Harden the code that does permission checks to protect against
re-opening of stdin.
Code that runs FFI is vulnerable to an attack where fd 0 is closed
during a permission check and re-opened with a file that contains a
positive response (ie: `y` or `A`). While FFI code is dangerous in
general, we can make it more difficult for FFI-enabled code to bypass
additional permission checks.
- Checks to see if the underlying file for stdin has changed from the
start to the end of the permission check (detects races)
- Checks to see if the message is excessively long (lowering the window
for races)
- Checks to see if stdin and stderr are still terminals at the end of
the function (making races more difficult)
Closes #14122.
Adds two extensions to `--allow-run` behaviour:
- When `--allow-run=foo` is specified and `foo` is found in the `PATH`
at startup, `RunDescriptor::Path(which("foo"))` is added to the
allowlist alongside `RunDescriptor::Name("foo")`. Currently only the
latter is.
- When run permission for `foo` is queried and `foo` is found in the
`PATH` at runtime, either `RunDescriptor::Path(which("foo"))` or
`RunDescriptor::Name("foo")` would qualify in the allowlist. Currently
only the latter does.
This patch adds a `remote` backend for `ext/kv`. This supports
connection to Deno Deploy and potentially other services compatible with
the KV Connect protocol.
This commit adds new "--deny-*" permission flags. These are complimentary to
"--allow-*" flags.
These flags can be used to restrict access to certain resources, even if they
were granted using "--allow-*" flags or the "--allow-all" ("-A") flag.
Eg. specifying "--allow-read --deny-read" will result in a permission error,
while "--allow-read --deny-read=/etc" will allow read access to all FS but the
"/etc" directory.
Runtime permissions APIs ("Deno.permissions") were adjusted as well, mainly
by adding, a new "PermissionStatus.partial" field. This field denotes that
while permission might be granted to requested resource, it's only partial (ie.
a "--deny-*" flag was specified that excludes some of the requested resources).
Eg. specifying "--allow-read=foo/ --deny-read=foo/bar" and then querying for
permissions like "Deno.permissions.query({ name: "read", path: "foo/" })"
will return "PermissionStatus { state: "granted", onchange: null, partial: true }",
denoting that some of the subpaths don't have read access.
Closes #18804.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
Co-authored-by: Nayeem Rahman <nayeemrmn99@gmail.com>
We can make `NodePermissions` rely on interior mutability (which the
`PermissionsContainer` is already doing) in order to not have to clone
everything all the time. This also reduces the chance of an accidental
`borrow` while `borrrow_mut`.
This commit abstracts out the specifics of the underlying system calls
FS operations behind a new `FileSystem` and `File` trait in the
`ext/fs` extension.
This allows other embedders to re-use ext/fs, but substituting in a
different FS backend.
This is likely not the final form of these traits. Eventually they will
be entirely `deno_core::Resource` agnostic, and will live in a seperate
crate.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
1. Rewrites the tests to be more back and forth rather than getting the
output all at once (which I believe was causing the hangs on linux and
maybe mac)
2. Runs the pty tests on the linux ci.
3. Fixes a bunch of tests that were just wrong.
4. Adds timeouts on the pty tests.
This commit adds unstable "Deno.openKv()" API that allows to open
a key-value database at a specified path.
---------
Co-authored-by: Luca Casonato <hello@lcas.dev>
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
These methods are confusing because the arguments are backwards. I feel
like they should have never been added to `Option<T>` and that clippy
should suggest rewriting to
`map(...).unwrap_or(...)`/`map(...).unwrap_or_else(|| ...)`
https://github.com/rust-lang/rfcs/issues/1025
This commit adds new "A" option to the interactive permission prompt, that will
allow all subsequent permissions for given group (domain). Ie. when querying for
permissions to access eg. env variables responding with "A" will allow access
to all environmental variables.
This works for all permission domains and should make permission prompts
more ergonomic for users.
Turns out we were cloning permissions which after prompting were discarded,
so the state of permissions was never preserved. To handle that we need to store
all permissions behind "Arc<Mutex<>>" (because there are situations where we
need to send them to other thread).
Testing and benching code still uses "Permissions" in most places - it's undesirable
to share the same permission set between various test/bench files - otherwise
granting or revoking permissions in one file would influence behavior of other test
files.
This commit changes permission prompt to show that "import()" API
is requesting permissions.
Given "dynamic_import.js" like so:
```
import("https://deno.land/std@0.170.0/version.ts");
```
Before:
```
deno run dynamic_import.js
⚠️ ┌ Deno requests net access to "deno.land".
├ Run again with --allow-net to bypass this prompt.
└ Allow? [y/n] (y = yes, allow; n = no, deny) >
```
After:
```
deno run dynamic_import.js
⚠️ ┌ Deno requests net access to "deno.land".
├ Requested by `import()` API
├ Run again with --allow-net to bypass this prompt.
└ Allow? [y/n] (y = yes, allow; n = no, deny) >
```
This commit refactors several things in "runtime/permissions" module:
- splits it into "mod.rs" and "prompter.rs"
- adds "PermissionPrompter" trait with two implementations:
* "TtyPrompter"
* "TestPrompter"
- adds "before" and "after" prompt callback which can be used to hide
progress bar in the CLI (to be done in a follow up)
- "permissions_prompt" API returns "PromptResponse" enum, instead
of a boolean; this allows to add "allow all"/"deny all" functionality
for the prompt
