d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

MFV 316894

Browse Source 
7252 7628 compressed zfs send / receive

illumos/illumos-gate@5602294fda
5602294fda

https://www.illumos.org/issues/7252
  This feature includes code to allow a system with compressed ARC enabled to
  send data in its compressed form straight out of the ARC, and receive data in
  its compressed form directly into the ARC.

https://www.illumos.org/issues/7628
  We should have longer, more readable versions of the ZFS send / recv options.

7628 create long versions of ZFS send / receive options

Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Reviewed by: David Quigley <dpquigl@davequigley.com>
Reviewed by: Thomas Caputi <tcaputi@datto.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Dan Kimmel <dan.kimmel@delphix.com>
This commit is contained in:
Josh Paetzel 2017-04-25 17:57:43 +00:00
commit c78abb8b50
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=317414
24 changed files with 1376 additions and 752 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

57
cddl/contrib/opensolaris/cmd/zfs/zfs.8
View File

@ -180,12 +180,12 @@
.Ar bookmark
.Nm
.Cm send
.Op Fl DnPpRveL
.Op Fl DLPRcenpv
.Op Fl i Ar snapshot | Fl I Ar snapshot
.Ar snapshot
.Nm
.Cm send
.Op Fl eL
.Op Fl Lce
.Op Fl i Ar snapshot Ns | Ns bookmark
.Ar filesystem Ns | Ns Ar volume Ns | Ns Ar snapshot
.Nm
@ -2535,7 +2535,7 @@ feature.
.It Xo
.Nm
.Cm send
.Op Fl DnPpRveL
.Op Fl DLPRcenpv
.Op Fl i Ar snapshot | Fl I Ar snapshot
.Ar snapshot
.Xc
@ -2580,7 +2580,7 @@ The incremental
source may be specified as with the
.Fl i
option.
.It Fl R
.It Fl R, -replicate
Generate a replication stream package, which will replicate the specified
filesystem, and all descendent file systems, up to the named snapshot. When
received, all properties, snapshots, descendent file systems, and clones are
@ -2598,7 +2598,7 @@ is received. If the
.Fl F
flag is specified when this stream is received, snapshots and file systems that
do not exist on the sending side are destroyed.
.It Fl D
.It Fl D, -dedup
Generate a deduplicated stream. Blocks which would have been sent multiple
times in the send stream will only be sent once. The receiving system must
also support this feature to receive a deduplicated stream. This flag can
@ -2607,7 +2607,7 @@ be used regardless of the dataset's
property, but performance will be much better if the filesystem uses a
dedup-capable checksum (eg.
.Sy sha256 ) .
.It Fl L
.It Fl L, -large-block
Generate a stream which may contain blocks larger than 128KB.
This flag
has no effect if the
@ -2623,7 +2623,7 @@ See
for details on ZFS feature flags and the
.Sy large_blocks
feature.
.It Fl e
.It Fl e, -embed
Generate a more compact stream by using WRITE_EMBEDDED records for blocks
which are stored more compactly on disk by the
.Sy embedded_data
@ -2646,11 +2646,25 @@ See
for details on ZFS feature flags and the
.Sy embedded_data
feature.
.It Fl p
.It Fl c, -compressed
Generate a more compact stream by using compressed WRITE records for blocks
which are compressed on disk and in memory (see the
.Sy compression property for details). If the
.Sy lz4_compress
feature is active on the sending system, then the receiving system must have that
feature enabled as well. If the
.Sy large_blocks
feature is enabled on the sending system but the
.Fl L
option is not supplied in conjunction with
.Fl c
then the data will be decompressed before sending so it can be split
into smaller block sizes.
.It Fl p, -props
Include the dataset's properties in the stream. This flag is implicit when
.Fl R
is specified. The receiving system must also support this feature.
.It Fl n
.It Fl n, -dryrun
Do a dry-run ("No-op") send. Do not generate any actual send data. This is
useful in conjunction with the
.Fl v
@ -2660,9 +2674,9 @@ flags to determine what data will be sent.
In this case, the verbose output will be written to
standard output (contrast with a non-dry-run, where the stream is written
to standard output and the verbose output goes to standard error).
.It Fl P
.It Fl P, -parsable
Print machine-parsable verbose information about the stream package generated.
.It Fl v
.It Fl v, -verbose
Print verbose information about the stream package generated.
This information includes a per-second report of how much data has been sent.
.El
@ -2673,7 +2687,7 @@ on future versions of
.It Xo
.Nm
.Cm send
.Op Fl eL
.Op Fl Lce
.Op Fl i Ar snapshot Ns | Ns Ar bookmark
.Ar filesystem Ns | Ns Ar volume Ns | Ns Ar snapshot
.Xc
@ -2699,7 +2713,7 @@ specified as the last component of the name
If the incremental target is a clone, the incremental source can
be the origin snapshot, or an earlier snapshot in the origin's filesystem,
or the origin's origin, etc.
.It Fl L
.It Fl L, -large-block
Generate a stream which may contain blocks larger than 128KB.
This flag
has no effect if the
@ -2715,7 +2729,22 @@ See
for details on ZFS feature flags and the
.Sy large_blocks
feature.
.It Fl e
.It Fl c, -compressed
Generate a more compact stream by using compressed WRITE records for blocks
which are compressed on disk and in memory (see the
.Sy compression
property for details). If the
.Sy lz4_compress
feature is active on the sending system, then the receiving system must have
that feature enabled as well. If the
.Sy large_blocks
feature is enabled on the sending system but the
.Fl L
option is not supplied in conjunction with
.Fl c
then the data will be decompressed before sending so it can be split
into smaller block sizes.
.It Fl e, -embed
Generate a more compact stream by using WRITE_EMBEDDED records for blocks
which are stored more compactly on disk by the
.Sy embedded_data

27
cddl/contrib/opensolaris/cmd/zfs/zfs_main.c
View File

@ -35,6 +35,7 @@
#include <assert.h>
#include <ctype.h>
#include <errno.h>
#include <getopt.h>
#include <libgen.h>
#include <libintl.h>
#include <libuutil.h>
@ -278,7 +279,7 @@ get_usage(zfs_help_t idx)
case HELP_ROLLBACK:
return (gettext("\trollback [-rRf] <snapshot>\n"));
case HELP_SEND:
return (gettext("\tsend [-DnPpRvLe] [-[iI] snapshot] "
return (gettext("\tsend [-DnPpRvLec] [-[iI] snapshot] "
"<snapshot>\n"
"\tsend [-Le] [-i snapshot|bookmark] "
"<filesystem|volume|snapshot>\n"
@ -3771,8 +3772,23 @@ zfs_do_send(int argc, char **argv)
nvlist_t *dbgnv = NULL;
boolean_t extraverbose = B_FALSE;
struct option long_options[] = {
{"replicate", no_argument, NULL, 'R'},
{"props", no_argument, NULL, 'p'},
{"parsable", no_argument, NULL, 'P'},
{"dedup", no_argument, NULL, 'D'},
{"verbose", no_argument, NULL, 'v'},
{"dryrun", no_argument, NULL, 'n'},
{"large-block", no_argument, NULL, 'L'},
{"embed", no_argument, NULL, 'e'},
{"resume", required_argument, NULL, 't'},
{"compressed", no_argument, NULL, 'c'},
{0, 0, 0, 0}
};
/* check options */
while ((c = getopt(argc, argv, ":i:I:RDpvnPLet:")) != -1) {
while ((c = getopt_long(argc, argv, ":i:I:RbDpvnPLet:c", long_options,
NULL)) != -1) {
switch (c) {
case 'i':
if (fromname)
@ -3816,12 +3832,17 @@ zfs_do_send(int argc, char **argv)
case 't':
resume_token = optarg;
break;
case 'c':
flags.compress = B_TRUE;
break;
case ':':
(void) fprintf(stderr, gettext("missing argument for "
"'%c' option\n"), optopt);
usage(B_FALSE);
break;
case '?':
/*FALLTHROUGH*/
default:
(void) fprintf(stderr, gettext("invalid option '%c'\n"),
optopt);
usage(B_FALSE);
@ -3892,6 +3913,8 @@ zfs_do_send(int argc, char **argv)
lzc_flags |= LZC_SEND_FLAG_LARGE_BLOCK;
if (flags.embed_data)
lzc_flags |= LZC_SEND_FLAG_EMBED_DATA;
if (flags.compress)
lzc_flags |= LZC_SEND_FLAG_COMPRESS;
if (fromname != NULL &&
(fromname[0] == '#' || fromname[0] == '@')) {

30
cddl/contrib/opensolaris/cmd/zstreamdump/zstreamdump.c
View File

@ -25,8 +25,8 @@
*/
/*
* Copyright (c) 2013, 2014 by Delphix. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
* Copyright (c) 2013, 2015 by Delphix. All rights reserved.
*/
#include <ctype.h>
@ -39,6 +39,7 @@
#include <sys/dmu.h>
#include <sys/zfs_ioctl.h>
#include <sys/zio.h>
#include <zfs_fletcher.h>
/*
@ -251,6 +252,7 @@ main(int argc, char *argv[])
(void) fprintf(stderr, "invalid option '%c'\n",
optopt);
usage();
break;
}
}
@ -453,38 +455,50 @@ main(int argc, char *argv[])
drrw->drr_object = BSWAP_64(drrw->drr_object);
drrw->drr_type = BSWAP_32(drrw->drr_type);
drrw->drr_offset = BSWAP_64(drrw->drr_offset);
drrw->drr_length = BSWAP_64(drrw->drr_length);
drrw->drr_logical_size =
BSWAP_64(drrw->drr_logical_size);
drrw->drr_toguid = BSWAP_64(drrw->drr_toguid);
drrw->drr_key.ddk_prop =
BSWAP_64(drrw->drr_key.ddk_prop);
drrw->drr_compressed_size =
BSWAP_64(drrw->drr_compressed_size);
}
uint64_t payload_size = DRR_WRITE_PAYLOAD_SIZE(drrw);
/*
* If this is verbose and/or dump output,
* print info on the modified block
*/
if (verbose) {
(void) printf("WRITE object = %llu type = %u "
"checksum type = %u\n"
" offset = %llu length = %llu "
"checksum type = %u compression type = %u\n"
" offset = %llu logical_size = %llu "
"compressed_size = %llu "
"payload_size = %llu "
"props = %llx\n",
(u_longlong_t)drrw->drr_object,
drrw->drr_type,
drrw->drr_checksumtype,
drrw->drr_compressiontype,
(u_longlong_t)drrw->drr_offset,
(u_longlong_t)drrw->drr_length,
(u_longlong_t)drrw->drr_logical_size,
(u_longlong_t)drrw->drr_compressed_size,
(u_longlong_t)payload_size,
(u_longlong_t)drrw->drr_key.ddk_prop);
}
/*
* Read the contents of the block in from STDIN to buf
*/
(void) ssread(buf, drrw->drr_length, &zc);
(void) ssread(buf, payload_size, &zc);
/*
* If in dump mode
*/
if (dump) {
print_block(buf, drrw->drr_length);
print_block(buf, payload_size);
}
total_write_size += drrw->drr_length;
total_write_size += payload_size;
break;
case DRR_WRITE_BYREF:

