Add zstd support to zfs

Signed-off-by: Allan Jude <[email protected]>
Signed-off-by: Sebastian Gottschall <[email protected]>
Signed-off-by: Kjeld Schouten-Lebbing <[email protected]>
Signed-off-by: Michael Niewöhner <[email protected]>

cmd/dbufstat/dbufstat

@@ -343,7 +343,7 @@ def get_compstring(c):
                     "ZIO_COMPRESS_GZIP_6",  "ZIO_COMPRESS_GZIP_7",
                     "ZIO_COMPRESS_GZIP_8",  "ZIO_COMPRESS_GZIP_9",
                     "ZIO_COMPRESS_ZLE",     "ZIO_COMPRESS_LZ4",
-                    "ZIO_COMPRESS_FUNCTION"]
+                    "ZIO_COMPRESS_ZSTD",    "ZIO_COMPRESS_FUNCTION"]
 
     # If "-rr" option is used, don't convert to string representation
     if raw > 1:

configure.ac

@@ -173,6 +173,7 @@ AC_CONFIG_FILES([
 	module/unicode/Makefile
 	module/zcommon/Makefile
 	module/zfs/Makefile
+	module/zstd/Makefile
 	rpm/Makefile
 	rpm/generic/Makefile
 	rpm/generic/zfs-dkms.spec

contrib/zstd/.gitignore

@@ -0,0 +1 @@
+*.o.cmd

copy-builtin

@@ -41,6 +41,8 @@ scripts/make_gitrev.sh || true
 rm -rf "$KERNEL_DIR/include/zfs" "$KERNEL_DIR/fs/zfs"
 cp --recursive include "$KERNEL_DIR/include/zfs"
 cp --recursive module "$KERNEL_DIR/fs/zfs"
+mkdir "$KERNEL_DIR/fs/zfs/contrib"
+cp --recursive contrib/zstd "$KERNEL_DIR/fs/zfs/contrib"
 cp zfs_config.h "$KERNEL_DIR/include/zfs/"
 
 for MODULE in "${MODULES[@]}"
@@ -72,10 +74,16 @@ EOF
 	ZFS_MODULE_CFLAGS += -I$(srctree)/include/zfs/os/linux/zfs
 	ZFS_MODULE_CFLAGS += -I$(srctree)/include/zfs/os/linux/kernel
 	ZFS_MODULE_CFLAGS += -include $(srctree)/include/zfs/zfs_config.h
+	ZFS_MODULE_CFLAGS += -I$(srctree)/fs/zfs/contrib/zstd
+	ZFS_MODULE_CFLAGS += -I$(srctree)/fs/zfs/contrib/zstd/common
+	ZFS_MODULE_CFLAGS += -I$(srctree)/fs/zfs/contrib/zstd/compress
+	ZFS_MODULE_CFLAGS += -I$(srctree)/fs/zfs/contrib/zstd/decompress
+	ZFS_MODULE_CFLAGS += -I$(srctree)/include/zfs/sys/zstd
 	ZFS_MODULE_CFLAGS += -std=gnu99 -Wno-declaration-after-statement
 	ZFS_MODULE_CPPFLAGS  = -D_KERNEL
 	ZFS_MODULE_CPPFLAGS += -UDEBUG -DNDEBUG
-	export ZFS_MODULE_CFLAGS ZFS_MODULE_CPPFLAGS
+	ZSTD_PATH = ../contrib/zstd
+	export ZFS_MODULE_CFLAGS ZFS_MODULE_CPPFLAGS ZSTD_PATH
 
 	obj-$(CONFIG_ZFS) :=
 	EOF

include/sys/zfs_ioctl.h

@@ -105,7 +105,7 @@ typedef enum drr_headertype {
 #define	DMU_BACKUP_FEATURE_COMPRESSED		(1 << 22)
 #define	DMU_BACKUP_FEATURE_LARGE_DNODE		(1 << 23)
 #define	DMU_BACKUP_FEATURE_RAW			(1 << 24)
-/* flag #25 is reserved for the ZSTD compression feature */
+#define	DMU_BACKUP_FEATURE_ZSTD			(1 << 25)
 #define	DMU_BACKUP_FEATURE_HOLDS		(1 << 26)
 
 /*
@@ -117,7 +117,7 @@ typedef enum drr_headertype {
     DMU_BACKUP_FEATURE_RESUMING | DMU_BACKUP_FEATURE_LARGE_BLOCKS | \
     DMU_BACKUP_FEATURE_COMPRESSED | DMU_BACKUP_FEATURE_LARGE_DNODE | \
     DMU_BACKUP_FEATURE_RAW | DMU_BACKUP_FEATURE_HOLDS | \
-	DMU_BACKUP_FEATURE_REDACTED)
+    DMU_BACKUP_FEATURE_REDACTED | DMU_BACKUP_FEATURE_ZSTD)
 
