Andrea Galbusera [Tue, 15 May 2018 06:29:28 +0000 (08:29 +0200)]
u-boot: add support for devicetree
Likewise the corresponding support for syslinux introduced by commit c5112c25e4519835c4cd53f4350c1b2f2a477746, this one enables writing devicetree
filename into the uEnv.txt environment file for u-boot.
Since u-boot does not strictly defines variable names, here 'fdt_file' was
chosen as it appear to be one the most frequently adopted names in u-boot
default environments. Outer boot logic should of course comply with this choice
and use $fdt_file as the device tree file name to pass to boot commands.
This was tested on a custom board booting with u-boot.
Jonathan Lebon [Fri, 11 May 2018 19:03:58 +0000 (15:03 -0400)]
ci: Add .papr-ex.yaml
I'd like to start working towards migrating the OSTree CI towards the
new PAPR model running in PACI. In order to take it for a test drive
(pun intended) without yet replacing our current CI, the new PAPR now
knows to look for a `.papr-ex.yaml` file. So as to not conflict with the
current PAPR, no GitHub statuses will be output other than a
`required-ex` label, which plays the same role as `required`. For more
information, see:
The `.papr-ex.yaml` version is essentially the same except for:
1. the `FAH27-insttests` suite runs directly in the scheduled container
(notice the new `kvm: true` key)
2. the `host:` based contexts (rpm-ostree and flatpak) are missing since
we don't support provisioning a host just yet
Keeping them in sync could get annoying, though building confidence in
the new approach will allow us to migrate faster.
I feel like I'm drowning in a pile of experimental-but-almost-stable
features...
Anyways, since we made the feature opt-in in rpm-ostree in
https://github.com/projectatomic/rpm-ostree/pull/1352
let's mirror that a bit here with an environment variable so people
can play with it more easily.
The tests needed some tweaks; specifically we need to reload the
status fact after making changes. I'm still a bit uncertain
about the Ansible-as-tests.
But we add an upgrade test that uses the new environment variable.
Jonathan Lebon [Mon, 7 May 2018 13:41:27 +0000 (09:41 -0400)]
docs: Add "Hello World" example
Let's get practical faster in the manual and have a simple "Hello World"
example right off the bat to hopefully make it easier to grok how OSTree
works.
Also some minor tweaks on wording around comparisons to git.
Tobias Mueller [Sat, 5 May 2018 20:00:06 +0000 (22:00 +0200)]
repo: handle GPG_ERR_AMBIGUOUS_NAME in sign_data
This should give a more insightful error message if the user provides
a UID which is present on multiple keys.
This happens if you have an old key in your keyring which you are not
actively using any more, e.g. because it is too old. You still have
your old keys in your keyring, because you want to read old email
encrypted for that key, though.
The gpgme function used by ostree right now complains if a UID is found
on multiple keys:
https://www.gnupg.org/documentation/manuals/gpgme/Listing-Keys.html#index-gpgme_005fget_005fkey
The used API is too simple for that use case.
Note that it would be nicer if ostree picked the only valid signing key out
of the available keys rather than using the simplistic gpgme_get_key
function. It be nicer, of course, if there was such a gpgme function.
Colin Walters [Fri, 4 May 2018 17:51:13 +0000 (13:51 -0400)]
tests: Fix installed tests more
OK so I noticed that something was failing and we were missing
`set -xeuo pipefail` in our shells. That of course revealed
the ansible tests didn't actually work - my only defense
here is spending so much time fighting to get it through CI
and trying something new.
Anyways, to make the staged-deploy tests work we need a task
that actually uses `rpm-ostree override` rather than `usroverlay`.
Let's make this a bit saner and have a clean split between
tests that are "shell-script+usroverlay" and "ansible+override".
Colin Walters [Wed, 2 May 2018 15:59:21 +0000 (11:59 -0400)]
deploy: Use fdatasync() for new kernel/initramfs by default
While we do a `syncfs()` plus `FIFREEZE/THAW` for `/boot`, that
only comes during deployment finalization.
