os,
+ drro->drr_object);
+ }
+ if (err != 0)
+ return (SET_ERROR(EINVAL));
+
+ txg_wait_synced(dmu_objset_pool(rwa->os), 0);
+ object = DMU_NEW_OBJECT;
+ }
+ } else if (err == EEXIST) {
+ /*
+ * The object requested is currently an interior slot of a
+ * multi-slot dnode. This will be resolved when the next txg
+ * is synced out, since the send stream will have told us
+ * to free this slot when we freed the associated dnode
+ * earlier in the stream.
+ */
+ txg_wait_synced(dmu_objset_pool(rwa->os), 0);
+ object = drro->drr_object;
+ } else {
+ /* object is free and we are about to allocate a new one */
+ object = DMU_NEW_OBJECT;
+ }
+
+ /*
+ * If this is a multi-slot dnode there is a chance that this
+ * object will expand into a slot that is already used by
+ * another object from the previous snapshot. We must free
+ * these objects before we attempt to allocate the new dnode.
+ */
+ if (drro->drr_dn_slots > 1) {
+ boolean_t need_sync = B_FALSE;
+
+ for (uint64_t slot = drro->drr_object + 1;
+ slot < drro->drr_object + drro->drr_dn_slots;
+ slot++) {
+ dmu_object_info_t slot_doi;
+
+ err = dmu_object_info(rwa->os, slot, &slot_doi);
+ if (err == ENOENT || err == EEXIST)
+ continue;
+ else if (err != 0)
+ return (err);
+
+ if (rwa->raw)
+ err = dmu_free_long_object_raw(rwa->os, slot);
+ else
+ err = dmu_free_long_object(rwa->os, slot);
+
+ if (err != 0)
+ return (err);
+
+ need_sync = B_TRUE;
+ }
+
+ if (need_sync)
+ txg_wait_synced(dmu_objset_pool(rwa->os), 0);
}
tx = dmu_tx_create(rwa->os);
dmu_tx_hold_bonus(tx, object);
+ dmu_tx_hold_write(tx, object, 0, 0);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err != 0) {
dmu_tx_abort(tx);
if (object == DMU_NEW_OBJECT) {
/* currently free, want to be allocated */
- err = dmu_object_claim(rwa->os, drro->drr_object,
+ err = dmu_object_claim_dnsize(rwa->os, drro->drr_object,
drro->drr_type, drro->drr_blksz,
- drro->drr_bonustype, drro->drr_bonuslen, tx);
+ drro->drr_bonustype, drro->drr_bonuslen,
+ drro->drr_dn_slots << DNODE_SHIFT, tx);
} else if (drro->drr_type != doi.doi_type ||
drro->drr_blksz != doi.doi_data_block_size ||
drro->drr_bonustype != doi.doi_bonus_type ||
return (SET_ERROR(EINVAL));
}
+ if (rwa->raw) {
+ /*
+ * Convert the buffer associated with this range of dnodes
+ * to a raw buffer. This ensures that it will be written out
+ * as a raw buffer when we fill in the dnode object. Since we
+ * are committing this tx now, it is possible for the dnode
+ * block to end up on-disk with the incorrect MAC. Despite
+ * this, the dataset is marked as inconsistent so no other
+ * code paths (apart from scrubs) will attempt to read this
+ * data. Scrubs will not be effected by this either since
+ * scrubs only read raw data and do not attempt to check
+ * the MAC.
+ */
+ err = dmu_convert_mdn_block_to_raw(rwa->os, rwa->or_firstobj,
+ rwa->or_byteorder, rwa->or_salt, rwa->or_iv, rwa->or_mac,
+ tx);
+ if (err != 0) {
+ dmu_tx_commit(tx);
+ return (SET_ERROR(EINVAL));
+ }
+ }
+
dmu_object_set_checksum(rwa->os, drro->drr_object,
drro->drr_checksumtype, tx);
dmu_object_set_compress(rwa->os, drro->drr_object,
drro->drr_compress, tx);
+ /* handle more restrictive dnode structuring for raw recvs */
+ if (rwa->raw) {
+ /*
+ * Set the indirect block shift and nlevels. This will not fail
+ * because we ensured all of the blocks were free earlier if
+ * this is a new object.
+ */
+ VERIFY0(dmu_object_set_blocksize(rwa->os, drro->drr_object,
+ drro->drr_blksz, drro->drr_indblkshift, tx));
+ VERIFY0(dmu_object_set_nlevels(rwa->os, drro->drr_object,
+ drro->drr_nlevels, tx));
+ VERIFY0(dmu_object_set_maxblkid(rwa->os, drro->drr_object,
+ drro->drr_maxblkid, tx));
+ }
+
if (data != NULL) {
dmu_buf_t *db;
+ uint32_t flags = DMU_READ_NO_PREFETCH;
- VERIFY0(dmu_bonus_hold(rwa->os, drro->drr_object, FTAG, &db));
+ if (rwa->raw)
+ flags |= DMU_READ_NO_DECRYPT;
+
+ VERIFY0(dmu_bonus_hold_impl(rwa->os, drro->drr_object,
+ FTAG, flags, &db));
dmu_buf_will_dirty(db, tx);
ASSERT3U(db->db_size, >=, drro->drr_bonuslen);
- bcopy(data, db->db_data, drro->drr_bonuslen);
- if (rwa->byteswap) {
+ bcopy(data, db->db_data, DRR_OBJECT_PAYLOAD_SIZE(drro));
+
+ /*
+ * Raw bonus buffers have their byteorder determined by the
+ * DRR_OBJECT_RANGE record.
+ */
+ if (rwa->byteswap && !rwa->raw) {
dmu_object_byteswap_t byteswap =
DMU_OT_BYTESWAP(drro->drr_bonustype);
dmu_ot_byteswap[byteswap].ob_func(db->db_data,
- drro->drr_bonuslen);
+ DRR_OBJECT_PAYLOAD_SIZE(drro));
}
dmu_buf_rele(db, FTAG);
}
dmu_tx_commit(tx);
+
return (0);
}
struct drr_freeobjects *drrfo)
{
uint64_t obj;
+ int next_err = 0;
if (drrfo->drr_firstobj + drrfo->drr_numobjs < drrfo->drr_firstobj)
return (SET_ERROR(EINVAL));
- for (obj = drrfo->drr_firstobj;
- obj < drrfo->drr_firstobj + drrfo->drr_numobjs;
- (void) dmu_object_next(rwa->os, &obj, FALSE, 0)) {
+ for (obj = drrfo->drr_firstobj == 0 ? 1 : drrfo->drr_firstobj;
+ obj < drrfo->drr_firstobj + drrfo->drr_numobjs && next_err == 0;
+ next_err = dmu_object_next(rwa->os, &obj, FALSE, 0)) {
+ dmu_object_info_t doi;
int err;
- if (dmu_object_info(rwa->os, obj, NULL) != 0)
+ err = dmu_object_info(rwa->os, obj, &doi);
+ if (err == ENOENT)
continue;
+ else if (err != 0)
+ return (err);
+
+ if (rwa->raw)
+ err = dmu_free_long_object_raw(rwa->os, obj);
+ else
+ err = dmu_free_long_object(rwa->os, obj);
- err = dmu_free_long_object(rwa->os, obj);
if (err != 0)
return (err);
+
+ if (obj > rwa->max_object)
+ rwa->max_object = obj;
}
+ if (next_err != ESRCH)
+ return (next_err);
return (0);
}
noinline static int
receive_write(struct receive_writer_arg *rwa, struct drr_write *drrw,
- arc_buf_t *abuf)
+ arc_buf_t *abuf)
{
- dmu_tx_t *tx;
- dmu_buf_t *bonus;
int err;
+ dmu_tx_t *tx;
+ dnode_t *dn;
- if (drrw->drr_offset + drrw->drr_length < drrw->drr_offset ||
+ if (drrw->drr_offset + drrw->drr_logical_size < drrw->drr_offset ||
!DMU_OT_IS_VALID(drrw->drr_type))
return (SET_ERROR(EINVAL));
+ /*
+ * For resuming to work, records must be in increasing order
+ * by (object, offset).
