We implemented many more MallocSizeOf tests (even if some such as
channels are zero).
Meaning we can not ignore more of it making the code cleaner.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Testing: Compilation is the test.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
This inserts a cache in painter that keeps data for animating images in
Painter.
We then can send cheap ImageUpdate::UpdateAnimation to just get the
required bytes from the cache.
There are some warts with the current design:
- ImageUpdate::AddImage only adds images in the cache that have the data
in SerializableImageData::Raw. Potentially there could be image
animations in other ways which then for a following
ImageUpdate::UpdateAnimation we do not have the data. We can probably
fix this on the sender side.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Testing: Tested on the usecase in
https://github.com/servo/servo/issues/41818 and WPT test here:
https://github.com/Narfinger/servo/actions/runs/21063721427
Fixes: https://github.com/servo/servo/issues/41818
---------
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Parsed and vectorized representations of SVGs in the image cache are
never evicted when SVGs are updated, leading to stale SVGs staying in
the image cache alongside updated ones. This PR:
- adds a stringified uuid field to SVGSVGElement
- adds a map of `SVGSVGElement` uuid to `PendingImageId` in
`ImageCacheImpl`
- adds an optional argument for the uuid to
`ImageCacheImpl::rasterize_vector_image`, which, if provided, removes
any existing parsed svgs associated with the uuid and stores a new entry
of uuid -> `PendingImageId`
- implements `unbind_from_tree` on SVGSVGElement, which clears the
associated images/representations from the image_cache (from
`vector_images`, `rasterized_vector_images`, and `completed_loads`) as
well as from the layout image resolver image cache.
Testing: Each of the following documents should display their (very
flashy) content correctly and memory usage should stay constant
<details>
<summary>Changing SVGSVGElement</summary>
```
<svg height="100" width="100" xmlns="http://www.w3.org/2000/svg">
<circle id="c" r="1" cx="50" cy="50" fill="red" />
</svg>
<script>
let c = document.querySelector("#c");
let r = 1;
setInterval(() => {
r += 1;
c.setAttribute("r", r.toString());
let tmp = document.createTextNode("ignored");
c.parentNode.appendChild(tmp);
tmp.remove();
}, 100);
</script>
```
</details>
<details>
<summary>Unbinding SVGSVGElements</summary>
```
<div id="parent_div">
<svg id="test" height="100" width="100" xmlns="http://www.w3.org/2000/svg">
<circle id="c" r="10" cx="50" cy="50" fill="red" />
</svg>
</div>
<script>
let div = document.querySelector("#parent_div");
let svg_html_string = div.innerHTML;
let svg = document.querySelector("#test");
let r = 10;
setInterval(() => {
svg.remove();
div.innerHTML = svg_html_string;
svg = document.querySelector("#test");
let circle = document.querySelector("#c");
r += 1;
circle.setAttribute("r", r.toString());
}, 100);
</script>
```
</details>
<details>
<summary>Unbinding SVGSVGElements (and rebinding the same SVG)</summary>
This didn't work until I also evicted the associated image from the
layout image resolver image cache and the image cache's
`completed_loads` on SVGSVGElement unbind, so it seems like a useful, if
a bit redundant, test.
```
<div id="parent_div">
<svg id="test" height="100" width="100" xmlns="http://www.w3.org/2000/svg">
<circle id="c" r="10" cx="50" cy="50" fill="red" />
</svg>
</div>
<script>
let div = document.querySelector("#parent_div");
let svg_html_string = div.innerHTML;
let svg = document.querySelector("#test");
let r = 10;
setInterval(() => {
svg.remove();
div.innerHTML = svg_html_string;
svg = document.querySelector("#test");
}, 100);
</script>
```
</details>
Fixes: #41070
---------
Signed-off-by: Tom Cummings <cummings.t287@gmail.com>
This change finishes the big rename associated with the old
`compositing` crates. Long ago, these crates managed a compositor, like
you might find in a traditional web engine. These days, compositing is
done in WebRender so the name has stopped making much sense. Various
structs inside the crates have already been renamed and this is the
final big change necessary for the rename
Testing: This is just a rename so existing tests should cover it.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
- make `ImageCache::memory_report` return `Vec<Report>` and rename
`memory_report` to `memory_reports`
- update `ImageCacheImpl::memory_report` to return a new Report for the
fontdb member
- add an implementation of `MallocSizeOf` for `fontdb::FaceInfo` which
sums the result of calling malloc_size_of on each of the string/vector
members
- add an implementation of `MallocSizeOf` for `fontdb::Database` which
sums the result of malloc_size_of on each of the faces
Testing: `./mach check` `./mach clippy`
`grep fontdb::Database::load_fonts_from_file untracked | wc -l` results:
2390 -> 545
- before:
<img width="1536" height="108" alt="截图 2025-12-14 13-53-32"
src="https://github.com/user-attachments/assets/422ce9d4-e962-45af-a072-7d1c0c9e80aa"
/>
- now:
<img width="1536" height="108" alt="截图 2025-12-14 13-58-26"
src="https://github.com/user-attachments/assets/363f933d-8ab0-4354-b8ae-89f6de46ca7b"
/>
Fixes: #38730
Signed-off-by: WaterWhisperer <waterwhisperer24@qq.com>
For a long time, the "Compositor" hasn't done any compositing. This is
handled by WebRender. In addition the "Compositor" does many other
tasks. This change renames `IOCompositor` to `Paint`.
