Go through the ZFS layer instead; given a BE, we can derive the dataset,
zfs_open it, then zfs_unmount. ZFS takes care of the dirty details and
likely gets it more correct than we did for more interesting setups.
MFC after: 3 days
libbe(3) currently uses zfs_be_root and locates which of its children is
currently mounted at "/". This is reasonable, but not correct in the case of
a chroot, for two reasons:
- chroot root may be of a different zpool than zfs_be_root
- chroot root will not show up as mounted at "/"
Fix both of these by rewriting libbe_init to work from the rootfs down.
zfs_path_to_zhandle on / will resolve to the dataset mounted at the new
root, rather than the real root. From there, we can derive the BE root/pool
and grab the bootfs off of the new pool. This does no harm in the average
case, and opens up bectl to operating on different pools for scenarios where
one may be, for instance, updating a pool that generally gets re-rooted into
from a separate UFS root or zfs bootpool.
While here, I've also:
- Eliminated the check for /boot and / to be on the same partition. This
leaves one open to a setup where /boot (and consequently, kernel/modules)
are not included in the boot environment. This may very well be an
intentional setup done by someone that knows what they're doing, we should
not kill BE usage because of it.
- Eliminated the validation bits of BEs and snapshots that enforced
'mountpoint' to be "/" -- this broke when trying to operate on an imported
pool with an altroot, but we need not be this picky.
Reported by: philip
Reviewed by: philip, allanjude (previous version)
Tested by: philip
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D18012
The previously activated BE should have canmount=noauto set on it upon
activation of the new BE, but we previously did not touch canmount on either
old or new BE.
PR: 233113
MFC after: 3 days
Most easily reproducible by attempting to activate the currently activated
BE, one would get a "not a cloned filesystem" error instead of success or a
sane message.
PR: 232488
MFC after: 3 days
This allows older BEs to be destroyed as they become replaced by a BE
created from them: e.g.
bectl create -e brokenworld fixedworld
bectl activate fixedworld
bectl destroy brokenworld
Submitted by: Shawn Webb
Approved by: re (gjb)
Obtained from: HardenedBSD (5948c0581e)
Some paths through be_exists will set the error state, others will not
There are multiple reasons that a call can fail, so clean it up a bit: all
paths now return an appropriate error code so the caller can attempt to
distinguish between a BE legitimately not existing and just having the wrong
mountpoint. The caller is expected to bubble the error through to the
internal error handler as needed.
This fixes some unfriendliness with bectl(8)'s activate subcommand, where
it might fail due to a bad mountpoint but the only message output is a
generic "failed to activate" message.
Approved by: re (gjb)
vermaden (maintainer of beadm) points out the following inconsistencies:
- "missing command" is not printed prior to usage if the error is simply a
missing command; this should be obvious from the context
- "bectl rename" isn't using the "don't unmount" flag (zfs rename -u), so
the active BE can't be renamed. It doesn't make sense in our context to
*not* use -u, so use it.
Documentation updates reflect the above and note an inconsistency with the
'destroy' command that is consistent with other parts of the base system.
A fix for libbe(3) not properly being installed to /lib is included.
SHLIBDIR should have been added when it was moved in r337995.
Approved by: re (kib)
Some background: in the GSoC project, libbe/Makefile lived in lib/libbe. I
created projects/bectl branch, maintained the above for all of five
minutes before I misread Makefile.inc1 and decided that it couldn't possibly
build outside of cddl/, so I kicked the Makefile out into the cddl/ build
and all was good. The misreading was of the bit where .WAIT is added to
SUBDIR after lib, libexec but prior to building bin and cddl *only during
the install targets*, which is the critical part.
Fast forward- buildworld was still broken in my branch unbeknownst to me
because I didn't nuke my OBJDIR. Combing through Makefile.inc1 eventually
revealed the necessary magic to make sure that libbe's dependencies are
specified well enough, and it becomes clear what needs done to make a
non-cddl/ build work. This is an interesting prospect, because the build
split is kind of annoying to work with.
