zfs label bootenv should store data as nvlist

nvlist does allow us to support different data types and systems.

To encapsulate user data to/from nvlist, the libzfsbootenv library is
provided.

Reviewed-by: Arvind Sankar <[email protected]>
Reviewed-by: Allan Jude <[email protected]>
Reviewed-by: Paul Dagnelie <[email protected]>
Reviewed-by: Igor Kozhukhov <[email protected]>
Signed-off-by: Toomas Soome <[email protected]>
Closes #10774 

2 files changed, 100 insertions, 36 deletions

diff --git a/module/zfs/vdev_label.c b/module/zfs/vdev_label.c
index 8c7468255..7fab7d0d7 100644
--- a/module/zfs/vdev_label.c
+++ b/module/zfs/vdev_label.c
@@ -149,6 +149,8 @@
 #include <sys/dsl_scan.h>
 #include <sys/abd.h>
 #include <sys/fs/zfs.h>
+#include <sys/byteorder.h>
+#include <sys/zfs_bootenv.h>
 
 /*
  * Basic routines to read and write from a vdev label.
@@ -1233,13 +1235,9 @@ vdev_label_read_bootenv_impl(zio_t *zio, vdev_t *vd, int flags)
 	 * bootloader should have rewritten them all to be the same on boot,
 	 * and any changes we made since boot have been the same across all
 	 * labels.
-	 *
-	 * While grub supports writing to all four labels, other bootloaders
-	 * don't, so we only use the first two labels to store boot
-	 * information.
 	 */
 	if (vd->vdev_ops->vdev_op_leaf && vdev_readable(vd)) {
-		for (int l = 0; l < VDEV_LABELS / 2; l++) {
+		for (int l = 0; l < VDEV_LABELS; l++) {
 			vdev_label_read(zio, vd, l,
 			    abd_alloc_linear(VDEV_PAD_SIZE, B_FALSE),
 			    offsetof(vdev_label_t, vl_be), VDEV_PAD_SIZE,
@@ -1249,14 +1247,15 @@ vdev_label_read_bootenv_impl(zio_t *zio, vdev_t *vd, int flags)
 }
 
 int
-vdev_label_read_bootenv(vdev_t *rvd, nvlist_t *command)
+vdev_label_read_bootenv(vdev_t *rvd, nvlist_t *bootenv)
 {
+	nvlist_t *config;
 	spa_t *spa = rvd->vdev_spa;
 	abd_t *abd = NULL;
 	int flags = ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_CANFAIL |
 	    ZIO_FLAG_SPECULATIVE | ZIO_FLAG_TRYHARD;
 
-	ASSERT(command);
+	ASSERT(bootenv);
 	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
 
 	zio_t *zio = zio_root(spa, NULL, &abd, flags);
@@ -1264,39 +1263,81 @@ vdev_label_read_bootenv(vdev_t *rvd, nvlist_t *command)
 	int err = zio_wait(zio);
 
 	if (abd != NULL) {
+		char *buf;
 		vdev_boot_envblock_t *vbe = abd_to_buf(abd);
-		if (vbe->vbe_version != VB_RAW) {
-			abd_free(abd);
-			return (SET_ERROR(ENOTSUP));
+
+		vbe->vbe_version = ntohll(vbe->vbe_version);
+		switch (vbe->vbe_version) {
+		case VB_RAW:
+			/*
+			 * if we have textual data in vbe_bootenv, create nvlist
+			 * with key "envmap".
+			 */
+			fnvlist_add_uint64(bootenv, BOOTENV_VERSION, VB_RAW);
+			vbe->vbe_bootenv[sizeof (vbe->vbe_bootenv) - 1] = '\0';
+			fnvlist_add_string(bootenv, GRUB_ENVMAP,
+			    vbe->vbe_bootenv);
+			break;
+
+		case VB_NVLIST:
+			err = nvlist_unpack(vbe->vbe_bootenv,
+			    sizeof (vbe->vbe_bootenv), &config, 0);
+			if (err == 0) {
+				fnvlist_merge(bootenv, config);
+				nvlist_free(config);
+				break;
+			}
+			/* FALLTHROUGH */
+		default:
+			/* Check for FreeBSD zfs bootonce command string */
+			buf = abd_to_buf(abd);
+			if (*buf == '\0') {
+				fnvlist_add_uint64(bootenv, BOOTENV_VERSION,
+				    VB_NVLIST);
+				break;
+			}
+			fnvlist_add_string(bootenv, FREEBSD_BOOTONCE, buf);
 		}
-		vbe->vbe_bootenv[sizeof (vbe->vbe_bootenv) - 1] = '\0';
-		fnvlist_add_string(command, "envmap", vbe->vbe_bootenv);
-		/* abd was allocated in vdev_label_read_bootenv_impl() */
+
+		/*
+		 * abd was allocated in vdev_label_read_bootenv_impl()
+		 */
 		abd_free(abd);
-		/* If we managed to read any successfully, return success. */
+		/*
+		 * If we managed to read any successfully,
+		 * return success.
+		 */
 		return (0);
 	}
 	return (err);
 }
 
 int
-vdev_label_write_bootenv(vdev_t *vd, char *envmap)
+vdev_label_write_bootenv(vdev_t *vd, nvlist_t *env)
 {
 	zio_t *zio;
 	spa_t *spa = vd->vdev_spa;
 	vdev_boot_envblock_t *bootenv;
 	int flags = ZIO_FLAG_CONFIG_WRITER | ZIO_FLAG_CANFAIL;
-	int error = ENXIO;
+	int error;
+	size_t nvsize;
+	char *nvbuf;
+
+	error = nvlist_size(env, &nvsize, NV_ENCODE_XDR);
+	if (error != 0)
+		return (SET_ERROR(error));
 