`Paint` is Servo's paint subsystem and contains multiple `Painter`s.
This change does not rename the crate; that will be done in a
followup change.
Testing: This just renames types and updates comments, so no new tests
are necessary.
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Currently, only the first `Painter` is used for all operations in the
`Compositor`. This change modifies the compositor API and message
handling to allow routing the operations to the correct `Painter` via a
provided `PainterId` or `WebViewId`.
This change is to enable support for per-`WebView`s `RenderingContext`s.
Testing: This change shouldn't change behavior, so existing WPT tests
should cover it.
---------
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
The current logic fails when the `Content-Type` header includes the
character encoding other parameters like `qs`.
Fixes: #40630
---------
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
RasterImage is now not seralizeable and backed by Vec<u8> storage. Add a
new struct `SharedRasterImage` that allows serializeing. This can be
constructed from RasterImage. This introduces some more copying but I
think the tradeoff is ok. Newly introduced copies are for favicon
response to the embedder and EmbedderNotifications.
Testing: Should not change functionality, so covered by existing tests.
Fixes: https://github.com/servo/servo/issues/39626 and related issues.
---------
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Previously, the `ImageCache` would inform consumers of changes to image
availability by messaging an IPC channel. This isn't great, because it
requires serialization and deserialization. Nowadays, `ImageCache`s are
always owned by the same `Global` that uses them. This change replaces
the IPC channel with callback that implements `Send`.
For the case of normal HTML document images, the results are still sent
over the pre-existing Crossbeam channel that was connected via an IPC
route. A followup change might eliminate that channel entirely.
Testing: This should not change observable behavior so is covered by
existing tests.
Fixes: #24338.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Instead of having the `ImageCache` return the broken image icon for
failed loads, have `HTMLImageElement` explicitly request it. This means
that the image is loaded on demand (reducing the usage of resources) and
also simplifying the interface of the `ImageCache` greatly.
In addition, the display of the broken image icon is improved, more
closely matching other browsers. A test for this display (which was
falsely passing before) is updated to reflect the new display of the
broken image icon.
Testing: There is a Servo-specific test for this change. Some WPT tests
start to fail as well. Before these were not properly loading the broken
image icon so they were failing before, just in a hidden way.
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
This change makes it so that the `HTMLCanvasElement` is responsible for
managing the `ImageKey` for associated `RenderingContext`s. Only
canvases display their contents into WebRender directly, so this makes
it so that keys are not generated for `OffscreenCanvas`.
The main goal here is that `ImageKey`s are always associated with a
particular `WebView`, which isn't possible in the various canvas
backends yet. This is important because each `WebView` may soon have a
different WebRender instance entirely with its own set of `ImageKey`s.
This also allows for clearing `ImageKey`s when canvases are disconnected
from the DOM in a future change. One tricky thing here is placeholder
canvases, which are meant to be driven from workers.
It seems that the implementation isn't correct for these at the moment
as they need to be updated to the specification. Instead, what is
happening is that any existing context / image is completely lost when
converting to an `OffscreenCanvas`.
Testing: This shouldn't change observable behavior, so is covered by
existing tests.
Fixes: This is part of #40261.