+ */
+ if (drrw->drr_object < rwa->last_object ||
+ (drrw->drr_object == rwa->last_object &&
+ drrw->drr_offset < rwa->last_offset)) {
+ return (SET_ERROR(EINVAL));
+ }
+ rwa->last_object = drrw->drr_object;
+ rwa->last_offset = drrw->drr_offset;
+
+ if (rwa->last_object > rwa->max_object)
+ rwa->max_object = rwa->last_object;
+
if (dmu_object_info(rwa->os, drrw->drr_object, NULL) != 0)
return (SET_ERROR(EINVAL));
tx = dmu_tx_create(rwa->os);
-
dmu_tx_hold_write(tx, drrw->drr_object,
- drrw->drr_offset, drrw->drr_length);
+ drrw->drr_offset, drrw->drr_logical_size);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err != 0) {
dmu_tx_abort(tx);
return (err);
}
- if (rwa->byteswap) {
+
+ if (rwa->raw)
+ VERIFY0(dmu_object_dirty_raw(rwa->os, drrw->drr_object, tx));
+
+ if (rwa->byteswap && !arc_is_encrypted(abuf) &&
+ arc_get_compression(abuf) == ZIO_COMPRESS_OFF) {
dmu_object_byteswap_t byteswap =
DMU_OT_BYTESWAP(drrw->drr_type);
dmu_ot_byteswap[byteswap].ob_func(abuf->b_data,
- drrw->drr_length);
+ DRR_WRITE_PAYLOAD_SIZE(drrw));
}
- if (dmu_bonus_hold(rwa->os, drrw->drr_object, FTAG, &bonus) != 0)
- return (SET_ERROR(EINVAL));
- dmu_assign_arcbuf(bonus, drrw->drr_offset, abuf, tx);
+ VERIFY0(dnode_hold(rwa->os, drrw->drr_object, FTAG, &dn));
+ dmu_assign_arcbuf_by_dnode(dn, drrw->drr_offset, abuf, tx);
+ dnode_rele(dn, FTAG);
+
+ /*
+ * Note: If the receive fails, we want the resume stream to start
+ * with the same record that we last successfully received (as opposed
+ * to the next record), so that we can verify that we are
+ * resuming from the correct location.
+ */
+ save_resume_state(rwa, drrw->drr_object, drrw->drr_offset, tx);
dmu_tx_commit(tx);
- dmu_buf_rele(bonus, FTAG);
+
return (0);
}
guid_map_entry_t *gmep;
avl_index_t where;
objset_t *ref_os = NULL;
+ int flags = DMU_READ_PREFETCH;
dmu_buf_t *dbp;
if (drrwbr->drr_offset + drrwbr->drr_length < drrwbr->drr_offset)
ref_os = rwa->os;
}
+ if (drrwbr->drr_object > rwa->max_object)
+ rwa->max_object = drrwbr->drr_object;
+
+ if (rwa->raw)
+ flags |= DMU_READ_NO_DECRYPT;
+
+ /* may return either a regular db or an encrypted one */
err = dmu_buf_hold(ref_os, drrwbr->drr_refobject,
- drrwbr->drr_refoffset, FTAG, &dbp, DMU_READ_PREFETCH);
+ drrwbr->drr_refoffset, FTAG, &dbp, flags);
if (err != 0)
return (err);
dmu_tx_abort(tx);
return (err);
}
- dmu_write(rwa->os, drrwbr->drr_object,
- drrwbr->drr_offset, drrwbr->drr_length, dbp->db_data, tx);
+
+ if (rwa->raw) {
+ VERIFY0(dmu_object_dirty_raw(rwa->os, drrwbr->drr_object, tx));
+ dmu_copy_from_buf(rwa->os, drrwbr->drr_object,
+ drrwbr->drr_offset, dbp, tx);
+ } else {
+ dmu_write(rwa->os, drrwbr->drr_object,
+ drrwbr->drr_offset, drrwbr->drr_length, dbp->db_data, tx);
+ }
dmu_buf_rele(dbp, FTAG);
+
+ /* See comment in restore_write. */
+ save_resume_state(rwa, drrwbr->drr_object, drrwbr->drr_offset, tx);
dmu_tx_commit(tx);
return (0);
}
static int
receive_write_embedded(struct receive_writer_arg *rwa,
- struct drr_write_embedded *drrwnp, void *data)
+ struct drr_write_embedded *drrwe, void *data)
{
dmu_tx_t *tx;
int err;
- if (drrwnp->drr_offset + drrwnp->drr_length < drrwnp->drr_offset)
- return (EINVAL);
+ if (drrwe->drr_offset + drrwe->drr_length < drrwe->drr_offset)
+ return (SET_ERROR(EINVAL));
+
+ if (drrwe->drr_psize > BPE_PAYLOAD_SIZE)
+ return (SET_ERROR(EINVAL));
- if (drrwnp->drr_psize > BPE_PAYLOAD_SIZE)
- return (EINVAL);
+ if (drrwe->drr_etype >= NUM_BP_EMBEDDED_TYPES)
+ return (SET_ERROR(EINVAL));
+ if (drrwe->drr_compression >= ZIO_COMPRESS_FUNCTIONS)
+ return (SET_ERROR(EINVAL));
+ if (rwa->raw)
+ return (SET_ERROR(EINVAL));
- if (drrwnp->drr_etype >= NUM_BP_EMBEDDED_TYPES)
- return (EINVAL);
- if (drrwnp->drr_compression >= ZIO_COMPRESS_FUNCTIONS)
- return (EINVAL);
+ if (drrwe->drr_object > rwa->max_object)
+ rwa->max_object = drrwe->drr_object;
tx = dmu_tx_create(rwa->os);
- dmu_tx_hold_write(tx, drrwnp->drr_object,
- drrwnp->drr_offset, drrwnp->drr_length);
+ dmu_tx_hold_write(tx, drrwe->drr_object,
+ drrwe->drr_offset, drrwe->drr_length);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err != 0) {
dmu_tx_abort(tx);
return (err);
}
- dmu_write_embedded(rwa->os, drrwnp->drr_object,
- drrwnp->drr_offset, data, drrwnp->drr_etype,
- drrwnp->drr_compression, drrwnp->drr_lsize, drrwnp->drr_psize,
+ dmu_write_embedded(rwa->os, drrwe->drr_object,
+ drrwe->drr_offset, data, drrwe->drr_etype,
+ drrwe->drr_compression, drrwe->drr_lsize, drrwe->drr_psize,
rwa->byteswap ^ ZFS_HOST_BYTEORDER, tx);
+ /* See comment in restore_write. */
+ save_resume_state(rwa, drrwe->drr_object, drrwe->drr_offset, tx);
dmu_tx_commit(tx);
return (0);
}
static int