-	if (strlen(envmap) >= sizeof (bootenv->vbe_bootenv)) {
+	if (nvsize >= sizeof (bootenv->vbe_bootenv)) {
 		return (SET_ERROR(E2BIG));
 	}
 
 	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
 
+	error = ENXIO;
 	for (int c = 0; c < vd->vdev_children; c++) {
-		int child_err = vdev_label_write_bootenv(vd->vdev_child[c],
-		    envmap);
+		int child_err;
+
+		child_err = vdev_label_write_bootenv(vd->vdev_child[c], env);
 		/*
 		 * As long as any of the disks managed to write all of their
 		 * labels successfully, return success.
@@ -1312,16 +1353,41 @@ vdev_label_write_bootenv(vdev_t *vd, char *envmap)
 	ASSERT3U(sizeof (*bootenv), ==, VDEV_PAD_SIZE);
 	abd_t *abd = abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE);
 	abd_zero(abd, VDEV_PAD_SIZE);
+
 	bootenv = abd_borrow_buf_copy(abd, VDEV_PAD_SIZE);
+	nvbuf = bootenv->vbe_bootenv;
+	nvsize = sizeof (bootenv->vbe_bootenv);
+
+	bootenv->vbe_version = fnvlist_lookup_uint64(env, BOOTENV_VERSION);
+	switch (bootenv->vbe_version) {
+	case VB_RAW:
+		if (nvlist_lookup_string(env, GRUB_ENVMAP, &nvbuf) == 0) {
+			(void) strlcpy(bootenv->vbe_bootenv, nvbuf, nvsize);
+		}
+		error = 0;
+		break;
 
-	char *buf = bootenv->vbe_bootenv;
-	(void) strlcpy(buf, envmap, sizeof (bootenv->vbe_bootenv));
-	bootenv->vbe_version = VB_RAW;
-	abd_return_buf_copy(abd, bootenv, VDEV_PAD_SIZE);
+	case VB_NVLIST:
+		error = nvlist_pack(env, &nvbuf, &nvsize, NV_ENCODE_XDR,
+		    KM_SLEEP);
+		break;
+
+	default:
+		error = EINVAL;
+		break;
+	}
+
+	if (error == 0) {
+		bootenv->vbe_version = htonll(bootenv->vbe_version);
+		abd_return_buf_copy(abd, bootenv, VDEV_PAD_SIZE);
+	} else {
+		abd_free(abd);
+		return (SET_ERROR(error));
+	}
 
 retry:
 	zio = zio_root(spa, NULL, NULL, flags);
-	for (int l = 0; l < VDEV_LABELS / 2; l++) {
+	for (int l = 0; l < VDEV_LABELS; l++) {
 		vdev_label_write(zio, vd, l, abd,
 		    offsetof(vdev_label_t, vl_be),
 		    VDEV_PAD_SIZE, NULL, NULL, flags);
diff --git a/module/zfs/zfs_ioctl.c b/module/zfs/zfs_ioctl.c
index c9322a826..eff66b32f 100644
--- a/module/zfs/zfs_ioctl.c
+++ b/module/zfs/zfs_ioctl.c
@@ -3511,30 +3511,29 @@ zfs_ioc_log_history(const char *unused, nvlist_t *innvl, nvlist_t *outnvl)
 /*
  * This ioctl is used to set the bootenv configuration on the current
  * pool. This configuration is stored in the second padding area of the label,
- * and it is used by the GRUB bootloader used on Linux to store the contents
- * of the grubenv file.  The file is stored as raw ASCII, and is protected by
- * an embedded checksum.  By default, GRUB will check if the boot filesystem
- * supports storing the environment data in a special location, and if so,
- * will invoke filesystem specific logic to retrieve it. This can be overridden
- * by a variable, should the user so desire.
+ * and it is used by the bootloader(s) to store the bootloader and/or system
+ * specific data.
+ * The data is stored as nvlist data stream, and is protected by
+ * an embedded checksum.
+ * The version can have two possible values:
+ * VB_RAW: nvlist should have key GRUB_ENVMAP, value DATA_TYPE_STRING.
+ * VB_NVLIST: nvlist with arbitrary <key, value> pairs.
  */
-/* ARGSUSED */
 static const zfs_ioc_key_t zfs_keys_set_bootenv[] = {
-	{"envmap",	DATA_TYPE_STRING,	0},
+	{"version",	DATA_TYPE_UINT64,	0},
+	{"<keys>",	DATA_TYPE_ANY, ZK_OPTIONAL | ZK_WILDCARDLIST},
 };
 
 static int
 zfs_ioc_set_bootenv(const char *name, nvlist_t *innvl, nvlist_t *outnvl)
 {
-	char *envmap;
 	int error;
 	spa_t *spa;
 
-	envmap = fnvlist_lookup_string(innvl, "envmap");
 	if ((error = spa_open(name, &spa, FTAG)) != 0)
 		return (error);
 	spa_vdev_state_enter(spa, SCL_ALL);
-	error = vdev_label_write_bootenv(spa->spa_root_vdev, envmap);
+	error = vdev_label_write_bootenv(spa->spa_root_vdev, innvl);
 	(void) spa_vdev_state_exit(spa, NULL, 0);
 	spa_close(spa, FTAG);
 	return (error);
@@ -3544,7 +3543,6 @@ static const zfs_ioc_key_t zfs_keys_get_bootenv[] = {
 	/* no nvl keys */
 };
 
-/* ARGSUSED */
 static int
 zfs_ioc_get_bootenv(const char *name, nvlist_t *innvl, nvlist_t *outnvl)
 {