I was seeing an `EPERM` here which was confusing.
It turned out the real error was `EEXIST`.
Since we're referring to the original error, but we do a
lot of computation in the middle, we need to save errno.
Alternative to https://github.com/ostreedev/ostree/pull/2197
Python's (usually) zero-sized `__init__.py` files can provoke
us hitting the hardlink limits on some filesystems (`EMLINK`).
At least one Fedora rpm-ostree user hit this.
The benefits of hardlinking here are quite marginal; lots
of hardlinks can behave suboptimally in particular filesystems
like BTRFS too.
This builds on prior code which made this an option, introduced
in 673cacd633
Now we just do it uncondtionally.
Also this provoked a different bug in a very obscure user mode checkout
case; when the "real" permissions were different from the "physical"
permissions, we would still hardlink. Fix the test case for this.
The extreme special case of "zero mode" files like `/etc/shadow`
comes up again. What we want is for "user mode" checkouts to
override it to make the file readable; otherwise when operating
as non-root without `CAP_DAC_OVERRIDE` it becomes very difficult
to work with.
Previously, we were hardlinking these files, but then it intersects
with *another* special case around zero sized files, which is
*also* true for `/etc/shadow`.
Trying to avoid hardlinking there unveiled this bug - when
we go to do a copy checkout, we need to override the mode.
For repo structure directories like `objects`, `refs`, etc... we should
be more permissive and let the system's `umask` narrow down the
permission bits as wanted.
This came up in a context where we want to be able to have read/write
access on an OSTree repo on NFS from two separate OpenShift apps by
using supplemental groups[1] so we don't require SCCs for running as the
same UID (supplemental groups are part of the default restricted SCC).
[1] https://docs.openshift.com/container-platform/3.11/install_config/persistent_storage/persistent_storage_nfs.html#nfs-supplemental-groups
Got this error when trying to rebase libostree in RHEL:
```
Error: CLANG_WARNING: [#def1]
libostree-2019.2/src/libostree/ostree-repo-checkout.c:375:21: warning: Access to field 'disable_xattrs' results in a dereference of a null pointer (loaded from variable 'repo')
```
I think what's happening is it sees us effectively testing
`if (repo == NULL)` via the `while (current_repo)`. Let's
tell it we're sure it's non-null right after the loop.
if a file ".wh..wh..opq" is present in a directory, delete anything
from lower layers that is already in that directory.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Closes: #1486
Approved by: cgwalters
Actually testing the patch to add `--force-copy-zerosized` to
rpm-ostree tripped over the fact that it uses `--union-identical`,
and we just hit an assertion failure with that combination.
Fix this by copying over the logic we have for the hardlink case.
Closes: #1753
Approved by: jlebon
In rpm-ostree we've hit a few cases where hardlinking zero-sized
files causes us problems. The most prominent is lock files in
`/usr/etc`, such as `/usr/etc/selinux/semanage.LOCK`. If there
are two zero-sized lock files to grab, but they're hardlinked,
then locking will fail.
Another case here is if one is using ostree inside a container
and don't have access to FUSE (i.e. `rofiles-fuse`), then the
ostree hardlinking can cause files that aren't ordinarily hardlinked
to become so, and mutation of one mutates all. An example where
this is concerning is Python `__init__.py` files.
Now, these lock files should clearly not be in the tree to begin
with, but - we're not gaining a huge amount by hardlinking these
files either, so let's add an option to disable it.
Closes: #1752
Approved by: jlebon
Having the `uncompressed-object-cache` directory in `archive` repos by default
is clutter; the functionality should be considered deprecated.
Now we only create the directory if we're doing a checkout with the cache
enabled.
Closes: #1446
Approved by: jlebon
This is analogous to the filtering support for the commit API: we allow
library users to skip over checking out specific files. This is useful
in some tricky situations where we *know* that the files to be checked
out will conflict with existing files in subtle ways.
