* Copyright (c) 2014, 2016 Joyent, Inc. All rights reserved.
* Copyright 2016 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2014, Joyent, Inc. All rights reserved.
- * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2020 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2013 Steven Hartland. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
* Copyright 2017 RackTop Systems.
* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
* Copyright (c) 2019 Datto Inc.
+ * Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved.
+ * Copyright (c) 2019, Klara Inc.
+ * Copyright (c) 2019, Allan Jude
*/
/*
#include <sys/cmn_err.h>
#include <sys/stat.h>
#include <sys/zfs_ioctl.h>
+#include <sys/zfs_quota.h>
#include <sys/zfs_vfsops.h>
#include <sys/zfs_znode.h>
#include <sys/zap.h>
#include <sys/zone.h>
#include <sys/nvpair.h>
#include <sys/pathname.h>
-#include <sys/sdt.h>
#include <sys/fs/zfs.h>
#include <sys/zfs_ctldir.h>
#include <sys/zfs_dir.h>
#include <sys/fm/util.h>
#include <sys/dsl_crypt.h>
#include <sys/rrwlock.h>
+#include <sys/zfs_file.h>
#include <sys/dmu_recv.h>
#include <sys/dmu_send.h>
kmutex_t zfsdev_state_lock;
zfsdev_state_t *zfsdev_state_list;
-extern void zfs_init(void);
-extern void zfs_fini(void);
-
/*
* Limit maximum nvlist size. We don't want users passing in insane values
* for zc->zc_nvlist_src_size, since we will need to allocate that much memory.
+ * Defaults to 0=auto which is handled by platform code.
*/
-#define MAX_NVLIST_SRC_SIZE KMALLOC_MAX_SIZE
+unsigned long zfs_max_nvlist_src_size = 0;
uint_t zfs_fsyncer_key;
uint_t zfs_allow_log_key;
return (buf);
}
-/*
- * Check to see if the named dataset is currently defined as bootable
- */
-static boolean_t
-zfs_is_bootfs(const char *name)
-{
- objset_t *os;
-
- if (dmu_objset_hold(name, FTAG, &os) == 0) {
- boolean_t ret;
- ret = (dmu_objset_id(os) == spa_bootfs(dmu_objset_spa(os)));
- dmu_objset_rele(os, FTAG);
- return (ret);
- }
- return (B_FALSE);
-}
-
/*
* Return non-zero if the spa version is less than requested version.
*/
ZFS_DELEG_PERM_SEND, cr));
}
-int
+static int
zfs_secpolicy_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr)
{
return (SET_ERROR(ENOTSUP));
}
-int
+static int
zfs_secpolicy_smb_acl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr)
{
return (SET_ERROR(ENOTSUP));
{
rrm_exit(&zfsvfs->z_teardown_lock, tag);
- if (zfsvfs->z_sb) {
- deactivate_super(zfsvfs->z_sb);
+ if (zfs_vfs_held(zfsvfs)) {
+ zfs_vfs_rele(zfsvfs);
} else {
dmu_objset_disown(zfsvfs->z_os, B_TRUE, zfsvfs);
zfsvfs_free(zfsvfs);
zfs_ioc_vdev_attach(zfs_cmd_t *zc)
{
spa_t *spa;
- int replacing = zc->zc_cookie;
nvlist_t *config;
+ int replacing = zc->zc_cookie;
+ int rebuild = zc->zc_simple;
int error;
if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
zc->zc_iflags, &config)) == 0) {
- error = spa_vdev_attach(spa, zc->zc_guid, config, replacing);
+ error = spa_vdev_attach(spa, zc->zc_guid, config, replacing,
+ rebuild);
nvlist_free(config);
}
error = dmu_objset_hold(zc->zc_name, FTAG, &os);
if (error != 0) {
- return (error == ENOENT ? ESRCH : error);
+ return (error == ENOENT ? SET_ERROR(ESRCH) : error);
}
/*
case ZFS_PROP_REFRESERVATION:
err = dsl_dataset_set_refreservation(dsname, source, intval);
break;
+ case ZFS_PROP_COMPRESSION:
+ err = dsl_dataset_set_compression(dsname, source, intval);
+ /*
+ * Set err to -1 to force the zfs_set_prop_nvlist code down the
+ * default path to set the value in the nvlist.
+ */
+ if (err == 0)
+ err = -1;
+ break;
case ZFS_PROP_VOLSIZE:
err = zvol_set_volsize(dsname, intval);
break;
zfs_cmd_t *zc;
zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP);
- (void) strcpy(zc->zc_name, dsname);
+ (void) strlcpy(zc->zc_name, dsname,
+ sizeof (zc->zc_name));
(void) zfs_ioc_userspace_upgrade(zc);
(void) zfs_ioc_id_quota_upgrade(zc);
kmem_free(zc, sizeof (zfs_cmd_t));
case PROP_TYPE_INDEX:
if (zfs_prop_index_to_string(prop,
intval, &unused) != 0)
- err = SET_ERROR(EINVAL);
+ err =
+ SET_ERROR(ZFS_ERR_BADPROP);
break;
default:
cmn_err(CE_PANIC,
/*
* Volumes will return EBUSY and cannot be destroyed
- * until all asynchronous minor handling has completed.