IGNORE_PRAGMA is added to avoid dropping WARNS by one more. This was
previously pulled in via cddl/Makefile.inc.
Previously, we only validated names for character restrictions. This is
helpful, but we should've also checked length restrictions- dataset names
must be restricted to MAXNAMELEN.
While here, move validation before doing a bunch of concatenations and fix
error handling in be_rename. It was previously setting the error state based
on return value from a libzfs function, which is wrong: libzfs errors don't
necessarily match cleanly to libbe errors. This would cause the assertion in
be_error to hit when the error was printed.
Deleting the temp snapshot isn't immediately possible because it's the
origin of the newly imported boot environment. However, this is trivially
solved by opening the new boot environment and promoting it. The roles are
now reversed and the temp snapshot/dataset may be completely destroyed.
Remove the BUGS from libbe(3) and bectl(8).
The mostly-undocumented 'add' functionality, from initial read-through, is
intended for construction of deep ("bdrewery style") boot environments.
However, it's mostly broken at this point. `#if SOON` it out on both sides
so that we're not exposing a broken API/feature.
Work will resume on it in due time.
be_add_child functionality gets split out into separate places as a bonus.
A lot of places here we'll gloss over libzfs errors, because they shouldn't
be happening given the conditions that we're operating under. "Unknown
error" is what I'm intending to use for the moment to indicate an
exceptional circumstance- exceptional enough that we can't tell the consumer
did because we're not so certain that they did anything.
- File names don't necessarily need to be repeated
- Add SPDX tags
- Add a missing copyright for Kyle Kneitinger in bectl.8, originally written
by him in GSoC 2017; his standard copyright notice has been copied from
other files within the same directory to remain consistent with how he
clearly wished to portray it
This makes the be_exists behavior match the comments that assert that we've
already checked that the dataset derived from the BE name is set to mount at
/.
Other changes of note:
- bectl_list sees another change; changing mountpoint based on mount status
turns out to be a bad idea, so instead make the mounted property of the
returned nvlist the path that it's mounted at
- Always return the "mountpoint" property in "mountpoint" if it's ste
Loader is still relied upon at the beginning of libbe to specify the be
root, but we can derive from that the primary zpool and any vdevs that we
need to set nextboot bits on.
This lets me successfully `bectl activate -t test`, but UEFI loader doesn't
quite yet understand so it's effectively defunct.
be_get_dataset_snapshots has been added to libbe(3), effectively returning
the same information as be_get_bootenv_props but for snapshots of the given
dataset. The assumption is that one will have the BE dataset name before
wanting to grab snapshots.
This also accomplishes the following:
- Proxy through zfs_nicenum as be_nicenum, because it looks better than
humanize_number and would presumably be useful to other libbe consumers.
- Rename be_get_snapshot_props to be_get_dataset_props, make it more useful
At a bare minimum, this function will return 0 if a BE is mounted at the
given path or non-zero otherwise. If the optional 'details' nvlist is
supplied, it is filled with an nvpair containing just the information about
the BE mounted at the path. This nvpair is structured just as it is for
be_get_bootenv_props, except limited to just the single mount point.
Based on the idea that we shouldn't have all-new library and utility going
into base that need WARNS=1...
- Decent amount of constification
- Lots of parentheses
- Minor other nits
For the moment, this is a primitive nvlist dump of what we get back from
be_get_bootenv_props as a proof-of-concept and to make sure that we're
getting back the kind of information we want to see from list.
This makes us more resilient to a rename of the bootfs, but still wouldn't
withstand pool renames or guid renames. More importantly, this allows `bectl
create <foo>` work out of the box to create a boot environment based on the
currently booted one.
- Rename 'active' to 'rootfs', which is used in other places to describe the
currently booted (or about to be booted) BE.
- Add 'bootfs', which indicates the next boot environment to be booted. This
is pulled from the BOOTFS zpool property.
- Go ahead and keep an open handle to the active zpool. We might need to
enumerate datasets, get properties, and set properties (e.g. bootfs)
throughout other libbe bits, and a single handle isn't overly expensive.