Protobuf uses `dynamic_cast` on polymorphic types which cannot be
inferred properly at compile-time, which can lead to unexpected behavior.
As a code safety measure, enable C++ RTTI for protobuf.
In `OnStreamCompleteInternal` an incorrect workerthread CSP enabled check
was added. This was a duplicate check and this is actually on the main
thread so asserts the wrong thread for a `WorkerPref`.
This now uses `if constexpr (...)` which is a lot more readable, and still compiles to almost no assembly instructions, as expected. Floating point casting assert when casting an integer that's too large to be represented exactly as a floating point (e.g. UINT64_MAX to double, since double have less than 64 bytes of mantissa), or when casting a double that's too large to be represented in a float.
Per the CSP specification, content injected by extensions is meant to
be exempt from page CSP. This patch takes care of the most common case
of content injected by extension content scripts, which always have
expanded principals which inherit from the page principal.
To make this easier, de-virtualize BasePrincipal::Kind(), using CTOR
initializers instead.
Use subject principal as triggering principal in <video>/<audio> "src"
attribute. We deviate from Mozilla here because we do not worry
about content isolation principals in Page Info.
Use subject principal as triggering principal in <img> "src" attribute.
Also get rid of the `BeforeMaybeChangeAttr`/`AfterMaybeChangeAttr` dance:
It makes more logical sense for these effects to happen _after_ the
attribute has actually been changed.
The imgLoader code consistently uses the term `loadingPrincipal` for
the principal that is called the `triggeringPrincipal` everywhere else.
This aligns the naming to avoid confusion in later changes.
In order to tailor certain security checks to the caller that is
attempting to load a particular piece of content, we need to be able to
attach an appropriate triggering principal to the corresponding requests.
Since most HTML content is loaded based on attribute values, this means
capturing the subject principal of the caller who sets those attributes,
which in turn means we need to make it available to `AfterSetAttr` hooks
on all relevant element types.
We're currently fairly vague and inconsistent about the values we provide to
content policy implementations for requestOrigin and requestPrincipal. In some
cases they're the triggering principal, sometimes the loading principal,
sometimes the channel principal.
Our existing content policy implementations which require or expect a
loading principal currently retrieve it from the context node.
Since no current callers require the principal to be the loading
principal, and some already expect it to be the triggering principal
(which there's currently no other way to retrieve), a choice was made
to pass the triggering principal whenever possible, but use the loading
principal to determine the origin URL.
See: https://github.com/whatwg/fetch/issues/274#issuecomment-2841053461
Considering the ubiquitous nature of gif, jpg, png, mp4, avi, etc. those
are left out, leaving just jxl, webp, x-matroska (mkv) and webm as
explicitly mentioned supported formats for conneg.
Note: it's not clear where network.http.accept.image is used in practice
because our normal place to use the image override (imgloader) uses
a different pref (image.http.accept).
Relying on the implicit conversion by `::FromContent` doesn't work for
us because we inherit that from `nsGenericHTMLElement` and isn't specific
to `HTMLAreaElement`, so it would accept other elements.
This also updates the Attribute changed event handler to exclude `<a>`
elements and cleans up an unused variable from the old code.
This was part of the HTML4 standard but only ever implemented by Mozilla
and removed again in HTML5 due to lack of adoption. Retaining this could
cause web compat issues and unspecified behavior in the layout engine.