The code here today generally assumes that if the file exists
it's been fully written. So let's do a `fdatasync()` before
we do the `rename()`.
This just came out of looking through the code while working
on deployment staging. In that scenario there's a much larger
window between when we copy the kernel/initramfs and when we
sync `/boot`.
Colin Walters [Wed, 2 May 2018 16:17:20 +0000 (12:17 -0400)]
lib: Use `Locking:` term in docs
This is easier to `git grep` etc. versus ad-hoc English. Although
we still have some English for the prepare_transaction/commit which
acquire/release in separate phases.
Colin Walters [Wed, 2 May 2018 14:33:32 +0000 (14:33 +0000)]
lib/deploy: Do post-ops when removing staged commit
These are further fixes based on running more of the rpm-ostree
test suite.
When dropping the staged deployment, we do need to do the
"post operations" such as bumping the sysroot mtime, so that
clients know something changed. We also need to regenerate
the deployment refs. And of course do a sysroot reload.
Also, add a "base cleanup" after creating a staged deployment
which also regenerates the refs.
Colin Walters [Mon, 30 Apr 2018 18:41:30 +0000 (14:41 -0400)]
deploy: Don't prune repo at finalization time by default
Doing so can break rpm-ostree, which wants to own the cleanup process
to ensure its baselayer refs are generated.
Further, doing the cleanup at shutdown time adds latency. It's also
going to be generally unnecessary as we expect repo pruning to have
been done when writing the refs.
Jonathan Lebon [Mon, 30 Apr 2018 12:53:01 +0000 (08:53 -0400)]
man/ostree: Document --repo option a bit more
This new information is already mostly part of `ostree.repo(5)`, though
let's put it in `ostree(1)` as well since that's where the switch is
officially documented.
Colin Walters [Mon, 23 Apr 2018 21:53:04 +0000 (17:53 -0400)]
lib/repo: Enable locking by default, but drop external API
The code has been sitting around for a while but since I disabled
it by default, I doubt anyone is really using it or relying on it.
This patch and turns on locking by default, and also drops the
API which was only public in the experimental API builds.
Conceptually these are two distinct things, and we
may actually want to split up the patches.
I don't think this will break anyone, but it's hard to say for sure.
It's also going to be hard to find out until we actually release
I suspect...
But anyone who is broken should be able to add `locking=false` into
their repo config. On the flip side Endless has been shipping with
this enabled and it is reported to help.
The reason to drop the APIs: I'm a bit concerned about the interactions over time
between libostree's use of the API and any apps that start using it.
For example, if an app specifies a SHARED lock in their code, then
later internally we decide to temporarily grab an `EXCLUSIVE`, but the
app had a second thread/process that was `EXCLUSIVE` already, and
that process was waiting on the first bit of code, then we could
deadlock. I can't think of a real world situation where this would happen
yet though.
We are likely to in the future have say `fsck` take an external lock,
`checkout` grab a shared one, etc.
Colin Walters [Fri, 27 Apr 2018 19:33:02 +0000 (15:33 -0400)]
deploy: Return staged deployment
Today rpm-ostree has some code to run a "sanitycheck" on a deployment.
I had initially deleted that when adapting it to use the staging code,
but I realized it should work fine; we just won't see the merged
config, but that's OK.
When I readded that code it started crashing because we didn't
actually return the new deployment object. We'll gain some coverage
here as I'll land the code to have rpm-ostree use staging, then bump
the rpm-ostree tests here.
The fact that `ostree admin deploy` always itself loaded the
merge kargs masked a bug in the core. Let's change our tests
to not pass any kernel arguments to ensure we cover this.
The new logic in the CLI is a bit subtle, but if you read
carefully is a lot clearer I believe. Basically we have one
of a few "starting points" in the first section, which can
then be further augmented.