- * Wait for the spa_zvol_taskq to drain then retry.
+ * until all asynchronous minor handling (e.g. from
+ * setting the volmode property) has completed. Wait for
+ * the spa_zvol_taskq to drain then retry.
*/
error2 = dsl_destroy_head(fsname);
while ((error2 == EBUSY) && (type == DMU_OST_ZVOL)) {
return (SET_ERROR(EINVAL));
(void) tsd_set(zfs_allow_log_key, NULL);
error = spa_open(poolname, &spa, FTAG);
- strfree(poolname);
+ kmem_strfree(poolname);
if (error != 0)
return (error);
return (error);
}
+/*
+ * 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.
+ */
+/* ARGSUSED */
+static const zfs_ioc_key_t zfs_keys_set_bootenv[] = {
+ {"envmap", DATA_TYPE_STRING, 0},
+};
+
+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);
+ (void) spa_vdev_state_exit(spa, NULL, 0);
+ spa_close(spa, FTAG);
+ return (error);
+}
+
+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)
+{
+ spa_t *spa;
+ int error;
+
+ if ((error = spa_open(name, &spa, FTAG)) != 0)
+ return (error);
+ spa_vdev_state_enter(spa, SCL_ALL);
+ error = vdev_label_read_bootenv(spa->spa_root_vdev, outnvl);
+ (void) spa_vdev_state_exit(spa, NULL, 0);
+ spa_close(spa, FTAG);
+ return (error);
+}
+
/*
* The dp_config_rwlock must not be held when calling this, because the
* unmount may need to write out data.
}
/*
- * Create bookmarks. Bookmark names are of the form <fs>#<bmark>.
- * All bookmarks must be in the same pool.
+ * Create bookmarks. The bookmark names are of the form <fs>#<bmark>.
+ * All bookmarks and snapshots must be in the same pool.
+ * dsl_bookmark_create_nvl_validate describes the nvlist schema in more detail.
*
* innvl: {
- * bookmark1 -> snapshot1, bookmark2 -> snapshot2
+ * new_bookmark1 -> existing_snapshot,
+ * new_bookmark2 -> existing_bookmark,
* }
*
* outnvl: bookmark -> error code (int32)
static int
zfs_ioc_bookmark(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl)
{
- for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL);
- pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) {
- char *snap_name;
-
- /*
- * Verify the snapshot argument.
- */
- if (nvpair_value_string(pair, &snap_name) != 0)
- return (SET_ERROR(EINVAL));
-
-
- /* Verify that the keys (bookmarks) are unique */
- for (nvpair_t *pair2 = nvlist_next_nvpair(innvl, pair);
- pair2 != NULL; pair2 = nvlist_next_nvpair(innvl, pair2)) {
- if (strcmp(nvpair_name(pair), nvpair_name(pair2)) == 0)
- return (SET_ERROR(EINVAL));
- }
- }
-
return (dsl_bookmark_create(innvl, outnvl));
}
return (error);
}
+/*
+ * This ioctl waits for activity of a particular type to complete. If there is
+ * no activity of that type in progress, it returns immediately, and the
+ * returned value "waited" is false. If there is activity in progress, and no
+ * tag is passed in, the ioctl blocks until all activity of that type is
+ * complete, and then returns with "waited" set to true.
+ *
+ * If a thread waiting in the ioctl receives a signal, the call will return
+ * immediately, and the return value will be EINTR.
