/*
* Convert our list of pools into the definitive set of configurations. We
* start by picking the best config for each toplevel vdev. Once that's done,
* we assemble the toplevel vdevs into a full config for the pool. We make a
* pass to fix up any incorrect paths, and then add it to the main list to
* return to the user.
*/
static nvlist_t *
get_configs(libzfs_handle_t *hdl, pool_list_t *pl, boolean_t active_ok)
{
pool_entry_t *pe;
vdev_entry_t *ve;
config_entry_t *ce;
nvlist_t *ret = NULL, *config = NULL, *tmp = NULL, *nvtop, *nvroot;
nvlist_t **spares, **l2cache;
uint_t i, nspares, nl2cache;
boolean_t config_seen;
uint64_t best_txg;
char *name, *hostname = NULL;
uint64_t guid;
uint_t children = 0;
nvlist_t **child = NULL;
uint_t holes;
uint64_t *hole_array, max_id;
uint_t c;
boolean_t isactive;
uint64_t hostid;
nvlist_t *nvl;
boolean_t valid_top_config = B_FALSE;
if (nvlist_alloc(&ret, 0, 0) != 0)
goto nomem;
for (pe = pl->pools; pe != NULL; pe = pe->pe_next) {
uint64_t id, max_txg = 0;
if (nvlist_alloc(&config, NV_UNIQUE_NAME, 0) != 0)
goto nomem;
config_seen = B_FALSE;
/*
* Iterate over all toplevel vdevs. Grab the pool configuration
* from the first one we find, and then go through the rest and
* add them as necessary to the 'vdevs' member of the config.
*/
for (ve = pe->pe_vdevs; ve != NULL; ve = ve->ve_next) {
/*
* Determine the best configuration for this vdev by
* selecting the config with the latest transaction
* group.
*/
best_txg = 0;
for (ce = ve->ve_configs; ce != NULL;
ce = ce->ce_next) {
if (ce->ce_txg > best_txg) {
tmp = ce->ce_config;
best_txg = ce->ce_txg;
}
}
/*
* We rely on the fact that the max txg for the
* pool will contain the most up-to-date information
* about the valid top-levels in the vdev namespace.
*/
if (best_txg > max_txg) {
(void) nvlist_remove(config,
ZPOOL_CONFIG_VDEV_CHILDREN,
DATA_TYPE_UINT64);
(void) nvlist_remove(config,
ZPOOL_CONFIG_HOLE_ARRAY,
DATA_TYPE_UINT64_ARRAY);
max_txg = best_txg;
hole_array = NULL;
holes = 0;
max_id = 0;
valid_top_config = B_FALSE;
if (nvlist_lookup_uint64(tmp,
ZPOOL_CONFIG_VDEV_CHILDREN, &max_id) == 0) {
verify(nvlist_add_uint64(config,
ZPOOL_CONFIG_VDEV_CHILDREN,
max_id) == 0);
valid_top_config = B_TRUE;
}
if (nvlist_lookup_uint64_array(tmp,
ZPOOL_CONFIG_HOLE_ARRAY, &hole_array,
&holes) == 0) {
verify(nvlist_add_uint64_array(config,
ZPOOL_CONFIG_HOLE_ARRAY,
hole_array, holes) == 0);
}
}
if (!config_seen) {
/*
* Copy the relevant pieces of data to the pool
* configuration:
*
* version
* pool guid
* name
* comment (if available)
* pool state
* hostid (if available)
* hostname (if available)
*/
uint64_t state, version;
char *comment = NULL;
version = fnvlist_lookup_uint64(tmp,
ZPOOL_CONFIG_VERSION);
fnvlist_add_uint64(config,
ZPOOL_CONFIG_VERSION, version);
guid = fnvlist_lookup_uint64(tmp,
ZPOOL_CONFIG_POOL_GUID);
fnvlist_add_uint64(config,
ZPOOL_CONFIG_POOL_GUID, guid);
name = fnvlist_lookup_string(tmp,
ZPOOL_CONFIG_POOL_NAME);
fnvlist_add_string(config,
ZPOOL_CONFIG_POOL_NAME, name);
if (nvlist_lookup_string(tmp,
ZPOOL_CONFIG_COMMENT, &comment) == 0)
fnvlist_add_string(config,
ZPOOL_CONFIG_COMMENT, comment);
state = fnvlist_lookup_uint64(tmp,
ZPOOL_CONFIG_POOL_STATE);
fnvlist_add_uint64(config,
ZPOOL_CONFIG_POOL_STATE, state);
hostid = 0;
if (nvlist_lookup_uint64(tmp,
ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
fnvlist_add_uint64(config,
ZPOOL_CONFIG_HOSTID, hostid);
hostname = fnvlist_lookup_string(tmp,
ZPOOL_CONFIG_HOSTNAME);
fnvlist_add_string(config,
ZPOOL_CONFIG_HOSTNAME, hostname);
}
config_seen = B_TRUE;
}
/*
* Add this top-level vdev to the child array.
*/
verify(nvlist_lookup_nvlist(tmp,
ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0);
verify(nvlist_lookup_uint64(nvtop, ZPOOL_CONFIG_ID,
&id) == 0);
if (id >= children) {
nvlist_t **newchild;
newchild = zfs_alloc(hdl, (id + 1) *
sizeof (nvlist_t *));
if (newchild == NULL)
goto nomem;
for (c = 0; c < children; c++)
newchild[c] = child[c];
free(child);
child = newchild;
children = id + 1;
}
if (nvlist_dup(nvtop, &child[id], 0) != 0)
goto nomem;
}
/*
* If we have information about all the top-levels then
* clean up the nvlist which we've constructed. This
* means removing any extraneous devices that are
* beyond the valid range or adding devices to the end
* of our array which appear to be missing.
*/
if (valid_top_config) {
if (max_id < children) {
for (c = max_id; c < children; c++)
nvlist_free(child[c]);
children = max_id;
} else if (max_id > children) {
nvlist_t **newchild;
newchild = zfs_alloc(hdl, (max_id) *
sizeof (nvlist_t *));
if (newchild == NULL)
goto nomem;
for (c = 0; c < children; c++)
newchild[c] = child[c];
free(child);
child = newchild;
children = max_id;
}
}
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
&guid) == 0);
/*
* The vdev namespace may contain holes as a result of
* device removal. We must add them back into the vdev
* tree before we process any missing devices.
*/
if (holes > 0) {
ASSERT(valid_top_config);
for (c = 0; c < children; c++) {
nvlist_t *holey;
if (child[c] != NULL ||
!vdev_is_hole(hole_array, holes, c))
continue;
if (nvlist_alloc(&holey, NV_UNIQUE_NAME,
0) != 0)
goto nomem;
/*
* Holes in the namespace are treated as
* "hole" top-level vdevs and have a
* special flag set on them.
*/
if (nvlist_add_string(holey,
ZPOOL_CONFIG_TYPE,
VDEV_TYPE_HOLE) != 0 ||
nvlist_add_uint64(holey,
ZPOOL_CONFIG_ID, c) != 0 ||
nvlist_add_uint64(holey,
ZPOOL_CONFIG_GUID, 0ULL) != 0) {
nvlist_free(holey);
goto nomem;
}
child[c] = holey;
}
}
/*
* Look for any missing top-level vdevs. If this is the case,
* create a faked up 'missing' vdev as a placeholder. We cannot
* simply compress the child array, because the kernel performs
* certain checks to make sure the vdev IDs match their location
* in the configuration.
*/
for (c = 0; c < children; c++) {
if (child[c] == NULL) {
nvlist_t *missing;
if (nvlist_alloc(&missing, NV_UNIQUE_NAME,
0) != 0)
goto nomem;
if (nvlist_add_string(missing,
ZPOOL_CONFIG_TYPE,
VDEV_TYPE_MISSING) != 0 ||
nvlist_add_uint64(missing,
ZPOOL_CONFIG_ID, c) != 0 ||
nvlist_add_uint64(missing,
ZPOOL_CONFIG_GUID, 0ULL) != 0) {
nvlist_free(missing);
goto nomem;
}
child[c] = missing;
}
}
/*
* Put all of this pool's top-level vdevs into a root vdev.
*/
if (nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) != 0)
goto nomem;
if (nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE,
VDEV_TYPE_ROOT) != 0 ||
nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) != 0 ||
nvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, guid) != 0 ||
nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
child, children) != 0) {
nvlist_free(nvroot);
goto nomem;
}
for (c = 0; c < children; c++)
nvlist_free(child[c]);
free(child);
children = 0;
child = NULL;
/*
* Go through and fix up any paths and/or devids based on our
* known list of vdev GUID -> path mappings.
