Make boot code and loader check for unsupported ZFS feature flags

OpenZFS uses feature flags instead of a zpool version number to track
features since the split from Oracle. In addition to avoiding confusion
on ZFS vs OpenZFS version numbers, this also allows features to be added
to different operating systems that use OpenZFS in different order.

The previous zfs boot code (gptzfsboot) and loader (zfsloader) blindly
tries to read the pool, and if failed provided only a vague error message.

With this change, both the boot code and loader check the MOS features
list in the ZFS label and compare it against the list of features that
the loader supports. If any unsupported feature is active, the pool is
not considered as a candidate for booting, and a helpful diagnostic
message is printed to the screen. Features that are merely enabled via
zpool upgrade, but not in use, do not block booting from the pool.

Submitted by:	Toomas Soome <tsoome@me.com>
Reviewed by:	delphij, mav
Relnotes:	yes
Differential Revision:	https://reviews.freebsd.org/D6857
This commit is contained in:
allanjude 2016-08-01 19:37:43 +00:00
parent fc8db6a083
commit ca9d82b266
4 changed files with 134 additions and 14 deletions

View File

@ -65,7 +65,7 @@ int zfs_probe_dev(const char *devname, uint64_t *pool_guid);
int zfs_list(const char *name);
void init_zfs_bootenv(char *currdev);
int zfs_bootenv(const char *name);
int zfs_belist_add(const char *name);
int zfs_belist_add(const char *name, uint64_t __unused);
int zfs_set_env(void);
extern struct devsw zfs_dev;

View File

@ -801,7 +801,7 @@ zfs_bootenv(const char *name)
}
int
zfs_belist_add(const char *name)
zfs_belist_add(const char *name, uint64_t value __unused)
{
/* Skip special datasets that start with a $ character */

View File

@ -1473,12 +1473,12 @@ zap_lookup(const spa_t *spa, const dnode_phys_t *dnode, const char *name, uint64
* the directory contents.
*/
static int
mzap_list(const dnode_phys_t *dnode, int (*callback)(const char *))
mzap_list(const dnode_phys_t *dnode, int (*callback)(const char *, uint64_t))
{
const mzap_phys_t *mz;
const mzap_ent_phys_t *mze;
size_t size;
int chunks, i;
int chunks, i, rc;
/*
* Microzap objects use exactly one block. Read the whole
@ -1490,9 +1490,11 @@ mzap_list(const dnode_phys_t *dnode, int (*callback)(const char *))
for (i = 0; i < chunks; i++) {
mze = &mz->mz_chunk[i];
if (mze->mze_name[0])
//printf("%-32s 0x%jx\n", mze->mze_name, (uintmax_t)mze->mze_value);
callback(mze->mze_name);
if (mze->mze_name[0]) {
rc = callback(mze->mze_name, mze->mze_value);
if (rc != 0)
return (rc);
}
}
return (0);
@ -1503,12 +1505,12 @@ mzap_list(const dnode_phys_t *dnode, int (*callback)(const char *))
* the directory header.
*/
static int
fzap_list(const spa_t *spa, const dnode_phys_t *dnode, int (*callback)(const char *))
fzap_list(const spa_t *spa, const dnode_phys_t *dnode, int (*callback)(const char *, uint64_t))
{
int bsize = dnode->dn_datablkszsec << SPA_MINBLOCKSHIFT;
zap_phys_t zh = *(zap_phys_t *) zap_scratch;
fat_zap_t z;
int i, j;
int i, j, rc;
if (zh.zap_magic != ZAP_MAGIC)
return (EIO);
@ -1566,14 +1568,16 @@ fzap_list(const spa_t *spa, const dnode_phys_t *dnode, int (*callback)(const cha
value = fzap_leaf_value(&zl, zc);
//printf("%s 0x%jx\n", name, (uintmax_t)value);
callback((const char *)name);
rc = callback((const char *)name, value);
if (rc != 0)
return (rc);
}
}
return (0);
}
static int zfs_printf(const char *name)
static int zfs_printf(const char *name, uint64_t value __unused)
{
printf("%s\n", name);
@ -1868,7 +1872,7 @@ zfs_list_dataset(const spa_t *spa, uint64_t objnum/*, int pos, char *entry*/)
}
int
zfs_callback_dataset(const spa_t *spa, uint64_t objnum, int (*callback)(const char *name))
zfs_callback_dataset(const spa_t *spa, uint64_t objnum, int (*callback)(const char *, uint64_t))
{
uint64_t dir_obj, child_dir_zapobj, zap_type;
dnode_phys_t child_dir_zap, dir, dataset;
@ -2008,9 +2012,67 @@ zfs_mount(const spa_t *spa, uint64_t rootobj, struct zfsmount *mount)
return (0);
}
/*
* callback function for feature name checks.
*/
static int
check_feature(const char *name, uint64_t value)
{
int i;
if (value == 0)
return (0);
if (name[0] == '\0')
return (0);
for (i = 0; features_for_read[i] != NULL; i++) {
if (strcmp(name, features_for_read[i]) == 0)
return (0);
}
printf("ZFS: unsupported feature: %s\n", name);
return (EIO);
}
/*
* Checks whether the MOS features that are active are supported.
*/
static int
check_mos_features(const spa_t *spa)
{
dnode_phys_t dir;
uint64_t objnum, zap_type;
size_t size;
int rc;
if ((rc = objset_get_dnode(spa, &spa->spa_mos, DMU_OT_OBJECT_DIRECTORY,
&dir)) != 0)
return (rc);
if ((rc = zap_lookup(spa, &dir, DMU_POOL_FEATURES_FOR_READ, &objnum)) != 0)
return (rc);
if ((rc = objset_get_dnode(spa, &spa->spa_mos, objnum, &dir)) != 0)
return (rc);
if (dir.dn_type != DMU_OTN_ZAP_METADATA)
return (EIO);
size = dir.dn_datablkszsec * 512;
if (dnode_read(spa, &dir, 0, zap_scratch, size))
return (EIO);
zap_type = *(uint64_t *) zap_scratch;
if (zap_type == ZBT_MICRO)
rc = mzap_list(&dir, check_feature);
else
rc = fzap_list(spa, &dir, check_feature);
return (rc);
}
static int
zfs_spa_init(spa_t *spa)
{
int rc;
if (zio_read(spa, &spa->spa_uberblock.ub_rootbp, &spa->spa_mos)) {
printf("ZFS: can't read MOS of pool %s\n", spa->spa_name);
@ -2020,7 +2082,13 @@ zfs_spa_init(spa_t *spa)
printf("ZFS: corrupted MOS of pool %s\n", spa->spa_name);
return (EIO);
}
return (0);
rc = check_mos_features(spa);
if (rc != 0) {
printf("ZFS: pool %s is not supported\n", spa->spa_name);
}
return (rc);
}
static int