+ *
+ * innvl: {
+ * "wait_activity" -> int32_t
+ * }
+ *
+ * outnvl: "waited" -> boolean_t
+ */
+static const zfs_ioc_key_t zfs_keys_fs_wait[] = {
+ {ZFS_WAIT_ACTIVITY, DATA_TYPE_INT32, 0},
+};
+
+static int
+zfs_ioc_wait_fs(const char *name, nvlist_t *innvl, nvlist_t *outnvl)
+{
+ int32_t activity;
+ boolean_t waited = B_FALSE;
+ int error;
+ dsl_pool_t *dp;
+ dsl_dir_t *dd;
+ dsl_dataset_t *ds;
+
+ if (nvlist_lookup_int32(innvl, ZFS_WAIT_ACTIVITY, &activity) != 0)
+ return (SET_ERROR(EINVAL));
+
+ if (activity >= ZFS_WAIT_NUM_ACTIVITIES || activity < 0)
+ return (SET_ERROR(EINVAL));
+
+ if ((error = dsl_pool_hold(name, FTAG, &dp)) != 0)
+ return (error);
+
+ if ((error = dsl_dataset_hold(dp, name, FTAG, &ds)) != 0) {
+ dsl_pool_rele(dp, FTAG);
+ return (error);
+ }
+
+ dd = ds->ds_dir;
+ mutex_enter(&dd->dd_activity_lock);
+ dd->dd_activity_waiters++;
+
+ /*
+ * We get a long-hold here so that the dsl_dataset_t and dsl_dir_t
+ * aren't evicted while we're waiting. Normally this is prevented by
+ * holding the pool, but we can't do that while we're waiting since
+ * that would prevent TXGs from syncing out. Some of the functionality
+ * of long-holds (e.g. preventing deletion) is unnecessary for this
+ * case, since we would cancel the waiters before proceeding with a
+ * deletion. An alternative mechanism for keeping the dataset around
+ * could be developed but this is simpler.
+ */
+ dsl_dataset_long_hold(ds, FTAG);
+ dsl_pool_rele(dp, FTAG);
+
+ error = dsl_dir_wait(dd, ds, activity, &waited);
+
+ dsl_dataset_long_rele(ds, FTAG);
+ dd->dd_activity_waiters--;
+ if (dd->dd_activity_waiters == 0)
+ cv_signal(&dd->dd_activity_cv);
+ mutex_exit(&dd->dd_activity_lock);
+
+ dsl_dataset_rele(ds, FTAG);
+
+ if (error == 0)
+ fnvlist_add_boolean_value(outnvl, ZFS_WAIT_WAITED, waited);
+
+ return (error);
+}
+
/*
* fsname is name of dataset to rollback (to most recent snapshot)
*
resume_err = zfs_resume_fs(zfsvfs, ds);
error = error ? error : resume_err;
}
- deactivate_super(zfsvfs->z_sb);
+ zfs_vfs_rele(zfsvfs);
} else if ((zv = zvol_suspend(fsname)) != NULL) {
error = dsl_dataset_rollback(fsname, target, zvol_tag(zv),
outnvl);
fullname = kmem_asprintf("%s@%s", fsname, snapname);
zfs_unmount_snap(fullname);
- strfree(fullname);
+ kmem_strfree(fullname);
return (0);
}
* snapname is the snapshot to redact.
* innvl: {
* "bookname" -> (string)
- * name of the redaction bookmark to generate
+ * shortname of the redaction bookmark to generate
* "snapnv" -> (nvlist, values ignored)
* snapshots to redact snapname with respect to
* }
const char *propname = nvpair_name(pair);
boolean_t issnap = (strchr(dsname, '@') != NULL);
zfs_prop_t prop = zfs_name_to_prop(propname);
- uint64_t intval;
+ uint64_t intval, compval;
int err;
if (prop == ZPROP_INVAL) {
* we'll catch them later.
*/
if (nvpair_value_uint64(pair, &intval) == 0) {
- if (intval >= ZIO_COMPRESS_GZIP_1 &&
- intval <= ZIO_COMPRESS_GZIP_9 &&
+ compval = ZIO_COMPRESS_ALGO(intval);
+ if (compval >= ZIO_COMPRESS_GZIP_1 &&
+ compval <= ZIO_COMPRESS_GZIP_9 &&
zfs_earlier_version(dsname,
SPA_VERSION_GZIP_COMPRESSION)) {
return (SET_ERROR(ENOTSUP));
}
- if (intval == ZIO_COMPRESS_ZLE &&
+ if (compval == ZIO_COMPRESS_ZLE &&
zfs_earlier_version(dsname,
SPA_VERSION_ZLE_COMPRESSION))
return (SET_ERROR(ENOTSUP));
- if (intval == ZIO_COMPRESS_LZ4) {
+ if (compval == ZIO_COMPRESS_LZ4) {
spa_t *spa;
if ((err = spa_open(dsname, &spa, FTAG)) != 0)
spa_close(spa, FTAG);
}
- /*
- * If this is a bootable dataset then
- * verify that the compression algorithm
- * is supported for booting. We must return
- * something other than ENOTSUP since it
- * implies a downrev pool version.
- */
- if (zfs_is_bootfs(dsname) &&
- !BOOTFS_COMPRESS_VALID(intval)) {
- return (SET_ERROR(ERANGE));
+ if (compval == ZIO_COMPRESS_ZSTD) {
+ spa_t *spa;
+
+ if ((err = spa_open(dsname, &spa, FTAG)) != 0)
+ return (err);
+
+ if (!spa_feature_is_enabled(spa,
+ SPA_FEATURE_ZSTD_COMPRESS)) {
+ spa_close(spa, FTAG);
+ return (SET_ERROR(ENOTSUP));
+ }
+ spa_close(spa, FTAG);
}
}
break;
intval != ZFS_DNSIZE_LEGACY) {
spa_t *spa;
- /*
- * If this is a bootable dataset then
- * we don't allow large (>512B) dnodes,
- * because GRUB doesn't support them.