3
cddl/contrib/opensolaris/lib/libzfs/common/libzfs.h
View File

@ -616,6 +616,9 @@ typedef struct sendflags {
/* WRITE_EMBEDDED records of type DATA are permitted */
boolean_t embed_data;
/* compressed WRITE records are permitted */
boolean_t compress;
} sendflags_t;
typedef boolean_t (snapfilter_cb_t)(zfs_handle_t *, void *);

51
cddl/contrib/opensolaris/lib/libzfs/common/libzfs_sendrecv.c
View File

@ -354,8 +354,10 @@ cksummer(void *arg)
{
struct drr_write *drrw = &drr->drr_u.drr_write;
dataref_t dataref;
uint64_t payload_size;
(void) ssread(buf, drrw->drr_length, ofp);
payload_size = DRR_WRITE_PAYLOAD_SIZE(drrw);
(void) ssread(buf, payload_size, ofp);
/*
* Use the existing checksum if it's dedup-capable,
@ -369,7 +371,7 @@ cksummer(void *arg)
zio_cksum_t tmpsha256;
SHA256Init(&ctx);
SHA256Update(&ctx, buf, drrw->drr_length);
SHA256Update(&ctx, buf, payload_size);
SHA256Final(&tmpsha256, &ctx);
drrw->drr_key.ddk_cksum.zc_word[0] =
BE_64(tmpsha256.zc_word[0]);
@ -399,7 +401,7 @@ cksummer(void *arg)
wbr_drrr->drr_object = drrw->drr_object;
wbr_drrr->drr_offset = drrw->drr_offset;
wbr_drrr->drr_length = drrw->drr_length;
wbr_drrr->drr_length = drrw->drr_logical_size;
wbr_drrr->drr_toguid = drrw->drr_toguid;
wbr_drrr->drr_refguid = dataref.ref_guid;
wbr_drrr->drr_refobject =
@ -421,7 +423,7 @@ cksummer(void *arg)
goto out;
} else {
/* block not previously seen */
if (dump_record(drr, buf, drrw->drr_length,
if (dump_record(drr, buf, payload_size,
&stream_cksum, outfd) != 0)
goto out;
}
@ -924,7 +926,7 @@ typedef struct send_dump_data {
uint64_t prevsnap_obj;
boolean_t seenfrom, seento, replicate, doall, fromorigin;
boolean_t verbose, dryrun, parsable, progress, embed_data, std_out;
boolean_t large_block;
boolean_t large_block, compress;
int outfd;
boolean_t err;
nvlist_t *fss;
@ -940,7 +942,7 @@ typedef struct send_dump_data {
static int
estimate_ioctl(zfs_handle_t *zhp, uint64_t fromsnap_obj,
boolean_t fromorigin, uint64_t *sizep)
boolean_t fromorigin, enum lzc_send_flags flags, uint64_t *sizep)
{
zfs_cmd_t zc = { 0 };
libzfs_handle_t *hdl = zhp->zfs_hdl;
@ -953,6 +955,7 @@ estimate_ioctl(zfs_handle_t *zhp, uint64_t fromsnap_obj,
zc.zc_sendobj = zfs_prop_get_int(zhp, ZFS_PROP_OBJSETID);
zc.zc_fromobj = fromsnap_obj;
zc.zc_guid = 1; /* estimate flag */
zc.zc_flags = flags;
if (zfs_ioctl(zhp->zfs_hdl, ZFS_IOC_SEND, &zc) != 0) {
char errbuf[1024];
@ -1192,6 +1195,7 @@ dump_snapshot(zfs_handle_t *zhp, void *arg)
progress_arg_t pa = { 0 };
pthread_t tid;
char *thissnap;
enum lzc_send_flags flags = 0;
int err;
boolean_t isfromsnap, istosnap, fromorigin;
boolean_t exclude = B_FALSE;
@ -1220,6 +1224,13 @@ dump_snapshot(zfs_handle_t *zhp, void *arg)
if (istosnap)
sdd->seento = B_TRUE;
if (sdd->large_block)
flags |= LZC_SEND_FLAG_LARGE_BLOCK;
if (sdd->embed_data)
flags |= LZC_SEND_FLAG_EMBED_DATA;
if (sdd->compress)
flags |= LZC_SEND_FLAG_COMPRESS;
if (!sdd->doall && !isfromsnap && !istosnap) {
if (sdd->replicate) {
char *snapname;
@ -1266,7 +1277,7 @@ dump_snapshot(zfs_handle_t *zhp, void *arg)
if (sdd->verbose) {
uint64_t size = 0;
(void) estimate_ioctl(zhp, sdd->prevsnap_obj,
fromorigin, &size);
fromorigin, flags, &size);
send_print_verbose(fout, zhp->zfs_name,
sdd->prevsnap[0] ? sdd->prevsnap : NULL,
@ -1291,12 +1302,6 @@ dump_snapshot(zfs_handle_t *zhp, void *arg)
}
}
enum lzc_send_flags flags = 0;
if (sdd->large_block)
flags |= LZC_SEND_FLAG_LARGE_BLOCK;
if (sdd->embed_data)
flags |= LZC_SEND_FLAG_EMBED_DATA;
err = dump_ioctl(zhp, sdd->prevsnap, sdd->prevsnap_obj,
fromorigin, sdd->outfd, flags, sdd->debugnv);
@ -1602,8 +1607,12 @@ zfs_send_resume(libzfs_handle_t *hdl, sendflags_t *flags, int outfd,
fromguid = 0;
(void) nvlist_lookup_uint64(resume_nvl, "fromguid", &fromguid);
if (flags->largeblock || nvlist_exists(resume_nvl, "largeblockok"))
lzc_flags |= LZC_SEND_FLAG_LARGE_BLOCK;
if (flags->embed_data || nvlist_exists(resume_nvl, "embedok"))
lzc_flags |= LZC_SEND_FLAG_EMBED_DATA;
if (flags->compress || nvlist_exists(resume_nvl, "compressok"))
lzc_flags |= LZC_SEND_FLAG_COMPRESS;
if (guid_to_name(hdl, toname, toguid, B_FALSE, name) != 0) {
if (zfs_dataset_exists(hdl, toname, ZFS_TYPE_DATASET)) {
@ -1636,7 +1645,8 @@ zfs_send_resume(libzfs_handle_t *hdl, sendflags_t *flags, int outfd,
if (flags->verbose) {
uint64_t size = 0;
error = lzc_send_space(zhp->zfs_name, fromname, &size);
error = lzc_send_space(zhp->zfs_name, fromname,
lzc_flags, &size);
if (error == 0)
size = MAX(0, (int64_t)(size - bytes));
send_print_verbose(stderr, zhp->zfs_name, fromname,
@ -1866,6 +1876,7 @@ zfs_send(zfs_handle_t *zhp, const char *fromsnap, const char *tosnap,
sdd.dryrun = flags->dryrun;
sdd.large_block = flags->largeblock;
sdd.embed_data = flags->embed_data;
sdd.compress = flags->compress;
sdd.filter_cb = filter_func;
sdd.filter_cb_arg = cb_arg;
if (debugnvp)
@ -2960,11 +2971,17 @@ recv_skip(libzfs_handle_t *hdl, int fd, boolean_t byteswap)
case DRR_WRITE:
if (byteswap) {
drr->drr_u.drr_write.drr_length =
BSWAP_64(drr->drr_u.drr_write.drr_length);
drr->drr_u.drr_write.drr_logical_size =
BSWAP_64(
drr->drr_u.drr_write.drr_logical_size);
drr->drr_u.drr_write.drr_compressed_size =
BSWAP_64(
drr->drr_u.drr_write.drr_compressed_size);
}
uint64_t payload_size =
DRR_WRITE_PAYLOAD_SIZE(&drr->drr_u.drr_write);
(void) recv_read(hdl, fd, buf,
drr->drr_u.drr_write.drr_length, B_FALSE, NULL);
payload_size, B_FALSE, NULL);
break;
case DRR_SPILL:
if (byteswap) {

11
cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.c
View File

@ -534,6 +534,8 @@ lzc_send_resume(const char *snapname, const char *from, int fd,
fnvlist_add_boolean(args, "largeblockok");
if (flags & LZC_SEND_FLAG_EMBED_DATA)
fnvlist_add_boolean(args, "embedok");
if (flags & LZC_SEND_FLAG_COMPRESS)
fnvlist_add_boolean(args, "compressok");
if (resumeobj != 0 || resumeoff != 0) {
fnvlist_add_uint64(args, "resume_object", resumeobj);
fnvlist_add_uint64(args, "resume_offset", resumeoff);
@ -559,7 +561,8 @@ lzc_send_resume(const char *snapname, const char *from, int fd,
* an equivalent snapshot.
*/
int
lzc_send_space(const char *snapname, const char *from, uint64_t *spacep)
lzc_send_space(const char *snapname, const char *from,
enum lzc_send_flags flags, uint64_t *spacep)
{
nvlist_t *args;
nvlist_t *result;
@ -568,6 +571,12 @@ lzc_send_space(const char *snapname, const char *from, uint64_t *spacep)
args = fnvlist_alloc();
if (from != NULL)
fnvlist_add_string(args, "from", from);
if (flags & LZC_SEND_FLAG_LARGE_BLOCK)
fnvlist_add_boolean(args, "largeblockok");
if (flags & LZC_SEND_FLAG_EMBED_DATA)
fnvlist_add_boolean(args, "embedok");
if (flags & LZC_SEND_FLAG_COMPRESS)
fnvlist_add_boolean(args, "compressok");
err = lzc_ioctl(ZFS_IOC_SEND_SPACE, snapname, args, &result);
nvlist_free(args);
if (err == 0)