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Every `ImageCache` should be assigned a `Pipeline` and a `WebView`, so
that it knows what WebRender instance to create image keys in. This
change accomplishes that by splitting the `ImageCache` trait into
`ImageCache` and `ImageCacheFactory`. There should only be one
`ImageCacheFactory` per process.
Testing: This should not change observable behavior and is thus covered
by
existing tests.
Fixes: This is part of #40261.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Canvas expects input data to be in RGBA premultiplied format and
WebRender already supports RGBA and BGRA data as long as they are
premultiplied. Pre-multiplying up front allows:
- Avoiding many conversions when painting images to canvas.
- Passing the `RasterImage` IpcSharedMemory of the image instead of
creating
a new one with the premultiplied data every time we upload to
WebRender. This is a big deal for animated gifs, because before every
frame was creating a new shared memory segment.
It seems that for rasterized SVGs were were already putting
premultplied data into the `RasterImage` so it's quite likely SVGs were
being composited incorrectly.
Testing: This causes 8 tests to start passing and 2 tests to fail in the
WebGL conformance suite.
The failures are due to the fact that premultiplying alpha is lossy when
alpha is 0. In that case,
the resulting color of a blend operation might be wrong. This is
typically only a problem if you
use RGBA data as RGB data, which is pretty unusual. In the case that you
are blending with
RGBA the final color values will be 0 or close to 0 anyway. Gecko solves
this issue by having a
cacheable surface generation API that can fetch both premulitplied and
unpremulitplied data
from things like image elements. We do not have that yet, but I argue
that this change
is important anyway due to the amount that it reduces memory and file
descriptor usage
as well as the cost of copying image data so much in memory.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This splits `Snapshot` into two structs:
- `Snapshot`: A non-serializable owned image buffer that might contain
either shared memory or a `Vec<u8>`. When mutated the shared memory
version is converted into a `Vec<u8>` avoiding a copy of the data
until absolutely necessary.
- `SharedSnapshot`: A serialzable version of `Snapshot` that contains
only shared memory and is suitable for sending across IPC channels.
This version is cheaply convertible from and to owned `Snapshot`s as
long as the source is also a shared memory `Snapshot`.
In addition, it does copyless conversions of `RasterImage` into
`Snapshot`s (including for frame offsets in animated images). Finally,
there are a few minor changes to try harder not to have to do a
`transform()` operation.
Testing: This should not change observable behavior, so is covered by
existing
tests. It should come with a mild performance improvement and open up
the
opportunity for others.
Fixes: #36594.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
load_system_fonts is relatively slow, since it needs to scan every font
in the system font directories. There's no need to repeat this work for
every single SVG on a page; we can do it once when creating an image
cache in a process and reuse that font information for all derived image
caches in the same process.
Testing: Manually verified that pages with SVGs still load. There should
be no observable behaviour change in WPT tests.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
This should be the final PR for the Hash Function series that is
trivial.
Of note: I decided to transform `HashMapTracedValues<Atom,..>` to use
FxBuildHasher. This is likely not going to improve performance as Atom's
already have a unique u32 that is used as the Hash but it safes a few
bytes for the RandomState that is normally in the HashMap.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Testing: Hash function changes should not change functionality, we
slightly decrease the size and unit tests still work.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
FxHash is faster than FnvHash and SipHash for simple types up to at
least 64 bytes. The cryptographic guarantees are not needed for any
types changed here because they are simple ids.
This changes the types in script and net crates.
In a future PR we will change the remaining Fnv to be also Fx unless
there is a reason to keep them as Fnv.
Testing: Should not change functionality but unit test and wpt will find
it.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
This shows up sometimes in code reviews, so it makes sense that tidy
enforces it. `rustfmt` supports this via comment normalization, but it
does many other things and is still an unstable feature (with bugs).
Testing: There are new tidy tests for this change.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This patch adds support for rendering static inline SVG documents in the
DOM tree by serializing the SVGElement's subtree and leveraging the
existing resvg based SVG stack for rendering. Serialiing the subtree is
necessary as resvg's tree representation (roxmltree) is immutable, so we
can't construct the tree incrementally.
Few other design choices here:
1. The `SVGSVGElement` is now treated as a replaced element and the
layout code is responsible for plumbing the serialized SVG source
(encoded as a base64 data: url) into the image cache, much like how
background images are handled.