- */
- if (zfs_is_bootfs(dsname) &&
- intval != ZFS_DNSIZE_LEGACY) {
- return (SET_ERROR(EDOM));
- }
-
if ((err = spa_open(dsname, &spa, FTAG)) != 0)
return (err);
char *poolname = arg;
if (poolname != NULL)
- strfree(poolname);
+ kmem_strfree(poolname);
}
-#ifdef DEBUG
+#ifdef ZFS_DEBUG
static boolean_t zfs_ioc_recv_inject_err;
#endif
static int
zfs_ioc_recv_impl(char *tofs, char *tosnap, char *origin, nvlist_t *recvprops,
nvlist_t *localprops, nvlist_t *hidden_args, boolean_t force,
- boolean_t resumable, int input_fd, dmu_replay_record_t *begin_record,
- int cleanup_fd, uint64_t *read_bytes, uint64_t *errflags,
- uint64_t *action_handle, nvlist_t **errors)
+ boolean_t resumable, int input_fd,
+ dmu_replay_record_t *begin_record, uint64_t *read_bytes,
+ uint64_t *errflags, nvlist_t **errors)
{
dmu_recv_cookie_t drc;
int error = 0;
int props_error = 0;
- offset_t off;
+ offset_t off, noff;
nvlist_t *local_delayprops = NULL;
nvlist_t *recv_delayprops = NULL;
nvlist_t *origprops = NULL; /* existing properties */
nvlist_t *origrecvd = NULL; /* existing received properties */
boolean_t first_recvd_props = B_FALSE;
boolean_t tofs_was_redacted;
- file_t *input_fp;
+ zfs_file_t *input_fp;
*read_bytes = 0;
*errflags = 0;
*errors = fnvlist_alloc();
+ off = 0;
- input_fp = getf(input_fd);
- if (input_fp == NULL)
- return (SET_ERROR(EBADF));
+ if ((error = zfs_file_get(input_fd, &input_fp)))
+ return (error);
- off = input_fp->f_offset;
+ noff = off = zfs_file_off(input_fp);
error = dmu_recv_begin(tofs, tosnap, begin_record, force,
- resumable, localprops, hidden_args, origin, &drc, input_fp->f_vnode,
+ resumable, localprops, hidden_args, origin, &drc, input_fp,
&off);
if (error != 0)
goto out;
nvlist_free(xprops);
}
- error = dmu_recv_stream(&drc, cleanup_fd, action_handle, &off);
+ error = dmu_recv_stream(&drc, &off);
if (error == 0) {
zfsvfs_t *zfsvfs = NULL;
error = zfs_resume_fs(zfsvfs, ds);
}
error = error ? error : end_err;
- deactivate_super(zfsvfs->z_sb);
+ zfs_vfs_rele(zfsvfs);
} else if ((zv = zvol_suspend(tofs)) != NULL) {
error = dmu_recv_end(&drc, zvol_tag(zv));
zvol_resume(zv);
ASSERT(nvlist_merge(localprops, local_delayprops, 0) == 0);
nvlist_free(local_delayprops);
}
+ *read_bytes = off - noff;
- *read_bytes = off - input_fp->f_offset;
- if (VOP_SEEK(input_fp->f_vnode, input_fp->f_offset, &off, NULL) == 0)
- input_fp->f_offset = off;
-
-#ifdef DEBUG
+#ifdef ZFS_DEBUG
if (zfs_ioc_recv_inject_err) {
zfs_ioc_recv_inject_err = B_FALSE;
error = 1;
nvlist_free(inheritprops);
}
out:
- releasef(input_fd);
+ zfs_file_put(input_fd);
nvlist_free(origrecvd);
nvlist_free(origprops);
* zc_cookie file descriptor to recv from
* zc_begin_record the BEGIN record of the stream (not byteswapped)
* zc_guid force flag
- * zc_cleanup_fd cleanup-on-exit file descriptor
- * zc_action_handle handle for this guid/ds mapping (or zero on first call)
*
* outputs:
* zc_cookie number of bytes read
* zc_obj zprop_errflags_t
- * zc_action_handle handle for this guid/ds mapping
* zc_nvlist_dst{_size} error for each unapplied received property
*/
static int
error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvdprops, localprops,