One such example is in rpm-ostree support for multilib. There, we want
to allow checking out a package onto an existing tree, but skipping over
files that are not coloured to our preferred value (e.g. not overwriting
an i686 version of `ldconfig` if we already have the `x86_64` version).
See https://github.com/projectatomic/rpm-ostree/pull/1227 for details.
Closes: #1441
Approved by: cgwalters
SPDX License List is a list of (common) open source
licenses that can be referred to by a “short identifier”.
It has several advantages compared to the common "license header texts"
usually found in source files.
Some of the advantages:
* It is precise; there is no ambiguity due to variations in license header
text
* It is language neutral
* It is easy to machine process
* It is concise
* It is simple and can be used without much cost in interpreted
environments like java Script, etc.
* An SPDX license identifier is immutable.
* It provides simple guidance for developers who want to make sure the
license for their code is respected
See http://spdx.org for further reading.
Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
Closes: #1439
Approved by: cgwalters
While we do protect against path traversal during pull, let's also validate
during checkout; it's a cheap operation and provides good last-mile protection.
Closes: #1412
Approved by: jlebon
I was trying to use this with pygobject for an OCI+ostree project, and pygobject
rejected simply assigning to the field (understandably, since it can't bind the
lifetime together).
Add a wrapper function, which is still unsafe, but hides that unsafety
where most people shouldn't find it. And if they do...well, sorry,
Rust wasn't invented when ostree was started.
Closes: #1295
Approved by: pwithnall
I'm playing around with some ostree ⇔ OCI/Docker bits, and ran
into this while importing an OCI image that built from the Fedora
base image where `/home` is a regular directory, and I added a layer
that did the ostree bits of moving it to `/var` and leaving a symlink.
OCI/Docker supports this. Now since "process whiteouts" is really the
"enable OCI/Docker" mode, let's only replace dirs if that's enabled.
This leaves the `UNION_FILES` targeted for its original use case
which is unioning components/packages. (Although that use case itself
is now a bit superceded by `UNION_IDENTICAL`, but eh).
Closes: #1294
Approved by: jlebon
There's a subtle issue going on with the way we use `UNION_IDENTICAL`
now in rpm-ostree. Basically, the crux of the issue is that we checkout
the whole tree from the system repo, but then overlay packages by
checking out from the pkgcache repo. This is an easy way to break the
assumption that we will be merging hardlinks from the same repo.
This ends up causing issues like:
https://github.com/projectatomic/rpm-ostree/issues/1047
There, `vim-minimal` is already part of the host and has an object for
`/usr/share/man/man1/ex.1.gz`. `vim-common` has that same file, but
because it's unpacked in the pkgcache repo first, the hardlinks are not
the same.
There are a few ways we *could* work around this in rpm-ostree itself,
e.g. by re-establishing hardlinks when we do the content pull into the
system repo, but it still felt somewhat hacky. Let's just do this the
proper way and fall back to checksumming the target file if needed,
which is what librpm does as well in this case. Note that we only
checksum if they're not hard links, but they're the same size.
Closes: #1258
Approved by: cgwalters
This ends up a lot better IMO. This commit is *mostly* just
`s/glnx_close_fd/glnx_autofd`, but there's also a number of hunks like:
```
- if (self->sysroot_fd != -1)
- {
- (void) close (self->sysroot_fd);
- self->sysroot_fd = -1;
- }
+ glnx_close_fd (&self->sysroot_fd);
```
Update submodule: libglnx
Closes: #1259
Approved by: jlebon
For the old `OSTREE_REPO_MODE_ARCHIVE_Z2`. Use it mostly tree
wide except for the repo finder tests (to avoid conflicting with
some outstanding PRs).
Just noted another user coming in some of those tests and wanted to do a
cleanup.
Closes: #1209
Approved by: jlebon
We added a `.dir-locals.el` in commit: 9a77017d87
There's no need to have it per-file, with that people might think
to add other editors, which is the wrong direction.
Closes: #1206
Approved by: jlebon
I was looking at fixing an `rpm-ostree livefs` bug where we need to replace
`/usr/lib/passwd`. It's obviously bad if that temporarily disappears 😉. My plan
is to do a subpath checkout of just `/usr/lib/{passwd,group}`.