5
cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.h
View File

@ -62,13 +62,14 @@ int lzc_get_holds(const char *, nvlist_t **);
enum lzc_send_flags {
LZC_SEND_FLAG_EMBED_DATA = 1 << 0,
LZC_SEND_FLAG_LARGE_BLOCK = 1 << 1
LZC_SEND_FLAG_LARGE_BLOCK = 1 << 1,
LZC_SEND_FLAG_COMPRESS = 1 << 2
};
int lzc_send(const char *, const char *, int, enum lzc_send_flags);
int lzc_send_resume(const char *, const char *, int,
enum lzc_send_flags, uint64_t, uint64_t);
int lzc_send_space(const char *, const char *, uint64_t *);
int lzc_send_space(const char *, const char *, enum lzc_send_flags, uint64_t *);
struct dmu_replay_record;

1066
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c
View File

File diff suppressed because it is too large Load Diff

171
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c
View File

@ -850,7 +850,7 @@ dbuf_loan_arcbuf(dmu_buf_impl_t *db)
spa_t *spa = db->db_objset->os_spa;
mutex_exit(&db->db_mtx);
abuf = arc_loan_buf(spa, blksz);
abuf = arc_loan_buf(spa, B_FALSE, blksz);
bcopy(db->db.db_data, abuf->b_data, blksz);
} else {
abuf = db->db_buf;
@ -973,8 +973,8 @@ dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
BP_IS_HOLE(db->db_blkptr)))) {
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
dbuf_set_data(db, arc_alloc_buf(db->db_objset->os_spa,
db->db.db_size, db, type));
dbuf_set_data(db, arc_alloc_buf(db->db_objset->os_spa, db, type,
db->db.db_size));
bzero(db->db.db_data, db->db.db_size);
if (db->db_blkptr != NULL && db->db_level > 0 &&
@ -1023,6 +1023,68 @@ dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
&aflags, &zb);
}
/*
* This is our just-in-time copy function. It makes a copy of buffers that
* have been modified in a previous transaction group before we access them in
* the current active group.
*
* This function is used in three places: when we are dirtying a buffer for the
* first time in a txg, when we are freeing a range in a dnode that includes
* this buffer, and when we are accessing a buffer which was received compressed
* and later referenced in a WRITE_BYREF record.
*
* Note that when we are called from dbuf_free_range() we do not put a hold on
* the buffer, we just traverse the active dbuf list for the dnode.
*/
static void
dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg)
{
dbuf_dirty_record_t *dr = db->db_last_dirty;
ASSERT(MUTEX_HELD(&db->db_mtx));
ASSERT(db->db.db_data != NULL);
ASSERT(db->db_level == 0);
ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT);
if (dr == NULL ||
(dr->dt.dl.dr_data !=
((db->db_blkid == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf)))
return;
/*
* If the last dirty record for this dbuf has not yet synced
* and its referencing the dbuf data, either:
* reset the reference to point to a new copy,
* or (if there a no active holders)
* just null out the current db_data pointer.
*/
ASSERT(dr->dr_txg >= txg - 2);
if (db->db_blkid == DMU_BONUS_BLKID) {
/* Note that the data bufs here are zio_bufs */
dr->dt.dl.dr_data = zio_buf_alloc(DN_MAX_BONUSLEN);
arc_space_consume(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
bcopy(db->db.db_data, dr->dt.dl.dr_data, DN_MAX_BONUSLEN);
} else if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
int size = arc_buf_size(db->db_buf);
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
spa_t *spa = db->db_objset->os_spa;
enum zio_compress compress_type =
arc_get_compression(db->db_buf);
if (compress_type == ZIO_COMPRESS_OFF) {
dr->dt.dl.dr_data = arc_alloc_buf(spa, db, type, size);
} else {
ASSERT3U(type, ==, ARC_BUFC_DATA);
dr->dt.dl.dr_data = arc_alloc_compressed_buf(spa, db,
size, arc_buf_lsize(db->db_buf), compress_type);
}
bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size);
} else {
db->db_buf = NULL;
dbuf_clear_data(db);
}
}
int
dbuf_read(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
{
@ -1051,6 +1113,18 @@ dbuf_read(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags)
mutex_enter(&db->db_mtx);
if (db->db_state == DB_CACHED) {
/*
* If the arc buf is compressed, we need to decompress it to
* read the data. This could happen during the "zfs receive" of
* a stream which is compressed and deduplicated.
*/
if (db->db_buf != NULL &&
arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF) {
dbuf_fix_old_data(db,
spa_syncing_txg(dmu_objset_spa(db->db_objset)));
err = arc_decompress(db->db_buf);
dbuf_set_data(db, db->db_buf);
}
mutex_exit(&db->db_mtx);
if (prefetch)
dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE);
@ -1126,7 +1200,7 @@ dbuf_noread(dmu_buf_impl_t *db)
ASSERT(db->db_buf == NULL);
ASSERT(db->db.db_data == NULL);
dbuf_set_data(db, arc_alloc_buf(spa, db->db.db_size, db, type));
dbuf_set_data(db, arc_alloc_buf(spa, db, type, db->db.db_size));
db->db_state = DB_FILL;
} else if (db->db_state == DB_NOFILL) {
dbuf_clear_data(db);
@ -1136,60 +1210,6 @@ dbuf_noread(dmu_buf_impl_t *db)
mutex_exit(&db->db_mtx);
}
/*
* This is our just-in-time copy function. It makes a copy of
* buffers, that have been modified in a previous transaction
* group, before we modify them in the current active group.
*
* This function is used in two places: when we are dirtying a
* buffer for the first time in a txg, and when we are freeing
* a range in a dnode that includes this buffer.
*
* Note that when we are called from dbuf_free_range() we do
* not put a hold on the buffer, we just traverse the active
* dbuf list for the dnode.
*/
static void
dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg)
{
dbuf_dirty_record_t *dr = db->db_last_dirty;
ASSERT(MUTEX_HELD(&db->db_mtx));
ASSERT(db->db.db_data != NULL);
ASSERT(db->db_level == 0);
ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT);
if (dr == NULL ||
(dr->dt.dl.dr_data !=
((db->db_blkid == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf)))
return;
/*
* If the last dirty record for this dbuf has not yet synced
* and its referencing the dbuf data, either:
* reset the reference to point to a new copy,
* or (if there a no active holders)
* just null out the current db_data pointer.
*/
ASSERT(dr->dr_txg >= txg - 2);
if (db->db_blkid == DMU_BONUS_BLKID) {
/* Note that the data bufs here are zio_bufs */
dr->dt.dl.dr_data = zio_buf_alloc(DN_MAX_BONUSLEN);
arc_space_consume(DN_MAX_BONUSLEN, ARC_SPACE_OTHER);
bcopy(db->db.db_data, dr->dt.dl.dr_data, DN_MAX_BONUSLEN);
} else if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
int size = db->db.db_size;
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
spa_t *spa = db->db_objset->os_spa;
dr->dt.dl.dr_data = arc_alloc_buf(spa, size, db, type);
bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size);
} else {
db->db_buf = NULL;
dbuf_clear_data(db);
}
}
void
dbuf_unoverride(dbuf_dirty_record_t *dr)
{
@ -1390,7 +1410,7 @@ dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx)
dmu_buf_will_dirty(&db->db, tx);
/* create the data buffer for the new block */
buf = arc_alloc_buf(dn->dn_objset->os_spa, size, db, type);
buf = arc_alloc_buf(dn->dn_objset->os_spa, db, type, size);
/* copy old block data to the new block */
obuf = db->db_buf;
@ -1984,9 +2004,9 @@ dbuf_assign_arcbuf(dmu_buf_impl_t *db, arc_buf_t *buf, dmu_tx_t *tx)
ASSERT(!refcount_is_zero(&db->db_holds));
ASSERT(db->db_blkid != DMU_BONUS_BLKID);
ASSERT(db->db_level == 0);
ASSERT(DBUF_GET_BUFC_TYPE(db) == ARC_BUFC_DATA);
ASSERT3U(dbuf_is_metadata(db), ==, arc_is_metadata(buf));
ASSERT(buf != NULL);
ASSERT(arc_buf_size(buf) == db->db.db_size);
ASSERT(arc_buf_lsize(buf) == db->db.db_size);
ASSERT(tx->tx_txg != 0);
arc_return_buf(buf, db);
@ -2583,8 +2603,8 @@ dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid,
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
dbuf_set_data(db,
arc_alloc_buf(dn->dn_objset->os_spa,
db->db.db_size, db, type));
arc_alloc_buf(dn->dn_objset->os_spa, db, type,
db->db.db_size));
bcopy(dr->dt.dl.dr_data->b_data, db->db.db_data,
db->db.db_size);
}
@ -3133,10 +3153,19 @@ dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx)
* objects only modified in the syncing context (e.g.
* DNONE_DNODE blocks).
*/
int blksz = arc_buf_size(*datap);
int psize = arc_buf_size(*datap);
arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
*datap = arc_alloc_buf(os->os_spa, blksz, db, type);
bcopy(db->db.db_data, (*datap)->b_data, blksz);
enum zio_compress compress_type = arc_get_compression(*datap);
if (compress_type == ZIO_COMPRESS_OFF) {
*datap = arc_alloc_buf(os->os_spa, db, type, psize);
} else {
ASSERT3U(type, ==, ARC_BUFC_DATA);
int lsize = arc_buf_lsize(*datap);
*datap = arc_alloc_compressed_buf(os->os_spa, db,
psize, lsize, compress_type);
}
bcopy(db->db.db_data, (*datap)->b_data, psize);
}
db->db_data_pending = dr;
@ -3541,7 +3570,9 @@ dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx)
wp_flag = WP_SPILL;
wp_flag |= (db->db_state == DB_NOFILL) ? WP_NOFILL : 0;
dmu_write_policy(os, dn, db->db_level, wp_flag, &zp);
dmu_write_policy(os, dn, db->db_level, wp_flag,
(data != NULL && arc_get_compression(data) != ZIO_COMPRESS_OFF) ?
arc_get_compression(data) : ZIO_COMPRESS_INHERIT, &zp);
DB_DNODE_EXIT(db);
/*
@ -3560,8 +3591,8 @@ dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx)
*/
void *contents = (data != NULL) ? data->b_data : NULL;
dr->dr_zio = zio_write(zio, os->os_spa, txg,
&dr->dr_bp_copy, contents, db->db.db_size, &zp,
dr->dr_zio = zio_write(zio, os->os_spa, txg, &dr->dr_bp_copy,
contents, db->db.db_size, db->db.db_size, &zp,
dbuf_write_override_ready, NULL, NULL,
dbuf_write_override_done,
dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
@ -3574,7 +3605,7 @@ dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx)
ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF ||
zp.zp_checksum == ZIO_CHECKSUM_NOPARITY);
dr->dr_zio = zio_write(zio, os->os_spa, txg,
&dr->dr_bp_copy, NULL, db->db.db_size, &zp,
&dr->dr_bp_copy, NULL, db->db.db_size, db->db.db_size, &zp,
dbuf_write_nofill_ready, NULL, NULL,
dbuf_write_nofill_done, db,
ZIO_PRIORITY_ASYNC_WRITE,