receive_spill(struct receive_writer_arg *rwa, struct drr_spill *drrs,
- void *data)
+ arc_buf_t *abuf)
{
dmu_tx_t *tx;
dmu_buf_t *db, *db_spill;
drrs->drr_length > spa_maxblocksize(dmu_objset_spa(rwa->os)))
return (SET_ERROR(EINVAL));
+ if (rwa->raw) {
+ if (!DMU_OT_IS_VALID(drrs->drr_type) ||
+ drrs->drr_compressiontype >= ZIO_COMPRESS_FUNCTIONS ||
+ drrs->drr_compressed_size == 0)
+ return (SET_ERROR(EINVAL));
+ }
+
if (dmu_object_info(rwa->os, drrs->drr_object, NULL) != 0)
return (SET_ERROR(EINVAL));
+ if (drrs->drr_object > rwa->max_object)
+ rwa->max_object = drrs->drr_object;
+
VERIFY0(dmu_bonus_hold(rwa->os, drrs->drr_object, FTAG, &db));
if ((err = dmu_spill_hold_by_bonus(db, FTAG, &db_spill)) != 0) {
dmu_buf_rele(db, FTAG);
dmu_tx_abort(tx);
return (err);
}
- dmu_buf_will_dirty(db_spill, tx);
+
+ if (rwa->raw) {
+ VERIFY0(dmu_object_dirty_raw(rwa->os, drrs->drr_object, tx));
+ dmu_buf_will_change_crypt_params(db_spill, tx);
+ } else {
+ dmu_buf_will_dirty(db_spill, tx);
+ }
if (db_spill->db_size < drrs->drr_length)
VERIFY(0 == dbuf_spill_set_blksz(db_spill,
drrs->drr_length, tx));
- bcopy(data, db_spill->db_data, drrs->drr_length);
+
+ if (rwa->byteswap && !arc_is_encrypted(abuf) &&
+ arc_get_compression(abuf) == ZIO_COMPRESS_OFF) {
+ dmu_object_byteswap_t byteswap =
+ DMU_OT_BYTESWAP(drrs->drr_type);
+ dmu_ot_byteswap[byteswap].ob_func(abuf->b_data,
+ DRR_SPILL_PAYLOAD_SIZE(drrs));
+ }
+
+ dbuf_assign_arcbuf((dmu_buf_impl_t *)db_spill, abuf, tx);
dmu_buf_rele(db, FTAG);
dmu_buf_rele(db_spill, FTAG);
{
int err;
- if (drrf->drr_length != -1ULL &&
+ if (drrf->drr_length != DMU_OBJECT_END &&
drrf->drr_offset + drrf->drr_length < drrf->drr_offset)
return (SET_ERROR(EINVAL));
if (dmu_object_info(rwa->os, drrf->drr_object, NULL) != 0)
return (SET_ERROR(EINVAL));
- err = dmu_free_long_range(rwa->os, drrf->drr_object,
- drrf->drr_offset, drrf->drr_length);
+ if (drrf->drr_object > rwa->max_object)
+ rwa->max_object = drrf->drr_object;
+
+ if (rwa->raw) {
+ err = dmu_free_long_range_raw(rwa->os, drrf->drr_object,
+ drrf->drr_offset, drrf->drr_length);
+ } else {
+ err = dmu_free_long_range(rwa->os, drrf->drr_object,
+ drrf->drr_offset, drrf->drr_length);
+ }
return (err);
}
+static int
+receive_object_range(struct receive_writer_arg *rwa,
+ struct drr_object_range *drror)
+{
+ /*
+ * By default, we assume this block is in our native format
+ * (ZFS_HOST_BYTEORDER). We then take into account whether
+ * the send stream is byteswapped (rwa->byteswap). Finally,
+ * we need to byteswap again if this particular block was
+ * in non-native format on the send side.
+ */
+ boolean_t byteorder = ZFS_HOST_BYTEORDER ^ rwa->byteswap ^
+ !!DRR_IS_RAW_BYTESWAPPED(drror->drr_flags);
+
+ /*
+ * Since dnode block sizes are constant, we should not need to worry
+ * about making sure that the dnode block size is the same on the
+ * sending and receiving sides for the time being. For non-raw sends,
+ * this does not matter (and in fact we do not send a DRR_OBJECT_RANGE
+ * record at all). Raw sends require this record type because the
+ * encryption parameters are used to protect an entire block of bonus
+ * buffers. If the size of dnode blocks ever becomes variable,
+ * handling will need to be added to ensure that dnode block sizes
+ * match on the sending and receiving side.
+ */
+ if (drror->drr_numslots != DNODES_PER_BLOCK ||
+ P2PHASE(drror->drr_firstobj, DNODES_PER_BLOCK) != 0 ||
+ !rwa->raw)
+ return (SET_ERROR(EINVAL));
+
+ if (drror->drr_firstobj > rwa->max_object)
+ rwa->max_object = drror->drr_firstobj;
+
+ /*
+ * The DRR_OBJECT_RANGE handling must be deferred to receive_object()
+ * so that the encryption parameters are set with each object that is
+ * written into that block.
+ */
+ rwa->or_firstobj = drror->drr_firstobj;
+ rwa->or_numslots = drror->drr_numslots;
+ bcopy(drror->drr_salt, rwa->or_salt, ZIO_DATA_SALT_LEN);
+ bcopy(drror->drr_iv, rwa->or_iv, ZIO_DATA_IV_LEN);
+ bcopy(drror->drr_mac, rwa->or_mac, ZIO_DATA_MAC_LEN);
+ rwa->or_byteorder = byteorder;
+
+ return (0);
+}
+
/* used to destroy the drc_ds on error */
static void
dmu_recv_cleanup_ds(dmu_recv_cookie_t *drc)
{
- char name[MAXNAMELEN];
- dsl_dataset_name(drc->drc_ds, name);
- dsl_dataset_disown(drc->drc_ds, dmu_recv_tag);
- (void) dsl_destroy_head(name);
+ dsl_dataset_t *ds = drc->drc_ds;
+ ds_hold_flags_t dsflags = (drc->drc_raw) ? 0 : DS_HOLD_FLAG_DECRYPT;
+
+ /*
+ * Wait for the txg sync before cleaning up the receive. For
+ * resumable receives, this ensures that our resume state has
+ * been written out to disk. For raw receives, this ensures
+ * that the user accounting code will not attempt to do anything
+ * after we stopped receiving the dataset.