2. The serialization is done on the script thread after an initial
layout pass. This is necessary because the serialization code asserts
that it is invoked from script thread i.e we can't call it from layout
workers.
3. The serialized SVG data: url is cached to avoid recomputing it on
subsequent layouts. The cache is invalidated when the SVGSVGElement's
subtree is mutated.
The original SVGSVGElement code was behind the `dom_svg_enabled` pref.
This patch also removes the preference and make SVG support using resvg
available unconditionally.
Below is the analysis of the new test failures:
These tests use inline SVG but used to pass by accident.
They now fail because they contain SVG with no intrinsic
sizing which is not handled by resvg in a way that would
allows us to distinguish it from the sized case. The same
limitation applies to non-inline SVG.
- /css/CSS2/positioning/absolute-replaced-width-003a.xht
- /css/CSS2/positioning/absolute-replaced-width-003b.xht
- /css/CSS2/positioning/absolute-replaced-width-003c.xht
These tests employ CSS styles in the HTML that
target the elements in inline SVG, which is not currently
supported.
-
/css/compositing/mix-blend-mode/mix-blend-mode-plus-lighter-svg-basic.html
- /css/compositing/mix-blend-mode/mix-blend-mode-plus-lighter-svg.html
This is a tentative test that uses the unsupported 'border-shape' CSS
property. The ref uses SVG, so it used to pass accidentally. The ref
still doesn't render correctly since it also relies on styling SVG
elements using CSS classes in the HTML (instead of inline in SVG).
- /css/css-borders/tentative/border-shape/border-shape-stroke.html
These tests use the attribute 'clip-path=circle(...)' in the
test, but this doesn't seem to work in resvg.
- /css/css-masking/clip-path/clip-path-borderBox-1b.html
- /css/css-masking/clip-path/clip-path-contentBox-1b.html
- /css/css-masking/clip-path/clip-path-contentBox-1c.html
- /css/css-masking/clip-path/clip-path-fillBox-1b.html
- /css/css-masking/clip-path/clip-path-marginBox-1a.html
- /css/css-masking/clip-path/clip-path-paddingBox-1b.html
- /css/css-masking/clip-path/clip-path-strokeBox-1b.html
- /css/css-masking/clip-path/clip-path-strokeBox-1c.html
- /css/css-masking/clip-path/clip-path-viewBox-1a.html
- /css/css-masking/clip-path/clip-path-viewBox-1b.html
- /css/css-masking/clip-path/clip-path-viewBox-1d.html
- /css/css-masking/clip-path/svg-clip-path-circle-offset.html
- /css/css-masking/clip-path/svg-clip-path-ellipse-offset.html
Additionally, the below two tests use a `foreignObject` SVG element
which
embeds a html div fragment. This is also not supported by resvg.
- /css/css-masking/clip-path/clip-path-viewBox-1d.html
- /css/css-masking/clip-path/clip-path-fillBox-1b.html
The following test fails because of apparent pixel differences
between a circle rendered purely using CSS clip-path vs a circle
rendered in SVG using resvg.
- /css/css-masking/clip-path/clip-path-contentBox-1c.html
These tests style the inline SVG elements using CSS in the HTML or
separate stylesheet. This is not supported by this implementation.
- /css-transforms/document-styles/svg-document-styles-{001..004}.html
- /css-transforms/document-styles/svg-document-styles-012.html
- /css-transforms/external-styles/svg-external-styles-{001..004}.html
- /css-transforms/external-styles/svg-external-styles-014.html
These tests seem like they should pass, but they fail because of what
seems like an anti-aliasing issue in the rendering engine. The
transformed element has a thin outline which is causing pixel difference
with the ref:
- /css/css-transforms/group/svg-transform-group-008.html
- /css/css-transforms/group/svg-transform-group-009.html
- /css/css-transforms/group/svg-transform-nested-009.html
- /css/css-transforms/group/svg-transform-nested-013.html
- /css/css-transforms/group/svg-transform-nested-014.html
- /css/css-transforms/group/svg-transform-nested-018.html
- /css/css-transforms/group/svg-transform-nested-019.html
- /css/css-transforms/group/svg-transform-nested-008.html
The below tests fail because resvg is calculating the wrong size for the
'rect' inside the SVG. The dimensions of the SVG are established via the
CSS in the HTML, so it seems resvg is using incorrect coordinates for
the children of the svg when explict width/height are not specified in
the root svg element.