42
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c
View File

@ -1069,7 +1069,7 @@ dmu_xuio_add(xuio_t *xuio, arc_buf_t *abuf, offset_t off, size_t n)
int i = priv->next++;
ASSERT(i < priv->cnt);
ASSERT(off + n <= arc_buf_size(abuf));
ASSERT(off + n <= arc_buf_lsize(abuf));
iov = uio->uio_iov + i;
iov->iov_base = (char *)abuf->b_data + off;
iov->iov_len = n;
@ -1487,7 +1487,7 @@ dmu_request_arcbuf(dmu_buf_t *handle, int size)
{
dmu_buf_impl_t *db = (dmu_buf_impl_t *)handle;
return (arc_loan_buf(db->db_objset->os_spa, size));
return (arc_loan_buf(db->db_objset->os_spa, B_FALSE, size));
}
/*
@ -1512,7 +1512,7 @@ dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf,
dmu_buf_impl_t *dbuf = (dmu_buf_impl_t *)handle;
dnode_t *dn;
dmu_buf_impl_t *db;
uint32_t blksz = (uint32_t)arc_buf_size(buf);
uint32_t blksz = (uint32_t)arc_buf_lsize(buf);
uint64_t blkid;
DB_DNODE_ENTER(dbuf);
@ -1525,12 +1525,9 @@ dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf,
/*
* We can only assign if the offset is aligned, the arc buf is the
* same size as the dbuf, and the dbuf is not metadata. It
* can't be metadata because the loaned arc buf comes from the
* user-data kmem arena.
* same size as the dbuf, and the dbuf is not metadata.
*/
if (offset == db->db.db_offset && blksz == db->db.db_size &&
DBUF_GET_BUFC_TYPE(db) == ARC_BUFC_DATA) {
if (offset == db->db.db_offset && blksz == db->db.db_size) {
#ifdef _KERNEL
curthread->td_ru.ru_oublock++;
#ifdef RACCT
@ -1548,6 +1545,10 @@ dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, arc_buf_t *buf,
objset_t *os;
uint64_t object;
/* compressed bufs must always be assignable to their dbuf */
ASSERT3U(arc_get_compression(buf), ==, ZIO_COMPRESS_OFF);
ASSERT(!(buf->b_flags & ARC_BUF_FLAG_COMPRESSED));
DB_DNODE_ENTER(dbuf);
dn = DB_DNODE(dbuf);
os = dn->dn_objset;
@ -1697,8 +1698,8 @@ dmu_sync_late_arrival(zio_t *pio, objset_t *os, dmu_sync_cb_t *done, zgd_t *zgd,
dsa->dsa_zgd = zgd;
dsa->dsa_tx = tx;
zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx),
zgd->zgd_bp, zgd->zgd_db->db_data, zgd->zgd_db->db_size,
zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx), zgd->zgd_bp,
zgd->zgd_db->db_data, zgd->zgd_db->db_size, zgd->zgd_db->db_size,
zp, dmu_sync_late_arrival_ready, NULL,
NULL, dmu_sync_late_arrival_done, dsa, ZIO_PRIORITY_SYNC_WRITE,
ZIO_FLAG_CANFAIL, zb));
@ -1752,7 +1753,8 @@ dmu_sync(zio_t *pio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd)
DB_DNODE_ENTER(db);
dn = DB_DNODE(db);
dmu_write_policy(os, dn, db->db_level, WP_DMU_SYNC, &zp);
dmu_write_policy(os, dn, db->db_level, WP_DMU_SYNC,
ZIO_COMPRESS_INHERIT, &zp);
DB_DNODE_EXIT(db);
/*
@ -1924,7 +1926,8 @@ SYSCTL_INT(_vfs_zfs, OID_AUTO, mdcomp_disable, CTLFLAG_RWTUN,
int zfs_redundant_metadata_most_ditto_level = 2;
void
dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, zio_prop_t *zp)
dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp,
enum zio_compress override_compress, zio_prop_t *zp)
{
dmu_object_type_t type = dn ? dn->dn_type : DMU_OT_OBJSET;
boolean_t ismd = (level > 0 || DMU_OT_IS_METADATA(type) ||
@ -1936,6 +1939,10 @@ dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, zio_prop_t *zp)
boolean_t nopwrite = B_FALSE;
boolean_t dedup_verify = os->os_dedup_verify;
int copies = os->os_copies;
boolean_t lz4_ac = spa_feature_is_active(os->os_spa,
SPA_FEATURE_LZ4_COMPRESS);
IMPLY(override_compress == ZIO_COMPRESS_LZ4, lz4_ac);
/*
* We maintain different write policies for each of the following
@ -2023,7 +2030,16 @@ dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, zio_prop_t *zp)
}
zp->zp_checksum = checksum;
zp->zp_compress = compress;
/*
* If we're writing a pre-compressed buffer, the compression type we use
* must match the data. If it hasn't been compressed yet, then we should
* use the value dictated by the policies above.
*/
zp->zp_compress = override_compress != ZIO_COMPRESS_INHERIT
? override_compress : compress;
ASSERT3U(zp->zp_compress, !=, ZIO_COMPRESS_INHERIT);
zp->zp_type = (wp & WP_SPILL) ? dn->dn_bonustype : type;
zp->zp_level = level;
zp->zp_copies = MIN(copies, spa_max_replication(os->os_spa));

11
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_objset.c
View File

@ -339,9 +339,8 @@ dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
/* Increase the blocksize if we are permitted. */
if (spa_version(spa) >= SPA_VERSION_USERSPACE &&
arc_buf_size(os->os_phys_buf) < sizeof (objset_phys_t)) {
arc_buf_t *buf = arc_alloc_buf(spa,
sizeof (objset_phys_t), &os->os_phys_buf,
ARC_BUFC_METADATA);
arc_buf_t *buf = arc_alloc_buf(spa, &os->os_phys_buf,
ARC_BUFC_METADATA, sizeof (objset_phys_t));
bzero(buf->b_data, sizeof (objset_phys_t));
bcopy(os->os_phys_buf->b_data, buf->b_data,
arc_buf_size(os->os_phys_buf));
@ -354,8 +353,8 @@ dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, blkptr_t *bp,
} else {
int size = spa_version(spa) >= SPA_VERSION_USERSPACE ?
sizeof (objset_phys_t) : OBJSET_OLD_PHYS_SIZE;
os->os_phys_buf = arc_alloc_buf(spa, size,
&os->os_phys_buf, ARC_BUFC_METADATA);
os->os_phys_buf = arc_alloc_buf(spa, &os->os_phys_buf,
ARC_BUFC_METADATA, size);
os->os_phys = os->os_phys_buf->b_data;
bzero(os->os_phys, size);
}
@ -1138,7 +1137,7 @@ dmu_objset_sync(objset_t *os, zio_t *pio, dmu_tx_t *tx)
ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
arc_release(os->os_phys_buf, &os->os_phys_buf);
dmu_write_policy(os, NULL, 0, 0, &zp);
dmu_write_policy(os, NULL, 0, 0, ZIO_COMPRESS_INHERIT, &zp);
zio = arc_write(pio, os->os_spa, tx->tx_txg,
blkptr_copy, os->os_phys_buf, DMU_OS_IS_L2CACHEABLE(os),

