zfsboot/zfsloader: support accessing filesystems within a pool

In zfs loader zfs device name format now is "zfs:pool/fs",
fully qualified file path is "zfs:pool/fs:/path/to/file"
loader allows accessing files from various pools and filesystems as well
as changing currdev to a different pool/filesystem.

zfsboot accepts kernel/loader name in a format pool:fs:path/to/file or,
as before, pool:path/to/file; in the latter case a default filesystem
is used (pool root or bootfs).  zfsboot passes guids of the selected
pool and dataset to zfsloader to be used as its defaults.

zfs support should be architecture independent and is provided
in a separate library, but architectures wishing to use this zfs support
still have to provide some glue code and their devdesc should be
compatible with zfs_devdesc.
arch_zfs_probe method is used to discover all disk devices that may
be part of ZFS pool(s).

libi386 unconditionally includes zfs support, but some zfs-specific
functions are stubbed out as weak symbols.  The strong definitions
are provided in libzfsboot.
This change mean that the size of i386_devspec becomes larger
to match zfs_devspec.

Backward-compatibility shims are provided for recently added sparc64
zfs boot support.  Currently that architecture still works the old
way and does not support the new features.

TODO:
- clear up pool root filesystem vs pool bootfs filesystem distinction
- update sparc64 support
- set vfs.root.mountfrom based on currdev (for zfs)

Mid-future TODO:
- loader sub-menu for selecting alternative boot environment

Distant future TODO:
- support accessing snapshots, using a snapshot as readonly root

Reviewed by:	marius (sparc64),
		Gavin Mu <gavin.mu@gmail.com> (sparc64)
Tested by:	Florian Wagner <florian@wagner-flo.net> (x86),
		marius (sparc64)
No objections:	fs@, hackers@
MFC after:	1 month

1 files changed, 191 insertions, 107 deletions

diff --git a/sys/boot/zfs/zfs.c b/sys/boot/zfs/zfs.c
index 97f924f..c90e1c7 100644
--- a/sys/boot/zfs/zfs.c
+++ b/sys/boot/zfs/zfs.c
@@ -43,9 +43,9 @@ __FBSDID("$FreeBSD$");
 #include <stand.h>
 #include <bootstrap.h>
 
-#include "zfsimpl.c"
+#include "libzfs.h"
 
-#define	MAXBDDEV	31
+#include "zfsimpl.c"
 
 static int	zfs_open(const char *path, struct open_file *f);
 static int	zfs_write(struct open_file *f, void *buf, size_t size, size_t *resid);
@@ -56,6 +56,7 @@ static int	zfs_stat(struct open_file *f, struct stat *sb);
 static int	zfs_readdir(struct open_file *f, struct dirent *d);
 
 struct devsw zfs_dev;
+struct devsw zfs_dev_compat;
 
 struct fs_ops zfs_fsops = {
 	"zfs",
@@ -85,35 +86,20 @@ struct file {
 static int
 zfs_open(const char *upath, struct open_file *f)
 {
-	spa_t *spa = (spa_t *) f->f_devdata;
+	struct zfsmount *mount = (struct zfsmount *)f->f_devdata;
 	struct file *fp;
 	int rc;
 
-	if (f->f_dev != &zfs_dev)
+	if (f->f_dev != &zfs_dev && f->f_dev != &zfs_dev_compat)
 		return (EINVAL);
 
-	rc = zfs_mount_pool(spa);
-	if (rc)
-		return (rc);
-
 	/* allocate file system specific data structure */
 	fp = malloc(sizeof(struct file));
 	bzero(fp, sizeof(struct file));
 	f->f_fsdata = (void *)fp;
 