- /css/css-transforms/group/svg-transform-group-011.html
- /css/css-transforms/group/svg-transform-nested-021.html
- /css/css-transforms/group/svg-transform-nested-029.html
All these tests use an SVG that doesn't have width nor height attributes
and this causes resvg to use incorrect coordinates for the SVG's
children. In addition, the following tests use the CSS syntax for
transforms inside the SVG (using style attribute) which is not supported
by resvg (it only supports the SVG 1.1 transform syntax).
- /css/css-transforms/inline-styles/svg-inline-styles-{001..004}.html
- /css/css-transforms/inline-styles/svg-inline-styles-012.html
In the case of these four tests, the `style` attribute specifies an
invalid transform, but resvg doesn't fallback to the transform specified
via the `transform` attribute on the same element.
- /css/css-transforms/inline-styles/svg-inline-styles-005.html
- /css/css-transforms/inline-styles/svg-inline-styles-006.html
- /css/css-transforms/inline-styles/svg-inline-styles-010.html
- /css/css-transforms/inline-styles/svg-inline-styles-013.html
The following test fails because of the lack of width/height in SVG as
described above but it also exposes gaps in our CSS tranform
implementation.
- /css/css-transforms/preserve3d-and-filter-with-perspective.html
These tests failure because resvg doesn't handle the SVG without
explicit width and height, but specified via CSS in the HTML. In
addition, there are pixel differences between the ref due to
antialiasing issues.
- /css/css-transforms/matrix/svg-matrix-{005...008}.html
- /css/css-transforms/matrix/svg-matrix-010.html
- /css/css-transforms/matrix/svg-matrix-012.html
- /css/css-transforms/matrix/svg-matrix-{015..069}.html
- /css/css-transforms/rotate/svg-rotate-angle-45-001.html
- /css/css-transforms/rotate/svg-rotate-angle-45-011.html
- /css/css-transforms/rotate/svg-rotate-angle-45-022.html
- /css/css-transforms/scale/svg-scale-006.html
- /css/css-transforms/scale/svg-scale-007.html
These tests seem to be failing due to some sort of antialiasing issue,
where a transformed SVG element has a thin border that causes pixel
differences compared to the solid colored reference.
- /css/css-transforms/skewX/svg-skewx-001.html
- /css/css-transforms/skewX/svg-skewx-006.html
- /css/css-transforms/skewX/svg-skewx-011.html
- /css/css-transforms/skewX/svg-skewx-016.html
- /css/css-transforms/skewX/svg-skewx-021.html
- /css/css-transforms/skewX/svg-skewxy-001.html
- /css/css-transforms/skewY/svg-skewy-001.html
- /css/css-transforms/skewY/svg-skewy-006.html
- /css/css-transforms/skewY/svg-skewy-011.html
- /css/css-transforms/skewY/svg-skewy-016.html
- /css/css-transforms/skewY/svg-skewy-021.html
These tests specify several SVG attributes such as transform,
vector-effect etc via CSS in the HTML (rather than inline in SVG). The
current implementation doesn't support this.
- /css/css-transforms/transform-box/stroke-box-mutation-001.html
- /css/css-transforms/transform-box/stroke-box-mutation-002.html
- /css/css-transforms/transform-box/stroke-box-mutation-003.html
- /css/css-transforms/transform-box/stroke-box-mutation-004.html
- /css/css-transforms/transform-box/svgbox-stroke-box-002.html
- /css/css-transforms/transform-box/svgbox-stroke-box-003.html
- /css/css-transforms/transform-box/svgbox-stroke-box-004.html
- /css/css-transforms/transform-box/svgbox-stroke-box-005.html
These tests depend on 'transform-origin' specified on an element inside
an SVG, but this transform is influenced by the 'tranform-box' set via
CSS in the HTML itself (not the SVG). The current implementation doesn't
support styling the SVG using document styles, so these tests just fail.
- /css/css-transforms/transform-origin/svg-origin-relative-length-*.html
These tests check the fallback behaviour when invalid syntax is
encountered in the 'transform-origin' value. resvg doesn't correctly
fallback to 0,0 causing the tests to fail.