231
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
View File

@ -274,8 +274,10 @@ dump_free(dmu_sendarg_t *dsp, uint64_t object, uint64_t offset,
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)
uint64_t object, uint64_t offset, int lsize, int psize, const blkptr_t *bp,
void *data)
{
uint64_t payload_size;
struct drr_write *drrw = &(dsp->dsa_drr->drr_u.drr_write);
/*
@ -286,7 +288,7 @@ dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type,
(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
@ -305,8 +307,26 @@ dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type,
drrw->drr_object = object;
drrw->drr_type = type;
drrw->drr_offset = offset;
drrw->drr_length = blksz;
drrw->drr_toguid = dsp->dsa_toguid;
drrw->drr_logical_size = lsize;
/* only set the compression fields if the buf is compressed */
if (lsize != psize) {
ASSERT(dsp->dsa_featureflags & DMU_BACKUP_FEATURE_COMPRESSED);
ASSERT(!BP_IS_EMBEDDED(bp));
ASSERT(!BP_SHOULD_BYTESWAP(bp));
ASSERT(!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)));
ASSERT3U(BP_GET_COMPRESS(bp), !=, ZIO_COMPRESS_OFF);
ASSERT3S(psize, >, 0);
ASSERT3S(lsize, >=, psize);
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)) {
/*
* There's no pre-computed checksum for partial-block
@ -326,7 +346,7 @@ dump_write(dmu_sendarg_t *dsp, dmu_object_type_t type,
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);
}
@ -501,7 +521,7 @@ 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);
/*
@ -665,18 +685,49 @@ do_dump(dmu_sendarg_t *dsa, struct send_block_record *data)
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);
/*
* 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));
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));
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_compressed)
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) {
if (zfs_send_corrupt_data) {
/* Send a block filled with 0x"zfs badd bloc" */
abuf = arc_alloc_buf(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;
@ -689,21 +740,21 @@ do_dump(dmu_sendarg_t *dsa, struct send_block_record *data)
offset = zb->zb_blkid * blksz;
if (!(dsa->dsa_featureflags &
DMU_BACKUP_FEATURE_LARGE_BLOCKS) &&
blksz > SPA_OLD_MAXBLOCKSIZE) {
if (split_large_blocks) {
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);
}
arc_buf_destroy(abuf, &abuf);
}
@ -730,9 +781,9 @@ get_next_record(bqueue_t *bq, struct send_block_record *data)
*/
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,
uint64_t resumeobj, uint64_t resumeoff,
zfs_bookmark_phys_t *ancestor_zb, boolean_t is_clone,
boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
int outfd, uint64_t resumeobj, uint64_t resumeoff,
#ifdef illumos
vnode_t *vp, offset_t *off)
#else
@ -779,7 +830,15 @@ dmu_send_impl(void *tag, dsl_pool_t *dp, dsl_dataset_t *to_ds,
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;
featureflags |= DMU_BACKUP_FEATURE_LZ4;
}
if (compressok) {
featureflags |= DMU_BACKUP_FEATURE_COMPRESSED;
}
if ((featureflags &
(DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_COMPRESSED)) !=
0 && spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) {
featureflags |= DMU_BACKUP_FEATURE_LZ4;
}
if (resumeobj != 0 || resumeoff != 0) {
@ -929,7 +988,7 @@ dmu_send_impl(void *tag, dsl_pool_t *dp, dsl_dataset_t *to_ds,
int
dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
boolean_t embedok, boolean_t large_block_ok,
boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
#ifdef illumos
int outfd, vnode_t *vp, offset_t *off)
#else
@ -970,10 +1029,10 @@ dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
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, 0, 0, fp, off);
embedok, large_block_ok, compressok, outfd, 0, 0, fp, off);
} else {
err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE,
embedok, large_block_ok, outfd, 0, 0, fp, off);
embedok, large_block_ok, compressok, outfd, 0, 0, fp, off);
}
dsl_dataset_rele(ds, FTAG);
return (err);
@ -981,7 +1040,8 @@ dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
int
dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok,
boolean_t large_block_ok, int outfd, uint64_t resumeobj, uint64_t resumeoff,
boolean_t large_block_ok, boolean_t compressok, int outfd,
uint64_t resumeobj, uint64_t resumeoff,
#ifdef illumos
vnode_t *vp, offset_t *off)
#else
@ -1053,11 +1113,11 @@ dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok,
return (err);
}
err = dmu_send_impl(FTAG, dp, ds, &zb, is_clone,
embedok, large_block_ok,
embedok, large_block_ok, compressok,
outfd, resumeobj, resumeoff, fp, off);
} else {
err = dmu_send_impl(FTAG, dp, ds, NULL, B_FALSE,
embedok, large_block_ok,
embedok, large_block_ok, compressok,
outfd, resumeobj, resumeoff, fp, off);
}
if (owned)
@ -1068,33 +1128,45 @@ dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok,
}
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;
/* Assume all (uncompressed) blocks are recordsize. */
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;
@ -1102,11 +1174,12 @@ dmu_adjust_send_estimate_for_indirects(dsl_dataset_t *ds, uint64_t 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)
{
dsl_pool_t *dp = ds->ds_dir->dd_pool;
int err;
uint64_t size;
uint64_t uncomp, comp;
ASSERT(dsl_pool_config_held(dp));
@ -1125,33 +1198,41 @@ dmu_send_estimate(dsl_dataset_t *ds, dsl_dataset_t *fromds, uint64_t *sizep)
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);
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);
}
@ -1163,16 +1244,16 @@ dmu_calculate_send_traversal(spa_t *spa, zilog_t *zilog, const blkptr_t *bp,
*/
int
dmu_send_estimate_from_txg(dsl_dataset_t *ds, uint64_t from_txg,
uint64_t *sizep)
boolean_t stream_compressed, uint64_t *sizep)
{
dsl_pool_t *dp = ds->ds_dir->dd_pool;
int err;
uint64_t size = 0;
struct calculate_send_arg size = { 0 };
ASSERT(dsl_pool_config_held(dp));
/* tosnap must be a snapshot */
if (!dsl_dataset_is_snapshot(ds))
if (!ds->ds_is_snapshot)
return (SET_ERROR(EINVAL));
/* verify that from_txg is before the provided snapshot was taken */
@ -1189,7 +1270,8 @@ dmu_send_estimate_from_txg(dsl_dataset_t *ds, uint64_t from_txg,
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);
}
@ -1320,14 +1402,14 @@ dmu_recv_begin_check(void *arg, dmu_tx_t *tx)
/*
* 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));
@ -1496,11 +1578,21 @@ dmu_recv_begin_sync(void *arg, dmu_tx_t *tx)
8, 1, &zero, tx));
VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_BYTES,
8, 1, &zero, tx));
if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) &
DMU_BACKUP_FEATURE_LARGE_BLOCKS) {
VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_LARGEBLOCK,
8, 1, &one, tx));
}
if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) &
DMU_BACKUP_FEATURE_EMBED_DATA) {
VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_EMBEDOK,
8, 1, &one, tx));
}
if (DMU_GET_FEATUREFLAGS(drrb->drr_versioninfo) &
DMU_BACKUP_FEATURE_COMPRESSED) {
VERIFY0(zap_add(mos, dsobj, DS_FIELD_RESUME_COMPRESSOK,
8, 1, &one, tx));
}
}
dmu_buf_will_dirty(newds->ds_dbuf, tx);
@ -1556,7 +1648,7 @@ dmu_recv_resume_begin_check(void *arg, dmu_tx_t *tx)
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));
@ -1763,7 +1855,7 @@ struct receive_objnode {
uint64_t object;
};
struct receive_arg {
struct receive_arg {
objset_t *os;
kthread_t *td;
struct file *fp;
@ -1916,10 +2008,11 @@ byteswap_record(dmu_replay_record_t *drr)
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);
@ -2149,7 +2242,7 @@ receive_write(struct receive_writer_arg *rwa, struct drr_write *drrw,
dmu_tx_t *tx;
int err;
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));
@ -2171,7 +2264,7 @@ receive_write(struct receive_writer_arg *rwa, struct drr_write *drrw,
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);
@ -2181,9 +2274,10 @@ receive_write(struct receive_writer_arg *rwa, struct drr_write *drrw,
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));
}
/* use the bonus buf to look up the dnode in dmu_assign_arcbuf */
dmu_buf_t *bonus;
if (dmu_bonus_hold(rwa->os, drrw->drr_object, FTAG, &bonus) != 0)
return (SET_ERROR(EINVAL));
@ -2596,18 +2690,31 @@ receive_read_record(struct receive_arg *ra)
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 (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;
receive_read_prefetch(ra, drrw->drr_object, drrw->drr_offset,
drrw->drr_length);
drrw->drr_logical_size);
return (err);
}
case DRR_WRITE_BYREF:

8
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dataset.c
View File

@ -1875,10 +1875,18 @@ get_receive_resume_stats(dsl_dataset_t *ds, nvlist_t *nv)
DS_FIELD_RESUME_TONAME, 1, sizeof (buf), buf) == 0) {
fnvlist_add_string(token_nv, "toname", buf);
}
if (zap_contains(dp->dp_meta_objset, ds->ds_object,
DS_FIELD_RESUME_LARGEBLOCK) == 0) {
fnvlist_add_boolean(token_nv, "largeblockok");
}
if (zap_contains(dp->dp_meta_objset, ds->ds_object,
DS_FIELD_RESUME_EMBEDOK) == 0) {
fnvlist_add_boolean(token_nv, "embedok");
}
if (zap_contains(dp->dp_meta_objset, ds->ds_object,
DS_FIELD_RESUME_COMPRESSOK) == 0) {
fnvlist_add_boolean(token_nv, "compressok");
}
packed = fnvlist_pack(token_nv, &packed_size);
fnvlist_free(token_nv);
compressed = kmem_alloc(packed_size, KM_SLEEP);

2
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/lz4.c
View File

@ -86,7 +86,7 @@ lz4_decompress(void *s_start, void *d_start, size_t s_len, size_t d_len, int n)
/*
* Returns 0 on success (decompression function returned non-negative)
* and non-zero on failure (decompression function returned negative.
* and non-zero on failure (decompression function returned negative).
*/
return (LZ4_uncompress_unknownOutputSize(&src[sizeof (bufsiz)],
d_start, bufsiz, d_len) < 0);

20
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/arc.h
View File

@ -122,11 +122,17 @@ typedef enum arc_flags
} arc_flags_t;
typedef enum arc_buf_flags {
ARC_BUF_FLAG_SHARED = 1 << 0,
ARC_BUF_FLAG_COMPRESSED = 1 << 1
} arc_buf_flags_t;
struct arc_buf {
arc_buf_hdr_t *b_hdr;
arc_buf_t *b_next;
kmutex_t b_evict_lock;
void *b_data;
arc_buf_flags_t b_flags;
};
typedef enum arc_buf_contents {
@ -150,13 +156,21 @@ typedef enum arc_space_type {
void arc_space_consume(uint64_t space, arc_space_type_t type);
void arc_space_return(uint64_t space, arc_space_type_t type);
arc_buf_t *arc_alloc_buf(spa_t *spa, int32_t size, void *tag,
arc_buf_contents_t type);
arc_buf_t *arc_loan_buf(spa_t *spa, int size);
boolean_t arc_is_metadata(arc_buf_t *buf);
enum zio_compress arc_get_compression(arc_buf_t *buf);
int arc_decompress(arc_buf_t *buf);
arc_buf_t *arc_alloc_buf(spa_t *spa, void *tag, arc_buf_contents_t type,
int32_t size);
arc_buf_t *arc_alloc_compressed_buf(spa_t *spa, void *tag,
uint64_t psize, uint64_t lsize, enum zio_compress compression_type);
arc_buf_t *arc_loan_buf(spa_t *spa, boolean_t is_metadata, int size);
arc_buf_t *arc_loan_compressed_buf(spa_t *spa, uint64_t psize, uint64_t lsize,
enum zio_compress compression_type);
void arc_return_buf(arc_buf_t *buf, void *tag);
void arc_loan_inuse_buf(arc_buf_t *buf, void *tag);
void arc_buf_destroy(arc_buf_t *buf, void *tag);
int arc_buf_size(arc_buf_t *buf);
int arc_buf_lsize(arc_buf_t *buf);
void arc_release(arc_buf_t *buf, void *tag);
int arc_released(arc_buf_t *buf);
void arc_buf_freeze(arc_buf_t *buf);

3
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h
View File

@ -47,6 +47,7 @@
#include <sys/zfs_context.h>
#include <sys/cred.h>
#include <sys/fs/zfs.h>
#include <sys/zio_compress.h>
#include <sys/zio_priority.h>
#ifdef __cplusplus
@ -421,7 +422,7 @@ dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset,
#define WP_SPILL 0x4
void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp,
struct zio_prop *zp);
enum zio_compress compress_override, struct zio_prop *zp);
/*
* The bonus data is accessed more or less like a regular buffer.
* You must dmu_bonus_hold() to get the buffer, which will give you a