*/
if (fix_paths(nvroot, pl->names) != 0) {
nvlist_free(nvroot);
goto nomem;
}
/*
* Add the root vdev to this pool's configuration.
*/
if (nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
nvroot) != 0) {
nvlist_free(nvroot);
goto nomem;
}
nvlist_free(nvroot);
/*
* zdb uses this path to report on active pools that were
* imported or created using -R.
*/
if (active_ok)
goto add_pool;
/*
* Determine if this pool is currently active, in which case we
* can't actually import it.
*/
verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
&name) == 0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
&guid) == 0);
if (pool_active(hdl, name, guid, &isactive) != 0)
goto error;
if (isactive) {
nvlist_free(config);
config = NULL;
continue;
}
if ((nvl = refresh_config(hdl, config)) == NULL) {
nvlist_free(config);
config = NULL;
continue;
}
nvlist_free(config);
config = nvl;
/*
* Go through and update the paths for spares, now that we have
* them.
*/
verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvroot) == 0);
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
&spares, &nspares) == 0) {
for (i = 0; i < nspares; i++) {
if (fix_paths(spares[i], pl->names) != 0)
goto nomem;
}
}
/*
* Update the paths for l2cache devices.
*/
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
&l2cache, &nl2cache) == 0) {
for (i = 0; i < nl2cache; i++) {
if (fix_paths(l2cache[i], pl->names) != 0)
goto nomem;
}
}
/*
* Restore the original information read from the actual label.
*/
(void) nvlist_remove(config, ZPOOL_CONFIG_HOSTID,
DATA_TYPE_UINT64);
(void) nvlist_remove(config, ZPOOL_CONFIG_HOSTNAME,
DATA_TYPE_STRING);
if (hostid != 0) {
verify(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
hostid) == 0);
verify(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
hostname) == 0);
}
add_pool:
/*
* Add this pool to the list of configs.
*/
verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
&name) == 0);
if (nvlist_add_nvlist(ret, name, config) != 0)
goto nomem;
nvlist_free(config);
config = NULL;
}
return (ret);
nomem:
(void) no_memory(hdl);
error:
nvlist_free(config);
nvlist_free(ret);
for (c = 0; c < children; c++)
nvlist_free(child[c]);
free(child);
return (NULL);
}
/*
* Sort datasets by specified columns.
*
* o Numeric types sort in ascending order.
* o String types sort in alphabetical order.
* o Types inappropriate for a row sort that row to the literal
* bottom, regardless of the specified ordering.
*
* If no sort columns are specified, or two datasets compare equally
* across all specified columns, they are sorted alphabetically by name
* with snapshots grouped under their parents.
*/
static int
zfs_sort(const void *larg, const void *rarg, void *data)
{
zfs_handle_t *l = ((zfs_node_t *)larg)->zn_handle;
zfs_handle_t *r = ((zfs_node_t *)rarg)->zn_handle;
zfs_sort_column_t *sc = (zfs_sort_column_t *)data;
zfs_sort_column_t *psc;
for (psc = sc; psc != NULL; psc = psc->sc_next) {
char lbuf[ZFS_MAXPROPLEN], rbuf[ZFS_MAXPROPLEN];
char *lstr, *rstr;
uint64_t lnum, rnum;
boolean_t lvalid, rvalid;
int ret = 0;
/*
* We group the checks below the generic code. If 'lstr' and
* 'rstr' are non-NULL, then we do a string based comparison.
* Otherwise, we compare 'lnum' and 'rnum'.
*/
lstr = rstr = NULL;
if (psc->sc_prop == ZPROP_INVAL) {
nvlist_t *luser, *ruser;
nvlist_t *lval, *rval;
luser = zfs_get_user_props(l);
ruser = zfs_get_user_props(r);
lvalid = (nvlist_lookup_nvlist(luser,
psc->sc_user_prop, &lval) == 0);
rvalid = (nvlist_lookup_nvlist(ruser,
psc->sc_user_prop, &rval) == 0);
if (lvalid)
verify(nvlist_lookup_string(lval,
ZPROP_VALUE, &lstr) == 0);
if (rvalid)
verify(nvlist_lookup_string(rval,
ZPROP_VALUE, &rstr) == 0);
} else if (psc->sc_prop == ZFS_PROP_NAME) {
lvalid = rvalid = B_TRUE;
(void) strlcpy(lbuf, zfs_get_name(l), sizeof (lbuf));
(void) strlcpy(rbuf, zfs_get_name(r), sizeof (rbuf));
lstr = lbuf;
rstr = rbuf;
} else if (zfs_prop_is_string(psc->sc_prop)) {
lvalid = (zfs_prop_get(l, psc->sc_prop, lbuf,
sizeof (lbuf), NULL, NULL, 0, B_TRUE) == 0);
rvalid = (zfs_prop_get(r, psc->sc_prop, rbuf,
sizeof (rbuf), NULL, NULL, 0, B_TRUE) == 0);
lstr = lbuf;
rstr = rbuf;
} else {
lvalid = zfs_prop_valid_for_type(psc->sc_prop,
zfs_get_type(l));
rvalid = zfs_prop_valid_for_type(psc->sc_prop,
zfs_get_type(r));
if (lvalid)
(void) zfs_prop_get_numeric(l, psc->sc_prop,
&lnum, NULL, NULL, 0);
if (rvalid)
(void) zfs_prop_get_numeric(r, psc->sc_prop,
&rnum, NULL, NULL, 0);
}
if (!lvalid && !rvalid)
continue;
else if (!lvalid)
return (1);
else if (!rvalid)
return (-1);
if (lstr)
ret = strcmp(lstr, rstr);
else if (lnum < rnum)
ret = -1;
else if (lnum > rnum)
ret = 1;
if (ret != 0) {
if (psc->sc_reverse == B_TRUE)
ret = (ret < 0) ? 1 : -1;
return (ret);
}
}
return (zfs_compare(larg, rarg, NULL));
}
/*
* Determines if the pool is in use. If so, it returns true and the state of
* the pool as well as the name of the pool. Both strings are allocated and
* must be freed by the caller.
*/
int
zpool_in_use(libzfs_handle_t *hdl, int fd, pool_state_t *state, char **namestr,
boolean_t *inuse)
{
nvlist_t *config;
char *name;
boolean_t ret;
uint64_t guid, vdev_guid;
zpool_handle_t *zhp;
nvlist_t *pool_config;
uint64_t stateval, isspare;
aux_cbdata_t cb = { 0 };
boolean_t isactive;
*inuse = B_FALSE;
if (zpool_read_label(fd, &config, NULL) != 0) {
(void) no_memory(hdl);
return (-1);
}
if (config == NULL)
return (0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
&stateval) == 0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID,
&vdev_guid) == 0);
if (stateval != POOL_STATE_SPARE && stateval != POOL_STATE_L2CACHE) {
verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
&name) == 0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
&guid) == 0);
}
switch (stateval) {
case POOL_STATE_EXPORTED:
/*
* A pool with an exported state may in fact be imported
* read-only, so check the in-core state to see if it's
* active and imported read-only. If it is, set
* its state to active.
*/
if (pool_active(hdl, name, guid, &isactive) == 0 && isactive &&
(zhp = zpool_open_canfail(hdl, name)) != NULL) {
if (zpool_get_prop_int(zhp, ZPOOL_PROP_READONLY, NULL))
stateval = POOL_STATE_ACTIVE;
/*
* All we needed the zpool handle for is the
* readonly prop check.
*/
zpool_close(zhp);
}
ret = B_TRUE;
break;
case POOL_STATE_ACTIVE:
/*
* For an active pool, we have to determine if it's really part
* of a currently active pool (in which case the pool will exist
* and the guid will be the same), or whether it's part of an
* active pool that was disconnected without being explicitly
* exported.
*/
if (pool_active(hdl, name, guid, &isactive) != 0) {
nvlist_free(config);
return (-1);
}
if (isactive) {
/*
* Because the device may have been removed while
* offlined, we only report it as active if the vdev is
* still present in the config. Otherwise, pretend like
* it's not in use.
*/
if ((zhp = zpool_open_canfail(hdl, name)) != NULL &&
(pool_config = zpool_get_config(zhp, NULL))
!= NULL) {
nvlist_t *nvroot;
verify(nvlist_lookup_nvlist(pool_config,
ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
ret = find_guid(nvroot, vdev_guid);
} else {
ret = B_FALSE;
}
/*
* If this is an active spare within another pool, we
* treat it like an unused hot spare. This allows the
* user to create a pool with a hot spare that currently
* in use within another pool. Since we return B_TRUE,
* libdiskmgt will continue to prevent generic consumers
* from using the device.