-	if (spa->spa_root_objset.os_type != DMU_OST_ZFS) {
-		printf("Unexpected object set type %llu\n",
-		    spa->spa_root_objset.os_type);
-		rc = EIO;
-		goto out;
-	}
-
-	rc = zfs_lookup(spa, upath, &fp->f_dnode);
-	if (rc)
-		goto out;
-
+	rc = zfs_lookup(mount, upath, &fp->f_dnode);
 	fp->f_seekp = 0;
-out:
 	if (rc) {
 		f->f_fsdata = NULL;
 		free(fp);
@@ -142,7 +128,7 @@ zfs_close(struct open_file *f)
 static int
 zfs_read(struct open_file *f, void *start, size_t size, size_t *resid	/* out */)
 {
-	spa_t *spa = (spa_t *) f->f_devdata;
+	spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
 	struct file *fp = (struct file *)f->f_fsdata;
 	struct stat sb;
 	size_t n;
@@ -216,7 +202,7 @@ zfs_seek(struct open_file *f, off_t offset, int where)
 static int
 zfs_stat(struct open_file *f, struct stat *sb)
 {
-	spa_t *spa = (spa_t *) f->f_devdata;
+	spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
 	struct file *fp = (struct file *)f->f_fsdata;
 
 	return (zfs_dnode_stat(spa, &fp->f_dnode, sb));
@@ -225,7 +211,7 @@ zfs_stat(struct open_file *f, struct stat *sb)
 static int
 zfs_readdir(struct open_file *f, struct dirent *d)
 {
-	spa_t *spa = (spa_t *) f->f_devdata;
+	spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
 	struct file *fp = (struct file *)f->f_fsdata;
 	mzap_ent_phys_t mze;
 	struct stat sb;
@@ -381,68 +367,33 @@ vdev_read(vdev_t *vdev, void *priv, off_t offset, void *buf, size_t size)
 	}
 }
 
-/*
- * Convert a pool guid to a 'unit number' suitable for use with zfs_dev_open.
- */
-int
-zfs_guid_to_unit(uint64_t guid)
+static int
+zfs_dev_init(void)
 {
-	spa_t *spa;
-	int unit;
-
-	unit = 0;
-	STAILQ_FOREACH(spa, &zfs_pools, spa_link) {
-		if (spa->spa_guid == guid)
-			return unit;
-		unit++;
-	}
-	return (-1);
+	zfs_init();
+	if (archsw.arch_zfs_probe == NULL)
+		return (ENXIO);
+	archsw.arch_zfs_probe();
+	return (0);
 }
 
-#if defined(__amd64__) || defined(__i386__)
-static int
-zfs_dev_init(void) 
+int
+zfs_probe_dev(const char *devname, uint64_t *pool_guid)
 {
-	char devname[512];
-	int unit, slice;
+	spa_t *spa;
 	int fd;
+	int ret;
 
-	/*
-	 * Open all the disks we can find and see if we can reconstruct
-	 * ZFS pools from them. Bogusly assumes that the disks are named
-	 * diskN, diskNpM or diskNsM.
-	 */
-	zfs_init();
-	for (unit = 0; unit < MAXBDDEV; unit++) {
-		sprintf(devname, "disk%d:", unit);
-		fd = open(devname, O_RDONLY);
-		if (fd == -1)
-			continue;
-
-		/*
-		 * If we find a vdev, the zfs code will eat the fd, otherwise
-		 * we close it.
-		 */
-		if (vdev_probe(vdev_read, (void*) (uintptr_t) fd, 0))
-			close(fd);
-
-		for (slice = 1; slice <= 128; slice++) {
-			sprintf(devname, "disk%dp%d:", unit, slice);
-			fd = open(devname, O_RDONLY);
-			if (fd == -1) {
-				sprintf(devname, "disk%ds%d:", unit, slice);
-				fd = open(devname, O_RDONLY);
-				if (fd == -1)
-					continue;
-			}
-			if (vdev_probe(vdev_read, (void*) (uintptr_t) fd, 0))
-				close(fd);
-		}
-	}
-
+	fd = open(devname, O_RDONLY);
+	if (fd == -1)
+		return (ENXIO);
+	ret = vdev_probe(vdev_read, (void *)(uintptr_t)fd, &spa);
+	if (ret != 0)
+		close(fd);
+	else if (pool_guid != NULL)
+		*pool_guid = spa->spa_guid;
 	return (0);
 }
-#endif
 