9
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu_send.h
View File

@ -41,18 +41,19 @@ struct dmu_replay_record;
extern const char *recv_clone_name;
int dmu_send(const char *tosnap, const char *fromsnap, boolean_t embedok,
boolean_t large_block_ok, int outfd, uint64_t resumeobj, uint64_t resumeoff,
boolean_t large_block_ok, boolean_t compressok, int outfd,
uint64_t resumeobj, uint64_t resumeoff,
#ifdef illumos
struct vnode *vp, offset_t *off);
#else
struct file *fp, offset_t *off);
#endif
int dmu_send_estimate(struct dsl_dataset *ds, struct dsl_dataset *fromds,
uint64_t *sizep);
boolean_t stream_compressed, uint64_t *sizep);
int dmu_send_estimate_from_txg(struct dsl_dataset *ds, uint64_t fromtxg,
uint64_t *sizep);
boolean_t stream_compressed, uint64_t *sizep);
int dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
boolean_t embedok, boolean_t large_block_ok,
boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
#ifdef illumos
int outfd, struct vnode *vp, offset_t *off);
#else

2
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dsl_dataset.h
View File

@ -96,7 +96,9 @@ struct dsl_pool;
#define DS_FIELD_RESUME_OBJECT "com.delphix:resume_object"
#define DS_FIELD_RESUME_OFFSET "com.delphix:resume_offset"
#define DS_FIELD_RESUME_BYTES "com.delphix:resume_bytes"
#define DS_FIELD_RESUME_LARGEBLOCK "com.delphix:resume_largeblockok"
#define DS_FIELD_RESUME_EMBEDOK "com.delphix:resume_embedok"
#define DS_FIELD_RESUME_COMPRESSOK "com.delphix:resume_compressok"
/*
* DS_FLAG_CI_DATASET is set if the dataset contains a file system whose

2
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/refcount.h
View File

@ -105,7 +105,7 @@ typedef struct refcount {
atomic_add_64(&(src)->rc_count, -__tmp); \
atomic_add_64(&(dst)->rc_count, __tmp); \
}
#define refcount_transfer_ownership(rc, current_holder, new_holder)
#define refcount_transfer_ownership(rc, current_holder, new_holder) (void)0
#define refcount_held(rc, holder) ((rc)->rc_count > 0)
#define refcount_not_held(rc, holder) (B_TRUE)

182
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zfs_ioctl.h
View File

@ -87,19 +87,22 @@ typedef enum drr_headertype {
#define DMU_BACKUP_FEATURE_SA_SPILL (1 << 2)
/* flags #3 - #15 are reserved for incompatible closed-source implementations */
#define DMU_BACKUP_FEATURE_EMBED_DATA (1 << 16)
#define DMU_BACKUP_FEATURE_EMBED_DATA_LZ4 (1 << 17)
#define DMU_BACKUP_FEATURE_LZ4 (1 << 17)
/* flag #18 is reserved for a Delphix feature */
#define DMU_BACKUP_FEATURE_LARGE_BLOCKS (1 << 19)
#define DMU_BACKUP_FEATURE_RESUMING (1 << 20)
/* flag #21 is reserved for a Delphix feature */
#define DMU_BACKUP_FEATURE_COMPRESSED (1 << 22)
/*
* Mask of all supported backup features
*/
#define DMU_BACKUP_FEATURE_MASK (DMU_BACKUP_FEATURE_DEDUP | \
DMU_BACKUP_FEATURE_DEDUPPROPS | DMU_BACKUP_FEATURE_SA_SPILL | \
DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_EMBED_DATA_LZ4 | \
DMU_BACKUP_FEATURE_EMBED_DATA | DMU_BACKUP_FEATURE_LZ4 | \
DMU_BACKUP_FEATURE_RESUMING | \
DMU_BACKUP_FEATURE_LARGE_BLOCKS)
DMU_BACKUP_FEATURE_LARGE_BLOCKS | \
DMU_BACKUP_FEATURE_COMPRESSED)
/* Are all features in the given flag word currently supported? */
#define DMU_STREAM_SUPPORTED(x) (!((x) & ~DMU_BACKUP_FEATURE_MASK))
@ -152,89 +155,11 @@ typedef enum dmu_send_resume_token_version {
#define DRR_IS_DEDUP_CAPABLE(flags) ((flags) & DRR_CHECKSUM_DEDUP)
/*
* zfs ioctl command structure
*/
struct drr_begin {
uint64_t drr_magic;
uint64_t drr_versioninfo; /* was drr_version */
uint64_t drr_creation_time;
dmu_objset_type_t drr_type;
uint32_t drr_flags;
uint64_t drr_toguid;
uint64_t drr_fromguid;
char drr_toname[MAXNAMELEN];
};
struct drr_end {
zio_cksum_t drr_checksum;
uint64_t drr_toguid;
};
struct drr_object {
uint64_t drr_object;
dmu_object_type_t drr_type;
dmu_object_type_t drr_bonustype;
uint32_t drr_blksz;
uint32_t drr_bonuslen;
uint8_t drr_checksumtype;
uint8_t drr_compress;
uint8_t drr_pad[6];
uint64_t drr_toguid;
/* bonus content follows */
};
struct drr_freeobjects {
uint64_t drr_firstobj;
uint64_t drr_numobjs;
uint64_t drr_toguid;
};
struct drr_write {
uint64_t drr_object;
dmu_object_type_t drr_type;
uint32_t drr_pad;
uint64_t drr_offset;
uint64_t drr_length;
uint64_t drr_toguid;
uint8_t drr_checksumtype;
uint8_t drr_checksumflags;
uint8_t drr_pad2[6];
ddt_key_t drr_key; /* deduplication key */
/* content follows */
};
struct drr_free {
uint64_t drr_object;
uint64_t drr_offset;
uint64_t drr_length;
uint64_t drr_toguid;
};
struct drr_write_byref {
/* where to put the data */
uint64_t drr_object;
uint64_t drr_offset;
uint64_t drr_length;
uint64_t drr_toguid;
/* where to find the prior copy of the data */
uint64_t drr_refguid;
uint64_t drr_refobject;
uint64_t drr_refoffset;
/* properties of the data */
uint8_t drr_checksumtype;
uint8_t drr_checksumflags;
uint8_t drr_pad2[6];
ddt_key_t drr_key; /* deduplication key */
};
struct drr_spill {
uint64_t drr_object;
uint64_t drr_length;
uint64_t drr_toguid;
uint64_t drr_pad[4]; /* needed for crypto */
/* spill data follows */
};
/* deal with compressed drr_write replay records */
#define DRR_WRITE_COMPRESSED(drrw) ((drrw)->drr_compressiontype != 0)
#define DRR_WRITE_PAYLOAD_SIZE(drrw) \
(DRR_WRITE_COMPRESSED(drrw) ? (drrw)->drr_compressed_size : \
(drrw)->drr_logical_size)
typedef struct dmu_replay_record {
enum {
@ -244,14 +169,83 @@ typedef struct dmu_replay_record {
} drr_type;
uint32_t drr_payloadlen;
union {
struct drr_begin drr_begin;
struct drr_end drr_end;
struct drr_object drr_object;
struct drr_freeobjects drr_freeobjects;
struct drr_write drr_write;
struct drr_free drr_free;
struct drr_write_byref drr_write_byref;
struct drr_spill drr_spill;
struct drr_begin {
uint64_t drr_magic;
uint64_t drr_versioninfo; /* was drr_version */
uint64_t drr_creation_time;
dmu_objset_type_t drr_type;
uint32_t drr_flags;
uint64_t drr_toguid;
uint64_t drr_fromguid;
char drr_toname[MAXNAMELEN];
} drr_begin;
struct drr_end {
zio_cksum_t drr_checksum;
uint64_t drr_toguid;
} drr_end;
struct drr_object {
uint64_t drr_object;
dmu_object_type_t drr_type;
dmu_object_type_t drr_bonustype;
uint32_t drr_blksz;
uint32_t drr_bonuslen;
uint8_t drr_checksumtype;
uint8_t drr_compress;
uint8_t drr_pad[6];
uint64_t drr_toguid;
/* bonus content follows */
} drr_object;
struct drr_freeobjects {
uint64_t drr_firstobj;
uint64_t drr_numobjs;
uint64_t drr_toguid;
} drr_freeobjects;
struct drr_write {
uint64_t drr_object;
dmu_object_type_t drr_type;
uint32_t drr_pad;
uint64_t drr_offset;
uint64_t drr_logical_size;
uint64_t drr_toguid;
uint8_t drr_checksumtype;
uint8_t drr_checksumflags;
uint8_t drr_compressiontype;
uint8_t drr_pad2[5];
/* deduplication key */
ddt_key_t drr_key;
/* only nonzero if drr_compressiontype is not 0 */
uint64_t drr_compressed_size;
/* content follows */
} drr_write;
struct drr_free {
uint64_t drr_object;
uint64_t drr_offset;
uint64_t drr_length;
uint64_t drr_toguid;
} drr_free;
struct drr_write_byref {
/* where to put the data */
uint64_t drr_object;
uint64_t drr_offset;
uint64_t drr_length;
uint64_t drr_toguid;
/* where to find the prior copy of the data */
uint64_t drr_refguid;
uint64_t drr_refobject;
uint64_t drr_refoffset;
/* properties of the data */
uint8_t drr_checksumtype;
uint8_t drr_checksumflags;
uint8_t drr_pad2[6];
ddt_key_t drr_key; /* deduplication key */
} drr_write_byref;
struct drr_spill {
uint64_t drr_object;
uint64_t drr_length;
uint64_t drr_toguid;
uint64_t drr_pad[4]; /* needed for crypto */
/* spill data follows */
} drr_spill;
struct drr_write_embedded {
uint64_t drr_object;
uint64_t drr_offset;