*/
if (ret && nvlist_lookup_uint64(config,
ZPOOL_CONFIG_IS_SPARE, &isspare) == 0 && isspare)
stateval = POOL_STATE_SPARE;
if (zhp != NULL)
zpool_close(zhp);
} else {
stateval = POOL_STATE_POTENTIALLY_ACTIVE;
ret = B_TRUE;
}
break;
case POOL_STATE_SPARE:
/*
* For a hot spare, it can be either definitively in use, or
* potentially active. To determine if it's in use, we iterate
* over all pools in the system and search for one with a spare
* with a matching guid.
*
* Due to the shared nature of spares, we don't actually report
* the potentially active case as in use. This means the user
* can freely create pools on the hot spares of exported pools,
* but to do otherwise makes the resulting code complicated, and
* we end up having to deal with this case anyway.
*/
cb.cb_zhp = NULL;
cb.cb_guid = vdev_guid;
cb.cb_type = ZPOOL_CONFIG_SPARES;
if (zpool_iter(hdl, find_aux, &cb) == 1) {
name = (char *)zpool_get_name(cb.cb_zhp);
ret = B_TRUE;
} else {
ret = B_FALSE;
}
break;
case POOL_STATE_L2CACHE:
/*
* Check if any pool is currently using this l2cache device.
*/
cb.cb_zhp = NULL;
cb.cb_guid = vdev_guid;
cb.cb_type = ZPOOL_CONFIG_L2CACHE;
if (zpool_iter(hdl, find_aux, &cb) == 1) {
name = (char *)zpool_get_name(cb.cb_zhp);
ret = B_TRUE;
} else {
ret = B_FALSE;
}
break;
default:
ret = B_FALSE;
}
if (ret) {
if ((*namestr = zfs_strdup(hdl, name)) == NULL) {
if (cb.cb_zhp)
zpool_close(cb.cb_zhp);
nvlist_free(config);
return (-1);
}
*state = (pool_state_t)stateval;
}
if (cb.cb_zhp)
zpool_close(cb.cb_zhp);
nvlist_free(config);
*inuse = ret;
return (0);
}
/*
* Linux adds ZFS_IOC_RECV_NEW for resumable streams and preserves the legacy
* ZFS_IOC_RECV user/kernel interface. The new interface supports all stream
* options but is currently only used for resumable streams. This way updated
* user space utilities will interoperate with older kernel modules.
*
* Non-Linux OpenZFS platforms have opted to modify the legacy interface.
*/
static int
recv_impl(const char *snapname, nvlist_t *props, const char *origin,
boolean_t force, boolean_t resumable, int input_fd,
const dmu_replay_record_t *begin_record, int cleanup_fd,
uint64_t *read_bytes, uint64_t *errflags, uint64_t *action_handle,
nvlist_t **errors)
{
dmu_replay_record_t drr;
char fsname[MAXPATHLEN];
char *atp;
int error;
ASSERT3S(g_refcount, >, 0);
VERIFY3S(g_fd, !=, -1);
/* Set 'fsname' to the name of containing filesystem */
(void) strlcpy(fsname, snapname, sizeof (fsname));
atp = strchr(fsname, '@');
if (atp == NULL)
return (EINVAL);
*atp = '\0';
/* If the fs does not exist, try its parent. */
if (!lzc_exists(fsname)) {
char *slashp = strrchr(fsname, '/');
if (slashp == NULL)
return (ENOENT);
*slashp = '\0';
}
/*
* The begin_record is normally a non-byteswapped BEGIN record.
* For resumable streams it may be set to any non-byteswapped
* dmu_replay_record_t.
*/
if (begin_record == NULL) {
error = recv_read(input_fd, &drr, sizeof (drr));
if (error != 0)
return (error);
} else {
drr = *begin_record;
}
if (resumable) {
nvlist_t *outnvl = NULL;
nvlist_t *innvl = fnvlist_alloc();
fnvlist_add_string(innvl, "snapname", snapname);
if (props != NULL)
fnvlist_add_nvlist(innvl, "props", props);
if (origin != NULL && strlen(origin))
fnvlist_add_string(innvl, "origin", origin);
fnvlist_add_byte_array(innvl, "begin_record",
(uchar_t *)&drr, sizeof (drr));
fnvlist_add_int32(innvl, "input_fd", input_fd);
if (force)
fnvlist_add_boolean(innvl, "force");
if (resumable)
fnvlist_add_boolean(innvl, "resumable");
if (cleanup_fd >= 0)
fnvlist_add_int32(innvl, "cleanup_fd", cleanup_fd);
if (action_handle != NULL)
fnvlist_add_uint64(innvl, "action_handle",
*action_handle);
error = lzc_ioctl(ZFS_IOC_RECV_NEW, fsname, innvl, &outnvl);
if (error == 0 && read_bytes != NULL)
error = nvlist_lookup_uint64(outnvl, "read_bytes",
read_bytes);
if (error == 0 && errflags != NULL)
error = nvlist_lookup_uint64(outnvl, "error_flags",
errflags);
if (error == 0 && action_handle != NULL)
error = nvlist_lookup_uint64(outnvl, "action_handle",
action_handle);
if (error == 0 && errors != NULL) {
nvlist_t *nvl;
error = nvlist_lookup_nvlist(outnvl, "errors", &nvl);
if (error == 0)
*errors = fnvlist_dup(nvl);
}
fnvlist_free(innvl);
fnvlist_free(outnvl);
} else {
zfs_cmd_t zc = {"\0"};
char *packed = NULL;
size_t size;
ASSERT3S(g_refcount, >, 0);
(void) strlcpy(zc.zc_name, fsname, sizeof (zc.zc_value));
(void) strlcpy(zc.zc_value, snapname, sizeof (zc.zc_value));
if (props != NULL) {
packed = fnvlist_pack(props, &size);
zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed;
zc.zc_nvlist_src_size = size;
}
if (origin != NULL)
(void) strlcpy(zc.zc_string, origin,
sizeof (zc.zc_string));
ASSERT3S(drr.drr_type, ==, DRR_BEGIN);
zc.zc_begin_record = drr.drr_u.drr_begin;
zc.zc_guid = force;
zc.zc_cookie = input_fd;
zc.zc_cleanup_fd = -1;
zc.zc_action_handle = 0;
if (cleanup_fd >= 0)
zc.zc_cleanup_fd = cleanup_fd;
if (action_handle != NULL)
zc.zc_action_handle = *action_handle;
zc.zc_nvlist_dst_size = 128 * 1024;
zc.zc_nvlist_dst = (uint64_t)(uintptr_t)
malloc(zc.zc_nvlist_dst_size);
error = ioctl(g_fd, ZFS_IOC_RECV, &zc);
if (error != 0) {
error = errno;
} else {
if (read_bytes != NULL)
*read_bytes = zc.zc_cookie;
if (errflags != NULL)
*errflags = zc.zc_obj;
if (action_handle != NULL)
*action_handle = zc.zc_action_handle;
if (errors != NULL)
VERIFY0(nvlist_unpack(
(void *)(uintptr_t)zc.zc_nvlist_dst,
zc.zc_nvlist_dst_size, errors, KM_SLEEP));
}
if (packed != NULL)
fnvlist_pack_free(packed, size);
free((void *)(uintptr_t)zc.zc_nvlist_dst);
}
return (error);
}
int
main(int argc, char **argv)
{
int fd_pool;
int fd_log;
vdev_label_t vl_pool;
vdev_label_t vl_log;
nvlist_t *config_pool;
nvlist_t *config_log;
uint64_t guid; // ZPOOL_CONFIG_GUID
uint64_t is_log; // ZPOOL_CONFIG_IS_LOG
nvlist_t *vdev_tree; // ZPOOL_CONFIG_VDEV_TREE
char *buf;
size_t buflen;
VERIFY(argc == 4);
VERIFY((fd_pool = open(argv[1], O_RDWR)) != -1);
VERIFY((fd_log = open(argv[2], O_RDWR)) != -1);
VERIFY(sscanf(argv[3], "%" SCNu64 , &guid) == 1);
//guid = 9851295902337437618ULL;
VERIFY(pread64(fd_pool, &vl_pool, sizeof (vdev_label_t), 0) ==
sizeof (vdev_label_t));
VERIFY(nvlist_unpack(vl_pool.vl_vdev_phys.vp_nvlist,
sizeof (vl_pool.vl_vdev_phys.vp_nvlist), &config_pool, 0) == 0);
VERIFY(pread64(fd_log, &vl_log, sizeof (vdev_label_t), 0) ==
sizeof (vdev_label_t));
VERIFY(nvlist_unpack(vl_log.vl_vdev_phys.vp_nvlist,
sizeof (vl_log.vl_vdev_phys.vp_nvlist), &config_log, 0) == 0);
// save what we want from config_log -- is_log, vdev_tree
VERIFY(nvlist_lookup_uint64(config_log, ZPOOL_CONFIG_IS_LOG, &is_log) == 0);
VERIFY(nvlist_lookup_nvlist(config_log, ZPOOL_CONFIG_VDEV_TREE, &vdev_tree) == 0);
// fix guid for vdev_log
VERIFY(nvlist_remove_all(vdev_tree, ZPOOL_CONFIG_GUID) == 0);
VERIFY(nvlist_add_uint64(vdev_tree, ZPOOL_CONFIG_GUID, guid) == 0);
// remove what we are going to replace on config_pool
VERIFY(nvlist_remove_all(config_pool, ZPOOL_CONFIG_TOP_GUID) == 0);
VERIFY(nvlist_remove_all(config_pool, ZPOOL_CONFIG_GUID) == 0);
VERIFY(nvlist_remove_all(config_pool, ZPOOL_CONFIG_VDEV_TREE) == 0);
// add back what we want
VERIFY(nvlist_add_uint64(config_pool, ZPOOL_CONFIG_TOP_GUID, guid) == 0);
VERIFY(nvlist_add_uint64(config_pool, ZPOOL_CONFIG_GUID, guid) == 0);
VERIFY(nvlist_add_uint64(config_pool, ZPOOL_CONFIG_IS_LOG, is_log) == 0);
VERIFY(nvlist_add_nvlist(config_pool, ZPOOL_CONFIG_VDEV_TREE, vdev_tree) == 0);
buf = vl_pool.vl_vdev_phys.vp_nvlist;
buflen = sizeof (vl_pool.vl_vdev_phys.vp_nvlist);
VERIFY(nvlist_pack(config_pool, &buf, &buflen, NV_ENCODE_XDR, 0) == 0);
label_write(fd_log, offsetof(vdev_label_t, vl_vdev_phys),
VDEV_PHYS_SIZE, &vl_pool.vl_vdev_phys);
fsync(fd_log);
return (0);
}
/*
* Active pool health status.