 /*
  * Print information about ZFS pools
@@ -452,54 +403,85 @@ zfs_dev_print(int verbose)
 {
 	spa_t *spa;
 	char line[80];
-	int unit;
 
 	if (verbose) {
 		spa_all_status();
 		return;
 	}
-	unit = 0;
 	STAILQ_FOREACH(spa, &zfs_pools, spa_link) {
-		sprintf(line, "    zfs%d:   %s\n", unit, spa->spa_name);
+		sprintf(line, "    zfs:%s\n", spa->spa_name);
 		pager_output(line);
-		unit++;
 	}
 }
 
 /*
  * Attempt to open the pool described by (dev) for use by (f).
  */
-static int 
+static int
+zfs_dev_open_spa(struct open_file *f, spa_t *spa, uint64_t root_guid)
+{
+	struct zfsmount	*mount;
+	int		rv;
+
+	rv = zfs_spa_init(spa);
+	if (rv != 0)
+		return (rv);
+	mount = malloc(sizeof(*mount));
+	rv = zfs_mount(spa, root_guid, mount);
+	if (rv != 0) {
+		free(mount);
+		return (rv);
+	}
+	if (mount->objset.os_type != DMU_OST_ZFS) {
+		printf("Unexpected object set type %llu\n",
+		    mount->objset.os_type);
+		free(mount);
+		return (EIO);
+	}
+	f->f_devdata = mount;
+	return (0);
+}
+
+static int
 zfs_dev_open(struct open_file *f, ...)
 {
+	va_list			args;
+	struct zfs_devdesc	*dev;
+	spa_t			*spa;
+	int			rv;
+
+	va_start(args, f);
+	dev = va_arg(args, struct zfs_devdesc *);
+	va_end(args);
+
+	spa = spa_find_by_guid(dev->pool_guid);
+	if (!spa)
+		return (ENXIO);
+	rv = zfs_dev_open_spa(f, spa, dev->root_guid);
+	if (rv != 0)
+		return (rv);
+	free(dev);
+	return (0);
+}
+
+static int
+zfs_dev_open_compat(struct open_file *f, ...)
+{
 	va_list		args;
 	struct devdesc	*dev;
-	int		unit, i;
 	spa_t		*spa;
+	int		rv;
 
 	va_start(args, f);
-	dev = va_arg(args, struct devdesc*);
+	dev = va_arg(args, struct devdesc *);
 	va_end(args);
 
-	/*
-	 * We mostly ignore the stuff that devopen sends us. For now,
-	 * use the unit to find a pool - later we will override the
-	 * devname parsing so that we can name a pool and a fs within
-	 * the pool.
-	 */
-	unit = dev->d_unit;
-	
-	i = 0;
-	STAILQ_FOREACH(spa, &zfs_pools, spa_link) {
-		if (i == unit)
-			break;
-		i++;
-	}
-	if (!spa) {
+	spa = spa_find_by_unit(dev->d_unit);
+	if (!spa)
 		return (ENXIO);
-	}
-
-	f->f_devdata = spa;
+	rv = zfs_dev_open_spa(f, spa, 0);
+	if (rv != 0)
+		return (rv);
 	free(dev);
 	return (0);
 }
@@ -508,6 +490,7 @@ static int
 zfs_dev_close(struct open_file *f)
 {
 