24
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio.h
View File

@ -106,26 +106,6 @@ enum zio_checksum {
#define ZIO_DEDUPCHECKSUM ZIO_CHECKSUM_SHA256
#define ZIO_DEDUPDITTO_MIN 100
enum zio_compress {
ZIO_COMPRESS_INHERIT = 0,
ZIO_COMPRESS_ON,
ZIO_COMPRESS_OFF,
ZIO_COMPRESS_LZJB,
ZIO_COMPRESS_EMPTY,
ZIO_COMPRESS_GZIP_1,
ZIO_COMPRESS_GZIP_2,
ZIO_COMPRESS_GZIP_3,
ZIO_COMPRESS_GZIP_4,
ZIO_COMPRESS_GZIP_5,
ZIO_COMPRESS_GZIP_6,
ZIO_COMPRESS_GZIP_7,
ZIO_COMPRESS_GZIP_8,
ZIO_COMPRESS_GZIP_9,
ZIO_COMPRESS_ZLE,
ZIO_COMPRESS_LZ4,
ZIO_COMPRESS_FUNCTIONS
};
/*
* The number of "legacy" compression functions which can be set on individual
* objects.
@ -455,6 +435,8 @@ struct zio {
void *io_orig_data;
uint64_t io_size;
uint64_t io_orig_size;
/* io_lsize != io_orig_size iff this is a raw write */
uint64_t io_lsize;
/* Stuff for the vdev stack */
vdev_t *io_vd;
@ -516,7 +498,7 @@ extern zio_t *zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp, void *data,
zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb);
extern zio_t *zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp,
void *data, uint64_t size, const zio_prop_t *zp,
void *data, uint64_t size, uint64_t psize, const zio_prop_t *zp,
zio_done_func_t *ready, zio_done_func_t *children_ready,
zio_done_func_t *physdone, zio_done_func_t *done,
void *priv, zio_priority_t priority, enum zio_flag flags,

23
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio_compress.h
View File

@ -25,17 +25,36 @@
*/
/*
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2015 by Delphix. All rights reserved.
*/
#ifndef _SYS_ZIO_COMPRESS_H
#define _SYS_ZIO_COMPRESS_H
#include <sys/zio.h>
#ifdef __cplusplus
extern "C" {
#endif
enum zio_compress {
ZIO_COMPRESS_INHERIT = 0,
ZIO_COMPRESS_ON,
ZIO_COMPRESS_OFF,
ZIO_COMPRESS_LZJB,
ZIO_COMPRESS_EMPTY,
ZIO_COMPRESS_GZIP_1,
ZIO_COMPRESS_GZIP_2,
ZIO_COMPRESS_GZIP_3,
ZIO_COMPRESS_GZIP_4,
ZIO_COMPRESS_GZIP_5,
ZIO_COMPRESS_GZIP_6,
ZIO_COMPRESS_GZIP_7,
ZIO_COMPRESS_GZIP_8,
ZIO_COMPRESS_GZIP_9,
ZIO_COMPRESS_ZLE,
ZIO_COMPRESS_LZ4,
ZIO_COMPRESS_FUNCTIONS
};
/* Common signature for all zio compress functions. */
typedef size_t zio_compress_func_t(void *src, void *dst,
size_t s_len, size_t d_len, int);

37
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c
View File

@ -4582,6 +4582,7 @@ zfs_ioc_send(zfs_cmd_t *zc)
boolean_t estimate = (zc->zc_guid != 0);
boolean_t embedok = (zc->zc_flags & 0x1);
boolean_t large_block_ok = (zc->zc_flags & 0x2);
boolean_t compressok = (zc->zc_flags & 0x4);
if (zc->zc_obj != 0) {
dsl_pool_t *dp;
@ -4629,7 +4630,7 @@ zfs_ioc_send(zfs_cmd_t *zc)
}
}
error = dmu_send_estimate(tosnap, fromsnap,
error = dmu_send_estimate(tosnap, fromsnap, compressok,
&zc->zc_objset_type);
if (fromsnap != NULL)
@ -4651,7 +4652,7 @@ zfs_ioc_send(zfs_cmd_t *zc)
off = fp->f_offset;
error = dmu_send_obj(zc->zc_name, zc->zc_sendobj,
zc->zc_fromobj, embedok, large_block_ok,
zc->zc_fromobj, embedok, large_block_ok, compressok,
#ifdef illumos
zc->zc_cookie, fp->f_vnode, &off);
#else
@ -5632,6 +5633,8 @@ zfs_ioc_unjail(zfs_cmd_t *zc)
* indicates that blocks > 128KB are permitted
* (optional) "embedok" -> (value ignored)
* presence indicates DRR_WRITE_EMBEDDED records are permitted
* (optional) "compressok" -> (value ignored)
* presence indicates compressed DRR_WRITE records are permitted
* (optional) "resume_object" and "resume_offset" -> (uint64)
* if present, resume send stream from specified object and offset.
* }
@ -5650,6 +5653,7 @@ zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
int fd;
boolean_t largeblockok;
boolean_t embedok;
boolean_t compressok;
uint64_t resumeobj = 0;
uint64_t resumeoff = 0;
@ -5661,6 +5665,7 @@ zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
largeblockok = nvlist_exists(innvl, "largeblockok");
embedok = nvlist_exists(innvl, "embedok");
compressok = nvlist_exists(innvl, "compressok");
(void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj);
(void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff);
@ -5674,11 +5679,11 @@ zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
return (SET_ERROR(EBADF));
off = fp->f_offset;
error = dmu_send(snapname, fromname, embedok, largeblockok, fd,
error = dmu_send(snapname, fromname, embedok, largeblockok, compressok,
#ifdef illumos
resumeobj, resumeoff, fp->f_vnode, &off);
fd, resumeobj, resumeoff, fp->f_vnode, &off);
#else
resumeobj, resumeoff, fp, &off);
fd, resumeobj, resumeoff, fp, &off);
#endif
#ifdef illumos
@ -5699,6 +5704,12 @@ zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
* innvl: {
* (optional) "from" -> full snap or bookmark name to send an incremental
* from
* (optional) "largeblockok" -> (value ignored)
* indicates that blocks > 128KB are permitted
* (optional) "embedok" -> (value ignored)
* presence indicates DRR_WRITE_EMBEDDED records are permitted
* (optional) "compressok" -> (value ignored)
* presence indicates compressed DRR_WRITE records are permitted
* }
*
* outnvl: {
@ -5712,6 +5723,11 @@ zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
dsl_dataset_t *tosnap;
int error;
char *fromname;
/* LINTED E_FUNC_SET_NOT_USED */
boolean_t largeblockok;
/* LINTED E_FUNC_SET_NOT_USED */
boolean_t embedok;
boolean_t compressok;
uint64_t space;
error = dsl_pool_hold(snapname, FTAG, &dp);
@ -5724,6 +5740,10 @@ zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
return (error);
}
largeblockok = nvlist_exists(innvl, "largeblockok");
embedok = nvlist_exists(innvl, "embedok");
compressok = nvlist_exists(innvl, "compressok");
error = nvlist_lookup_string(innvl, "from", &fromname);
if (error == 0) {
if (strchr(fromname, '@') != NULL) {
@ -5736,7 +5756,8 @@ zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
error = dsl_dataset_hold(dp, fromname, FTAG, &fromsnap);
if (error != 0)
goto out;
error = dmu_send_estimate(tosnap, fromsnap, &space);
error = dmu_send_estimate(tosnap, fromsnap, compressok,
&space);
dsl_dataset_rele(fromsnap, FTAG);
} else if (strchr(fromname, '#') != NULL) {
/*
@ -5751,7 +5772,7 @@ zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
if (error != 0)
goto out;
error = dmu_send_estimate_from_txg(tosnap,
frombm.zbm_creation_txg, &space);
frombm.zbm_creation_txg, compressok, &space);
} else {
/*
* from is not properly formatted as a snapshot or
@ -5762,7 +5783,7 @@ zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl)
}
} else {
// If estimating the size of a full send, use dmu_send_estimate
error = dmu_send_estimate(tosnap, NULL, &space);
error = dmu_send_estimate(tosnap, NULL, compressok, &space);
}
fnvlist_add_uint64(outnvl, "space", space);

