zfs: add an option to the bootloader to rewind the ZFS checkpoint

The checkpoints are another way of keeping the state of ZFS.
During the rewind, the pool has to be exported.
This makes checkpoints unusable when using ZFS as root.
Add the option to rewind the ZFS checkpoint at the boot time.
If checkpoint exists, a new option for rewinding a checkpoint will appear in
the bootloader menu.
We fully support boot environments.
If the rewind option is selected, the boot loader will show a list of
boot environments that existed before the checkpoint.

Reviewed by:	tsoome, allanjude, kevans (ok with high-level overview)
Differential Revision:	https://reviews.freebsd.org/D24920

4 files changed, 28 insertions, 5 deletions

diff --git a/sys/cddl/boot/zfs/zfsimpl.h b/sys/cddl/boot/zfs/zfsimpl.h
index 9ad244d083b9..6a6331450e8b 100644
--- a/sys/cddl/boot/zfs/zfsimpl.h
+++ b/sys/cddl/boot/zfs/zfsimpl.h
@@ -1351,6 +1351,7 @@ typedef struct dsl_dataset_phys {
 #define	DMU_POOL_REMOVING		"com.delphix:removing"
 #define	DMU_POOL_OBSOLETE_BPOBJ		"com.delphix:obsolete_bpobj"
 #define	DMU_POOL_CONDENSING_INDIRECT	"com.delphix:condensing_indirect"
+#define	DMU_POOL_ZPOOL_CHECKPOINT       "com.delphix:zpool_checkpoint"
 
 #define	ZAP_MAGIC 0x2F52AB2ABULL
 
@@ -1814,12 +1815,17 @@ typedef struct spa {
 	char		*spa_name;	/* pool name */
 	uint64_t	spa_guid;	/* pool guid */
 	uint64_t	spa_txg;	/* most recent transaction */
-	struct uberblock spa_uberblock;	/* best uberblock so far */
+	struct uberblock *spa_uberblock;	/* best uberblock so far */
 	vdev_t		*spa_root_vdev;	/* toplevel vdev container */
-	objset_phys_t	spa_mos;	/* MOS for this pool */
+	objset_phys_t	*spa_mos;	/* MOS for this pool */
 	zio_cksum_salt_t spa_cksum_salt;	/* secret salt for cksum */
 	void		*spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
 	boolean_t	spa_with_log;	/* this pool has log */
+
+	struct uberblock spa_uberblock_master;  /* best uberblock so far */
+	objset_phys_t    spa_mos_master;        /* MOS for this pool */
+	struct uberblock spa_uberblock_checkpoint;      /* checkpoint uberblock */
+	objset_phys_t    spa_mos_checkpoint;    /* Checkpoint MOS */
 } spa_t;
 
 /* IO related arguments. */
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
index 1012f1908d02..65ae7904047b 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
@@ -5719,7 +5719,7 @@ spa_generate_rootconf(const char *name)
 }
 
 int
-spa_import_rootpool(const char *name)
+spa_import_rootpool(const char *name, bool checkpointrewind)
 {
 	spa_t *spa;
 	vdev_t *rvd, *bvd, *avd = NULL;
@@ -5778,6 +5778,9 @@ spa_import_rootpool(const char *name)
 	}
 	spa->spa_is_root = B_TRUE;
 	spa->spa_import_flags = ZFS_IMPORT_VERBATIM;
+	if (checkpointrewind) {
+		spa->spa_import_flags |= ZFS_IMPORT_CHECKPOINT;
+	}
 
 	/*
 	 * Build up a vdev tree based on the boot device's label config.
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h
index 925358febd7c..5bdc4feb3d5e 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h
@@ -643,7 +643,7 @@ extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
 #ifdef illumos
 extern int spa_import_rootpool(char *devpath, char *devid);
 #else
-extern int spa_import_rootpool(const char *name);
+extern int spa_import_rootpool(const char *name, bool checkpointrewind);
 #endif
 extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
     uint64_t flags);
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c
index fb6a40ac49b4..1ce186d0862e 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c
@@ -1783,6 +1783,18 @@ getpoolname(const char *osname, char *poolname)
 	return (0);
 }
 
+static void
+fetch_osname_options(char *name, bool *checkpointrewind)
+{
+
+	if (name[0] == '!') {
+		*checkpointrewind = true;
+		memmove(name, name + 1, strlen(name));
+	} else {
+		*checkpointrewind = false;
+	}
+}
+
 /*ARGSUSED*/
 static int
 zfs_mount(vfs_t *vfsp)
@@ -1793,6 +1805,7 @@ zfs_mount(vfs_t *vfsp)
 	char		*osname;
 	int		error = 0;
 	int		canwrite;
+	bool		checkpointrewind;
 
 #ifdef illumos
 	if (mvp->v_type != VDIR)
@@ -1836,6 +1849,7 @@ zfs_mount(vfs_t *vfsp)
 		secpolicy_fs_mount_clearopts(cr, vfsp);
 	}
 #endif	/* illumos */
+	fetch_osname_options(osname, &checkpointrewind);
 
 	/*
 	 * Check for mount privilege?
@@ -1921,7 +1935,7 @@ zfs_mount(vfs_t *vfsp)
 
 		error = getpoolname(osname, pname);
 		if (error == 0)
-			error = spa_import_rootpool(pname);
+			error = spa_import_rootpool(pname, checkpointrewind);
 		if (error)
 			goto out;
 	}