Colin Walters [Fri, 27 Apr 2018 15:53:43 +0000 (11:53 -0400)]
lib/deploy: Fix staged deployments with no kargs
Testing out the staged API with rpm-ostree, ostree-prepare-root.service
in the initramfs was failing. Turned out that was because we didn't
have a `root=` kernel argument. Which was because we didn't have
any kernel arguments at all except `ostree=`.
That in turn was because we weren't loading the bootloader config
from the merge deployment.
The serialized deployment data holds the unique identity of
(osname, checksum, deployserial) - look for the real merge deployment
in our deployment list which has the bootloader arguments we need.
This issue was entirely masked by the `ostree admin deploy` command
which itself explicitly loads the merge deployment's kernel arguments
in every case - it never passes the `NULL` default down. A followup
patch will fix that.
Jonathan Lebon [Tue, 24 Apr 2018 16:51:38 +0000 (12:51 -0400)]
lib/deploy: Also compare deployment csum versions
When comparing deployments to determine whether we need a new
bootversion, we should also check whether the commit "version" metadata
is the same. Otherwise, we may end up with the a bootconfig whose
`title` includes a version that doesn't match the one from the
deployment checksum.
Jonathan Lebon [Thu, 19 Apr 2018 13:58:51 +0000 (09:58 -0400)]
tests: Port the last python2 script to python3
This is not used by any test, nor is it packaged. Though let's just port
it over to py3 to certify our codebase completely py2-free. I've
manually checked that the script is still functional.
Jonathan Lebon [Thu, 19 Apr 2018 14:16:35 +0000 (10:16 -0400)]
ci: Temporary libgcrypt workaround for older kernels
I'm trying to get ostree tests to pass in OpenShift as part of our CI
move but I've been seeing lots of failures related to GPG tests. It
finally turned out to be because libgcrypt doesn't behave well on older
kernels that don't have `getrandom()` (the cluster is running on RHEL7).
Thankfully, there's a new build with a fix for this. Pull that in
manually until it gets into stable.
For more information, see:
https://bugzilla.redhat.com/show_bug.cgi?id=1542453
Colin Walters [Thu, 19 Apr 2018 19:53:58 +0000 (15:53 -0400)]
tests/installed: Make reboot task less racy
This took a whole lot of experimentation. I hit upon the idea
of doing a `systemctl stop sshd` to avoid the situation where we
might ssh back into the system while it's in the process of shutting
down.
Ultimately the other fix is disabling `ControlMaster`; see
for example: https://github.com/ansible/ansible/issues/17935
Matthew Leeds [Mon, 26 Mar 2018 07:01:39 +0000 (00:01 -0700)]
create-usb: Update summary in destination repo
Currently the create-usb command only generates a summary file in the
destination repo if one doesn't already exist, which means if one does
exist it becomes out of date after the new refs are pulled. This commit
makes ostree regenerate the summary regardless of whether it exists, so
that consumers such as ostree_repo_find_remotes_async() (and at a higher
level, GNOME Software) get an accurate picture of the refs available on
the mount. This commit also updates one of the unit tests to check that
the summary is accurate after a second pull into the same repo.
Since any user of the create-usb command is using collection IDs they
are new enough to be using the unsigned summary support. While it would
technically be possible to use summary signatures on a repo and use the
create-usb command on it (a scenario broken by this commit), the
create-usb command is designed for P2P distribution of refs, which
requires use of unsigned summary support. So this is a legitimate
narrowing of the tool.
Sam Spilsbury [Tue, 16 Jan 2018 01:23:11 +0000 (09:23 +0800)]
avahi: Don't complain with g_warning if the daemon wasn't running
This is a normal case when running unit tests in client code
on continuous integration infrastructure. When those tests are
running they will set G_DEBUG=fatal-warnings which will cause
the program to abort if a warning is emitted. Instead, emit
a debug message if the problem was that we couldn't connect to
the daemon.
Colin Walters [Thu, 12 Apr 2018 16:40:08 +0000 (12:40 -0400)]
lib/sysroot: Move staged into deployment list, rework handling
Followup to: https://github.com/ostreedev/ostree/pull/1503
After starting some more work on on this in rpm-ostree, it is
actually simpler if the staged deployment just shows up in the list.