 /* Are all features in the given flag word currently supported? */
 #define	DMU_STREAM_SUPPORTED(x)	(!((x) & ~DMU_BACKUP_FEATURE_MASK))

include/sys/zio.h

@@ -156,6 +156,7 @@ enum zio_encrypt {
 	(compress) == ZIO_COMPRESS_GZIP_8 ||		\
 	(compress) == ZIO_COMPRESS_GZIP_9 ||		\
 	(compress) == ZIO_COMPRESS_ZLE ||		\
+	(compress) == ZIO_COMPRESS_ZSTD ||		\
 	(compress) == ZIO_COMPRESS_ON ||		\
 	(compress) == ZIO_COMPRESS_OFF)
 

include/sys/zio_compress.h

@@ -51,12 +51,68 @@ enum zio_compress {
 	ZIO_COMPRESS_GZIP_9,
 	ZIO_COMPRESS_ZLE,
 	ZIO_COMPRESS_LZ4,
+	ZIO_COMPRESS_ZSTD,
 	ZIO_COMPRESS_FUNCTIONS
 };
 
 #define	ZIO_COMPLEVEL_INHERIT	0
 #define	ZIO_COMPLEVEL_DEFAULT	255
 
+enum zio_zstd_levels {
+	ZIO_ZSTDLVL_INHERIT = 0,
+	ZIO_ZSTDLVL_1,
+#define	ZIO_ZSTD_LEVEL_MIN	ZIO_ZSTDLVL_1
+	ZIO_ZSTDLVL_2,
+	ZIO_ZSTDLVL_3,
+#define	ZIO_ZSTD_LEVEL_DEFAULT	ZIO_ZSTDLVL_3
+	ZIO_ZSTDLVL_4,
+	ZIO_ZSTDLVL_5,
+	ZIO_ZSTDLVL_6,
+	ZIO_ZSTDLVL_7,
+	ZIO_ZSTDLVL_8,
+	ZIO_ZSTDLVL_9,
+	ZIO_ZSTDLVL_10,
+	ZIO_ZSTDLVL_11,
+	ZIO_ZSTDLVL_12,
+	ZIO_ZSTDLVL_13,
+	ZIO_ZSTDLVL_14,
+	ZIO_ZSTDLVL_15,
+	ZIO_ZSTDLVL_16,
+	ZIO_ZSTDLVL_17,
+	ZIO_ZSTDLVL_18,
+	ZIO_ZSTDLVL_19,
+#define	ZIO_ZSTD_LEVEL_MAX	ZIO_ZSTDLVL_19
+#define	ZIO_ZSTDLVL_MAX		ZIO_ZSTDLVL_19
+	ZIO_ZSTDLVL_RESERVE = 101, /* Leave room for new positive levels */
+	ZIO_ZSTDLVL_FAST, /* Fast levels are negative */
+	ZIO_ZSTDLVL_FAST_1,
+#define	ZIO_ZSTD_FAST_LEVEL_DEFAULT	ZIO_ZSTDLVL_FAST_1
+	ZIO_ZSTDLVL_FAST_2,
+	ZIO_ZSTDLVL_FAST_3,
+	ZIO_ZSTDLVL_FAST_4,
+	ZIO_ZSTDLVL_FAST_5,
+	ZIO_ZSTDLVL_FAST_6,
+	ZIO_ZSTDLVL_FAST_7,
+	ZIO_ZSTDLVL_FAST_8,
+	ZIO_ZSTDLVL_FAST_9,
+	ZIO_ZSTDLVL_FAST_10,
+	ZIO_ZSTDLVL_FAST_20,
+	ZIO_ZSTDLVL_FAST_30,
+	ZIO_ZSTDLVL_FAST_40,
+	ZIO_ZSTDLVL_FAST_50,
+	ZIO_ZSTDLVL_FAST_60,
+	ZIO_ZSTDLVL_FAST_70,
+	ZIO_ZSTDLVL_FAST_80,
+	ZIO_ZSTDLVL_FAST_90,
+	ZIO_ZSTDLVL_FAST_100,
+	ZIO_ZSTDLVL_FAST_500,
+	ZIO_ZSTDLVL_FAST_1000,
+#define	ZIO_ZSTDLVL_FAST_MAX	ZIO_ZSTDLVL_FAST_1000
+	ZIO_ZSTDLVL_DEFAULT = 250, /* Allow the default level to change */
+	ZIO_ZSTDLVL_AUTO = 251, /* Reserved for future use */
+	ZIO_ZSTDLVL_LEVELS
+};
+
 /* Forward Declaration to avoid visibility problems */
 struct zio_prop;
 