*
* To determine the status for a pool, we make several passes over the config,
* picking the most egregious error we find. In order of importance, we do the
* following:
*
* - Check for a complete and valid configuration
* - Look for any faulted or missing devices in a non-replicated config
* - Check for any data errors
* - Check for any faulted or missing devices in a replicated config
* - Look for any devices showing errors
* - Check for any resilvering devices
*
* There can obviously be multiple errors within a single pool, so this routine
* only picks the most damaging of all the current errors to report.
*/
static zpool_status_t
check_status(nvlist_t *config, boolean_t isimport)
{
nvlist_t *nvroot;
vdev_stat_t *vs;
pool_scan_stat_t *ps = NULL;
uint_t vsc, psc;
uint64_t nerr;
uint64_t version;
uint64_t stateval;
uint64_t suspended;
uint64_t hostid = 0;
unsigned long system_hostid = gethostid() & 0xffffffff;
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
&version) == 0);
verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvroot) == 0);
verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS,
(uint64_t **)&vs, &vsc) == 0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
&stateval) == 0);
/*
* Currently resilvering a vdev
*/
(void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS,
(uint64_t **)&ps, &psc);
if (ps && ps->pss_func == POOL_SCAN_RESILVER &&
ps->pss_state == DSS_SCANNING)
return (ZPOOL_STATUS_RESILVERING);
/*
* Pool last accessed by another system.
*/
(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid);
if (hostid != 0 && (unsigned long)hostid != system_hostid &&
stateval == POOL_STATE_ACTIVE)
return (ZPOOL_STATUS_HOSTID_MISMATCH);
/*
* Newer on-disk version.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_VERSION_NEWER)
return (ZPOOL_STATUS_VERSION_NEWER);
/*
* Unsupported feature(s).
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_UNSUP_FEAT) {
nvlist_t *nvinfo;
verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO,
&nvinfo) == 0);
if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY))
return (ZPOOL_STATUS_UNSUP_FEAT_WRITE);
return (ZPOOL_STATUS_UNSUP_FEAT_READ);
}
/*
* Check that the config is complete.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_BAD_GUID_SUM)
return (ZPOOL_STATUS_BAD_GUID_SUM);
/*
* Check whether the pool has suspended due to failed I/O.
*/
if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED,
&suspended) == 0) {
if (suspended == ZIO_FAILURE_MODE_CONTINUE)
return (ZPOOL_STATUS_IO_FAILURE_CONTINUE);
return (ZPOOL_STATUS_IO_FAILURE_WAIT);
}
/*
* Could not read a log.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_BAD_LOG) {
return (ZPOOL_STATUS_BAD_LOG);
}
/*
* Bad devices in non-replicated config.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_faulted))
return (ZPOOL_STATUS_FAULTED_DEV_NR);
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_missing))
return (ZPOOL_STATUS_MISSING_DEV_NR);
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_broken))
return (ZPOOL_STATUS_CORRUPT_LABEL_NR);
/*
* Corrupted pool metadata
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_CORRUPT_DATA)
return (ZPOOL_STATUS_CORRUPT_POOL);
/*
* Persistent data errors.
*/
if (!isimport) {
if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT,
&nerr) == 0 && nerr != 0)
return (ZPOOL_STATUS_CORRUPT_DATA);
}
/*
* Missing devices in a replicated config.
*/
if (find_vdev_problem(nvroot, vdev_faulted))
return (ZPOOL_STATUS_FAULTED_DEV_R);
if (find_vdev_problem(nvroot, vdev_missing))
return (ZPOOL_STATUS_MISSING_DEV_R);
if (find_vdev_problem(nvroot, vdev_broken))
return (ZPOOL_STATUS_CORRUPT_LABEL_R);
/*
* Devices with errors
*/
if (!isimport && find_vdev_problem(nvroot, vdev_errors))
return (ZPOOL_STATUS_FAILING_DEV);
/*
* Offlined devices
*/
if (find_vdev_problem(nvroot, vdev_offlined))
return (ZPOOL_STATUS_OFFLINE_DEV);
/*
* Removed device
*/
if (find_vdev_problem(nvroot, vdev_removed))
return (ZPOOL_STATUS_REMOVED_DEV);
/*
* Outdated, but usable, version
*/
if (SPA_VERSION_IS_SUPPORTED(version) && version != SPA_VERSION)
return (ZPOOL_STATUS_VERSION_OLDER);
/*
* Usable pool with disabled features
*/
if (version >= SPA_VERSION_FEATURES) {
int i;
nvlist_t *feat;
if (isimport) {
feat = fnvlist_lookup_nvlist(config,
ZPOOL_CONFIG_LOAD_INFO);
feat = fnvlist_lookup_nvlist(feat,
ZPOOL_CONFIG_ENABLED_FEAT);
} else {
feat = fnvlist_lookup_nvlist(config,
ZPOOL_CONFIG_FEATURE_STATS);
}
for (i = 0; i < SPA_FEATURES; i++) {
zfeature_info_t *fi = &spa_feature_table[i];
if (!nvlist_exists(feat, fi->fi_guid))
return (ZPOOL_STATUS_FEAT_DISABLED);
}
}
return (ZPOOL_STATUS_OK);
}
/*
* Active pool health status.
*
* To determine the status for a pool, we make several passes over the config,
* picking the most egregious error we find. In order of importance, we do the
* following:
*
* - Check for a complete and valid configuration
* - Look for any faulted or missing devices in a non-replicated config
* - Check for any data errors
* - Check for any faulted or missing devices in a replicated config
* - Look for any devices showing errors
* - Check for any resilvering devices
*
* There can obviously be multiple errors within a single pool, so this routine
* only picks the most damaging of all the current errors to report.
*/
static zpool_status_t
check_status(zpool_handle_t *zhp, nvlist_t *config, boolean_t isimport)
{
nvlist_t *nvroot;
vdev_stat_t *vs;
uint_t vsc;
uint64_t nerr;
uint64_t version;
uint64_t stateval;
uint64_t hostid = 0;
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
&version) == 0);
verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
&nvroot) == 0);
verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_STATS,
(uint64_t **)&vs, &vsc) == 0);
verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
&stateval) == 0);
(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid);
/*
* Pool last accessed by another system.