It's effectively opt-in today; down the line we may make it the default,
but I worry about breaking things that e.g. assume they can mutate
the deployment before rebooting and have `/etc` already merged.
There's not that many things in libostree that iterate over the deployment
list. The biggest change here is around the
`ostree_sysroot_write_deployments_with_options` API. I initially
tried hard to support a use case like "push a rollback" while retaining
the staged deployment, but everything gets very messy because that
function truly is operating on the bootloader list.
For now what I settled on is to just discard the staged deployment;
down the line we can enhance things.
Where we then have some new gymnastics is around implementing
the finalization; we need to go to some effort to pull the staged
deployment out of the list and mark it as unstaged, and then pass
it down to `write_deployments()`.
This is a version of ostree_repo_traverse_commit_union that also
remembers where the objects came from, by recording the parent
relationships in a hashtable. This can be used to later find which
commits each object was from, which we want to use in fsck.
Differently than the original implementation, now we don't attempt any
test for reflinks support on the parent repository, since the test
requires write access to the repository.
Additionally, also check that the two repositories are on the same
device before attempting any reflink.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com> Closes: #1525
Approved by: cgwalters
Colin Walters [Tue, 3 Apr 2018 20:22:31 +0000 (16:22 -0400)]
switchroot: Don't log if running as pid1, minor code style cleanups
If we're running as pid1, avoid printing anything in the normal
success paths as we don't want to affect the physical console by
default; the device may be using a splash screen, etc.
Also cleanup the code a bit to use a single variable
`running_as_pid1`, declare-and-initialize, use the
`bool` type, etc.
Colin Walters [Thu, 12 Apr 2018 16:53:17 +0000 (12:53 -0400)]
deploy: Silently do nothing if passed same set of deployments
Prep for handling staged deployments better; if we're not passed
the staged one back, then we just want to delete it but not
touch the bootloader config.
Colin Walters [Thu, 12 Apr 2018 16:43:39 +0000 (12:43 -0400)]
deploy: Clean up bootserial assignment function
The reason we were returning a hashtable is a bit lost to history,
there's no reason to do so now anyways. Also port to declare-and-initialize
style and add more comments.
Colin Walters [Thu, 22 Feb 2018 20:27:59 +0000 (15:27 -0500)]
Add concept of "staged" deployment
Add API to write a deployment state to `/run/ostree/staged-deployment`,
along with a systemd service which runs at shutdown time.
This is a big change to the ostree model for hosts,
but it closes a longstanding set of bugs; many, many people have
hit the "losing changes in /etc" problem. It also avoids
the other problem of racing with programs that modify `/etc`
such as LVM backups:
https://bugzilla.redhat.com/show_bug.cgi?id=1365297
We need this in particular to go to a full-on model for
automatically updated host systems where (like a dual-partition model)
everything is fully prepared and the reboot can be taken
asynchronously.
Jonathan Lebon [Wed, 11 Apr 2018 20:45:40 +0000 (16:45 -0400)]
ci: turn gating back on for most testsuites
Somehow, this slipped through in #1513. We weren't inheriting anymore,
so `branches` defaulted back to just `master`. This means we weren't
gating on most of the containerized builds anymore. Ouch!
It'd make sense to teach PAPR to allow some defaults across all
testsuites, even in the `inherit: false` case. Though it's tempting to
also just change the hardcoded PAPR default to those branches since our
use of Homu + PAPR at this point is pretty ubiquitous, and it doesn't
really hurt for the ones that don't use it.
Owen W. Taylor [Wed, 11 Apr 2018 22:06:39 +0000 (18:06 -0400)]
Don't scan uncompressed_objects_dir if it doesn't exist
A newly created archive-mode repository won't have a uncompressed-objects-cache
directory, and uncompressed_objects_dir is -1 to flag that. The special meaning of
-1 meaning "cwd" for libglnx means that the current directory was scanned as
if it was an objects directory, producing unexpected results, especially if there
were any two-letter files/subdirs in the current directory.