NULL, zc->zc_guid, B_FALSE, zc->zc_cookie, &begin_record,
- zc->zc_cleanup_fd, &zc->zc_cookie, &zc->zc_obj,
- &zc->zc_action_handle, &errors);
+ &zc->zc_cookie, &zc->zc_obj, &errors);
nvlist_free(recvdprops);
nvlist_free(localprops);
* "input_fd" -> file descriptor to read stream from (int32)
* (optional) "force" -> force flag (value ignored)
* (optional) "resumable" -> resumable flag (value ignored)
- * (optional) "cleanup_fd" -> cleanup-on-exit file descriptor
- * (optional) "action_handle" -> handle for this guid/ds mapping
+ * (optional) "cleanup_fd" -> unused
+ * (optional) "action_handle" -> unused
* (optional) "hidden_args" -> { "wkeydata" -> value }
* }
*
* outnvl: {
* "read_bytes" -> number of bytes read
* "error_flags" -> zprop_errflags_t
- * "action_handle" -> handle for this guid/ds mapping
* "errors" -> error for each unapplied received property (nvlist)
* }
*/
char tofs[ZFS_MAX_DATASET_NAME_LEN];
boolean_t force;
boolean_t resumable;
- uint64_t action_handle = 0;
uint64_t read_bytes = 0;
uint64_t errflags = 0;
int input_fd = -1;
- int cleanup_fd = -1;
int error;
snapname = fnvlist_lookup_string(innvl, "snapname");
strchr(snapname, '%'))
return (SET_ERROR(EINVAL));
- (void) strcpy(tofs, snapname);
+ (void) strlcpy(tofs, snapname, sizeof (tofs));
tosnap = strchr(tofs, '@');
*tosnap++ = '\0';
force = nvlist_exists(innvl, "force");
resumable = nvlist_exists(innvl, "resumable");
- error = nvlist_lookup_int32(innvl, "cleanup_fd", &cleanup_fd);
- if (error && error != ENOENT)
- return (error);
-
- error = nvlist_lookup_uint64(innvl, "action_handle", &action_handle);
- if (error && error != ENOENT)
- return (error);
-
/* we still use "props" here for backwards compatibility */
error = nvlist_lookup_nvlist(innvl, "props", &recvprops);
if (error && error != ENOENT)
return (error);
error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvprops, localprops,
- hidden_args, force, resumable, input_fd, begin_record, cleanup_fd,
- &read_bytes, &errflags, &action_handle, &errors);
+ hidden_args, force, resumable, input_fd, begin_record,
+ &read_bytes, &errflags, &errors);
fnvlist_add_uint64(outnvl, "read_bytes", read_bytes);
fnvlist_add_uint64(outnvl, "error_flags", errflags);
- fnvlist_add_uint64(outnvl, "action_handle", action_handle);
fnvlist_add_nvlist(outnvl, "errors", errors);
nvlist_free(errors);
}
typedef struct dump_bytes_io {
- vnode_t *dbi_vp;
- void *dbi_buf;
+ zfs_file_t *dbi_fp;
+ caddr_t dbi_buf;
int dbi_len;
int dbi_err;
} dump_bytes_io_t;
dump_bytes_cb(void *arg)
{
dump_bytes_io_t *dbi = (dump_bytes_io_t *)arg;
- ssize_t resid; /* have to get resid to get detailed errno */
+ zfs_file_t *fp;
+ caddr_t buf;
+
+ fp = dbi->dbi_fp;
+ buf = dbi->dbi_buf;
- dbi->dbi_err = vn_rdwr(UIO_WRITE, dbi->dbi_vp,
- (caddr_t)dbi->dbi_buf, dbi->dbi_len,
- 0, UIO_SYSSPACE, FAPPEND, RLIM64_INFINITY, CRED(), &resid);
+ dbi->dbi_err = zfs_file_write(fp, buf, dbi->dbi_len, NULL);
}
static int
{
dump_bytes_io_t dbi;
- dbi.dbi_vp = arg;
+ dbi.dbi_fp = arg;
dbi.dbi_buf = buf;
dbi.dbi_len = len;
boolean_t large_block_ok = (zc->zc_flags & 0x2);
boolean_t compressok = (zc->zc_flags & 0x4);