*/
if (hostid != 0 && (unsigned long)hostid != gethostid() &&
stateval == POOL_STATE_ACTIVE)
return (ZPOOL_STATUS_HOSTID_MISMATCH);
/*
* Newer on-disk version.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_VERSION_NEWER)
return (ZPOOL_STATUS_VERSION_NEWER);
/*
* Check that the config is complete.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_BAD_GUID_SUM)
return (ZPOOL_STATUS_BAD_GUID_SUM);
/*
* Pool has experienced failed I/O.
*/
if (stateval == POOL_STATE_IO_FAILURE) {
zpool_handle_t *tmp_zhp = NULL;
libzfs_handle_t *hdl = NULL;
char property[ZPOOL_MAXPROPLEN];
char *failmode = NULL;
if (zhp == NULL) {
char *poolname;
verify(nvlist_lookup_string(config,
ZPOOL_CONFIG_POOL_NAME, &poolname) == 0);
if ((hdl = libzfs_init()) == NULL)
return (ZPOOL_STATUS_IO_FAILURE_WAIT);
tmp_zhp = zpool_open_canfail(hdl, poolname);
if (tmp_zhp == NULL) {
libzfs_fini(hdl);
return (ZPOOL_STATUS_IO_FAILURE_WAIT);
}
}
if (zpool_get_prop(zhp ? zhp : tmp_zhp, ZPOOL_PROP_FAILUREMODE,
property, sizeof (property), NULL) == 0)
failmode = property;
if (tmp_zhp != NULL)
zpool_close(tmp_zhp);
if (hdl != NULL)
libzfs_fini(hdl);
if (failmode == NULL)
return (ZPOOL_STATUS_IO_FAILURE_WAIT);
if (strncmp(failmode, "continue", strlen("continue")) == 0)
return (ZPOOL_STATUS_IO_FAILURE_CONTINUE);
else
return (ZPOOL_STATUS_IO_FAILURE_WAIT);
}
/*
* Could not read a log.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_BAD_LOG) {
return (ZPOOL_STATUS_BAD_LOG);
}
/*
* Bad devices in non-replicated config.
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_faulted))
return (ZPOOL_STATUS_FAULTED_DEV_NR);
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_missing))
return (ZPOOL_STATUS_MISSING_DEV_NR);
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
find_vdev_problem(nvroot, vdev_broken))
return (ZPOOL_STATUS_CORRUPT_LABEL_NR);
/*
* Corrupted pool metadata
*/
if (vs->vs_state == VDEV_STATE_CANT_OPEN &&
vs->vs_aux == VDEV_AUX_CORRUPT_DATA)
return (ZPOOL_STATUS_CORRUPT_POOL);
/*
* Persistent data errors.
*/
if (!isimport) {
if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT,
&nerr) == 0 && nerr != 0)
return (ZPOOL_STATUS_CORRUPT_DATA);
}
/*
* Missing devices in a replicated config.
*/
if (find_vdev_problem(nvroot, vdev_faulted))
return (ZPOOL_STATUS_FAULTED_DEV_R);
if (find_vdev_problem(nvroot, vdev_missing))
return (ZPOOL_STATUS_MISSING_DEV_R);
if (find_vdev_problem(nvroot, vdev_broken))
return (ZPOOL_STATUS_CORRUPT_LABEL_R);
/*
* Devices with errors
*/
if (!isimport && find_vdev_problem(nvroot, vdev_errors))
return (ZPOOL_STATUS_FAILING_DEV);
/*
* Offlined devices
*/
if (find_vdev_problem(nvroot, vdev_offlined))
return (ZPOOL_STATUS_OFFLINE_DEV);
/*
* Currently resilvering
*/
if (!vs->vs_scrub_complete && vs->vs_scrub_type == POOL_SCRUB_RESILVER)
return (ZPOOL_STATUS_RESILVERING);
/*
* Outdated, but usable, version
*/
if (version < SPA_VERSION)
return (ZPOOL_STATUS_VERSION_OLDER);
return (ZPOOL_STATUS_OK);
}
static char *
find_root_device_from_libzfs (const char *dir)
{
char *device = NULL;
char *poolname;
char *poolfs;
grub_find_zpool_from_dir (dir, &poolname, &poolfs);
if (! poolname)
return NULL;
{
zpool_handle_t *zpool;
libzfs_handle_t *libzfs;
nvlist_t *config, *vdev_tree;
nvlist_t **children, **path;
unsigned int nvlist_count;
unsigned int i;
libzfs = grub_get_libzfs_handle ();
if (! libzfs)
return NULL;
zpool = zpool_open (libzfs, poolname);
config = zpool_get_config (zpool, NULL);
if (nvlist_lookup_nvlist (config, "vdev_tree", &vdev_tree) != 0)
error (1, errno, "nvlist_lookup_nvlist (\"vdev_tree\")");
if (nvlist_lookup_nvlist_array (vdev_tree, "children", &children, &nvlist_count) != 0)
error (1, errno, "nvlist_lookup_nvlist_array (\"children\")");
assert (nvlist_count > 0);
while (nvlist_lookup_nvlist_array (children[0], "children",
&children, &nvlist_count) == 0)
assert (nvlist_count > 0);
for (i = 0; i < nvlist_count; i++)
{
if (nvlist_lookup_string (children[i], "path", &device) != 0)
error (1, errno, "nvlist_lookup_string (\"path\")");
struct stat st;
if (stat (device, &st) == 0)
{
device = xstrdup (device);
break;
}
device = NULL;
}
zpool_close (zpool);
}
free (poolname);
if (poolfs)
free (poolfs);
return device;
}
/*
* Attach new_disk (fully described by nvroot) to old_disk.
* If 'replacing' is specified, tne new disk will replace the old one.
*/
int
zpool_vdev_attach(zpool_handle_t *zhp,
const char *old_disk, const char *new_disk, nvlist_t *nvroot, int replacing)
{
zfs_cmd_t zc = { 0 };
char msg[1024];
int ret;
nvlist_t *tgt;
boolean_t avail_spare;
uint64_t val;
char *path;
nvlist_t **child;
uint_t children;
nvlist_t *config_root;
libzfs_handle_t *hdl = zhp->zpool_hdl;
if (replacing)
(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
"cannot replace %s with %s"), old_disk, new_disk);
else
(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
"cannot attach %s to %s"), new_disk, old_disk);
(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
if ((tgt = zpool_find_vdev(zhp, old_disk, &avail_spare)) == 0)
return (zfs_error(hdl, EZFS_NODEVICE, msg));
if (avail_spare)
return (zfs_error(hdl, EZFS_ISSPARE, msg));
verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0);
zc.zc_cookie = replacing;
if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
&child, &children) != 0 || children != 1) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"new device must be a single disk"));
return (zfs_error(hdl, EZFS_INVALCONFIG, msg));
}
verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL),
ZPOOL_CONFIG_VDEV_TREE, &config_root) == 0);
/*
* If the target is a hot spare that has been swapped in, we can only
* replace it with another hot spare.
*/
if (replacing &&
nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_IS_SPARE, &val) == 0 &&
nvlist_lookup_string(child[0], ZPOOL_CONFIG_PATH, &path) == 0 &&
(zpool_find_vdev(zhp, path, &avail_spare) == NULL ||
!avail_spare) && is_replacing_spare(config_root, tgt, 1)) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"can only be replaced by another hot spare"));
return (zfs_error(hdl, EZFS_BADTARGET, msg));
}
/*
* If we are attempting to replace a spare, it canot be applied to an
* already spared device.
*/
if (replacing &&
nvlist_lookup_string(child[0], ZPOOL_CONFIG_PATH, &path) == 0 &&
zpool_find_vdev(zhp, path, &avail_spare) != NULL && avail_spare &&
is_replacing_spare(config_root, tgt, 0)) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"device has already been replaced with a spare"));
return (zfs_error(hdl, EZFS_BADTARGET, msg));
}
if (zcmd_write_src_nvlist(hdl, &zc, nvroot, NULL) != 0)
return (-1);
ret = ioctl(zhp->zpool_hdl->libzfs_fd, ZFS_IOC_VDEV_ATTACH, &zc);
zcmd_free_nvlists(&zc);
if (ret == 0)
return (0);
switch (errno) {
case ENOTSUP:
/*
* Can't attach to or replace this type of vdev.
*/
if (replacing)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"cannot replace a replacing device"));
else
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"can only attach to mirrors and top-level "
"disks"));
(void) zfs_error(hdl, EZFS_BADTARGET, msg);
break;
case EINVAL:
/*
* The new device must be a single disk.