Make this atomic (i.e. file always exists) by changing the logic to create a
temporary link in repo/tmp, then rename() it into place.
A bonus here is we kill one of the very few (only?) non-error-cleanup i.e.
"non-linear" `goto`s in the ostree codebase.
Closes: #1171
Approved by: jlebon
It turns out that librpm automatically merges identical files between
distinct packages, and this occurs in practice with Fedora today between
`chkconfig` and `initscripts` for exmaple.
Since we added this for rpm-ostree, we basically want to do what librpm does,
let's change the semantics to do a merge. While we're here rename
to `UNION_IDENTICAL`.
Closes: #1156
Approved by: jlebon
We have `ot_ensure_unlinked_at()` for the "ignore ENOENT" case, and
`glnx_unlinkat()` otherwise. Port all in-tree callers to one or the other as
appropriate.
Just noticed an unprefixed error in the refs case and decided to do a tree-wide
check.
Closes: #1142
Approved by: jlebon
This is for issue projectatomic/rpm-ostree#365,
an extra option of overwrite mode is added to the checkout command
so that when there is "non-directory" file already exist
during checkout, the error will be handled.
Some tests are added for regression
Closes: #1116
Approved by: cgwalters
This is mostly the `copy_file_range` changes plus the Coverity files.
```
Colin Walters (4):
localalloc: Abort on EBADF from close() by default
local-alloc: Remove almost all macros like glnx_free, glnx_unref_variant
console: Fix Coverity NULL deref warning
fdio: Merge systemd code to use copy_file_range(), use FICLONE
Jonathan Lebon (1):
console: trim useless check
Matthew Leeds (1):
dirfd: Fix typo in comment
Philip Withnall (1):
glnx-console: Add missing NULL check before writing out text
```
Update submodule: libglnx
Closes: #1081
Approved by: jlebon
Using the error prefixing in the delta processing allows us to
do new code style. Also strip trailing whitespace.
Use error prefixing in a few other random places. I didn't
hunt for all of them, just testing out the new API.
Use `glnx_fchmod()`. Also note I dropped one `fchmod (tmpf, 0600)`
which is no longer necessary.
Update submodule: libglnx
Closes: #1011
Approved by: jlebon
When we [switched to using checkout + force_copy](e8efd1c8dc),
a side effect that went unnoticed at the time is that we started
setting directory mtimes to zero.
See the below bug where we long ago set the file times to zero, which got fixed,
so let's not regress things by setting the directory times to zero either. (Even
though AFAICS GNU tar doesn't complain about those)
This semantic is somewhat "overloaded" onto `force_copy`, but it avoids adding
yet another boolean; we don't have that many reserved boolean slots left. I
can't really think of many good use cases for `force_copy` *other* than the
`/etc` merge anyways.
https://bugzilla.redhat.com/show_bug.cgi?id=1229160
Closes: https://github.com/ostreedev/ostree/issues/995Closes: #997
Approved by: jlebon
These are regression from #971. We were stuffing a pointer size inside a
variable of integer size. So the assignment was spilling over into other
variables' storage space. Actually use a gpointer and GPOINTER_TO_[U]INT
as was done originally.
Also bump libglnx which has static checks for this error in the future.
Update submodule: libglnx
Closes: #990
Approved by: cgwalters
The `-z2` is annoying now since it's really a legacy; we've long
since supported typing `archive`. Convert the docs fully and
explain that.
Also do some (but not all) of the tests just to encourage newer tests to use
`archive` too.
Closes: #980
Approved by: jlebon
Use the new macros introduced recently in libglnx to make iterating over
hash tables cleaner. This is just a start, it does not migrate the whole
tree.
Update submodule: libglnx
Closes: #971
Approved by: cgwalters
There's lots of mechanically replacing `OtTmpFile` with `GLnxTmpfile`;
the biggest changes are in the commit path. Symlink commits are now
very clearly separated from regular files. Symlinks are `OtCleanupUnlinkat`,
and regular files are `GLnxTmpfile`.
