Rather than having each callside specifying the relevant
information from the GlobalScope, do this via a trait instead.
This would have saved us quite a bit of test debugging
since we would often forget to set relevant information
from the global context for a request.
Now, in the future when we need additional information from
the globalscope for a request, we only need to update this
method to make that happen.
Previously it would also sometimes use `document`, but
calling the relevant information on either `document` or
`globalscope` doesn't matter, since the `globalscope`
defers to the value from the `document` anyways.
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
There's no naming conflict, and nowhere else the name NetRequestInit.
All it does is increase confusion.
Testing: No runtime changes.
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
This implements waiting for pending preloads, where the preload request
is still fetching the result when the second "real" request is started.
It is
implemented by storing responses in the `SharedPreloadedResources`
which is communicated via `PreloadId` send to the `CoreResourceManager`.
Part of #35035
---------
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
The goal of this change is to prevent having to copy so much data out of
listeners when a fetch completes, which will be particularly important
for off-the-main thread parsing of CSS (see #22478). This change has
pros and cons:
Pros:
- This makes the design of the `FetchResponseListener` a great deal
simpler.
They no longer individually store a dummy `ResourceFetchTiming` that is
only replaced right before `process_response_eof`.
- The creation of the `Arc<Mutex<FetchResponseListener>>` in the
`NetworkListener` is abstracted away from clients and now they just
pass the `FetchResponseListener` to the fetch methods in the global.
Cons:
- Now each `FetchResponseListener` must explicitly call `submit_timing`
instead of having the `NetworkListener` do it. This is arguably a bit
easier to follow in the code.
- Since the internal data of the `NetworkListener` is now an
`Arc<Mutex<Option<FetchResponseListener>>>`, when the fetching code
needs to share state with the `NetworkListener` it either needs to
share an `Option` or some sort of internal state. In one case I've
stored the `Option` and in another case, I've stored a new inner
shared value.
Testing: This should not change observable behavior and is thus covered
by existing tests.
Fixes: #22550
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Adds a wrapper type for vectors that can contain large response bodies
to prevent flooding debug logs with the contents of those bodies.
Testing: Can't test debug log output.
Fixes: #37769
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
`FetchReponseListener` has traditionally lived in `net` even though it
is only used in `script` currently. Because of the two way dependency,
it has also use a lot of templating to implement something pretty basic
(call methods on a trait object).
This change moves the trait to `script` and removes several levels of
templating, making the code quite a bit shorter and easier to
understand.
This change is preparation for fixing #22550 and implementing
off-the-main-thread CSS parsing.
Testing: This should not change any behavior so is covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Moves interfaces defined by the performance spec to the
`script/dom/performance/` module from `script/dom/`.
Testing: Just a refactor shouldn't need any testing
Fixes: Partially #38901
Signed-off-by: WaterWhisperer <waterwhisperer24@qq.com>
The scripted_caller only has information if the context is coming
from a script. If an element fetch listener processes CSP
violations, then this information doesn't exist. Instead, we should
use the global URL and the line number. WPT tests don't appear
to expect a column number, as they are all zero. Not all elements
are updated, as I am not actually sure all of them need it.
The source position remains an Option, since there are also code
paths that don't correspond to element or script sources. Maybe
in the future we can always determine the source position, but
let's take small steps towards that.
Part of #4577
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
This refactoring moves various CSP-related methods away from GlobalScope
and Document into a dedicated entrypoint. It also reduces the amount of
imports of the CSP crate, so that types are consolidated into this one
entrypoint. That way, we control how CSP code interacts with the script
crate.
For reviewing purposes, I split up the refactoring into separate
distinct commits that all move 1 method(group) into the new file.
Testing: no change in behavior, only a build improvement + code cleanup
---------
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Signed-off-by: Tim van der Lippe <TimvdLippe@users.noreply.github.com>
All logic is implemented in `report_csp_violations` to avoid
pulling in various element-logic into SecurityManager.
Update the `icon-blocked.sub.html` WPT test to ensure that
the document is the correct target (verified in Firefox and Chrome).
Fixes#36806
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
This PR updates the `fetch_image_for_layout` function to include missing
security-related request settings:
- `insecure_requests_policy`
- `has_trustworthy_ancestor_origin`
- `policy_container`
---
<!-- Thank you for contributing to Servo! Please replace each `[ ]` by
`[X]` when the step is complete, and replace `___` with appropriate
data: -->
- [X] `./mach build -d` does not report any errors
- [X] `./mach test-tidy` does not report any errors
- [X] These changes fix#36591
<!-- Either: -->
- [X] There are tests for these changes
Signed-off-by: Emmanuel Elom <elomemmanuel007@gmail.com>
It also updates the FetchResponseListener to process CSP violations to
ensure that iframe elements (amongst others) properly generate the CSP
events. These iframe elements are used in the Trusted Types tests
themselves and weren't propagating the violations before.
However, the tests themselves are still not passing since they also use
Websockets, which currently aren't using the fetch machinery itself.
That is fixed as part of [1].
[1]: https://github.com/servo/servo/issues/35028
---------
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
`EmbedderMsg` was previously paired with an implicit
`Option<WebViewId>`, even though almost all variants were either always
`Some` or always `None`, depending on whether there was a `WebView
involved.