*/
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"new device must be a single disk"));
(void) zfs_error(hdl, EZFS_INVALCONFIG, msg);
break;
case EBUSY:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s is busy"),
new_disk);
(void) zfs_error(hdl, EZFS_BADDEV, msg);
break;
case EOVERFLOW:
/*
* The new device is too small.
*/
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"device is too small"));
(void) zfs_error(hdl, EZFS_BADDEV, msg);
break;
case EDOM:
/*
* The new device has a different alignment requirement.
*/
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"devices have different sector alignment"));
(void) zfs_error(hdl, EZFS_BADDEV, msg);
break;
case ENAMETOOLONG:
/*
* The resulting top-level vdev spec won't fit in the label.
*/
(void) zfs_error(hdl, EZFS_DEVOVERFLOW, msg);
break;
default:
(void) zpool_standard_error(hdl, errno, msg);
}
return (-1);
}
/*ARGSUSED*/
static int
sw_fmri_nvl2str(topo_mod_t *mod, tnode_t *node, topo_version_t version,
nvlist_t *nvl, nvlist_t **out)
{
nvlist_t *object, *site = NULL, *anvl = NULL;
char *file, *func, *token;
uint8_t scheme_version;
char *path, *root;
nvlist_t *fmristr;
size_t buflen = 0;
int linevalid = 0;
char *buf = NULL;
ssize_t size = 0;
char linebuf[32];
int64_t line;
int pass;
int err;
if (version > TOPO_METH_NVL2STR_VERSION)
return (topo_mod_seterrno(mod, EMOD_VER_NEW));
if (nvlist_lookup_uint8(nvl, FM_VERSION, &scheme_version) != 0 ||
scheme_version > FM_SW_SCHEME_VERSION)
return (topo_mod_seterrno(mod, EMOD_FMRI_NVL));
/* Get authority, if present */
err = nvlist_lookup_nvlist(nvl, FM_FMRI_AUTHORITY, &anvl);
if (err != 0 && err != ENOENT)
return (topo_mod_seterrno(mod, EMOD_FMRI_NVL));
/*
* The 'object' nvlist is required. It must include the path,
* but the root is optional.
*/
if (nvlist_lookup_nvlist(nvl, FM_FMRI_SW_OBJ, &object) != 0 ||
!lookup_string(object, FM_FMRI_SW_OBJ_PATH, &path, B_TRUE) ||
!lookup_string(object, FM_FMRI_SW_OBJ_ROOT, &root, B_FALSE))
return (topo_mod_seterrno(mod, EMOD_FMRI_NVL));
/* The 'site' nvlist is optional */
file = func = token = NULL;
linevalid = 0;
if ((err = nvlist_lookup_nvlist(nvl, FM_FMRI_SW_SITE, &site)) == 0) {
/*
* Prefer 'token' to file/func/line
*/
if (lookup_string(site, FM_FMRI_SW_SITE_TOKEN, &token,
B_FALSE) <= 0) {
/*
* If no token then try file, func, line - but
* func and line are meaningless without file.
*/
if (lookup_string(site, FM_FMRI_SW_SITE_FILE,
&file, B_FALSE) == 1) {
(void) lookup_string(site, FM_FMRI_SW_SITE_FUNC,
&func, B_FALSE);
if (nvlist_lookup_int64(site,
FM_FMRI_SW_SITE_LINE, &line) == 0)
linevalid = 1;
}
}
} else if (err != ENOENT) {
return (topo_mod_seterrno(mod, EMOD_FMRI_NVL));
}
/* On the first pass buf is NULL and size and buflen are 0 */
pass = 1;
again:
/*
* sw://[<authority>]/
* [:root=<object.root]
* :path=<object.path>
* [#<fragment-identifier>]
*
* <fragment-identifier> is one of
*
* :token=<site.token>
* or
* :file=<site.file>[:func=<site.func>][:line=<site.line>]
*/
/* sw:// */
topo_fmristr_build(&size, buf, buflen, FM_FMRI_SCHEME_SW,
NULL, "://");
/* authority, if any */
if (anvl != NULL) {
nvpair_t *apair;
char *aname, *aval;
for (apair = nvlist_next_nvpair(anvl, NULL);
apair != NULL; apair = nvlist_next_nvpair(anvl, apair)) {
if (nvpair_type(apair) != DATA_TYPE_STRING ||
nvpair_value_string(apair, &aval) != 0)
continue;
aname = nvpair_name(apair);
topo_fmristr_build(&size, buf, buflen, ":", NULL, NULL);
topo_fmristr_build(&size, buf, buflen, "=",
aname, aval);
}
}
/* separating slash */
topo_fmristr_build(&size, buf, buflen, "/", NULL, NULL);
/* :root=... */
if (root) {
topo_fmristr_build(&size, buf, buflen, root,
":" FM_FMRI_SW_OBJ_ROOT "=", NULL);
}
/* :path=... */
topo_fmristr_build(&size, buf, buflen, path,
":" FM_FMRI_SW_OBJ_PATH "=", NULL);
if (token) {
/* #:token=... */
topo_fmristr_build(&size, buf, buflen, token,
"#:" FM_FMRI_SW_SITE_TOKEN "=", NULL);
} else if (file) {
/* #:file=... */
topo_fmristr_build(&size, buf, buflen, file,
"#:" FM_FMRI_SW_SITE_FILE "=", NULL);
/* :func=... */
if (func) {
topo_fmristr_build(&size, buf, buflen, func,
":" FM_FMRI_SW_SITE_FUNC "=", NULL);
}
/* :line=... */
if (linevalid) {
if (pass == 1)
(void) snprintf(linebuf, sizeof (linebuf),
"%lld", line);
topo_fmristr_build(&size, buf, buflen, linebuf,
":" FM_FMRI_SW_SITE_LINE "=", NULL);
}
}
if (buf == NULL) {
if ((buf = topo_mod_alloc(mod, size + 1)) == NULL)
return (topo_mod_seterrno(mod, EMOD_NOMEM));
buflen = size + 1;
size = 0;
pass = 2;
goto again;
}
/*
* Construct the nvlist to return as the result.
*/
if (topo_mod_nvalloc(mod, &fmristr, NV_UNIQUE_NAME) != 0) {
topo_mod_strfree(mod, buf);
return (topo_mod_seterrno(mod, EMOD_NOMEM));
}
if (nvlist_add_string(fmristr, "fmri-string", buf) != 0) {
topo_mod_strfree(mod, buf);
nvlist_free(fmristr);
return (topo_mod_seterrno(mod, EMOD_NOMEM));
}
topo_mod_strfree(mod, buf);
*out = fmristr;
return (0);
}
/*ARGSUSED*/
static int
sw_fmri_create(topo_mod_t *mod, tnode_t *node, topo_version_t version,
nvlist_t *in, nvlist_t **out)
{
nvlist_t *args, *fmri = NULL, *obj = NULL, *site = NULL, *ctxt = NULL;
topo_mod_errno_t moderr;
int err = 0;
char *obj_path, *obj_root;
nvlist_t *obj_pkg;
char *site_token, *site_module, *site_file, *site_func;
int64_t site_line;
char *ctxt_origin, *ctxt_execname, *ctxt_zone;
int64_t ctxt_pid, ctxt_ctid;
char **ctxt_stack;
uint_t ctxt_stackdepth;
if (version > TOPO_METH_FMRI_VERSION)
return (topo_mod_seterrno(mod, EMOD_VER_NEW));
if (nvlist_lookup_nvlist(in, TOPO_METH_FMRI_ARG_NVL, &args) != 0)
return (topo_mod_seterrno(mod, EMOD_METHOD_INVAL));
if (nvlist_lookup_string(args, "obj_path", &obj_path) != 0)
return (topo_mod_seterrno(mod, EMOD_NVL_INVAL));
err |= sw_get_optl_string(args, "obj_root", &obj_root);
err |= sw_get_optl_nvlist(args, "obj-pkg", &obj_pkg);
err |= sw_get_optl_string(args, "site_token", &site_token);
err |= sw_get_optl_string(args, "site_module", &site_module);
err |= sw_get_optl_string(args, "site_file", &site_file);
err |= sw_get_optl_string(args, "site_func", &site_func);
err |= sw_get_optl_int64(args, "site_line", &site_line);
err |= sw_get_optl_string(args, "ctxt_origin", &ctxt_origin);
err |= sw_get_optl_string(args, "ctxt_execname", &ctxt_execname);
err |= sw_get_optl_string(args, "ctxt_zone", &ctxt_zone);
err |= sw_get_optl_int64(args, "ctxt_pid", &ctxt_pid);
err |= sw_get_optl_int64(args, "ctxt_ctid", &ctxt_ctid);
if (nvlist_lookup_string_array(args, "stack", &ctxt_stack,
&ctxt_stackdepth) != 0) {
if (errno == ENOENT)
ctxt_stack = NULL;
else
err++;
}
if (err)
(void) topo_mod_seterrno(mod, EMOD_FMRI_NVL);
if (topo_mod_nvalloc(mod, &fmri, NV_UNIQUE_NAME) != 0 ||
topo_mod_nvalloc(mod, &obj, NV_UNIQUE_NAME) != 0) {
moderr = EMOD_NOMEM;
goto out;
}
/*
* Add standard FMRI members 'version' and 'scheme'.