111
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c
View File

@ -613,21 +613,23 @@ zio_timestamp_compare(const void *x1, const void *x2)
*/
static zio_t *
zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp,
void *data, uint64_t size, zio_done_func_t *done, void *private,
zio_type_t type, zio_priority_t priority, enum zio_flag flags,
vdev_t *vd, uint64_t offset, const zbookmark_phys_t *zb,
enum zio_stage stage, enum zio_stage pipeline)
void *data, uint64_t lsize, uint64_t psize, zio_done_func_t *done,
void *private, zio_type_t type, zio_priority_t priority,
enum zio_flag flags, vdev_t *vd, uint64_t offset,
const zbookmark_phys_t *zb, enum zio_stage stage, enum zio_stage pipeline)
{
zio_t *zio;
ASSERT3U(type == ZIO_TYPE_FREE || size, <=, SPA_MAXBLOCKSIZE);
ASSERT(P2PHASE(size, SPA_MINBLOCKSIZE) == 0);
ASSERT3U(type == ZIO_TYPE_FREE || psize, <=, SPA_MAXBLOCKSIZE);
ASSERT(P2PHASE(psize, SPA_MINBLOCKSIZE) == 0);
ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0);
ASSERT(!vd || spa_config_held(spa, SCL_STATE_ALL, RW_READER));
ASSERT(!bp || !(flags & ZIO_FLAG_CONFIG_WRITER));
ASSERT(vd || stage == ZIO_STAGE_OPEN);
IMPLY(lsize != psize, (flags & ZIO_FLAG_RAW) != 0);
zio = kmem_cache_alloc(zio_cache, KM_SLEEP);
bzero(zio, sizeof (zio_t));
@ -671,7 +673,8 @@ zio_create(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp,
zio->io_vd = vd;
zio->io_offset = offset;
zio->io_orig_data = zio->io_data = data;
zio->io_orig_size = zio->io_size = size;
zio->io_orig_size = zio->io_size = psize;
zio->io_lsize = lsize;
zio->io_orig_flags = zio->io_flags = flags;
zio->io_orig_stage = zio->io_stage = stage;
zio->io_orig_pipeline = zio->io_pipeline = pipeline;
@ -711,7 +714,7 @@ zio_null(zio_t *pio, spa_t *spa, vdev_t *vd, zio_done_func_t *done,
{
zio_t *zio;
zio = zio_create(pio, spa, 0, NULL, NULL, 0, done, private,
zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private,
ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL,
ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE);
@ -816,7 +819,7 @@ zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp,
zfs_blkptr_verify(spa, bp);
zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp,
data, size, done, private,
data, size, size, done, private,
ZIO_TYPE_READ, priority, flags, NULL, 0, zb,
ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ?
ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE);
@ -826,7 +829,7 @@ zio_read(zio_t *pio, spa_t *spa, const blkptr_t *bp,
zio_t *
zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp,
void *data, uint64_t size, const zio_prop_t *zp,
void *data, uint64_t lsize, uint64_t psize, const zio_prop_t *zp,
zio_done_func_t *ready, zio_done_func_t *children_ready,
zio_done_func_t *physdone, zio_done_func_t *done,
void *private, zio_priority_t priority, enum zio_flag flags,
@ -843,7 +846,7 @@ zio_write(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp,
zp->zp_copies > 0 &&
zp->zp_copies <= spa_max_replication(spa));
zio = zio_create(pio, spa, txg, bp, data, size, done, private,
zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private,
ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb,
ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ?
ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE);
@ -873,7 +876,7 @@ zio_rewrite(zio_t *pio, spa_t *spa, uint64_t txg, blkptr_t *bp, void *data,
{
zio_t *zio;
zio = zio_create(pio, spa, txg, bp, data, size, done, private,
zio = zio_create(pio, spa, txg, bp, data, size, size, done, private,
ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_IO_REWRITE, NULL, 0, zb,
ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE);
@ -959,8 +962,8 @@ zio_free_sync(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp,
flags |= ZIO_FLAG_DONT_QUEUE;
zio = zio_create(pio, spa, txg, bp, NULL, size,
NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW, flags,
NULL, 0, NULL, ZIO_STAGE_OPEN, stage);
BP_GET_PSIZE(bp), NULL, NULL, ZIO_TYPE_FREE, ZIO_PRIORITY_NOW,
flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage);
return (zio);
}
@ -993,8 +996,8 @@ zio_claim(zio_t *pio, spa_t *spa, uint64_t txg, const blkptr_t *bp,
ASSERT(!BP_GET_DEDUP(bp) || !spa_writeable(spa)); /* zdb(1M) */
zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp),
done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW, flags,
NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE);
BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW,
flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE);
ASSERT0(zio->io_queued_timestamp);
return (zio);
@ -1009,8 +1012,8 @@ zio_ioctl(zio_t *pio, spa_t *spa, vdev_t *vd, int cmd, uint64_t offset,
int c;
if (vd->vdev_children == 0) {
zio = zio_create(pio, spa, 0, NULL, NULL, size, done, private,
ZIO_TYPE_IOCTL, priority, flags, vd, offset, NULL,
zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private,
ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL,
ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE);
zio->io_cmd = cmd;
@ -1037,9 +1040,9 @@ zio_read_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size,
offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE);
ASSERT3U(offset + size, <=, vd->vdev_psize);
zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private,
ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset,
NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE);
zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done,
private, ZIO_TYPE_READ, priority, flags | ZIO_FLAG_PHYSICAL, vd,
offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE);
zio->io_prop.zp_checksum = checksum;
@ -1058,9 +1061,9 @@ zio_write_phys(zio_t *pio, vdev_t *vd, uint64_t offset, uint64_t size,
offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE);
ASSERT3U(offset + size, <=, vd->vdev_psize);
zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, done, private,
ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd, offset,
NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE);
zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done,
private, ZIO_TYPE_WRITE, priority, flags | ZIO_FLAG_PHYSICAL, vd,
offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE);
zio->io_prop.zp_checksum = checksum;
@ -1140,7 +1143,7 @@ zio_vdev_child_io(zio_t *pio, blkptr_t *bp, vdev_t *vd, uint64_t offset,
flags &= ~ZIO_FLAG_IO_ALLOCATING;
}
zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size,
zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, size,
done, private, type, priority, flags, vd, offset, &pio->io_bookmark,
ZIO_STAGE_VDEV_IO_START >> 1, pipeline);
ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV);
@ -1162,7 +1165,7 @@ zio_vdev_delegated_io(vdev_t *vd, uint64_t offset, void *data, uint64_t size,
ASSERT(vd->vdev_ops->vdev_op_leaf);
zio = zio_create(NULL, vd->vdev_spa, 0, NULL,
data, size, done, private, type, priority,
data, size, size, done, private, type, priority,
flags | ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_RETRY | ZIO_FLAG_DELEGATED,
vd, offset, NULL,
ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE);
@ -1184,7 +1187,7 @@ zio_trim(zio_t *zio, spa_t *spa, vdev_t *vd, uint64_t offset, uint64_t size)
ASSERT(vd->vdev_ops->vdev_op_leaf);
return (zio_create(zio, spa, 0, NULL, NULL, size, NULL, NULL,
return (zio_create(zio, spa, 0, NULL, NULL, size, size, NULL, NULL,
ZIO_TYPE_FREE, ZIO_PRIORITY_TRIM, ZIO_FLAG_DONT_AGGREGATE |
ZIO_FLAG_CANFAIL | ZIO_FLAG_DONT_PROPAGATE | ZIO_FLAG_DONT_RETRY,
vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_FREE_PHYS_PIPELINE));
@ -1203,8 +1206,11 @@ zio_shrink(zio_t *zio, uint64_t size)
* Note, BP_IS_RAIDZ() assumes no compression.
*/
ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF);
if (!BP_IS_RAIDZ(zio->io_bp))
zio->io_orig_size = zio->io_size = size;
if (!BP_IS_RAIDZ(zio->io_bp)) {
/* we are not doing a raw write */
ASSERT3U(zio->io_size, ==, zio->io_lsize);
zio->io_orig_size = zio->io_size = zio->io_lsize = size;
}
}
/*
@ -1313,10 +1319,12 @@ zio_write_compress(zio_t *zio)
zio_prop_t *zp = &zio->io_prop;
enum zio_compress compress = zp->zp_compress;
blkptr_t *bp = zio->io_bp;
uint64_t lsize = zio->io_size;
uint64_t psize = lsize;
uint64_t lsize = zio->io_lsize;
uint64_t psize = zio->io_size;
int pass = 1;
EQUIV(lsize != psize, (zio->io_flags & ZIO_FLAG_RAW) != 0);
/*
* If our children haven't all reached the ready stage,
* wait for them and then repeat this pipeline stage.
@ -1365,7 +1373,8 @@ zio_write_compress(zio_t *zio)
spa_max_replication(spa)) == BP_GET_NDVAS(bp));
}
if (compress != ZIO_COMPRESS_OFF) {
/* If it's a compressed write that is not raw, compress the buffer. */
if (compress != ZIO_COMPRESS_OFF && psize == lsize) {
void *cbuf = zio_buf_alloc(lsize);
psize = zio_compress_data(compress, zio->io_data, cbuf, lsize);
if (psize == 0 || psize == lsize) {
@ -1416,6 +1425,8 @@ zio_write_compress(zio_t *zio)
zio->io_bp_override = NULL;
*bp = zio->io_bp_orig;
zio->io_pipeline = zio->io_orig_pipeline;
} else {
ASSERT3U(psize, !=, 0);
}
/*
@ -2282,8 +2293,8 @@ zio_write_gang_block(zio_t *pio)
zp.zp_nopwrite = B_FALSE;
zio_t *cio = zio_write(zio, spa, txg, &gbh->zg_blkptr[g],
(char *)pio->io_data + (pio->io_size - resid), lsize, &zp,
zio_write_gang_member_ready, NULL, NULL, NULL,
(char *)pio->io_data + (pio->io_size - resid), lsize, lsize,
&zp, zio_write_gang_member_ready, NULL, NULL, NULL,
&gn->gn_child[g], pio->io_priority,
ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark);
@ -2488,6 +2499,10 @@ static boolean_t
zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde)
{
spa_t *spa = zio->io_spa;
boolean_t do_raw = (zio->io_flags & ZIO_FLAG_RAW);
/* We should never get a raw, override zio */
ASSERT(!(zio->io_bp_override && do_raw));
/*
* Note: we compare the original data, not the transformed data,
@ -2511,6 +2526,7 @@ zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde)
if (ddp->ddp_phys_birth != 0) {
arc_buf_t *abuf = NULL;
arc_flags_t aflags = ARC_FLAG_WAIT;
int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE;
blkptr_t blk = *zio->io_bp;
int error;
@ -2518,10 +2534,26 @@ zio_ddt_collision(zio_t *zio, ddt_t *ddt, ddt_entry_t *dde)
ddt_exit(ddt);
/*
* Intuitively, it would make more sense to compare
* io_data than io_orig_data in the raw case since you
* don't want to look at any transformations that have
* happened to the data. However, for raw I/Os the
* data will actually be the same in io_data and
* io_orig_data, so all we have to do is issue this as
* a raw ARC read.
*/
if (do_raw) {
zio_flags |= ZIO_FLAG_RAW;
ASSERT3U(zio->io_size, ==, zio->io_orig_size);
ASSERT0(bcmp(zio->io_data, zio->io_orig_data,
zio->io_size));
ASSERT3P(zio->io_transform_stack, ==, NULL);
}
error = arc_read(NULL, spa, &blk,
arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ,
ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE,
&aflags, &zio->io_bookmark);
zio_flags, &aflags, &zio->io_bookmark);
if (error == 0) {
if (arc_buf_size(abuf) != zio->io_orig_size ||
@ -2636,6 +2668,7 @@ zio_ddt_write(zio_t *zio)
ASSERT(BP_GET_DEDUP(bp));
ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum);
ASSERT(BP_IS_HOLE(bp) || zio->io_bp_override);
ASSERT(!(zio->io_bp_override && (zio->io_flags & ZIO_FLAG_RAW)));
ddt_enter(ddt);
dde = ddt_lookup(ddt, bp, B_TRUE);
@ -2656,7 +2689,9 @@ zio_ddt_write(zio_t *zio)
BP_ZERO(bp);
} else {
zp->zp_dedup = B_FALSE;
BP_SET_DEDUP(bp, B_FALSE);
}
ASSERT(!BP_GET_DEDUP(bp));
zio->io_pipeline = ZIO_WRITE_PIPELINE;
ddt_exit(ddt);
return (ZIO_PIPELINE_CONTINUE);
@ -2689,7 +2724,7 @@ zio_ddt_write(zio_t *zio)
}
dio = zio_write(zio, spa, txg, bp, zio->io_orig_data,
zio->io_orig_size, &czp, NULL, NULL,
zio->io_orig_size, zio->io_orig_size, &czp, NULL, NULL,
NULL, zio_ddt_ditto_write_done, dde, zio->io_priority,
ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark);
@ -2711,7 +2746,7 @@ zio_ddt_write(zio_t *zio)
ddt_phys_addref(ddp);
} else {
cio = zio_write(zio, spa, txg, bp, zio->io_orig_data,
zio->io_orig_size, zp,
zio->io_orig_size, zio->io_orig_size, zp,
zio_ddt_child_write_ready, NULL, NULL,
zio_ddt_child_write_done, dde, zio->io_priority,
ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark);