Jonathan Lebon [Fri, 22 Apr 2016 16:24:04 +0000 (12:24 -0400)]
ostree-repo-libarchive.c: major refactor
- Make hardlink handling more generic. The previous strategy worked for
tar archives, but not for cpio. It now works for both.
- Add support for SEL labeling (through the OstreeRepoCommitModifier)
- Add support for xattr_callback (through the OstreeRepoCommitModifier)
- Add support for filter (through the OstreeRepoCommitModifier)
- Add a use_ostree_convention option
Colin Walters [Tue, 3 May 2016 21:24:25 +0000 (17:24 -0400)]
libglnx porting: Migrate to glnx_stream_fstat()
I ended up deciding to move this one into libglnx, seems like
something other libglnx-using software might want to do, even though
xdg-app doesn't right now.
Colin Walters [Wed, 13 Jan 2016 16:33:54 +0000 (11:33 -0500)]
repo: Clean up staging directory for previous boot IDs
We had a policy of cleaning up all files in `$repo/tmp` older
than one day, but we should really clean up previous bootid staging
directories too, as they can potentially take up a lot of disk space.
Colin Walters [Tue, 19 Apr 2016 16:11:01 +0000 (12:11 -0400)]
repo: Add OSTREE_REPO_TEST_ERROR=pre-commit env var
Setting this causes commit to error out. There are other ways we
could do this in a more sophisticated fashion, such as via SystemTap
etc. But this has low-tech applicablity, works as non-root.
The reason I'm adding this is so that we can add test cases for
cleanup of the `tmp/staging-` directory.
Colin Walters [Sun, 1 May 2016 18:22:52 +0000 (14:22 -0400)]
libglnx porting: Migrate from GSConsole
To GLnxConsoleRef. There were some subtleties here, for example we
used to reference `GSConsole` inside the progress changed function,
which at first seems like an ABI hazard, because e.g. rpm-ostree or
xdg-app could still be passing a `GSConsole` instance there. Luckily,
it turns out to be compatible to just start calling libglnx here.
Another issue was that due to libglnx's use of the cleanup function,
we needed to ensure we always called `ostree_async_progress_finish()`
*before* the cleanup function was invoked.
Colin Walters [Mon, 25 Apr 2016 19:59:36 +0000 (15:59 -0400)]
deploy: Start hardlinking kernel/initramfs on single-part again
Commit
https://github.com/ostreedev/ostree/commit/1810de2b51680dbf35fdbd33d0a8d7e65eadc91f
lost an optimization where we would try hardlinks for the
kernel/initramfs in `/boot`. This would be a noticeable space savings
on single-partition systems.
repo: Handle parent repos with different remote configuration
In the case we have a repo with a parent, and the child repo has a
remote called "foo", but some option is unset. Then when we look up
the parent repo for a value before using the default we will fail due
to the parent not having the "foo" remote. As soon as we find the
requested remote at some point in the hierarchy we need to ignore further
errors and use the default value.
We looked for and locked old temporary directories so we can
reuse them if not in use. However, once we found one that
we can reuse we didn't stop iterating, and eventually we
reached the end. This means we can lock multiple dirs.
Colin Walters [Thu, 21 Apr 2016 19:14:51 +0000 (15:14 -0400)]
repo: Simplify internal has_object() lookup code
There was some leftover intermediate cruft here I noticed
while reviewing another patch:
- We had an output `GFile*` for that was never used
- We required the caller to allocate the loose pathbuf, but
none of them ever reused it
- We had an extra intermediate function
Also while looking at this, I'm now uncertain whether some of the
callers of `_ostree_repo_has_loose_object` should really be invoking
`ostree_repo_has_object()`, but let's leave that aside for now.
This lets you set a prefix for the resulting archive patsh.
Especially useful in combination with --subpath, for instance
--subpath=subdir --prefix=subdir to extract just subdir.
Colin Walters [Mon, 18 Apr 2016 14:25:32 +0000 (10:25 -0400)]
libtest: Skip FUSE if CAP_SYS_ADMIN not in bounding set
I'm not sure if something changed or there was a race condition , or
if extraterrestrials intervened to make the tests pass before, but
in the Docker container `/dev/fuse` does exist and since we appear
to be running as uid 0 it'll be writable.
We previously discussed that another requirement for FUSE (and most
setuid binaries) is having `CAP_SYS_ADMIN`, so let's check for that in
the bounding set too.
Simon McVittie [Sun, 17 Apr 2016 08:57:06 +0000 (09:57 +0100)]
libtest: show files' contents when assertions about them fail
I've seen an intermittent test failure in an autobuilder (sbuild)
environment where logs from failed builds cannot be retrieved,
but I can no longer reproduce it. Put the contents of the offending
file in the test's failing output so that if the failure comes back,
it can be debugged.
Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #264
Approved by: cgwalters
Colin Walters [Fri, 15 Apr 2016 19:49:05 +0000 (15:49 -0400)]
Rename test-abi to test-symbols, start verifying symbols are documented
We keep forgetting to update `apidoc/ostree-sections.txt`, so let's
start enforcing it. Of course it turns out we had some bugs here
like symbols marked as public but never implemented, etc. Those
are fixed in the prior commits.
Colin Walters [Fri, 15 Apr 2016 19:20:57 +0000 (15:20 -0400)]
core: Remove ostree_checksum_update_meta that was not implemented
If a symbol falls in a git merkle tree forest, but no one hears it,
did it ever exist?
I'm claiming this isn't an API/ABI break because nothing could
have actually used this. `git log -S checksum_update_meta` leads
me to 38ef75e6e04f7ec19844f9a1c178e6066682d13d which was before
we really had a stable shared library.
This allows you to replace the default
$sysroot/$sysconfdir/ostree/repos.d string value, and to use a similar
feature for repos that are not the system repo.
In particular, this allows us to support /etc/xdg-app/remotes.d for
xdg-app.
in write_directory_content_to_mtree_internal dfd_iter can be NULL,
for instance if commiting from --tree=ref=FOO. Don't blindly de-ref
it to avoid crashing.
Colin Walters [Fri, 8 Apr 2016 14:28:17 +0000 (10:28 -0400)]
build: Move grub2-15_ostree back to pkglibexecdir
It's not quite namespaced enough to have
`/usr/libexec/grub2-15_ostree`, and the Fedora spec file expects
things in `/usr/libexec/ostree`. Changing the spec file would be
annoying as we'd need conditionals.
Colin Walters [Tue, 5 Apr 2016 19:37:32 +0000 (15:37 -0400)]
pull: More consistently use remote_repo_local for local repos
I think it's cleaner if we use `remote_repo_local` to know
that we have a local repo. In reality, it might be nicest
if we didn't even create an `OstreeFetcher` for this case,
but untangling the code is tricky.
These are useful for ostree users (like xdg-app) that have custom
options for remotes. In particular they are useful when we later make them
all respect self->parent_repo.
Colin Walters [Wed, 6 Apr 2016 18:22:19 +0000 (14:22 -0400)]
build: Find grub2-mkconfig a bit more automagically
If one happens to not have grub2 installed, the previous commit
would cause us to fall back to `grub-mkconfig` with bad results.
We should likely just do dynamic detection and avoid pain, but
there's no harm in also allowing this to be statically determined.
The automagic here is that if `/etc/os-release` says `ID_LIKE=fedora`,
we know to use `grub2-mkconfig`. But distro scripts should set this
in the build configurations using `--with-grub2-mkconfig=`.
ostree-grub-generator can be used to customize
the generated grub.cfg file. Compile time
decision ostree-grub-generator vs grub2-mkconfig
can be overwritten with the OSTREE_GRUB2_EXEC
envvar - useful for auto tests and OS installers.
Why this alternative approach:
1) The current approach is less flexible than using a
custom 'ostree-grub-generator' script. Each system can
adjust this script for its needs, instead of using the
hardcoded values from ostree-bootloader-grub2.c.
2) Too much overhead on embedded to generate grub.cfg
via /etc/grub.d/ configuration files. It is still
possible to do so, even with this patch applied.
No need to install grub2 package on a target device.
3) The grub2-mkconfig code path has other issues:
https://bugzilla.gnome.org/show_bug.cgi?id=761180
Simon McVittie [Thu, 31 Mar 2016 16:52:57 +0000 (17:52 +0100)]
libtest.sh: only check whether $(pwd) is empty once
test-sysroot.js runs libtestExec() twice, one of which is after
creating non-hidden directories in $(pwd), so this check needs to be
skipped the second time.
Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #234
Approved by: cgwalters
Simon McVittie [Thu, 31 Mar 2016 17:29:44 +0000 (18:29 +0100)]
libtest.sh: use G_TEST_SRCDIR, G_TEST_BUILDDIR to find resources
This fixes the bug that in installed-tests that run testlib.sh under
"bash -c" (i.e. the C and JS tests), $(dirname $0) is "." and we can't do
the LD_PRELOAD correctly:
ERROR: ld.so: object './libreaddir-rand.so' from LD_PRELOAD cannot be
preloaded (cannot open shared object file): ignored.