The commit codepath separates those as `_ostree_repo_commit_path_final()` and
`_ostree_repo_commit_tmpf_final()`. A nice aspect of all of this is that they
both *consume* the temporary on success. This avoids an extra spurious
`unlink()` call.
One of the biggest bits of code motion is in `commit_loose_regfile_object()`,
which no longer needs to care about symlinks. For the most parth though it's
just removing conditionals.
Update submodule: libglnx
Closes: #958
Approved by: jlebon
The code here tried to truncate the string to the previous length,
but that doesn't work when recursing, since further calls change the
length.
What actually ended up happening was the string would get corrupted
after the first level of recursion.
Closes: #936
Approved by: jlebon
This is a continuation of https://github.com/ostreedev/ostree/pull/926
for directories instead of files.
See: https://github.com/flatpak/flatpak/issues/845
This option suppresses mode bits outside of `0775` for directory
checkouts. I think most people should start doing this by default,
and use explicit overrides for e.g. `/tmp` if doing a recommit based
on a checkout.
Closes: #927
Approved by: alexlarsson
See https://github.com/ostreedev/ostree/pull/909 for more information on the
rationale. Basically there's no reason for flatpak (which uses `bare-user-only`)
to have world-writable dirs. Particularly with the presence of the system
helper.
An approach I considered instead was to parse and validate directory metadata
objects at commit time. We still may do that in addition; for file objects we *had*
to do it that way because the actual files would be laid down suid. But directories
live only as inert `.dirmeta` objects until we do a checkout (i.e. `mkdir()`), so
we can solve the problem at checkout time.
Closes: #914
Approved by: alexlarsson
When falling back to copying, we previously would only chmod checked out
files in the non-user-checkout mode. Fix this by always doing chmod.
The file_mode was being prepared but never actually applied.
Add a basic test in the archive-z2 --> usermode checkout case in which
we're guaranteed to always fall back to copy mode.
Closes: #633Closes: #903
Approved by: cgwalters
It's hard right now to do a full port to the new libglnx tmpfile
API since there are complex cases in the commit path which deal
with symlinks as well.
Let's make things more gradual by introducing the important part (struct with
autocleanup) here in libotutil, port what we can. This will make a future
complete port easier.
Closes: #871
Approved by: jlebon
This is what caused the merge of
https://github.com/projectatomic/rpm-ostree/pull/652
to blow up, since https://github.com/ostreedev/ostree/pull/848
landed right before we tried to merge it.
When I was writing that PR I remember having an uncertain feeling
since we were doing a `mkdirat` above, but at the time I thought
we'd have test suite coverage...turns out we didn't.
For backwards compatibility, we need to continue to do a `mkdirat` here of the
parent. However...I can't think of a reason anyone would *want* that behavior.
Hence, let's add a special trick - if the destination name is `.`, we skip
`mkdirat()`. That way rpm-ostree for example can open a dfd for `/etc` and avoid
the `mkdir`.
Fold the subpath tests into `test-basic.sh` since it's not worth a separate
file. Add a test case for checking out a file.
Closes: #854
Approved by: jlebon
Looking at `perf record ostree checkout`, some things stand out; e.g.:
```
+ 27.63% 0.07% ostree libgio-2.0.so.0.5000.3 [.] g_file_enumerator_iterate
+ 22.74% 0.28% ostree libostree-1.so.1.0.0 [.] ostree_repo_file_tree_query_child
+ 13.74% 0.08% ostree libostree-1.so.1.0.0 [.] ot_variant_bsearch_str
```
The GIO abstractions are already fairly heavyweight, and `OstreeRepoFile` mallocs
a lot too.
Make things more efficient here by dropping the GIO bits for reading ostree data -
we just read from the variants directly and iterate over them. The end result
here is that according to perf we go from ~40% of our time in the kernel to
~70%, and things like `g_file_enumerator_iterate()` drop entirely out of the
hot set.