boolean_t rawok = (zc->zc_flags & 0x8);
+ boolean_t savedok = (zc->zc_flags & 0x10);
if (zc->zc_obj != 0) {
dsl_pool_t *dp;
}
error = dmu_send_estimate_fast(tosnap, fromsnap, NULL,
- compressok || rawok, &zc->zc_objset_type);
+ compressok || rawok, savedok, &zc->zc_objset_type);
if (fromsnap != NULL)
dsl_dataset_rele(fromsnap, FTAG);
dsl_dataset_rele(tosnap, FTAG);
dsl_pool_rele(dp, FTAG);
} else {
- file_t *fp = getf(zc->zc_cookie);
- if (fp == NULL)
- return (SET_ERROR(EBADF));
-
- off = fp->f_offset;
+ zfs_file_t *fp;
dmu_send_outparams_t out = {0};
+
+ if ((error = zfs_file_get(zc->zc_cookie, &fp)))
+ return (error);
+
+ off = zfs_file_off(fp);
out.dso_outfunc = dump_bytes;
- out.dso_arg = fp->f_vnode;
+ out.dso_arg = fp;
out.dso_dryrun = B_FALSE;
error = dmu_send_obj(zc->zc_name, zc->zc_sendobj,
- zc->zc_fromobj, embedok, large_block_ok, compressok, rawok,
- zc->zc_cookie, &off, &out);
+ zc->zc_fromobj, embedok, large_block_ok, compressok,
+ rawok, savedok, zc->zc_cookie, &off, &out);
- if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0)
- fp->f_offset = off;
- releasef(zc->zc_cookie);
+ zfs_file_put(zc->zc_cookie);
}
return (error);
}
for (dsp = list_head(&ds->ds_sendstreams); dsp != NULL;
dsp = list_next(&ds->ds_sendstreams, dsp)) {
if (dsp->dss_outfd == zc->zc_cookie &&
- dsp->dss_proc == curproc)
+ zfs_proc_is_caller(dsp->dss_proc))
break;
}
} else {
vd = spa_lookup_by_guid(spa, zc->zc_guid, B_TRUE);
if (vd == NULL) {
- (void) spa_vdev_state_exit(spa, NULL, ENODEV);
+ error = SET_ERROR(ENODEV);
+ (void) spa_vdev_state_exit(spa, NULL, error);
spa_close(spa, FTAG);
- return (SET_ERROR(ENODEV));
+ return (error);
}
}
}
if (error == 0)
error = dmu_objset_userspace_upgrade(zfsvfs->z_os);
- deactivate_super(zfsvfs->z_sb);
+ zfs_vfs_rele(zfsvfs);
} else {
/* XXX kind of reading contents without owning */
error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os);
if (error == 0)
(void) strlcpy(zc->zc_value, snap_name,
sizeof (zc->zc_value));
- strfree(snap_name);
- strfree(hold_name);
+ kmem_strfree(snap_name);
+ kmem_strfree(hold_name);
zfs_onexit_fd_rele(zc->zc_cleanup_fd);
return (error);
}
static int
zfs_ioc_diff(zfs_cmd_t *zc)
{
- file_t *fp;
+ zfs_file_t *fp;
offset_t off;
int error;
- fp = getf(zc->zc_cookie);
- if (fp == NULL)
- return (SET_ERROR(EBADF));
-
- off = fp->f_offset;
+ if ((error = zfs_file_get(zc->zc_cookie, &fp)))
+ return (error);
- error = dmu_diff(zc->zc_name, zc->zc_value, fp->f_vnode, &off);
+ off = zfs_file_off(fp);
+ error = dmu_diff(zc->zc_name, zc->zc_value, fp, &off);
- if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0)
- fp->f_offset = off;
- releasef(zc->zc_cookie);
+ zfs_file_put(zc->zc_cookie);
return (error);
}
* presence indicates compressed DRR_WRITE records are permitted
* (optional) "rawok" -> (value ignored)
* presence indicates raw encrypted records should be used.
+ * (optional) "savedok" -> (value ignored)
+ * presence indicates we should send a partially received snapshot
* (optional) "resume_object" and "resume_offset" -> (uint64)
* if present, resume send stream from specified object and offset.