-
/css/css-transforms/transform-origin/svg-origin-relative-length-invalid-001.html
-
/css/css-transforms/transform-origin/svg-origin-relative-length-invalid-002.html
-
/css/css-transforms/transform-origin/svg-origin-relative-length-invalid-003.html
-
/css/css-transforms/transform-origin/svg-origin-relative-length-invalid-004.html
These tests use unimplemented Canvas APIs like 'beginLayer' and
the 'CanvasFilter' constructor and hence fail at runtime.
-
/html/canvas/element/filters/2d.filter.canvasFilterObject.gaussianBlur.tentative.html
-
/html/canvas/element/filters/2d.filter.layers.gaussianBlur.tentative.html
-
/html/canvas/element/layers/2d.layer.anisotropic-blur.isotropic.tentative.html
-
/html/canvas/element/layers/2d.layer.anisotropic-blur.mostly-x.tentative.html
-
/html/canvas/element/layers/2d.layer.anisotropic-blur.mostly-y.tentative.html
-
/html/canvas/element/layers/2d.layer.anisotropic-blur.x-only.tentative.html
-
/html/canvas/element/layers/2d.layer.anisotropic-blur.y-only.tentative.html
-
/html/canvas/element/layers/2d.layer.css-filters.blur-and-shadow.tentative.html
- /html/canvas/element/layers/2d.layer.css-filters.blur.tentative.html
- /html/canvas/element/layers/2d.layer.css-filters.shadow.tentative.html
- /html/canvas/element/layers/2d.layer.ctm.layer-filter.tentative.html
-
/html/canvas/offscreen/filters/2d.filter.canvasFilterObject.dropShadow.tentative.html
-
/html/canvas/offscreen/filters/2d.filter.canvasFilterObject.gaussianBlur.tentative.html
-
/html/canvas/offscreen/filters/2d.filter.layers.dropShadow.tentative.html
-
/html/canvas/offscreen/filters/2d.filter.layers.gaussianBlur.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.anisotropic-blur.isotropic.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.anisotropic-blur.mostly-x.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.anisotropic-blur.mostly-y.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.anisotropic-blur.x-only.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.anisotropic-blur.y-only.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.css-filters.blur-and-shadow.tentative.html
- /html/canvas/offscreen/layers/2d.layer.css-filters.blur.tentative.html
-
/html/canvas/offscreen/layers/2d.layer.css-filters.shadow.tentative.html
- /html/canvas/offscreen/layers/2d.layer.ctm.layer-filter.tentative.html
These tests fail because resvg doesn't seem to honour the 'translate'
CSS property specified on an SVG element using an inline 'style'
attribute.
- /css/css-transforms/translate/svg-translate-with-units.html
-
/css/css-transforms/translate/translate-and-transform-attribute-in-svg.html
-
/css/css-transforms/translate/translate-and-transform-css-property-in-svg.html
- /css/css-transforms/translate/translate-in-svg.html
These tests seem to fail due to the filter effect implementation in
resvg either not being complete or spec compliant.
- /css/filter-effects/feconvolve-divisor.html
- /css/filter-effects/feconvolve-region-001.html
- /css/filter-effects/feconvolve-region-002.html
- /css/filter-effects/filter-subregion-01.html
- /css/filter-effects/svg-feimage-002.html
- /css/filter-effects/svg-feimage-003.html
- /css/filter-effects/svg-feimage-004.html
- /css/filter-effects/svg-feoffset-001.html
The test /css/filter-effects/svg-feimage-004.html should ideally PASS
but currently fails because we don't propagate height/width set using
CSS in HTML element to the root SVG, so resvg uses the wrong dimensions
when rendering the children of the SVG.
These failures are due to deficienies in our current implementation
i.e we don't support styling SVG elements using CSS in HTML.
-
/css/css-transforms/gradientTransform/svg-gradientTransform-combination-001.html
- /css/selectors/sharing-in-svg-use.html
The below test fails as our current implementation relies on resvg to
tell us the intrinsic ratio of the SVG, but this doesn't always work
correctly.
- /css/css-sizing/svg-intrinsic-size-005.html
This failure is due to lack of proper fallback to no-op transform in
resvg when the `rotate()` syntax is specified with an invalid list e.g
`rotate(90,)`.
- /css/css-transforms/rotate/svg-rotate-3args-invalid-002.html
This test only passes in CI and based on the raw log output, it seems
that no text inside the SVG is rendered in the CI. This could be an font
stack related issue.
