By default we look at `$FS_MISTRUST_DISABLE_PERMISSIONS_CHECKS`.
Optionally, the user can provide another variable as well, or
disable looking at the environment entirely.
These changes make sure that the errors conform to our preferred
style, and include a description of what exactly we were doing when
something went wrong.
According to doc/Errors.md, and in keeping with current best
practices, we should not include display an error's `source()` as
part of that error's display method. Instead, we should let the
caller decide to call source() and display that error in turn.
Part of #323.
To me, "Incorrect permissions on file or directory /path: g=w o=w"
implies that the current permissions on /path are 022.
Change the message to "Incorrect permissions: /path is
u=rwx,g=rwx,o=rwx; need g-w,o-w", which is closer to chmod syntax and is
more useful in non-interactive environments such as CI and support.
This fixes a unit test failure in weird environments (like some
containers) where the current effective GID is not included in the
list of current groups.
Closes#487.
Bug reported by @sjm217.
I followed the following procedure to make these changes:
* I used maint/changed_crates to find out which crates had changed
since 0.3.0.
* I used grep and maint/list_crates to sort those crates in
topological (dependency) order.
* I looked through semver_status to find which crates were listed as
having semver-relevant changes (new APIs and breaking changes).
* I scanned through the git logs of the crates with no
semver-relevant changes listed to confirm that, indeed, they had
no changes. For those crates, I incremented their patch-level
version _without_ changing the version that other crates depend on.
* I scanned through the git logs of the crates with no
semver-relevant changes listed to confirm that, indeed, they had
no obvious breaking changes.
* I treated all crates that depend on `arti` and/or `arti-client` as
having breaking changes.
* I identified crates that depend on crates that have changed, even
if they have not changed themselves, and identified them as having
a non-breaking change.
* For all of the crates, I used `cargo set-version -p $CRATE --bump
$STATUS` (where `STATUS` is `patch` or `minor`) to update the
versions, and the depended-upon versions.
It turns out that the `toml` crate can't handle OsString, since
`toml` doesn't support serialize_newtype_variant, and the `serde`
crate tries to serialize OsString using that method.
In this commit we document that limitation, and test that we can at
least round-trip through json.
Found by inspecting test coverage.
This change requires a little refactoring of TorClientBuilder: now,
instead of enabling or disabling mistrust, it enables or disables
the decision to _override_ the mistrust in the config.
We support all of the following (in TOML notation):
```
user = "rose" # by name
user = 413 # by ID
user = false # no user
user = ":current" # A 'special' user.
user = { name: "rose" }
user = { id: 413 }
user = { special: ":none" }
user = { special: ":current" }
```
If the target directory itself is unreadable by untrusted users,
then its contents can't be read[*] by them regardless of their
permissions. If the target directory _is_ readable, then _it_ will
be rejected if we are forbidding readable objects. (And if we
aren't we don't care if the contents are readable.)
A similar argument would apply to writable objects within an
unreadable target directory. We're not making that argument, since
such contents are likelier to be a mistake.
[*] Unless they're hard-linked; see comments in "Limitations"
section.
This includes:
* a CachedDir::join method.
* functions to read and write from provided filenames in a
CachedDir.
* a method to tell whether a fs-mistrust error is about bad file
permissions, or failure to inspect file permissions or some other
kind of IO problem.
This helps make it possible to use `SecureDir` (name pending) even
when we want to disable permissions checks. Otherwise, optional
permission checking would require users of this crate to maintain
separate code paths for the "check" and "don't check" cases.
This required a bit of poking through the `users` crate, to mess
with the user and group dbs. The original goal was to "trust the
group with the same name as us", but it turned into a bit of a
production, since:
* We want to take our own name from $USER, assuming that matches
our uid. (Otherwise we want to ask getpwuid_r().)
* We only want to trust the group if we are actually a member of
that group.
* We want to cache this information.
* We want to test this code.
Previously we would temporarily put self.resolved into an invalid
state by adding a path component that might be a symlink. With this
change, we create a new temporary path object (using Cow to avoid
unnecessary allocations) and only conditionally replace
self.resolved.
The only way to get a SecureDir is by having checked a directory.
Once you have one, it encourages you to open and create files and
directories with the right permissions, and checks them for you.
This crate is meant to solve #315 by giving a way to make sure that
a file or directory is only accessible by trusted users. I've tried
to explain carefully (in comments and documentation) what this crate
is doing and why, under the assumption that it will someday be read
by another person like me who does _not_ live and breathe unix file
permissions. The crate is still missing some key features, noted in
the TODO section.
It differs from the first version of the crate by taking a more
principled approach to directory checking: it emulates the path
lookup process (reading symlinks and all) one path change at a time,
thus ensuring that we check every directory which could enable
an untrusted user to get to our target file, _or_ which could
enable them to get to any symlink that would get them to the target
file.
The API is also slightly different: It separates the `Mistrust`
object (where you configure what you do or do not trust) from the
`Verifier` (where you set up a check that you want to perform on a
single object). Verifiers are set up to be a bit ephemeral,
so that it is hard to accidentally declare that _every_ object
is meant to be readable when you only mean that _some_ objects
may be readable.