* (optional) "redactbook" -> (string)
{"embedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
{"compressok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
{"rawok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
+ {"savedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
{"resume_object", DATA_TYPE_UINT64, ZK_OPTIONAL},
{"resume_offset", DATA_TYPE_UINT64, ZK_OPTIONAL},
{"redactbook", DATA_TYPE_STRING, ZK_OPTIONAL},
offset_t off;
char *fromname = NULL;
int fd;
- file_t *fp;
+ zfs_file_t *fp;
boolean_t largeblockok;
boolean_t embedok;
boolean_t compressok;
boolean_t rawok;
+ boolean_t savedok;
uint64_t resumeobj = 0;
uint64_t resumeoff = 0;
char *redactbook = NULL;
embedok = nvlist_exists(innvl, "embedok");
compressok = nvlist_exists(innvl, "compressok");
rawok = nvlist_exists(innvl, "rawok");
+ savedok = nvlist_exists(innvl, "savedok");
(void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj);
(void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff);
(void) nvlist_lookup_string(innvl, "redactbook", &redactbook);
- if ((fp = getf(fd)) == NULL)
- return (SET_ERROR(EBADF));
+ if ((error = zfs_file_get(fd, &fp)))
+ return (error);
+
+ off = zfs_file_off(fp);
- off = fp->f_offset;
dmu_send_outparams_t out = {0};
out.dso_outfunc = dump_bytes;
- out.dso_arg = fp->f_vnode;
+ out.dso_arg = fp;
out.dso_dryrun = B_FALSE;
- error = dmu_send(snapname, fromname, embedok, largeblockok, compressok,
- rawok, resumeobj, resumeoff, redactbook, fd, &off, &out);
+ error = dmu_send(snapname, fromname, embedok, largeblockok,
+ compressok, rawok, savedok, resumeobj, resumeoff,
+ redactbook, fd, &off, &out);
- if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0)
- fp->f_offset = off;
-
- releasef(fd);
+ zfs_file_put(fd);
return (error);
}
/* ARGSUSED */
-int
+static int
send_space_sum(objset_t *os, void *buf, int len, void *arg)
{
uint64_t *size = arg;
boolean_t embedok;
boolean_t compressok;
boolean_t rawok;
+ boolean_t savedok;
uint64_t space = 0;
boolean_t full_estimate = B_FALSE;
uint64_t resumeobj = 0;
embedok = nvlist_exists(innvl, "embedok");
compressok = nvlist_exists(innvl, "compressok");
rawok = nvlist_exists(innvl, "rawok");
+ savedok = nvlist_exists(innvl, "savedok");
boolean_t from = (nvlist_lookup_string(innvl, "from", &fromname) == 0);
boolean_t altbook = (nvlist_lookup_string(innvl, "redactbook",
&redactlist_book) == 0);
dsl_dataset_rele(tosnap, FTAG);
dsl_pool_rele(dp, FTAG);
error = dmu_send(snapname, fromname, embedok, largeblockok,
- compressok, rawok, resumeobj, resumeoff, redactlist_book,
- fd, &off, &out);
+ compressok, rawok, savedok, resumeobj, resumeoff,
+ redactlist_book, fd, &off, &out);
} else {
error = dmu_send_estimate_fast(tosnap, fromsnap,
(from && strchr(fromname, '#') != NULL ? &zbm : NULL),
- compressok || rawok, &space);
+ compressok || rawok, savedok, &space);
space -= resume_bytes;
if (fromsnap != NULL)
dsl_dataset_rele(fromsnap, FTAG);
DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY);
}
-void
+static void
zfs_ioctl_init(void)
{
zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT,
POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE,
zfs_keys_pool_wait, ARRAY_SIZE(zfs_keys_pool_wait));
+ zfs_ioctl_register("wait_fs", ZFS_IOC_WAIT_FS,
+ zfs_ioc_wait_fs, zfs_secpolicy_none, DATASET_NAME,
+ POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE,
+ zfs_keys_fs_wait, ARRAY_SIZE(zfs_keys_fs_wait));
+
+ zfs_ioctl_register("set_bootenv", ZFS_IOC_SET_BOOTENV,
+ zfs_ioc_set_bootenv, zfs_secpolicy_config, POOL_NAME,
+ POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE,
+ zfs_keys_set_bootenv, ARRAY_SIZE(zfs_keys_set_bootenv));
+
+ zfs_ioctl_register("get_bootenv", ZFS_IOC_GET_BOOTENV,
+ zfs_ioc_get_bootenv, zfs_secpolicy_none, POOL_NAME,
+ POOL_CHECK_SUSPENDED, B_FALSE, B_TRUE,
+ zfs_keys_get_bootenv, ARRAY_SIZE(zfs_keys_get_bootenv));
+
/* IOCTLS that use the legacy function signature */
zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze,
return (0);
}
-int
+static int
pool_status_check(const char *name, zfs_ioc_namecheck_t type,
zfs_ioc_poolcheck_t check)
{
return (error);
}
+int
+zfsdev_getminor(int fd, minor_t *minorp)
+{
+ zfsdev_state_t *zs, *fpd;
+ zfs_file_t *fp;
+ int rc;
+
+ ASSERT(!MUTEX_HELD(&zfsdev_state_lock));
+
+ if ((rc = zfs_file_get(fd, &fp)))
+ return (rc);
+
+ fpd = zfs_file_private(fp);
+ if (fpd == NULL)
+ return (SET_ERROR(EBADF));
+
+ mutex_enter(&zfsdev_state_lock);
+
+ for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) {
+
+ if (zs->zs_minor == -1)
+ continue;
+
+ if (fpd == zs) {
+ *minorp = fpd->zs_minor;
+ mutex_exit(&zfsdev_state_lock);
+ return (0);
+ }
+ }
+
+ mutex_exit(&zfsdev_state_lock);
+
+ return (SET_ERROR(EBADF));
+}
+
static void *
zfsdev_get_state_impl(minor_t minor, enum zfsdev_state_type which)
{
}
long
-zfsdev_ioctl_common(uint_t vecnum, unsigned long arg)
+zfsdev_ioctl_common(uint_t vecnum, zfs_cmd_t *zc, int flag)
{
- zfs_cmd_t *zc;
- int error, cmd, rc, flag = 0;
+ int error, cmd;
const zfs_ioc_vec_t *vec;
char *saved_poolname = NULL;
+ uint64_t max_nvlist_src_size;
+ size_t saved_poolname_len = 0;
nvlist_t *innvl = NULL;
fstrans_cookie_t cookie;
cmd = vecnum;
+ error = 0;
if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0]))
- return (-SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));
+ return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));
+
vec = &zfs_ioc_vec[vecnum];
/*
* a normal or legacy handler are registered.