*/
err |= nvlist_add_uint8(fmri, FM_VERSION, FM_SW_SCHEME_VERSION);
err |= nvlist_add_string(fmri, FM_FMRI_SCHEME, FM_FMRI_SCHEME_SW);
/*
* Build up the 'object' nvlist.
*/
err |= nvlist_add_string(obj, FM_FMRI_SW_OBJ_PATH, obj_path);
err |= sw_add_optl_string(obj, FM_FMRI_SW_OBJ_ROOT, obj_root);
if (obj_pkg)
err |= nvlist_add_nvlist(obj, FM_FMRI_SW_OBJ_PKG, obj_pkg);
/*
* Add 'object' to the fmri.
*/
if (err == 0)
err |= nvlist_add_nvlist(fmri, FM_FMRI_SW_OBJ, obj);
if (err) {
moderr = EMOD_NOMEM;
goto out;
}
/*
* Do we have anything for a 'site' nvlist?
*/
if (site_token == NULL && site_module == NULL && site_file == NULL &&
site_func == NULL && site_line == -1)
goto context;
/*
* Allocate and build 'site' nvlist.
*/
if (topo_mod_nvalloc(mod, &site, NV_UNIQUE_NAME) != 0) {
moderr = EMOD_NOMEM;
goto out;
}
err |= sw_add_optl_string(site, FM_FMRI_SW_SITE_TOKEN, site_token);
err |= sw_add_optl_string(site, FM_FMRI_SW_SITE_MODULE, site_module);
err |= sw_add_optl_string(site, FM_FMRI_SW_SITE_FILE, site_file);
err |= sw_add_optl_string(site, FM_FMRI_SW_SITE_FUNC, site_func);
if ((site_token || site_module || site_file || site_func) &&
site_line != -1)
err |= nvlist_add_int64(site, FM_FMRI_SW_SITE_LINE, site_line);
/*
* Add 'site' to the fmri.
*/
if (err == 0)
err |= nvlist_add_nvlist(fmri, FM_FMRI_SW_SITE, site);
if (err) {
moderr = EMOD_NOMEM;
goto out;
}
context:
/*
* Do we have anything for a 'context' nvlist?
*/
if (ctxt_origin || ctxt_execname || ctxt_zone ||
ctxt_pid != -1 || ctxt_ctid != -1 || ctxt_stack != NULL)
goto out;
/*
* Allocate and build 'context' nvlist.
*/
if (topo_mod_nvalloc(mod, &ctxt, NV_UNIQUE_NAME) != 0) {
moderr = EMOD_NOMEM;
goto out;
}
err |= sw_add_optl_string(ctxt, FM_FMRI_SW_CTXT_ORIGIN, ctxt_origin);
err |= sw_add_optl_string(ctxt, FM_FMRI_SW_CTXT_EXECNAME,
ctxt_execname);
err |= sw_add_optl_string(ctxt, FM_FMRI_SW_CTXT_ZONE, ctxt_zone);
if (ctxt_pid != -1)
err |= nvlist_add_int64(ctxt, FM_FMRI_SW_CTXT_PID, ctxt_pid);
if (ctxt_ctid != -1)
err |= nvlist_add_int64(ctxt, FM_FMRI_SW_CTXT_CTID, ctxt_ctid);
if (ctxt_stack != NULL)
err |= nvlist_add_string_array(ctxt, FM_FMRI_SW_CTXT_STACK,
ctxt_stack, ctxt_stackdepth);
/*
* Add 'context' to the fmri.
*/
if (err == 0)
err |= nvlist_add_nvlist(fmri, FM_FMRI_SW_CTXT, ctxt);
moderr = err ? EMOD_NOMEM : 0;
out:
if (moderr == 0)
*out = fmri;
if (moderr != 0 && fmri)
nvlist_free(fmri);
if (obj)
nvlist_free(obj);
if (site)
nvlist_free(site);
if (ctxt)
nvlist_free(ctxt);
return (moderr == 0 ? 0 : topo_mod_seterrno(mod, moderr));
}
/*
* Decodes share information in an nvlist format into a smb_kshare_t
* structure.
*
* This is a temporary function and will be replaced by functions
* provided by libsharev2 code after it's available.
*/
static smb_kshare_t *
smb_kshare_decode(nvlist_t *share)
{
smb_kshare_t tmp;
smb_kshare_t *shr;
nvlist_t *smb;
char *csc_name = NULL;
int rc;
ASSERT(share);
bzero(&tmp, sizeof (smb_kshare_t));
rc = nvlist_lookup_string(share, "name", &tmp.shr_name);
rc |= nvlist_lookup_string(share, "path", &tmp.shr_path);
(void) nvlist_lookup_string(share, "desc", &tmp.shr_cmnt);
ASSERT(tmp.shr_name && tmp.shr_path);
rc |= nvlist_lookup_nvlist(share, "smb", &smb);
if (rc != 0) {
cmn_err(CE_WARN, "kshare: failed looking up SMB properties"
" (%d)", rc);
return (NULL);
}
rc = nvlist_lookup_uint32(smb, "type", &tmp.shr_type);
if (rc != 0) {
cmn_err(CE_WARN, "kshare[%s]: failed getting the share type"
" (%d)", tmp.shr_name, rc);
return (NULL);
}
(void) nvlist_lookup_string(smb, SHOPT_AD_CONTAINER,
&tmp.shr_container);
(void) nvlist_lookup_string(smb, SHOPT_NONE, &tmp.shr_access_none);
(void) nvlist_lookup_string(smb, SHOPT_RO, &tmp.shr_access_ro);
(void) nvlist_lookup_string(smb, SHOPT_RW, &tmp.shr_access_rw);
tmp.shr_flags |= smb_kshare_decode_bool(smb, SHOPT_ABE, SMB_SHRF_ABE);
tmp.shr_flags |= smb_kshare_decode_bool(smb, SHOPT_CATIA,
SMB_SHRF_CATIA);
tmp.shr_flags |= smb_kshare_decode_bool(smb, SHOPT_GUEST,
SMB_SHRF_GUEST_OK);
tmp.shr_flags |= smb_kshare_decode_bool(smb, SHOPT_DFSROOT,
SMB_SHRF_DFSROOT);
tmp.shr_flags |= smb_kshare_decode_bool(smb, "Autohome",
SMB_SHRF_AUTOHOME);
if ((tmp.shr_flags & SMB_SHRF_AUTOHOME) == SMB_SHRF_AUTOHOME) {
rc = nvlist_lookup_uint32(smb, "uid", &tmp.shr_uid);
rc |= nvlist_lookup_uint32(smb, "gid", &tmp.shr_gid);
if (rc != 0) {
cmn_err(CE_WARN, "kshare: failed looking up uid/gid"
" (%d)", rc);
return (NULL);
}
}
(void) nvlist_lookup_string(smb, SHOPT_CSC, &csc_name);
smb_kshare_csc_flags(&tmp, csc_name);
shr = kmem_cache_alloc(smb_kshare_cache_share, KM_SLEEP);
bzero(shr, sizeof (smb_kshare_t));
shr->shr_magic = SMB_SHARE_MAGIC;
shr->shr_refcnt = 1;
shr->shr_name = smb_mem_strdup(tmp.shr_name);
shr->shr_path = smb_mem_strdup(tmp.shr_path);
if (tmp.shr_cmnt)
shr->shr_cmnt = smb_mem_strdup(tmp.shr_cmnt);
if (tmp.shr_container)
shr->shr_container = smb_mem_strdup(tmp.shr_container);
if (tmp.shr_access_none)
shr->shr_access_none = smb_mem_strdup(tmp.shr_access_none);
if (tmp.shr_access_ro)
shr->shr_access_ro = smb_mem_strdup(tmp.shr_access_ro);
if (tmp.shr_access_rw)
shr->shr_access_rw = smb_mem_strdup(tmp.shr_access_rw);
shr->shr_oemname = smb_kshare_oemname(shr->shr_name);
shr->shr_flags = tmp.shr_flags | smb_kshare_is_admin(shr->shr_name);
shr->shr_type = tmp.shr_type | smb_kshare_is_special(shr->shr_name);
shr->shr_uid = tmp.shr_uid;
shr->shr_gid = tmp.shr_gid;
if ((shr->shr_flags & SMB_SHRF_AUTOHOME) == SMB_SHRF_AUTOHOME)
shr->shr_autocnt = 1;
return (shr);
}
/*
* Synchronize pool configuration to disk. This must be called with the
* namespace lock held. Synchronizing the pool cache is typically done after
* the configuration has been synced to the MOS. This exposes a window where
* the MOS config will have been updated but the cache file has not. If
* the system were to crash at that instant then the cached config may not
* contain the correct information to open the pool and an explicity import
* would be required.