Matthew Leeds [Wed, 28 Mar 2018 07:31:05 +0000 (00:31 -0700)]
lib/repo: Add timestamps to OstreeRepoFinderResult
Currently OstreeRepoFinderResult, a data structure used by pull code
that supports P2P operations, has a hash table mapping refs to checksums
but doesn't include timestamp information. This means that clients have
no way of knowing just from the OstreeRepoFinderResult information if a
commit being offered by a peer remote is an update or downgrade until
they start pulling it. The client could check the summary or the commit
metadata for the timestamps, but this requires adding the temporary
remotes to the repo config, and ostree is already checking timestamps
before returning the results, so I think it makes more sense for them to
be returned rather than leaving it to the client. This limitation is
especially important for offline computers, because for online computers
the latest commit available from any remote is the latest commit,
period.
This commit adds a "ref_to_timestamp" hash table to
OstreeRepoFinderResult that is symmetric to "ref_to_checksum" in that it
shares the same keys. This is an API break, but it's part of the
experimental API, and none of the current users of that (flatpak,
eos-updater, and gnome-software) are affected. See the documentation for
more details on "ref_to_timestamp". One thing to note is the data
structure currently gets initialized in find_remotes_cb(), so only users
of ostree_repo_find_remotes_async() will get them, not users of, say,
ostree_repo_finder_resolve_all_async(). This is because the individual
OstreeRepoFinder implementations don't currently access the timestamps
(but I think this could be changed in the future if there's a need).
This commit will allow P2P support to be added to
flatpak_installation_list_installed_refs_for_update, which will allow
GNOME Software to update apps from USB drives while offline (it's
already possible online).
pull: Don't save summary to cache before validating signatures
In case of some kind of race or other weirdness we might be getting
non-matching versions of summary.sig and summary, where summary.sig
is the latest version. Currently we're saving them to the cache
directly after downloading them successfully, but they will then fail
to gpg validate. Then on the next run we'll keep using the cached files
even if they are incorrect, until summary.sig changes upstream.
This changes the order so that we verify the signatures before saving
to the cache, thus ensuring that we don't end up in a stuck state.
Matthew Leeds [Thu, 29 Mar 2018 07:03:43 +0000 (00:03 -0700)]
lib/repo-pull: Improve error message when no summary is found
In ostree_repo_remote_fetch_summary_with_options(), if no summary is
found on the server and summary verification is enabled, the error
message implies that it's the summary signature that's missing, which is
misleading. This commit adds a more specific error message for the case
of a missing summary, which has the side effect of explicitly checking
for the case that signatures != NULL && summary == NULL after
repo_remote_fetch_summary(), even though that should never happen.
One effect of this is that if you run "flatpak remote-add" with an
incorrect URL you get a more helpful error message, and similarly for
other flatpak operations and other users of ostree.
Matthew Leeds [Sat, 31 Mar 2018 01:11:45 +0000 (18:11 -0700)]
lib/repo-pull: Rename a variable for clarity
In libostree, the phrase "commit metadata" has two meanings-- one is the
first dictionary in a commit GVariant that stores metadata such as ref
bindings, and the other is the commit metadata in the summary file,
which stores the commit size, checksum, and timestamp. In
find_remotes_process_refs(), the entire commit GVariant was being
referred to as commit metadata, so this commit changes the variable
name and a comment to make things more consistent.
In _try_clone_from_payload_link, don't try to do the clone in the
parent repo, because we don't want to modify that. parent repos are
typically used when you want a shared, immutable base.
For example in flatpak, the parent repo is the system repo which you
don't have write access to, so any modification to it will fail with
EACCES, making it impossible to install via the system helper.