*/
if (vec->zvec_func == NULL && vec->zvec_legacy_func == NULL)
- return (-SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));
-
- zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP);
-
- error = ddi_copyin((void *)(uintptr_t)arg, zc, sizeof (zfs_cmd_t),
- flag);
- if (error != 0) {
- error = SET_ERROR(EFAULT);
- goto out;
- }
+ return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));
zc->zc_iflags = flag & FKIOCTL;
- if (zc->zc_nvlist_src_size > MAX_NVLIST_SRC_SIZE) {
+ max_nvlist_src_size = zfs_max_nvlist_src_size_os();
+ if (zc->zc_nvlist_src_size > max_nvlist_src_size) {
/*
* Make sure the user doesn't pass in an insane value for
* zc_nvlist_src_size. We have to check, since we will end
goto out;
/* legacy ioctls can modify zc_name */
- saved_poolname = strdup(zc->zc_name);
- if (saved_poolname == NULL) {
- error = SET_ERROR(ENOMEM);
- goto out;
- } else {
- saved_poolname[strcspn(saved_poolname, "/@#")] = '\0';
- }
+ /*
+ * Can't use kmem_strdup() as we might truncate the string and
+ * kmem_strfree() would then free with incorrect size.
+ */
+ saved_poolname_len = strlen(zc->zc_name) + 1;
+ saved_poolname = kmem_alloc(saved_poolname_len, KM_SLEEP);
+
+ strlcpy(saved_poolname, zc->zc_name, saved_poolname_len);
+ saved_poolname[strcspn(saved_poolname, "/@#")] = '\0';
if (vec->zvec_func != NULL) {
nvlist_t *outnvl;
out:
nvlist_free(innvl);
- rc = ddi_copyout(zc, (void *)(uintptr_t)arg, sizeof (zfs_cmd_t), flag);
- if (error == 0 && rc != 0)
- error = SET_ERROR(EFAULT);
if (error == 0 && vec->zvec_allow_log) {
char *s = tsd_get(zfs_allow_log_key);
if (s != NULL)
- strfree(s);
- (void) tsd_set(zfs_allow_log_key, saved_poolname);
- } else {
- if (saved_poolname != NULL)
- strfree(saved_poolname);
+ kmem_strfree(s);
+ (void) tsd_set(zfs_allow_log_key, kmem_strdup(saved_poolname));
}
+ if (saved_poolname != NULL)
+ kmem_free(saved_poolname, saved_poolname_len);
- kmem_free(zc, sizeof (zfs_cmd_t));
- return (-error);
+ return (error);
}
int
if ((error = zvol_init()) != 0)
return (error);
- spa_init(FREAD | FWRITE);
+ spa_init(SPA_MODE_READ | SPA_MODE_WRITE);
zfs_init();
zfs_ioctl_init();
void
zfs_kmod_fini(void)
{
- zfsdev_state_t *zs, *zsprev = NULL;
+ zfsdev_state_t *zs, *zsnext = NULL;
zfsdev_detach();
mutex_destroy(&zfsdev_state_lock);
- for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) {
- if (zsprev)
- kmem_free(zsprev, sizeof (zfsdev_state_t));
- zsprev = zs;
+ for (zs = zfsdev_state_list; zs != NULL; zs = zsnext) {
+ zsnext = zs->zs_next;
+ if (zs->zs_onexit)
+ zfs_onexit_destroy(zs->zs_onexit);
+ if (zs->zs_zevent)
+ zfs_zevent_destroy(zs->zs_zevent);
+ kmem_free(zs, sizeof (zfsdev_state_t));
}
- if (zsprev)
- kmem_free(zsprev, sizeof (zfsdev_state_t));
zfs_fini();
spa_fini();
tsd_destroy(&rrw_tsd_key);
tsd_destroy(&zfs_allow_log_key);
}
+
+/* BEGIN CSTYLED */
+ZFS_MODULE_PARAM(zfs, zfs_, max_nvlist_src_size, ULONG, ZMOD_RW,
+ "Maximum size in bytes allowed for src nvlist passed with ZFS ioctls");
+/* END CSTYLED */
projects / mirror_zfs.git / blobdiff