include/sys/zio_impl.h

@@ -73,9 +73,9 @@ extern "C" {
  * the supported transformations:
  *
  * Compression:
- * ZFS supports three different flavors of compression -- gzip, lzjb, and
- * zle. Compression occurs as part of the write pipeline and is performed
- * in the ZIO_STAGE_WRITE_BP_INIT stage.
+ * ZFS supports five different flavors of compression -- gzip, lzjb, lz4, zle,
+ * and zstd. Compression occurs as part of the write pipeline and is
+ * performed in the ZIO_STAGE_WRITE_BP_INIT stage.
  *
  * Dedup:
  * Dedup reads are handled by the ZIO_STAGE_DDT_READ_START and

include/sys/zstd/Makefile.am

@@ -5,6 +5,7 @@ COMMON_H = \
 	$(top_srcdir)/include/sys/zstd/stdio.h \
 	$(top_srcdir)/include/sys/zstd/stdlib.h \
 	$(top_srcdir)/include/sys/zstd/string.h \
+	$(top_srcdir)/include/sys/zstd/zstd.h \
 	$(top_srcdir)/include/sys/zstd/zstd_kfreebsd.c \
 	$(top_srcdir)/include/sys/zstd/zstd_kfreebsd.h \
 	$(top_srcdir)/include/sys/zstd/zstd_kmalloc.c

include/sys/zstd/zstd.h

@@ -0,0 +1,53 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License, Version 1.0 only
+ * (the "License").  You may not use this file except in compliance
+ * with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+
+/*
+ * Copyright (c) 2016-2018, Klara Systems Inc. All rights reserved.
+ * Copyright (c) 2016-2018, Allan Jude. All rights reserved.
+ * Copyright (c) 2018-2019, Sebastian Gottschall. All rights reserved.
+ * Copyright (c) 2019, Michael Niewöhner. All rights reserved.
+ */
+
+#ifndef	_ZFS_ZSTD_H
+#define	_ZFS_ZSTD_H
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/* (de)init for user space / kernel emulation */
+int zstd_init(void);
+void zstd_fini(void);
+
+size_t zstd_compress(void *s_start, void *d_start, size_t s_len, size_t d_len,
+    int n);
+int zstd_get_level(void *s_start, size_t s_len, uint8_t *level);
+int zstd_decompress_level(void *s_start, void *d_start, size_t s_len,
+    size_t d_len, uint8_t *level);
+int zstd_decompress(void *s_start, void *d_start, size_t s_len, size_t d_len,
+    int n);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif /* _ZFS_ZSTD_H */

include/zfeature_common.h

@@ -68,6 +68,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_SPACEMAP_V2,
 	SPA_FEATURE_ALLOCATION_CLASSES,
 	SPA_FEATURE_RESILVER_DEFER,
+	SPA_FEATURE_ZSTD_COMPRESS,
 	SPA_FEATURE_BOOKMARK_V2,
 	SPA_FEATURE_REDACTION_BOOKMARKS,
 	SPA_FEATURE_REDACTED_DATASETS,

module/Makefile.in

@@ -8,6 +8,7 @@ obj-m += unicode/
 obj-m += zcommon/
 obj-m += zfs/
 obj-m += os/linux/zfs/
+obj-m += zstd/
 
 INSTALL_MOD_DIR ?= extra
 
@@ -18,6 +19,7 @@ ZFS_MODULE_CFLAGS += -I@abs_top_srcdir@/include/os/linux/kernel
 ZFS_MODULE_CFLAGS += -I@abs_top_srcdir@/include/os/linux/spl
 ZFS_MODULE_CFLAGS += -I@abs_top_srcdir@/include/os/linux/zfs
 ZFS_MODULE_CFLAGS += -I@abs_top_srcdir@/include
+
 ZFS_MODULE_CPPFLAGS += -D_KERNEL
 ZFS_MODULE_CPPFLAGS += @KERNEL_DEBUG_CPPFLAGS@
 

module/zcommon/zfeature_common.c

@@ -559,6 +559,15 @@ zpool_feature_init(void)
 	    "com.datto:resilver_defer", "resilver_defer",
 	    "Support for deferring new resilvers when one is already running.",
 	    ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);
+
+	static const spa_feature_t zstd_deps[] = {
+		SPA_FEATURE_EXTENSIBLE_DATASET,
+		SPA_FEATURE_NONE
+	};
+	zfeature_register(SPA_FEATURE_ZSTD_COMPRESS,
+	    "org.freebsd:zstd_compress", "zstd_compress",
+	    "zstd compression algorithm support.",
+	    ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, zstd_deps);
 }
 