View File

@ -63,6 +63,8 @@
#define _NOTE(s)
typedef enum { B_FALSE, B_TRUE } boolean_t;
/* CRC64 table */
#define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */
@ -899,6 +901,41 @@ typedef struct dnode_phys {
blkptr_t dn_spill;
} dnode_phys_t;
typedef enum dmu_object_byteswap {
DMU_BSWAP_UINT8,
DMU_BSWAP_UINT16,
DMU_BSWAP_UINT32,
DMU_BSWAP_UINT64,
DMU_BSWAP_ZAP,
DMU_BSWAP_DNODE,
DMU_BSWAP_OBJSET,
DMU_BSWAP_ZNODE,
DMU_BSWAP_OLDACL,
DMU_BSWAP_ACL,
/*
* Allocating a new byteswap type number makes the on-disk format
* incompatible with any other format that uses the same number.
*
* Data can usually be structured to work with one of the
* DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types.
*/
DMU_BSWAP_NUMFUNCS
} dmu_object_byteswap_t;
#define DMU_OT_NEWTYPE 0x80
#define DMU_OT_METADATA 0x40
#define DMU_OT_BYTESWAP_MASK 0x3f
/*
* Defines a uint8_t object type. Object types specify if the data
* in the object is metadata (boolean) and how to byteswap the data
* (dmu_object_byteswap_t).
*/
#define DMU_OT(byteswap, metadata) \
(DMU_OT_NEWTYPE | \
((metadata) ? DMU_OT_METADATA : 0) | \
((byteswap) & DMU_OT_BYTESWAP_MASK))
typedef enum dmu_object_type {
DMU_OT_NONE,
/* general: */
@ -959,7 +996,21 @@ typedef enum dmu_object_type {
DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */
DMU_OT_SCAN_XLATE, /* ZAP */
DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */
DMU_OT_NUMTYPES
DMU_OT_NUMTYPES,
/*
* Names for valid types declared with DMU_OT().
*/
DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE),
DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE),
DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE),
DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE),
DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE),
DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE),
DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE),
DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE),
DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE),
DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE)
} dmu_object_type_t;
typedef enum dmu_objset_type {
@ -1097,6 +1148,7 @@ typedef struct dsl_dataset_phys {
*/
#define DMU_POOL_DIRECTORY_OBJECT 1
#define DMU_POOL_CONFIG "config"
#define DMU_POOL_FEATURES_FOR_READ "features_for_read"
#define DMU_POOL_ROOT_DATASET "root_dataset"
#define DMU_POOL_SYNC_BPLIST "sync_bplist"
#define DMU_POOL_ERRLOG_SCRUB "errlog_scrub"