Similarly, those tests can't copy gpghome correctly.
This also removes the confusing situation that SRCDIR in libtest.sh
(which is ${top_srcdir}/tests) does not mean the same thing as SRCDIR
in test-abi.sh (which is just ${top_srcdir}).
Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #234
Approved by: cgwalters
Colin Walters [Thu, 31 Mar 2016 17:46:16 +0000 (13:46 -0400)]
tests: Make failing to kill the GPG agent non-fatal
It's not working for me in `make check` on a RHEL 7 Workstation,
apparently because no GPG agent is spawned. I'm guessing this has
something to do with the GPG version?
The downside of this is we will be less likely to notice if GPG
changes again and we start leaking agents like we're in The Matrix
Reloaded. But the real solution to that is containers anyways.
Simon McVittie [Thu, 31 Mar 2016 13:32:55 +0000 (14:32 +0100)]
Probe for GNU parallel more accurately
moreutils parallel isn't compatible with the command-line syntax
used here. Because it doesn't implement GNU-style --help and exits 1
when that option is given, this test was correctly skipped when
using moreutils parallel, but only by mistake.
moreutils parallel might conceivably gain --help and --version in
future, but hopefully nothing incompatible with GNU parallel is going
to gain a --gnu option.
Also use the --gnu option to force the new command-line semantics;
some versions optionally supported an incompatible command-line syntax
taken from moreutils parallel.
Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #232
Approved by: cgwalters
Jonathan Lebon [Thu, 31 Mar 2016 13:10:27 +0000 (14:10 +0100)]
basic-test: commit with a non-empty subject
[smcv: split out from a larger commit, part of PR #231; add commit message] Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #232
Approved by: cgwalters
Jonathan Lebon [Thu, 31 Mar 2016 13:07:01 +0000 (14:07 +0100)]
test-pull-untrusted.sh: always corrupt a regular file, not a symlink
test-pull-untrusted.sh would pass when run as root, but fail when run
as testuser. It turned out that the way the files were stored in the
repo when running as a testuser were different, which meant that a
different .file object was chosen for corruption. Except that file
turned out to be a symlink, so the echo "broke" actually just wrote
to the no_such_file symlink target, thus keeping the actual symlink
file's checksum the same and causing the pull-local to pass when it
should have failed.
[smcv: split this out of a larger commit, part of PR #231] Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #232
Approved by: cgwalters
Simon McVittie [Thu, 31 Mar 2016 08:48:18 +0000 (09:48 +0100)]
tests/admin-test.sh: this is a bash script, not a POSIX sh script
The "function foo()" syntax is bash-specific, and Colin indicated in
PR #226 that he prefers to require bash rather than trying to support
every POSIX shell.
Signed-off-by: Simon McVittie <smcv@debian.org> Closes: #232
Approved by: cgwalters
Colin Walters [Tue, 29 Mar 2016 14:53:50 +0000 (10:53 -0400)]
commit: Support writing orphans
The API supports this, and it's not hard for us to do in the command
line as well. One possible use case is separating "content
generation" in a separate server.
Colin Walters [Fri, 25 Mar 2016 15:03:32 +0000 (11:03 -0400)]
commit: Support generating commits with no parent, or a custom one
When I'm doing local development builds, it's quite common for me not
to want to accumulate history. There are also use cases for this on
build servers as well.
In particular, using this, one could write a build system that didn't
necessarily need to have access to (a copy of) the OSTree repository.
Instead, the build system would determine the last commit ID on the
branch, and pass that to a worker node, then sync the generated
content back.
The API supported generating custom commits that don't necessarily
reference the previous commit on the same branch, let's just expose
this in the command line for convenience.
Dan Nicholson [Sat, 26 Mar 2016 15:35:30 +0000 (08:35 -0700)]
main: Set log handler for OSTree domain
Now that OSTree is used as G_LOG_DOMAIN, set the main handler to match
so the appropriate messages are filtered. It would probably be more
appropriate to spell out "OSTree" in the code, but since G_LOG_DOMAIN is
being defined globally in the project, might as well reuse it here.
Dan Nicholson [Mon, 21 Mar 2016 22:55:19 +0000 (15:55 -0700)]
core: Add debug messages for traversing
If you have a repo where a needed object has been inadvertantly removed,
all you'll get is a "No such metadata object" error with no clue about
where it was referenced from.
Add some debug messages to provide clues about which objects are being
traversed and found.
projects / ostree.git / log