*/
void
spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent)
{
spa_config_dirent_t *dp, *tdp;
nvlist_t *nvl;
boolean_t ccw_failure;
int error;
ASSERT(MUTEX_HELD(&spa_namespace_lock));
if (rootdir == NULL || !(spa_mode_global & FWRITE))
return;
/*
* Iterate over all cachefiles for the pool, past or present. When the
* cachefile is changed, the new one is pushed onto this list, allowing
* us to update previous cachefiles that no longer contain this pool.
*/
ccw_failure = B_FALSE;
for (dp = list_head(&target->spa_config_list); dp != NULL;
dp = list_next(&target->spa_config_list, dp)) {
spa_t *spa = NULL;
if (dp->scd_path == NULL)
continue;
/*
* Iterate over all pools, adding any matching pools to 'nvl'.
*/
nvl = NULL;
while ((spa = spa_next(spa)) != NULL) {
nvlist_t *nvroot = NULL;
/*
* Skip over our own pool if we're about to remove
* ourselves from the spa namespace or any pool that
* is readonly. Since we cannot guarantee that a
* readonly pool would successfully import upon reboot,
* we don't allow them to be written to the cache file.
*/
if ((spa == target && removing) ||
(spa_state(spa) == POOL_STATE_ACTIVE &&
!spa_writeable(spa)))
continue;
mutex_enter(&spa->spa_props_lock);
tdp = list_head(&spa->spa_config_list);
if (spa->spa_config == NULL ||
tdp->scd_path == NULL ||
strcmp(tdp->scd_path, dp->scd_path) != 0) {
mutex_exit(&spa->spa_props_lock);
continue;
}
if (nvl == NULL)
VERIFY(nvlist_alloc(&nvl, NV_UNIQUE_NAME,
KM_SLEEP) == 0);
VERIFY(nvlist_add_nvlist(nvl, spa->spa_name,
spa->spa_config) == 0);
mutex_exit(&spa->spa_props_lock);
if (nvlist_lookup_nvlist(nvl, spa->spa_name, &nvroot) == 0)
spa_config_clean(nvroot);
}
error = spa_config_write(dp, nvl);
if (error != 0)
ccw_failure = B_TRUE;
nvlist_free(nvl);
}
if (ccw_failure) {
/*
* Keep trying so that configuration data is
* written if/when any temporary filesystem
* resource issues are resolved.
*/
if (target->spa_ccw_fail_time == 0) {
zfs_ereport_post(FM_EREPORT_ZFS_CONFIG_CACHE_WRITE,
target, NULL, NULL, 0, 0);
}
target->spa_ccw_fail_time = gethrtime();
spa_async_request(target, SPA_ASYNC_CONFIG_UPDATE);
} else {
/*
* Do not rate limit future attempts to update
* the config cache.
*/
target->spa_ccw_fail_time = 0;
}
/*
* Remove any config entries older than the current one.
*/
dp = list_head(&target->spa_config_list);
while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) {
list_remove(&target->spa_config_list, tdp);
if (tdp->scd_path != NULL)
spa_strfree(tdp->scd_path);
kmem_free(tdp, sizeof (spa_config_dirent_t));
}
spa_config_generation++;
if (postsysevent)
spa_event_notify(target, NULL, ESC_ZFS_CONFIG_SYNC);
}
/*ARGSUSED*/
int
cma_page_retire(fmd_hdl_t *hdl, nvlist_t *nvl, nvlist_t *asru,
const char *uuid, boolean_t repair)
{
cma_page_t *page;
uint64_t pageaddr;
const char *action = repair ? "unretire" : "retire";
int rc;
nvlist_t *rsrc = NULL, *asrucp = NULL, *hcsp;
(void) nvlist_lookup_nvlist(nvl, FM_FAULT_RESOURCE, &rsrc);
if (nvlist_dup(asru, &asrucp, 0) != 0) {
fmd_hdl_debug(hdl, "page retire nvlist dup failed\n");
return (CMA_RA_FAILURE);
}
/* It should already be expanded, but we'll do it again anyway */
if (fmd_nvl_fmri_expand(hdl, asrucp) < 0) {
fmd_hdl_debug(hdl, "failed to expand page asru\n");
cma_stats.bad_flts.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_FAILURE);
}
if (!repair && !fmd_nvl_fmri_present(hdl, asrucp)) {
fmd_hdl_debug(hdl, "page retire overtaken by events\n");
cma_stats.page_nonent.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_SUCCESS);
}
/* Figure out physaddr from resource or asru */
if (rsrc == NULL ||
nvlist_lookup_nvlist(rsrc, FM_FMRI_HC_SPECIFIC, &hcsp) != 0 ||
(nvlist_lookup_uint64(hcsp, "asru-" FM_FMRI_HC_SPECIFIC_PHYSADDR,
&pageaddr) != 0 && nvlist_lookup_uint64(hcsp,
FM_FMRI_HC_SPECIFIC_PHYSADDR, &pageaddr) != 0)) {
if (nvlist_lookup_uint64(asrucp, FM_FMRI_MEM_PHYSADDR,
&pageaddr) != 0) {
fmd_hdl_debug(hdl, "mem fault missing 'physaddr'\n");
cma_stats.bad_flts.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_FAILURE);
}
}
if (repair) {
if (!cma.cma_page_dounretire) {
fmd_hdl_debug(hdl, "suppressed unretire of page %llx\n",
(u_longlong_t)pageaddr);
cma_stats.page_supp.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_SUCCESS);
}
/* If unretire via topo fails, we fall back to legacy way */
if (rsrc == NULL || (rc = fmd_nvl_fmri_unretire(hdl, rsrc)) < 0)
rc = cma_fmri_page_unretire(hdl, asrucp);
} else {
if (!cma.cma_page_doretire) {
fmd_hdl_debug(hdl, "suppressed retire of page %llx\n",
(u_longlong_t)pageaddr);
cma_stats.page_supp.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_FAILURE);
}
/* If retire via topo fails, we fall back to legacy way */
if (rsrc == NULL || (rc = fmd_nvl_fmri_retire(hdl, rsrc)) < 0)
rc = cma_fmri_page_retire(hdl, asrucp);
}
if (rc == FMD_AGENT_RETIRE_DONE) {
fmd_hdl_debug(hdl, "%sd page 0x%llx\n",
action, (u_longlong_t)pageaddr);
if (repair)
cma_stats.page_repairs.fmds_value.ui64++;
else
cma_stats.page_flts.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_SUCCESS);
} else if (repair || rc != FMD_AGENT_RETIRE_ASYNC) {
fmd_hdl_debug(hdl, "%s of page 0x%llx failed, will not "
"retry: %s\n", action, (u_longlong_t)pageaddr,
strerror(errno));
cma_stats.page_fails.fmds_value.ui64++;
nvlist_free(asrucp);
return (CMA_RA_FAILURE);
}
/*
* The page didn't immediately retire. We'll need to periodically
* check to see if it has been retired.
*/
fmd_hdl_debug(hdl, "page didn't retire - sleeping\n");
page = fmd_hdl_zalloc(hdl, sizeof (cma_page_t), FMD_SLEEP);
page->pg_addr = pageaddr;
if (rsrc != NULL)
(void) nvlist_dup(rsrc, &page->pg_rsrc, 0);
page->pg_asru = asrucp;
if (uuid != NULL)
page->pg_uuid = fmd_hdl_strdup(hdl, uuid, FMD_SLEEP);
page->pg_next = cma.cma_pages;
cma.cma_pages = page;
if (cma.cma_page_timerid != 0)
fmd_timer_remove(hdl, cma.cma_page_timerid);
cma.cma_page_curdelay = cma.cma_page_mindelay;
cma.cma_page_timerid =
fmd_timer_install(hdl, NULL, NULL, cma.cma_page_curdelay);
/* Don't free asrucp here. This FMRI will be needed for retry. */
return (CMA_RA_FAILURE);
}