Closes: #848
Approved by: jlebon
Since we now have a cleaner separation of "toplevel checkout prep"
versus "recursive checkout", handle the special case of checking out
a single file at first rather than later.
Prep for future work in optimizing this function more.
Closes: #848
Approved by: jlebon
Rather than `g_output_stream_splice()`, where the input is a regular
file.
See https://github.com/GNOME/libglnx/pull/44 for some more information.
I didn't try to measure the performance difference, but seeing the
read()/write() to/from userspace mixed in with the pointless `poll()` annoyed me
when reading strace.
As a bonus, we will again start using reflinks (if available) for `/etc`,
which is a regression from the https://github.com/ostreedev/ostree/pull/797
changes (which before used `glnx_file_copy_at()`).
Also, for the first time we'll use reflinks when doing commits from file-backed
content. This happens in `rpm-ostree compose tree` today for example.
Update submodule: libglnx
Closes: #817
Approved by: jlebon
This is a variant of the efforts in https://github.com/ostreedev/ostree/pull/741
Working on `rpm-ostree livefs`, I realized though I needed to just
check out *new* files directly into the live `/etc` (and possibly
delete obsolete files).
The way the current `/etc` merge works is fundamentally different from
that. So my plan currently is to probably do something like:
- Compute diff
- Check out each *new* file individually (as a copy)
- Optionally delete obsolete files
Also, a few other things become more important - in the current deploy code, we
copy all of the files, then relabel them. But we shouldn't expose to *live*
systems the race conditions of doing that, plus we should only relabel files we
checked out.
By converting the deploy's /etc code to use this, we fix the same TODO item
there around atomically having the label set up as we create files. And further,
if we kill the `/var` relabeling which I think is unnecessary since Anaconda
does it, we could delete large chunks of code there.
In the implementation, there are two types of things: regular files, and
symlinks. For regular files, in the `O_TMPFILE` case, we have the ability to
do *everything* atomically (including SELinux labeling) before linking it into
place. So let's just use that. For symlinks, we use `setfscreatecon()`.
Closes: #797
Approved by: jlebon
We really have an astonishing variety of similar functions which write files and
symlinks. I was working on a different PR and the duplication between the
union-mode and add-mode/none-mode checkout functions bothered me.
I realized that the "handle EEXIST" tri-state maps directly to the
`GLnxLinkTmpfileReplaceMode`, so deduping things makes even more sense.
Closes: #801
Approved by: jlebon
I was reading a strace the other day and noticed we were loading the same
`.dirmeta` object many times. Unlike the other object types, `.dirmeta` objects
don't accumulate much over time; there are only so many directory metadata types.
(Without SELinux involved it'd probably be 5-6 I'd guess offhand).
For `fedora-atomic/25/x86_64/docker-host` there are currently 34 `.dirmeta` in
the tree.
But how many times during a checkout did we load those 34 dirmeta objects?
With a quick strace:
```
$ strace -s 2048 -f -o strace.log ostree --repo=repo-build checkout -U fedora-atomic/25/x86_64/docker-host host-test-checkout
$ grep dirmeta strace.log | wc -l
7165
```
After, as you'd expect, we just loaded `34` from disk. We do
6 system calls (`openat+fstat+fstat+read+read+close`) per dirmeta,
so we dropped a total of 42780 system calls - which is about 20% of the total
system calls made.
`perf record` tells me that we're spending ~40 of our time in the kernel during
a checkout, so reducing syscall traffic helps. Though most of that appears to be
in the VFS and XFS layers for `linkat` (which isn't surprising).
So how much did perf improve? Well, on my workstation, I get a lot of
fluctuation in timing, sometimes by 30%, so this was well within the noise. But
it's well worth speeding up checkout, and I think this optimization will shine
more as we improve performance elsewhere.
Closes: #795
Approved by: jlebon
Colin Walters
Luca BRUNO
Jonathan Lebon
Felix Krull
Giuseppe Scrivano
Jonathan Lebon
Marcus Folkesson
Ruixin