- /css/css-display/display-contents-svg-elements.html
This test asserts that the CSP blocks loads triggered using `use`
elements in SVG. It used to TIMEOUT as without inline SVG support, no
CSP violation event was triggered. It fails now since the event is now
triggered for the load of the SVG itself (our current implementation
loads inline SVGs as serialized base64 data: urls). This doesn't match
the blocked URL in the use element though.
- /content-security-policy/img-src/svg-use-blocked.tentative.html
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
This creates a new method in shared/compositing/lib to generate image
keys that are send over the webview. This does not immediately return
the keys but goes over the constellation to receive the keys from the
IOCompositor. To make this more efficient, we now cache the keys in
image_cache in a simple FIFO order. The old blocking method stays intact
for now but got renamed to make the blocking clear.
The blocking calls that are left are in:
- `components/canvas/canvas_data.rs`
- `components/script/dom/htmlmediaelement.rs`
Testing: WPT tests should cover this as this doesn't change any
functionality.
Fixes: Was mentioned in
https://github.com/servo/servo/issues/37161#issuecomment-2915750051 and
part of https://github.com/servo/servo/issues/37086
---------
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
Co-authored-by: gterzian <2792687+gterzian@users.noreply.github.com>
Similar to #37638. It was using `threadpools_async_runtime_workers_max`,
and `threadpools_image_cache_workers_max` was unused, this PR fixes
that.
Signed-off-by: Ashwin Naren <arihant2math@gmail.com>
This makes servo use less file descriptors for animated images and
avoids the crash described in
https://github.com/servo/servo/issues/36792.
Doing this also forces the end users to be more explicit about whether
they want to deal with all image frames or just the first one.
Previously, `Image::bytes` silently returned only the data for the first
frame. With this change there's now a `frames` method which returns an
iterator over all frames in the image.
Testing: No tests - this simply reduces the number of fds used. Servo
doesn't currently display animated gifs anyways.
Fixes: https://github.com/servo/servo/issues/36792
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
This deletes images from WR when dropping the ImageCacheStore for a
WebView.
Testing: Run `./mach run --enable-experimental-web-platform-features
unsplash.com` and then open `about:memory` in a new tab. On Linux we end
up with ~30MB of WR images. Then close the unsplash.com tab and measure
memory again, it will down to ~1.25MB
Fixes: https://github.com/servo/servo/issues/25927
Signed-off-by: webbeef <me@webbeef.org>
Rather than sharing the full image cache in a script_thread, the image
cache is now unique per document. This ensures that CSP factors no
longer affect whether the image is retrieved from the cache incorrectly.
To do so, the thread_pool is shared across all caches, but the store is
fresh. Except for the place_holder{image,url}, which are cloned. That's
because the `rippy_data` is only available in the constellation and no
longer accessible at the point that we need to create the document in
the script_thread.
Contrary to the description in #36505, the script_thread still has an
image_cache for this reason: so it has access to the store and
thread_pool to clone it.
With these changes, the two CSP tests no longer flake. Confirmed with
running the following commmand:
```
./mach test-wpt tests/wpt/tests/content-security-policy/generic/ --rerun=10
```
Fixes#36505
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Added a simple check to only perform metadata extraction and listener
notification when we haven't already processed the metadata for an image
Testing: Existing tests should cover if we break decoding image metadata
complete.
Fixes: #36502
---------
Signed-off-by: Barigbue <barigbuenbira@gmail.com>
Co-authored-by: Josh Matthews <josh@joshmatthews.net>
This removes a bunch of duplicated code needed to support
ConditionalMallocSizeOf correctly, and fixes multiple places where that
code was subtly wrong (the seen pointers hashset was never cleared).
Testing: Measuring https://www.nist.gov/image-gallery lots of times.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
This allows removing a LazyLock around the resources.
We override the baked in resources unconditionally in servoshell
upon initialization anyway
([desktop](9f93ccd942/ports/servoshell/desktop/cli.rs (L15)),
[android](9f93ccd942/ports/servoshell/egl/android/simpleservo.rs (L49)),
[ohos](9f93ccd942/ports/servoshell/egl/ohos/simpleservo.rs (L43))
), meaning that the baked in resources
are unused in servoshell.
For 3rd-party embedders, we probably also want to the let them know
early that they should initialize the resources, instead of
restricting the panics to production mode.
