If you lchown("symlink") then we were incorrectly trying to chown the
symlink target, rather than the symlink itself. In particular, this cause
cp -a to fail for a broken symlink. Additionally, it was using the
symlink target when verifying writability, rather than the symlink
itself.
To fix this, we need pass AT_SYMLINK_NOFOLLOW in these cases.
In general, the kernel itself will always resolve any symlinks for us
before calling into the fuse backend, so we should really never do any
symlink following in the fuse fs itself. So, we pro-actively add
NOFOLLOW flags to a few other places:
truncate:
In reality this will never be hit, because
the kernel will resolve symlinks before calling us.
access:
It seems the current fuse implementation never calls this
(faccessat w/AT_SYMLINK_NOFOLLOW never reaches the fuse fs)
but if this ever is implemented this is the correct behaviour.
We would ideally do `chmod` but this is not implemented on current kernels.
Because we're not multi-threaded, this is OK anyways.
Further, our write verification wasn't correctly handling the case of hardlinked
symlinks, which can occur for `bare` checkouts but *not* `bare-user` which the
tests were using. Change to `bare` mode to verify that.
Closes: #1137
Approved by: alexlarsson
|
||
|---|---|---|
| apidoc | ||
| bash | ||
| bsdiff@1edf9f6568 | ||
| build-aux | ||
| buildutil | ||
| ci | ||
| coccinelle | ||
| docs | ||
| libglnx@e226ccf691 | ||
| man | ||
| manual-tests | ||
| rust | ||
| src | ||
| tests | ||
| .dir-locals.el | ||
| .gitmodules | ||
| .papr.yml | ||
| .travis.yml | ||
| CONTRIBUTING.md | ||
| COPYING | ||
| GNUmakefile | ||
| Makefile-bash.am | ||
| Makefile-boot.am | ||
| Makefile-decls.am | ||
| Makefile-libostree-defines.am | ||
| Makefile-libostree.am | ||
| Makefile-man.am | ||
| Makefile-ostree.am | ||
| Makefile-otutil.am | ||
| Makefile-switchroot.am | ||
| Makefile-tests.am | ||
| Makefile.am | ||
| README-historical.md | ||
| README.md | ||
| TODO | ||
| autogen.sh | ||
| cfg.mk | ||
| configure.ac | ||
| git.mk | ||
| maint.mk | ||
| mkdocs.yml | ||
| ostree.doap | ||
README.md
libostree
New! See the docs online at Read The Docs (OSTree)
This project is now known as "libostree", though it is still appropriate to use the previous name: "OSTree" (or "ostree"). The focus is on projects which use libostree's shared library, rather than users directly invoking the command line tools (except for build systems). However, in most of the rest of the documentation, we will use the term "OSTree", since it's slightly shorter, and changing all documentation at once is impractical. We expect to transition to the new name over time.
As implied above, libostree is both a shared library and suite of command line tools that combines a "git-like" model for committing and downloading bootable filesystem trees, along with a layer for deploying them and managing the bootloader configuration.
The core OSTree model is like git in that it checksums individual files and has a content-addressed-object store. It's unlike git in that it "checks out" the files via hardlinks, and they should thus be immutable. Therefore, another way to think of OSTree is that it's just a more polished version of Linux VServer hardlinks.
Features:
- Transactional upgrades and rollback for the system
- Replicating content incrementally over HTTP via GPG signatures and "pinned TLS" support
- Support for parallel installing more than just 2 bootable roots
- Binary history on the server side (and client)
- Introspectable shared library API for build and deployment systems
- Flexible support for multiple branches and repositories, supporting projects like flatpak which use libostree for applications, rather than hosts.
Projects using OSTree
meta-updater is a layer available for OpenEmbedded systems.
QtOTA is Qt's over-the-air update framework which uses libostree.
rpm-ostree is a next-generation hybrid package/image system for Fedora and CentOS, used by the Atomic Host project. By default it uses libostree to atomically replicate a base OS (all dependency resolution is done on the server), but it supports "package layering", where additional RPMs can be layered on top of the base. This brings a "best of both worlds"" model for image and package systems.
flatpak uses libostree for desktop application containers. Unlike most of the other systems here, flatpak does not use the "libostree host system" aspects (e.g. bootloader management), just the "git-like hardlink dedup". For example, flatpak supports a per-user OSTree repository.
Endless OS uses libostree for their host system as well as flatpak. See their eos-updater and deb-ostree-builder projects.
GNOME Continuous is where OSTree was born - as a high performance continuous delivery/testing system for GNOME.
Building
Releases are available as GPG signed git tags, and most recent versions support extended validation using git-evtag.
However, in order to build from a git clone, you must update the submodules. If you're packaging OSTree and want a tarball, I recommend using a "recursive git archive" script. There are several available online; this code in OSTree is an example.
Once you have a git clone or recursive archive, building is the same as almost every autotools project:
git submodule update --init
env NOCONFIGURE=1 ./autogen.sh
./configure --prefix=...
make
make install DESTDIR=/path/to/dest
More documentation
New! See the docs online at Read The Docs (OSTree)
Contributing
See Contributing.