 #if defined(_KERNEL)

module/zcommon/zfs_prop.c

@@ -126,6 +126,99 @@ zfs_prop_init(void)
 		{ "gzip-9",	ZIO_COMPRESS_GZIP_9 },
 		{ "zle",	ZIO_COMPRESS_ZLE },
 		{ "lz4",	ZIO_COMPRESS_LZ4 },
+		{ "zstd",	ZIO_COMPRESS_ZSTD },
+		{ "zstd-fast",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_1 << SPA_COMPRESSBITS) },
+		/*
+		 * ZSTD 1-19 are synthetic. We store the compression level in a
+		 * separate hidden property to avoid wasting a large amount of
+		 * space in the ZIO_COMPRESS enum. We do not need to know the
+		 * compression level at decompress time, so it does not need
+		 * to be stored on disk in the block pointer.
+		 */
+		{ "zstd-1",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_1 << SPA_COMPRESSBITS) },
+		{ "zstd-2",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_2 << SPA_COMPRESSBITS) },
+		{ "zstd-3",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_3 << SPA_COMPRESSBITS) },
+		{ "zstd-4",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_4 << SPA_COMPRESSBITS) },
+		{ "zstd-5",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_5 << SPA_COMPRESSBITS) },
+		{ "zstd-6",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_6 << SPA_COMPRESSBITS) },
+		{ "zstd-7",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_7 << SPA_COMPRESSBITS) },
+		{ "zstd-8",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_8 << SPA_COMPRESSBITS) },
+		{ "zstd-9",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_9 << SPA_COMPRESSBITS) },
+		{ "zstd-10",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_10 << SPA_COMPRESSBITS) },
+		{ "zstd-11",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_11 << SPA_COMPRESSBITS) },
+		{ "zstd-12",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_12 << SPA_COMPRESSBITS) },
+		{ "zstd-13",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_13 << SPA_COMPRESSBITS) },
+		{ "zstd-14",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_14 << SPA_COMPRESSBITS) },
+		{ "zstd-15",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_15 << SPA_COMPRESSBITS) },
+		{ "zstd-16",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_16 << SPA_COMPRESSBITS) },
+		{ "zstd-17",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_17 << SPA_COMPRESSBITS) },
+		{ "zstd-18",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_18 << SPA_COMPRESSBITS) },
+		{ "zstd-19",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_19 << SPA_COMPRESSBITS) },
+		/*
+		 * The ZSTD-Fast levels are also synthetic.
+		 */
+		{ "zstd-fast-1",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_1 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-2",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_2 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-3",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_3 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-4",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_4 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-5",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_5 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-6",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_6 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-7",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_7 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-8",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_8 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-9",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_9 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-10",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_10 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-20",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_20 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-30",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_30 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-40",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_40 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-50",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_50 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-60",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_60 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-70",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_70 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-80",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_80 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-90",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_90 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-100",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_100 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-500",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_500 << SPA_COMPRESSBITS) },
+		{ "zstd-fast-1000",	ZIO_COMPRESS_ZSTD |
+		    (ZIO_ZSTDLVL_FAST_1000 << SPA_COMPRESSBITS) },
 		{ NULL }
 	};
 
@@ -323,8 +416,10 @@ zfs_prop_init(void)
 	zprop_register_index(ZFS_PROP_COMPRESSION, "compression",
 	    ZIO_COMPRESS_DEFAULT, PROP_INHERIT,
 	    ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME,
-	    "on | off | lzjb | gzip | gzip-[1-9] | zle | lz4", "COMPRESS",
-	    compress_table);
+	    "on | off | lzjb | gzip | gzip-[1-9] | zle | lz4 | "
+	    "zstd | zstd-[1-19] | "
+	    "zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]",
+	    "COMPRESS", compress_table);
 	zprop_register_index(ZFS_PROP_SNAPDIR, "snapdir", ZFS_SNAPDIR_HIDDEN,
 	    PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
 	    "hidden | visible", "SNAPDIR", snapdir_table);

module/zfs/dmu_recv.c

@@ -591,6 +591,13 @@ dmu_recv_begin_check(void *arg, dmu_tx_t *tx)
 		dsflags |= DS_HOLD_FLAG_DECRYPT;
 	}
 
+	if ((featureflags & DMU_BACKUP_FEATURE_ZSTD) &&
+	    !spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_ZSTD_COMPRESS))
+		return (SET_ERROR(ENOTSUP));
+
+	if (!(DMU_STREAM_SUPPORTED(featureflags)))
+		return (SET_ERROR(ENOTSUP));
+
 	error = dsl_dataset_hold_flags(dp, tofs, dsflags, FTAG, &ds);
 	if (error == 0) {
 		/* target fs already exists; recv into temp clone */