Rippy is the only resource which was required. Since it is only
253 bytes large, we just bake that resource in as a fallback.
We do want to make using the resources easy from tests, so we add
some logic to keep the baked in resources for tests only and initialize
the resource reader on first access.
---
<!-- 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
- [ ] These changes fix #___ (GitHub issue number if applicable)
- [ ] There are tests for these changes OR
- [ ] These changes do not require tests because ___
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
These changes add a new report for image cache memory usage for each
script thread.
Testing: Looked at the numbers after browsing various stock photo sites
that show galleries of images.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
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>
These two traits both exposed different parts of the compositing API,
but now that the compositor doesn't depend directly on `script` any
longer and the `script_traits` crate has been split into the
`constellation_traits` crate, this can be finally be cleaned up without
causing circular dependencies. In addition, some unit tests for the
`IOPCompositor`'s scroll node tree are also moved into
`compositing_traits` as well.
Testing: This just combines two crates, so no new tests are necessary.
Fixes: #35984.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Instead of setting up a route for every image load in the DOM / Layout,
route all incoming image cache responses through the `ScriptThread`.
This avoids creating a set of file descriptor for every image that is
loaded.
This change requires having the `ImageCache` track the `PipelineId` of
the original the listener so that the `ScriptThread` can route it
properly to the correct `Window`.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Flatten and simplify Servo's preferences code. In addition, have both
preferences and options passed in as arguments to `Servo::new()` and
make sure not to use the globally set preferences in `servoshell` (as
much as possible now).
Instead of a complex procedural macro to generate preferences, just
expose a very simple derive macro that adds string based getters and
setters.
- All command-line parsing is moved to servoshell.
- There is no longer the concept of a missing preference.
- Preferences no longer have to be part of the resources bundle because
they now have reasonable default values.
- servoshell specific preferences are no longer part of the preferences
exposed by the Servo API.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* Add prefs to limit threadpool sizes
Add preferences to control the size of threadpools,
so that we can easily reduce the amount of runtime
threads and test which pools benefit from more
threads.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
* Add pref for Webrender threadpool
Add a preference to limit the size of the webrender threadpool.
Note: WebRender by default calls hooks which register the threads with
a profiler instance that the embedder can register with webrender.
Servo currently doesn't register such a profiler with webrender,
but in the future we might also want to profile the
webrender threadpool.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
---------
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.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>
Instead of exposing many different kinds of messages to the compositor
that are routed through the constellation, expose a single message type
which can be sent across IPC channels. In addition, this IPC channel and
the route to the crossbeam channel with the compositor is created along
with the `CompositorProxy`, simplifying what needs to be passed around
during pipeline initialization.
Previously, some image updates (from video) were sent over IPC with a
special serialization routine and some were sent via crossbeam channels
(canvas). Now all updates go over the IPC channel `IpcSharedMemory` is
used to avoid serialization penalties. This should improve performance
and reduce copies for video, but add a memory copy overhead for canvas.
This will improve in the future when canvas renders directly into a
texture.
All-in-all this is a simplification which opens the path toward having a
standard compositor API and reduces the number of duplicate messages and
proxying that had to happen in libservo.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Instead of exposing many different kinds of messages to the compositor
that are routed through the constellation, expose a single message type
which can be sent across IPC channels. In addition, this IPC channel and
the route to the crossbeam channel with the compositor is created along
with the `CompositorProxy`, simplifying what needs to be passed around
during pipeline initialization.
Previously, some image updates (from video) were sent over IPC with a
special serialization routine and some were sent via crossbeam channels
(canvas). Now all updates go over the IPC channel `IpcSharedMemory` is
used to avoid serialization penalties. This should improve performance
and reduce copies for video, but add a memory copy overhead for canvas.
This will improve in the future when canvas renders directly into a
texture.
All-in-all this is a simplification which opens the path toward having a
standard compositor API and reduces the number of duplicate messages and
proxying that had to happen in libservo.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* Move WebRender related types to `webrender_traits`
This refactor moves several WebRender related types
from `compositing_traits`, `script_traits` and `net_traits`
crates to the `webrender_traits` crate.
This change also moves the `Image` type and associated
function out of `net_traits` and into the `pixels` crate.
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* Move `script_traits::WebrenderIpcSender` to `webrender_traits::WebRenderScriptApi`
---------
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