+ */
+ txg_wait_synced(ds->ds_dir->dd_pool, 0);
+
+ rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG);
+ if (drc->drc_resumable && !BP_IS_HOLE(dsl_dataset_get_blkptr(ds))) {
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
+ dsl_dataset_disown(ds, dsflags, dmu_recv_tag);
+ } else {
+ char name[ZFS_MAX_DATASET_NAME_LEN];
+ rrw_exit(&ds->ds_bp_rwlock, FTAG);
+ dsl_dataset_name(ds, name);
+ dsl_dataset_disown(ds, dsflags, dmu_recv_tag);
+ (void) dsl_destroy_head(name);
+ }
}
static void
receive_cksum(struct receive_arg *ra, int len, void *buf)
{
if (ra->byteswap) {
- fletcher_4_incremental_byteswap(buf, len, &ra->cksum);
+ (void) fletcher_4_incremental_byteswap(buf, len, &ra->cksum);
} else {
- fletcher_4_incremental_native(buf, len, &ra->cksum);
+ (void) fletcher_4_incremental_native(buf, len, &ra->cksum);
}
}
if (len != 0) {
ASSERT3U(len, <=, SPA_MAXBLOCKSIZE);
- ra->rrd->payload = buf;
- ra->rrd->payload_size = len;
- err = receive_read(ra, len, ra->rrd->payload);
+ err = receive_read(ra, len, buf);
if (err != 0)
return (err);
- receive_cksum(ra, len, ra->rrd->payload);
+ receive_cksum(ra, len, buf);
+
+ /* note: rrd is NULL when reading the begin record's payload */
+ if (ra->rrd != NULL) {
+ ra->rrd->payload = buf;
+ ra->rrd->payload_size = len;
+ ra->rrd->bytes_read = ra->bytes_read;
+ }
}
ra->prev_cksum = ra->cksum;
ra->next_rrd = kmem_zalloc(sizeof (*ra->next_rrd), KM_SLEEP);
err = receive_read(ra, sizeof (ra->next_rrd->header),
&ra->next_rrd->header);
+ ra->next_rrd->bytes_read = ra->bytes_read;
+
if (err != 0) {
kmem_free(ra->next_rrd, sizeof (*ra->next_rrd));
ra->next_rrd = NULL;
return (0);
}
+static void
+objlist_create(struct objlist *list)
+{
+ list_create(&list->list, sizeof (struct receive_objnode),
+ offsetof(struct receive_objnode, node));
+ list->last_lookup = 0;
+}
+
+static void
+objlist_destroy(struct objlist *list)
+{
+ for (struct receive_objnode *n = list_remove_head(&list->list);
+ n != NULL; n = list_remove_head(&list->list)) {
+ kmem_free(n, sizeof (*n));
+ }
+ list_destroy(&list->list);
+}
+
+/*
+ * This function looks through the objlist to see if the specified object number
+ * is contained in the objlist. In the process, it will remove all object
+ * numbers in the list that are smaller than the specified object number. Thus,
+ * any lookup of an object number smaller than a previously looked up object
+ * number will always return false; therefore, all lookups should be done in
+ * ascending order.
+ */
+static boolean_t
+objlist_exists(struct objlist *list, uint64_t object)
+{
+ struct receive_objnode *node = list_head(&list->list);
+ ASSERT3U(object, >=, list->last_lookup);
+ list->last_lookup = object;
+ while (node != NULL && node->object < object) {
+ VERIFY3P(node, ==, list_remove_head(&list->list));
+ kmem_free(node, sizeof (*node));
+ node = list_head(&list->list);
+ }
+ return (node != NULL && node->object == object);
+}
+
+/*
+ * The objlist is a list of object numbers stored in ascending order. However,
+ * the insertion of new object numbers does not seek out the correct location to
+ * store a new object number; instead, it appends it to the list for simplicity.
+ * Thus, any users must take care to only insert new object numbers in ascending
+ * order.
+ */
+static void
+objlist_insert(struct objlist *list, uint64_t object)
+{
+ struct receive_objnode *node = kmem_zalloc(sizeof (*node), KM_SLEEP);
+ node->object = object;
+#ifdef ZFS_DEBUG
+ {
+ struct receive_objnode *last_object = list_tail(&list->list);
+ uint64_t last_objnum = (last_object != NULL ? last_object->object : 0);
+ ASSERT3U(node->object, >, last_objnum);
+ }
+#endif
+ list_insert_tail(&list->list, node);
+}
+
/*
* Issue the prefetch reads for any necessary indirect blocks.
*
receive_read_prefetch(struct receive_arg *ra,
uint64_t object, uint64_t offset, uint64_t length)
{
- struct receive_ign_obj_node *node = list_head(&ra->ignore_obj_list);
- while (node != NULL && node->object < object) {
- VERIFY3P(node, ==, list_remove_head(&ra->ignore_obj_list));
- kmem_free(node, sizeof (*node));
- node = list_head(&ra->ignore_obj_list);
- }
- if (node == NULL || node->object > object) {
+ if (!objlist_exists(&ra->ignore_objlist, object)) {
dmu_prefetch(ra->os, object, 1, offset, length,
ZIO_PRIORITY_SYNC_READ);
}
case DRR_OBJECT:
{
struct drr_object *drro = &ra->rrd->header.drr_u.drr_object;
- uint32_t size = P2ROUNDUP(drro->drr_bonuslen, 8);
+ uint32_t size = DRR_OBJECT_PAYLOAD_SIZE(drro);
void *buf = kmem_zalloc(size, KM_SLEEP);
dmu_object_info_t doi;
+
err = receive_read_payload_and_next_header(ra, size, buf);
if (err != 0) {
kmem_free(buf, size);
* See receive_read_prefetch for an explanation why we're
* storing this object in the ignore_obj_list.
*/
- if (err == ENOENT ||
+ if (err == ENOENT || err == EEXIST ||
(err == 0 && doi.doi_data_block_size != drro->drr_blksz)) {
- struct receive_ign_obj_node *node =
- kmem_zalloc(sizeof (*node),
- KM_SLEEP);
- node->object = drro->drr_object;
-#ifdef ZFS_DEBUG
- {
- struct receive_ign_obj_node *last_object =
- list_tail(&ra->ignore_obj_list);
- uint64_t last_objnum = (last_object != NULL ?