+	free(f->f_devdata);
 	f->f_devdata = NULL;
 	return (0);
 }
@@ -520,13 +503,114 @@ zfs_dev_strategy(void *devdata, int rw, daddr_t dblk, size_t size, char *buf, si
 }
 
 struct devsw zfs_dev = {
-	.dv_name = "zfs", 
-	.dv_type = DEVT_ZFS, 
+	.dv_name = "zfs",
+	.dv_type = DEVT_ZFS,
+	.dv_init = zfs_dev_init,
+	.dv_strategy = zfs_dev_strategy,
+	.dv_open = zfs_dev_open,
+	.dv_close = zfs_dev_close,
+	.dv_ioctl = noioctl,
+	.dv_print = zfs_dev_print,
+	.dv_cleanup = NULL
+};
+
+struct devsw zfs_dev_compat = {
+	.dv_name = "zfs",
+	.dv_type = DEVT_ZFS,
 	.dv_init = zfs_dev_init,
-	.dv_strategy = zfs_dev_strategy, 
-	.dv_open = zfs_dev_open, 
-	.dv_close = zfs_dev_close, 
+	.dv_strategy = zfs_dev_strategy,
+	.dv_open = zfs_dev_open_compat,
+	.dv_close = zfs_dev_close,
 	.dv_ioctl = noioctl,
 	.dv_print = zfs_dev_print,
 	.dv_cleanup = NULL
 };
+
+int
+zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
+{
+	static char	rootname[ZFS_MAXNAMELEN];
+	static char	poolname[ZFS_MAXNAMELEN];
+	spa_t		*spa;
+	const char	*end;
+	const char	*np;
+	const char	*sep;
+	int		rv;
+
+	np = devspec;
+	if (*np != ':')
+		return (EINVAL);
+	np++;
+	end = strchr(np, ':');
+	if (end == NULL)
+		return (EINVAL);
+	sep = strchr(np, '/');
+	if (sep == NULL || sep >= end)
+		sep = end;
+	memcpy(poolname, np, sep - np);
+	poolname[sep - np] = '\0';
+	if (sep < end) {
+		sep++;
+		memcpy(rootname, sep, end - sep);
+		rootname[end - sep] = '\0';
+	}
+	else
+		rootname[0] = '\0';
+
+	spa = spa_find_by_name(poolname);
+	if (!spa)
+		return (ENXIO);
+	rv = zfs_spa_init(spa);
+	if (rv != 0)
+		return (rv);
+	dev->pool_guid = spa->spa_guid;
+	if (rootname[0] != '\0') {
+		rv = zfs_lookup_dataset(spa, rootname, &dev->root_guid);
+		if (rv != 0)
+			return (rv);
+	} else
+		dev->root_guid = 0;
+	if (path != NULL)
+		*path = (*end == '\0') ? end : end + 1;
+	dev->d_dev = &zfs_dev;
+	dev->d_type = zfs_dev.dv_type;
+	return (0);
+}
+
+char *
+zfs_fmtdev(void *vdev)
+{
+	static char		rootname[ZFS_MAXNAMELEN];
+	static char		buf[2 * ZFS_MAXNAMELEN + 8];
+	struct zfs_devdesc	*dev = (struct zfs_devdesc *)vdev;
+	spa_t			*spa;
+
+	buf[0] = '\0';
+	if (dev->d_type != DEVT_ZFS)
+		return (buf);
+
+	spa = spa_find_by_guid(dev->pool_guid);
+	if (spa == NULL) {
+		printf("ZFS: can't find pool by guid\n");
+		return (buf);
+	}
+	if (zfs_spa_init(spa) != 0) {
+		printf("ZFS: can't init pool\n");
+		return (buf);
+	}
+	if (dev->root_guid == 0 && zfs_get_root(spa, &dev->root_guid)) {
+		printf("ZFS: can't find root filesystem\n");
+		return (buf);
+	}
+	if (zfs_rlookup(spa, dev->root_guid, rootname)) {
+		printf("ZFS: can't find filesystem by guid\n");
+		return (buf);
+	}
+
+	if (rootname[0] == '\0')
+		sprintf(buf, "%s:%s:", dev->d_dev->dv_name, spa->spa_name);
+	else
+		sprintf(buf, "%s:%s/%s:", dev->d_dev->dv_name, spa->spa_name,
+		    rootname);
+	return (buf);
+}