This patch adds the `WebViewId` to as many `EmbedderMsg` variants as
possible, so we can call their associated `WebView` delegate methods
without needing to check and unwrap the `Option`. In many cases, this
required more changes to plumb through the `WebViewId`.
Notably, all `Request`s now explicitly need a `WebView` or not, in order
to ensure that it is passed when appropriate.
Signed-off-by: Delan Azabani <dazabani@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Instead of creating an IPC channel for every fetch, allow cancelling
fetches based on the `RequestId` of the original request. This requires
that `RequestId`s be UUIDs so that they are unique between processes
that might communicating with the resource process.
In addition, the resource process loop now keeps a `HashMap` or `Weak`
handles to cancellers and cleans them up.
This allows for creating mutiple `FetchCanceller`s in `script` for a
single fetch request, allowing integration of the media and video
elements to integrate with the `Document` canceller list -- meaning
these fetches also get cancelled when the `Document` unloads.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This puts a few commonly used `Node` helpers into a trait (`NodeTraits`)
and gives them more descriptive names and documentation. The renames:
- `document_from_node` -> `NodeTraits::owner_document`
- `window_from_node` -> `NodeTraits::owner_window`
- `stylesheets_owner_from_node<T:` -> `NodeTraits::stylesheet_list_owner`
- `containing_shadow_root` -> `NodeTraits::containing_shadow_root`
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Instead of creating a `ROUTER` for each fetch, create a fetch thread
which handles all incoming and outcoming fetch requests. Now messages
involving fetches carry a "request id" which indicates which fetch is
being addressed by the message. This greatly reduces the number of file
descriptors used by fetch.
In addition, the interface for kicking off fetches is simplified when
using the `Listener` with `Document`s and the `GlobalScope`.
This does not fix all leaked file descriptors / mach ports, but greatly
eliminates the number used. Now tests can be run without limiting
procesess on modern macOS systems.
Followup work:
1. There are more instances where fetch is done using the old method.
Some of these require more changes in order to be converted to the
`FetchThread` approach.
2. Eliminate usage of IPC channels when doing redirects.
3. Also eliminate the IPC channel used for cancel handling.
4. This change opens up the possiblity of controlling the priority of
fetch requests.
Fixes#29834.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* script: Do not run layout in a thread
Instead of spawning a thread for layout that almost always runs
synchronously with script, simply run layout in the script thread.
This is a resurrection of #28708, taking just the bits that remove the
layout thread. It's a complex change and thus is just a first step
toward cleaning up the interface between script and layout. Messages are
still passed from script to layout via a `process()` method and script
proxies some messages to layout from other threads as well.
Big changes:
1. Layout is created in the script thread on Document load, thus every
live document is guaranteed to have a layout. This isn't completely
hidden in the interface, but we can safely `unwrap()` on a Document's
layout.
2. Layout configuration is abstracted away into a LayoutConfig struct
and the LayoutFactory is a struct passed around by the Constellation.
This is to avoid having to monomorphize the entire script thread
for each layout.
3. Instead of having the Constellation block on the layout thread to
figure out the current epoch and whether there are pending web fonts
loading, updates are sent synchronously to the Constellation when
rendering to a screenshot. This practically only used by the WPT.
A couple tests start to fail, which is probably inevitable since removing
the layout thread has introduced timing changes in "exit after load" and
screenshot behavior.
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
* Update test expectations
* Fix some issues found during review
* Clarify some comments
* Address review comments
---------
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
refactoring with ResourceFetchMetadata
implemented deprecated window.timing functionality
created ResourceTimingListener trait
fixed w3c links in navigation timing
updated include.ini to run resource timing tests on ci
I moved away from the `Window` struct all the logic to handle task
sources, into a new struct called `TaskManager`. In a happy world, I'd
be able to just have there two functions, of the types:
```rust
fn task_source<T: TaskSource>(&self, name: TaskSourceName) -> Box<T>
fn task_source_with_canceller<T: TaskSource>(&self, name: TaskSourceName)
-> (Box<T>, TaskSourceCanceller)
```
And not so much duplicated code. However, because TaskSource can't be a
trait object (because it has generic type parameters), that's not
possible. Instead, I decided to reduce duplicated logic through macros.
For reasons[1], I have to pass both the name of the function with
canceller and the name of the function without, as I'm not able to
concatenate them in the macro itself. I could probably use
`concat_idents` to create both types already defined and reduce the
amount of arguments by one, but that macro is nightly only. At the same
time, not being able to declare macros inside `impl` forces me to pass
`self` as an argument.
All this makes this solution more verbose than it would be ideally. It
does reduce duplication, but it doesn't reduce the size of the file.
[1](https://github.com/rust-lang/rust/issues/29599)
http://www.robohornet.org gives a score of 101.36 on master,
and 102.68 with this PR. The latter is slightly better,
but probably within noise level.
So it looks like this PR does not affect DOM performance.
This is expected since `Box::new` is defined as:
```rust
impl<T> Box<T> {
#[inline(always)]
pub fn new(x: T) -> Box<T> {
box x
}
}
```
With inlining, it should compile to the same as box syntax.