- last_object->object : 0);
- ASSERT3U(node->object, >, last_objnum);
- }
-#endif
- list_insert_tail(&ra->ignore_obj_list, node);
+ objlist_insert(&ra->ignore_objlist, drro->drr_object);
err = 0;
}
return (err);
case DRR_WRITE:
{
struct drr_write *drrw = &ra->rrd->header.drr_u.drr_write;
- arc_buf_t *abuf = arc_loan_buf(dmu_objset_spa(ra->os),
- drrw->drr_length);
+ arc_buf_t *abuf;
+ boolean_t is_meta = DMU_OT_IS_METADATA(drrw->drr_type);
+
+ if (ra->raw) {
+ boolean_t byteorder = ZFS_HOST_BYTEORDER ^
+ !!DRR_IS_RAW_BYTESWAPPED(drrw->drr_flags) ^
+ ra->byteswap;
+
+ abuf = arc_loan_raw_buf(dmu_objset_spa(ra->os),
+ drrw->drr_object, byteorder, drrw->drr_salt,
+ drrw->drr_iv, drrw->drr_mac, drrw->drr_type,
+ drrw->drr_compressed_size, drrw->drr_logical_size,
+ drrw->drr_compressiontype);
+ } else if (DRR_WRITE_COMPRESSED(drrw)) {
+ ASSERT3U(drrw->drr_compressed_size, >, 0);
+ ASSERT3U(drrw->drr_logical_size, >=,
+ drrw->drr_compressed_size);
+ ASSERT(!is_meta);
+ abuf = arc_loan_compressed_buf(
+ dmu_objset_spa(ra->os),
+ drrw->drr_compressed_size, drrw->drr_logical_size,
+ drrw->drr_compressiontype);
+ } else {
+ abuf = arc_loan_buf(dmu_objset_spa(ra->os),
+ is_meta, drrw->drr_logical_size);
+ }
err = receive_read_payload_and_next_header(ra,
- drrw->drr_length, abuf->b_data);
+ DRR_WRITE_PAYLOAD_SIZE(drrw), abuf->b_data);
if (err != 0) {
dmu_return_arcbuf(abuf);
return (err);
}
- ra->rrd->write_buf = abuf;
+ ra->rrd->arc_buf = abuf;
receive_read_prefetch(ra, drrw->drr_object, drrw->drr_offset,
- drrw->drr_length);
+ drrw->drr_logical_size);
return (err);
}
case DRR_WRITE_BYREF:
{
struct drr_end *drre = &ra->rrd->header.drr_u.drr_end;
if (!ZIO_CHECKSUM_EQUAL(ra->prev_cksum, drre->drr_checksum))
- return (SET_ERROR(EINVAL));
+ return (SET_ERROR(ECKSUM));
return (0);
}
case DRR_SPILL:
{
struct drr_spill *drrs = &ra->rrd->header.drr_u.drr_spill;
- void *buf = kmem_zalloc(drrs->drr_length, KM_SLEEP);
- err = receive_read_payload_and_next_header(ra, drrs->drr_length,
- buf);
- if (err != 0)
- kmem_free(buf, drrs->drr_length);
+ arc_buf_t *abuf;
+ int len = DRR_SPILL_PAYLOAD_SIZE(drrs);
+
+ /* DRR_SPILL records are either raw or uncompressed */
+ if (ra->raw) {
+ boolean_t byteorder = ZFS_HOST_BYTEORDER ^
+ !!DRR_IS_RAW_BYTESWAPPED(drrs->drr_flags) ^
+ ra->byteswap;
+
+ abuf = arc_loan_raw_buf(dmu_objset_spa(ra->os),
+ drrs->drr_object, byteorder, drrs->drr_salt,
+ drrs->drr_iv, drrs->drr_mac, drrs->drr_type,
+ drrs->drr_compressed_size, drrs->drr_length,
+ drrs->drr_compressiontype);
+ } else {
+ abuf = arc_loan_buf(dmu_objset_spa(ra->os),
+ DMU_OT_IS_METADATA(drrs->drr_type),
+ drrs->drr_length);
+ }
+
+ err = receive_read_payload_and_next_header(ra, len,
+ abuf->b_data);
+ if (err != 0) {
+ dmu_return_arcbuf(abuf);
+ return (err);
+ }
+ ra->rrd->arc_buf = abuf;
+ return (err);
+ }
+ case DRR_OBJECT_RANGE:
+ {
+ err = receive_read_payload_and_next_header(ra, 0, NULL);
return (err);
}
default:
}
}
+static void
+dprintf_drr(struct receive_record_arg *rrd, int err)
+{
+#ifdef ZFS_DEBUG
+ switch (rrd->header.drr_type) {
+ case DRR_OBJECT:
+ {
+ struct drr_object *drro = &rrd->header.drr_u.drr_object;
+ dprintf("drr_type = OBJECT obj = %llu type = %u "
+ "bonustype = %u blksz = %u bonuslen = %u cksumtype = %u "
+ "compress = %u dn_slots = %u err = %d\n",
+ drro->drr_object, drro->drr_type, drro->drr_bonustype,
+ drro->drr_blksz, drro->drr_bonuslen,
+ drro->drr_checksumtype, drro->drr_compress,
+ drro->drr_dn_slots, err);
+ break;
+ }
+ case DRR_FREEOBJECTS:
+ {
+ struct drr_freeobjects *drrfo =
+ &rrd->header.drr_u.drr_freeobjects;
+ dprintf("drr_type = FREEOBJECTS firstobj = %llu "
+ "numobjs = %llu err = %d\n",
+ drrfo->drr_firstobj, drrfo->drr_numobjs, err);
+ break;
+ }
+ case DRR_WRITE:
+ {
+ struct drr_write *drrw = &rrd->header.drr_u.drr_write;
+ dprintf("drr_type = WRITE obj = %llu type = %u offset = %llu "
+ "lsize = %llu cksumtype = %u cksumflags = %u "
+ "compress = %u psize = %llu err = %d\n",
+ drrw->drr_object, drrw->drr_type, drrw->drr_offset,
+ drrw->drr_logical_size, drrw->drr_checksumtype,
+ drrw->drr_flags, drrw->drr_compressiontype,
+ drrw->drr_compressed_size, err);
+ break;
+ }
+ case DRR_WRITE_BYREF:
+ {
+ struct drr_write_byref *drrwbr =
+ &rrd->header.drr_u.drr_write_byref;
+ dprintf("drr_type = WRITE_BYREF obj = %llu offset = %llu "
+ "length = %llu toguid = %llx refguid = %llx "
+ "refobject = %llu refoffset = %llu cksumtype = %u "
+ "cksumflags = %u err = %d\n",
+ drrwbr->drr_object, drrwbr->drr_offset,
+ drrwbr->drr_length, drrwbr->drr_toguid,
+ drrwbr->drr_refguid, drrwbr->drr_refobject,
+ drrwbr->drr_refoffset, drrwbr->drr_checksumtype,
+ drrwbr->drr_flags, err);
+ break;
+ }
+ case DRR_WRITE_EMBEDDED:
+ {
+ struct drr_write_embedded *drrwe =
+ &rrd->header.drr_u.drr_write_embedded;
+ dprintf("drr_type = WRITE_EMBEDDED obj = %llu offset = %llu "
+ "length = %llu compress = %u etype = %u lsize = %u "
+ "psize = %u err = %d\n",
+ drrwe->drr_object, drrwe->drr_offset, drrwe->drr_length,
+ drrwe->drr_compression, drrwe->drr_etype,
+ drrwe->drr_lsize, drrwe->drr_psize, err);
+ break;
+ }
+ case DRR_FREE:
+ {
+ struct drr_free *drrf = &rrd->header.drr_u.drr_free;
+ dprintf("drr_type = FREE obj = %llu offset = %llu "
+ "length = %lld err = %d\n",
+ drrf->drr_object, drrf->drr_offset, drrf->drr_length,
+ err);
+ break;
+ }
+ case DRR_SPILL:
+ {
+ struct drr_spill *drrs = &rrd->header.drr_u.drr_spill;
+ dprintf("drr_type = SPILL obj = %llu length = %llu "
+ "err = %d\n", drrs->drr_object, drrs->drr_length, err);
+ break;
+ }
+ default:
+ return;
+ }
+#endif
+}
+
/*
* Commit the records to the pool.