Matthew Leeds [Thu, 29 Mar 2018 06:11:50 +0000 (23:11 -0700)]
libotutil/checksum-utils: Fix memory management
Ostree uses the OtChecksum data structure as a wrapper around GChecksum
(depending on what libraries are available at compile time). According
to the docs for g_checksum_get_digest(), a GChecksum value can no longer
be updated after that function is called. Ostree enforces this by
setting "initialized" to FALSE after getting the digest, but this leads
to ot_checksum_clear() avoiding freeing any memory, leading to leaks. So
this commit adds a "closed" value that gets set when getting a digest
and checked when updating the value, so the initialized value can be
used only for memory management.
Colin Walters [Mon, 26 Mar 2018 17:23:25 +0000 (13:23 -0400)]
tests/installed: Fix TESTS= being empty
I broke this in https://github.com/ostreedev/ostree/pull/1509/commits/9b55aaea6f34b7094c44932a3c2e1cf2d54634fd
I'd obviously tested *setting* it locally worked, but I didn't test that
not having it set ran all the tests.
I don't understand why we were doing the `+ ` pattern before; let's
just check if it's empty.
Colin Walters [Fri, 23 Mar 2018 20:02:38 +0000 (16:02 -0400)]
lib/deploy: Port final bootconfig writing to new style
The main blocker for doing this before was the `goto out` handling
for remounting `/boot`. Handle that by factoring out the bits that
require it to a helper function, and do the C/GError equivalent of
"try/finally".
Not prep for anything right now, just decided to do this since I had the file
open.
Colin Walters [Fri, 23 Mar 2018 19:33:38 +0000 (15:33 -0400)]
lib/deploy: Split /etc merge into two stages
For staged deploy, we want to pay the cost of creating copies from
`/usr/etc` → `/etc` at stage time, since it can be expensive. (We
want to minimize time spent during shutdown).
Split it up into two functions; the logic is also simply clearer.
Colin Walters [Fri, 23 Mar 2018 19:25:14 +0000 (15:25 -0400)]
lib/deploy: Set kargs in one place
Prep for deployment staging. We had the code to hande "explicit kargs" in one
place, but the "use merge deployment" karg bits mixed in with the "/etc merge"
logic. Those are separate things, and it's better to have karg handling in one
place.
Colin Walters [Thu, 22 Mar 2018 20:52:04 +0000 (16:52 -0400)]
Release 2018.4
A quick turnaround to include one PR: https://github.com/ostreedev/ostree/pull/1508
"switchroot: Ensure /run/ostree-booted is created even without initramfs"
This fixes ostree when booting without an initramfs. Thanks to @akiernan for the
bug report and helping review the fix! I'm working on enhancing
the test suite, which will help in adding some coverage here.
Also for this release I'm going to avoid adding a "stub" symbol section
to the `-released.sym` file; I don't believe it's necessary.
Colin Walters [Thu, 22 Mar 2018 14:14:22 +0000 (10:14 -0400)]
switchroot: Ensure /run/ostree-booted is created even without initramfs
See https://mail.gnome.org/archives/ostree-list/2018-March/msg00012.html
If ostree-prepare-root is run as pid 1 (i.e we're not using an initramfs), then
anything we write outside the target sysroot (such as `/run/ostree-booted`) will
be lost.
Since `ostree-remount.service` runs fairly early in boot, and is triggered via
`ConditionKernelCommandLine=ostree`, we can just touch the file there in
addition.
Colin Walters [Thu, 22 Mar 2018 15:36:33 +0000 (11:36 -0400)]
tests/installed: Use temporary directories more consistently
This is prep for splitting off "nondestructive" tests which
we can run in parallel from the destructive/invasive ones which
e.g. change the host refspec, do deployments.
The `cd` invocation in `prepare_tmpdir` wasn't working because we were running
it in a subshell. Fix this by dropping the subshell.
Colin Walters [Wed, 21 Mar 2018 14:14:12 +0000 (10:14 -0400)]
Release 2018.3
It's been over a month since 2018.2; we have a few features and various fixes,
and the "stage" work pending which is pretty invasive. Time for a new release!
projects / ostree.git / log