*/
{
int err;
+ /* Processing in order, therefore bytes_read should be increasing. */
+ ASSERT3U(rrd->bytes_read, >=, rwa->bytes_read);
+ rwa->bytes_read = rrd->bytes_read;
+
switch (rrd->header.drr_type) {
case DRR_OBJECT:
{
err = receive_object(rwa, drro, rrd->payload);
kmem_free(rrd->payload, rrd->payload_size);
rrd->payload = NULL;
- return (err);
+ break;
}
case DRR_FREEOBJECTS:
{
struct drr_freeobjects *drrfo =
&rrd->header.drr_u.drr_freeobjects;
- return (receive_freeobjects(rwa, drrfo));
+ err = receive_freeobjects(rwa, drrfo);
+ break;
}
case DRR_WRITE:
{
struct drr_write *drrw = &rrd->header.drr_u.drr_write;
- err = receive_write(rwa, drrw, rrd->write_buf);
+ err = receive_write(rwa, drrw, rrd->arc_buf);
/* if receive_write() is successful, it consumes the arc_buf */
if (err != 0)
- dmu_return_arcbuf(rrd->write_buf);
- rrd->write_buf = NULL;
+ dmu_return_arcbuf(rrd->arc_buf);
+ rrd->arc_buf = NULL;
rrd->payload = NULL;
- return (err);
+ break;
}
case DRR_WRITE_BYREF:
{
struct drr_write_byref *drrwbr =
&rrd->header.drr_u.drr_write_byref;
- return (receive_write_byref(rwa, drrwbr));
+ err = receive_write_byref(rwa, drrwbr);
+ break;
}
case DRR_WRITE_EMBEDDED:
{
err = receive_write_embedded(rwa, drrwe, rrd->payload);
kmem_free(rrd->payload, rrd->payload_size);
rrd->payload = NULL;
- return (err);
+ break;
}
case DRR_FREE:
{
struct drr_free *drrf = &rrd->header.drr_u.drr_free;
- return (receive_free(rwa, drrf));
+ err = receive_free(rwa, drrf);
+ break;
}
case DRR_SPILL:
{
struct drr_spill *drrs = &rrd->header.drr_u.drr_spill;
- err = receive_spill(rwa, drrs, rrd->payload);
- kmem_free(rrd->payload, rrd->payload_size);
+ err = receive_spill(rwa, drrs, rrd->arc_buf);
+ /* if receive_spill() is successful, it consumes the arc_buf */
+ if (err != 0)
+ dmu_return_arcbuf(rrd->arc_buf);
+ rrd->arc_buf = NULL;
rrd->payload = NULL;
- return (err);
+ break;
+ }
+ case DRR_OBJECT_RANGE:
+ {
+ struct drr_object_range *drror =
+ &rrd->header.drr_u.drr_object_range;
+ return (receive_object_range(rwa, drror));
}
default:
return (SET_ERROR(EINVAL));
}
+
+ if (err != 0)
+ dprintf_drr(rrd, err);
+
+ return (err);
}
/*
{
struct receive_writer_arg *rwa = arg;
struct receive_record_arg *rrd;
+ fstrans_cookie_t cookie = spl_fstrans_mark();
+
for (rrd = bqueue_dequeue(&rwa->q); !rrd->eos_marker;
rrd = bqueue_dequeue(&rwa->q)) {
/*
*/
if (rwa->err == 0) {
rwa->err = receive_process_record(rwa, rrd);
- } else if (rrd->write_buf != NULL) {
- dmu_return_arcbuf(rrd->write_buf);
- rrd->write_buf = NULL;
+ } else if (rrd->arc_buf != NULL) {
+ dmu_return_arcbuf(rrd->arc_buf);
+ rrd->arc_buf = NULL;
rrd->payload = NULL;
} else if (rrd->payload != NULL) {
kmem_free(rrd->payload, rrd->payload_size);
rwa->done = B_TRUE;
cv_signal(&rwa->cv);
mutex_exit(&rwa->mutex);
+ spl_fstrans_unmark(cookie);
+ thread_exit();
+}
+
+static int
+resume_check(struct receive_arg *ra, nvlist_t *begin_nvl)
+{
+ uint64_t val;
+ objset_t *mos = dmu_objset_pool(ra->os)->dp_meta_objset;
+ uint64_t dsobj = dmu_objset_id(ra->os);
+ uint64_t resume_obj, resume_off;
+
+ if (nvlist_lookup_uint64(begin_nvl,
+ "resume_object", &resume_obj) != 0 ||
+ nvlist_lookup_uint64(begin_nvl,
+ "resume_offset", &resume_off) != 0) {
+ return (SET_ERROR(EINVAL));
+ }
+ VERIFY0(zap_lookup(mos, dsobj,
+ DS_FIELD_RESUME_OBJECT, sizeof (val), 1, &val));
+ if (resume_obj != val)
+ return (SET_ERROR(EINVAL));
+ VERIFY0(zap_lookup(mos, dsobj,
+ DS_FIELD_RESUME_OFFSET, sizeof (val), 1, &val));
+ if (resume_off != val)
+ return (SET_ERROR(EINVAL));
+
+ return (0);
}
/*
struct receive_arg *ra;
struct receive_writer_arg *rwa;
int featureflags;
- struct receive_ign_obj_node *n;
+ uint32_t payloadlen;
+ void *payload;
+ nvlist_t *begin_nvl = NULL;
ra = kmem_zalloc(sizeof (*ra), KM_SLEEP);
rwa = kmem_zalloc(sizeof (*rwa), KM_SLEEP);
ra->byteswap = drc->drc_byteswap;
+ ra->raw = drc->drc_raw;
ra->cksum = drc->drc_cksum;
ra->vp = vp;
ra->voff = *voffp;
- list_create(&ra->ignore_obj_list, sizeof (struct receive_ign_obj_node),
- offsetof(struct receive_ign_obj_node, node));
+
+ if (dsl_dataset_is_zapified(drc->drc_ds)) {
+ (void) zap_lookup(drc->drc_ds->ds_dir->dd_pool->dp_meta_objset,
+ drc->drc_ds->ds_object, DS_FIELD_RESUME_BYTES,
+ sizeof (ra->bytes_read), 1, &ra->bytes_read);
+ }
+
+ objlist_create(&ra->ignore_objlist);
/* these were verified in dmu_recv_begin */
ASSERT3U(DMU_GET_STREAM_HDRTYPE(drc->drc_drrb->drr_versioninfo), ==,
ASSERT(dsl_dataset_phys(drc->drc_ds)->ds_flags & DS_FLAG_INCONSISTENT);
featureflags = DMU_GET_FEATUREFLAGS(drc->drc_drrb->drr_versioninfo);
+ ra->featureflags = featureflags;
+
+ /* embedded data is incompatible with encrypted datasets */
+ if (ra->os->os_encrypted &&
+ (featureflags & DMU_BACKUP_FEATURE_EMBED_DATA)) {
+ err = SET_ERROR(EINVAL);
+ goto out;
+ }
/* if this stream is dedup'ed, set up the avl tree for guid mapping */
if (featureflags & DMU_BACKUP_FEATURE_DEDUP) {
minor_t minor;
if (cleanup_fd == -1) {
- ra->err = SET_ERROR(EBADF);
+ err = SET_ERROR(EBADF);
goto out;
}
- ra->err = zfs_onexit_fd_hold(cleanup_fd, &minor);
- if (ra->err != 0) {
+ err = zfs_onexit_fd_hold(cleanup_fd, &minor);
+ if (err != 0) {
cleanup_fd = -1;
goto out;
}
err = zfs_onexit_add_cb(minor,
free_guid_map_onexit, rwa->guid_to_ds_map,
action_handlep);
- if (ra->err != 0)
+ if (err != 0)
goto out;
} else {
err = zfs_onexit_cb_data(minor, *action_handlep,
(void **)&rwa->guid_to_ds_map);
- if (ra->err != 0)
+ if (err != 0)
goto out;
}
drc->drc_guid_to_ds_map = rwa->guid_to_ds_map;
}
- err = receive_read_payload_and_next_header(ra, 0, NULL);
- if (err)
+ payloadlen = drc->drc_drr_begin->drr_payloadlen;
+ payload = NULL;
+ if (payloadlen != 0)
+ payload = kmem_alloc(payloadlen, KM_SLEEP);
+
+ err = receive_read_payload_and_next_header(ra, payloadlen, payload);
+ if (err != 0) {
+ if (payloadlen != 0)
+ kmem_free(payload, payloadlen);
goto out;
+ }
+ if (payloadlen != 0) {
+ err = nvlist_unpack(payload, payloadlen, &begin_nvl, KM_SLEEP);
+ kmem_free(payload, payloadlen);
+ if (err != 0)
+ goto out;
+ }
+
+ /* handle DSL encryption key payload */
+ if (featureflags & DMU_BACKUP_FEATURE_RAW) {
+ nvlist_t *keynvl = NULL;
+
+ ASSERT(ra->os->os_encrypted);
+ ASSERT(drc->drc_raw);
+
+ err = nvlist_lookup_nvlist(begin_nvl, "crypt_keydata", &keynvl);
+ if (err != 0)
+ goto out;
+
+ /*
+ * If this is a new dataset we set the key immediately.
+ * Otherwise we don't want to change the key until we
+ * are sure the rest of the receive succeeded so we stash
+ * the keynvl away until then.
+ */
+ err = dsl_crypto_recv_raw(spa_name(ra->os->os_spa),
+ drc->drc_ds->ds_object, drc->drc_drrb->drr_type,
+ keynvl, drc->drc_newfs);
+ if (err != 0)
+ goto out;
+
+ if (!drc->drc_newfs)
+ drc->drc_keynvl = fnvlist_dup(keynvl);
+ }
+
+ if (featureflags & DMU_BACKUP_FEATURE_RESUMING) {
+ err = resume_check(ra, begin_nvl);
+ if (err != 0)
+ goto out;
+ }
- (void) bqueue_init(&rwa->q, zfs_recv_queue_length,
+ (void) bqueue_init(&rwa->q,
+ MAX(zfs_recv_queue_length, 2 * zfs_max_recordsize),
offsetof(struct receive_record_arg, node));
cv_init(&rwa->cv, NULL, CV_DEFAULT, NULL);
mutex_init(&rwa->mutex, NULL, MUTEX_DEFAULT, NULL);
rwa->os = ra->os;
rwa->byteswap = drc->drc_byteswap;
+ rwa->resumable = drc->drc_resumable;
+ rwa->raw = drc->drc_raw;
(void) thread_create(NULL, 0, receive_writer_thread, rwa, 0, curproc,
TS_RUN, minclsyspri);
* We can leave this loop in 3 ways: First, if rwa->err is
* non-zero. In that case, the writer thread will free the rrd we just
* pushed. Second, if we're interrupted; in that case, either it's the
- * first loop and ra->rrd was never allocated, or it's later, and ra.rrd
+ * first loop and ra->rrd was never allocated, or it's later and ra->rrd
* has been handed off to the writer thread who will free it. Finally,
* if receive_read_record fails or we're at the end of the stream, then
* we free ra->rrd and exit.
}
mutex_exit(&rwa->mutex);
+ /*
+ * If we are receiving a full stream as a clone, all object IDs which
+ * are greater than the maximum ID referenced in the stream are
+ * by definition unused and must be freed.
+ */
+ if (drc->drc_clone && drc->drc_drrb->drr_fromguid == 0) {
+ uint64_t obj = rwa->max_object + 1;
+ int free_err = 0;
+ int next_err = 0;
+
+ while (next_err == 0) {
+ if (drc->drc_raw) {
+ free_err = dmu_free_long_object_raw(rwa->os,
+ obj);
+ } else {
+ free_err = dmu_free_long_object(rwa->os, obj);
+ }
+ if (free_err != 0 && free_err != ENOENT)
+ break;
+
+ next_err = dmu_object_next(rwa->os, &obj, FALSE, 0);
+ }
+
+ if (err == 0) {
+ if (free_err != 0 && free_err != ENOENT)
+ err = free_err;
+ else if (next_err != ESRCH)
+ err = next_err;
+ }
+ }
+
cv_destroy(&rwa->cv);
mutex_destroy(&rwa->mutex);
bqueue_destroy(&rwa->q);
err = rwa->err;
out:
+ nvlist_free(begin_nvl);
if ((featureflags & DMU_BACKUP_FEATURE_DEDUP) && (cleanup_fd != -1))
zfs_onexit_fd_rele(cleanup_fd);
if (err != 0) {
/*
- * destroy what we created, so we don't leave it in the
- * inconsistent restoring state.
+ * Clean up references. If receive is not resumable,
+ * destroy what we created, so we don't leave it in
+ * the inconsistent state.
*/
dmu_recv_cleanup_ds(drc);
+ nvlist_free(drc->drc_keynvl);
}
*voffp = ra->voff;
-
- for (n = list_remove_head(&ra->ignore_obj_list); n != NULL;
- n = list_remove_head(&ra->ignore_obj_list)) {
- kmem_free(n, sizeof (*n));
- }
- list_destroy(&ra->ignore_obj_list);
+ objlist_destroy(&ra->ignore_objlist);
kmem_free(ra, sizeof (*ra));
kmem_free(rwa, sizeof (*rwa));
return (err);
return (error);
}
}
+ if (drc->drc_keynvl != NULL) {
+ error = dsl_crypto_recv_raw_key_check(drc->drc_ds,
+ drc->drc_keynvl, tx);
+ if (error != 0) {
+ dsl_dataset_rele(origin_head, FTAG);
+ return (error);
+ }
+ }
+
error = dsl_dataset_clone_swap_check_impl(drc->drc_ds,
origin_head, drc->drc_force, drc->drc_owner, tx);
if (error != 0) {
{
dmu_recv_cookie_t *drc = arg;
dsl_pool_t *dp = dmu_tx_pool(tx);
+ boolean_t encrypted = drc->drc_ds->ds_dir->dd_crypto_obj != 0;
spa_history_log_internal_ds(drc->drc_ds, "finish receiving",
tx, "snap=%s", drc->drc_tosnap);
dsl_dataset_rele(snap, FTAG);
}
}
- VERIFY3P(drc->drc_ds->ds_prev, ==,
- origin_head->ds_prev);
+ if (drc->drc_keynvl != NULL) {
+ dsl_crypto_recv_raw_key_sync(drc->drc_ds,
+ drc->drc_keynvl, tx);
+ nvlist_free(drc->drc_keynvl);
+ drc->drc_keynvl = NULL;
+ }
+
+ VERIFY3P(drc->drc_ds->ds_prev, ==, origin_head->ds_prev);
dsl_dataset_clone_swap_sync_impl(drc->drc_ds,
origin_head, tx);
dsl_dataset_phys(origin_head)->ds_flags &=
~DS_FLAG_INCONSISTENT;
+ drc->drc_newsnapobj =
+ dsl_dataset_phys(origin_head)->ds_prev_snap_obj;
+
dsl_dataset_rele(origin_head, FTAG);
dsl_destroy_head_sync_impl(drc->drc_ds, tx);
dmu_buf_will_dirty(ds->ds_dbuf, tx);
dsl_dataset_phys(ds)->ds_flags &= ~DS_FLAG_INCONSISTENT;
+ if (dsl_dataset_has_resume_receive_state(ds)) {
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_FROMGUID, tx);
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_OBJECT, tx);
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_OFFSET, tx);
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_BYTES, tx);
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_TOGUID, tx);
+ (void) zap_remove(dp->dp_meta_objset, ds->ds_object,
+ DS_FIELD_RESUME_TONAME, tx);
+ }
+ drc->drc_newsnapobj =
+ dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj;
}
- drc->drc_newsnapobj = dsl_dataset_phys(drc->drc_ds)->ds_prev_snap_obj;
zvol_create_minors(dp->dp_spa, drc->drc_tofs, B_TRUE);
+
/*
* Release the hold from dmu_recv_begin. This must be done before
- * we return to open context, so that when we free the dataset's dnode,
- * we can evict its bonus buffer.
+ * we return to open context, so that when we free the dataset's dnode
+ * we can evict its bonus buffer. Since the dataset may be destroyed
+ * at this point (and therefore won't have a valid pointer to the spa)
+ * we release the key mapping manually here while we do have a valid
+ * pointer, if it exists.
*/
- dsl_dataset_disown(drc->drc_ds, dmu_recv_tag);
+ if (!drc->drc_raw && encrypted) {
+ (void) spa_keystore_remove_mapping(dmu_tx_pool(tx)->dp_spa,
+ drc->drc_ds->ds_object, drc->drc_ds);
+ }
+ dsl_dataset_disown(drc->drc_ds, 0, dmu_recv_tag);
drc->drc_ds = NULL;
}
static int
-add_ds_to_guidmap(const char *name, avl_tree_t *guid_map, uint64_t snapobj)
+add_ds_to_guidmap(const char *name, avl_tree_t *guid_map, uint64_t snapobj,
+ boolean_t raw)
{
dsl_pool_t *dp;
dsl_dataset_t *snapds;
guid_map_entry_t *gmep;
+ ds_hold_flags_t dsflags = (raw) ? 0 : DS_HOLD_FLAG_DECRYPT;
int err;
ASSERT(guid_map != NULL);
if (err != 0)
return (err);
gmep = kmem_alloc(sizeof (*gmep), KM_SLEEP);
- err = dsl_dataset_hold_obj(dp, snapobj, gmep, &snapds);
+ err = dsl_dataset_hold_obj_flags(dp, snapobj, dsflags, gmep, &snapds);
if (err == 0) {
gmep->guid = dsl_dataset_phys(snapds)->ds_guid;
+ gmep->raw = raw;
gmep->gme_ds = snapds;
avl_add(guid_map, gmep);
dsl_dataset_long_hold(snapds, gmep);
static int
dmu_recv_existing_end(dmu_recv_cookie_t *drc)
{
- int error;
-
#ifdef _KERNEL
- char *name;
-
/*
* We will be destroying the ds; make sure its origin is unmounted if
* necessary.
*/
- name = kmem_alloc(MAXNAMELEN, KM_SLEEP);
+ char name[ZFS_MAX_DATASET_NAME_LEN];
dsl_dataset_name(drc->drc_ds, name);
zfs_destroy_unmount_origin(name);
- kmem_free(name, MAXNAMELEN);
#endif
- error = dsl_sync_task(drc->drc_tofs,
+ return (dsl_sync_task(drc->drc_tofs,
dmu_recv_end_check, dmu_recv_end_sync, drc,
- dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL);
-
- if (error != 0)
- dmu_recv_cleanup_ds(drc);
- return (error);
+ dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL));
}
static int
dmu_recv_new_end(dmu_recv_cookie_t *drc)
{
- int error;
-
- error = dsl_sync_task(drc->drc_tofs,
+ return (dsl_sync_task(drc->drc_tofs,
dmu_recv_end_check, dmu_recv_end_sync, drc,
- dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL);
-
- if (error != 0) {
- dmu_recv_cleanup_ds(drc);
- } else if (drc->drc_guid_to_ds_map != NULL) {
- (void) add_ds_to_guidmap(drc->drc_tofs,
- drc->drc_guid_to_ds_map,
- drc->drc_newsnapobj);
- }
- return (error);
+ dmu_recv_end_modified_blocks, ZFS_SPACE_CHECK_NORMAL));
}
int
dmu_recv_end(dmu_recv_cookie_t *drc, void *owner)
{
+ int error;
+
drc->drc_owner = owner;
if (drc->drc_newfs)
- return (dmu_recv_new_end(drc));
+ error = dmu_recv_new_end(drc);
else
- return (dmu_recv_existing_end(drc));
+ error = dmu_recv_existing_end(drc);
+
+ if (error != 0) {
+ dmu_recv_cleanup_ds(drc);
+ nvlist_free(drc->drc_keynvl);
+ } else if (drc->drc_guid_to_ds_map != NULL) {
+ (void) add_ds_to_guidmap(drc->drc_tofs, drc->drc_guid_to_ds_map,
+ drc->drc_newsnapobj, drc->drc_raw);
+ }
+ return (error);
}
/*
#if defined(_KERNEL)
module_param(zfs_send_corrupt_data, int, 0644);
MODULE_PARM_DESC(zfs_send_corrupt_data, "Allow sending corrupt data");
+
+module_param(zfs_send_queue_length, int, 0644);
+MODULE_PARM_DESC(zfs_send_queue_length, "Maximum send queue length");
+
+module_param(zfs_recv_queue_length, int, 0644);
+MODULE_PARM_DESC(zfs_recv_queue_length, "Maximum receive queue length");
#endif
projects / mirror_zfs.git / blobdiff