The wasm baseline and via-Ion compilers both generate numerous calls to
supporting functions ("instance functions"). To generate correct code, and,
shortly, correct stackmaps too, the compilers need to know argument and return
types for these functions. Currently that information is scattered around
somewhat:
* In baseline, emitInstanceCall() specifies both arg and return types
* In via-Ion, builtinInstanceMethodCall specifies the return type. Arg types
aren't (obviously) available, but will be needed for upcoming stack map
generation.
This bug makes the following main changes:
* Adds a new type, SymbolicAddressSignature, related to the existing
SymbolicAddress type. A SymbolicAddressSignature specifies the arity,
argument and return types for a SymbolicAddress.
* For all the SymbolicAddresses referred to in the baseline and via-Ion
compilers, a static SymbolicAddressSignature is provided, as readonly data.
* The compilers then pass around |const SymbolicAddressSignature&| where
before they just passed around a SymbolicAddress, to the extent possible.
* All ad-hoc specification of arg and return types in emitInstanceCall()
[baseline] and builtinInstanceMethodCall() [via Ion] have been removed.
* Not all SymbolicAddress values have an associated SymbolicAddressSignature
-- only those referred to in calls to emitInstanceCall() and
builtinInstanceMethodCall() do.
Smaller changes:
* WasmBaselineCompile.cpp:
- removed MIRTypeVector and SigP*_.
- pushReturnValueOfCall: overloaded so as to accept both ExprType and MIRType.
- emitInstanceCall now accepts a |const SymbolicAddressSignature&| and pulls
type info out of that. All uses changed accordingly.
- [drive-by fix]: emitInstanceCall: added a comment update that should have
been in bug 1528240 ("Fix inconsistent return type use in
BaseCompiler::emitInstanceCall")
* WasmIonCompile.cpp:
- passArg: overloaded so as to accept both MIRType and ValType for the node
type
- builtinInstanceMethodCall now accepts a |const SymbolicAddressSignature&|
and pulls type info out of that. All uses changed accordingly.
* WasmBuiltins.{cpp, h}: add the static SymbolicAddressSignature bundles.
* WasmGC.h: add an overload of StackArgAreaSizeUnaligned that can take a
Symbolic AddressSignature directly.
--HG--
extra : rebase_source : 00909db26977288401c708dfc73175c6f1882b3d
By this change, icons will never be shrunk to a smaller size. Windows will do
it appropriately.
Internet Explorer's icon handler will not paint the white background if the
icon is large enough. I'm also replicating the behavior.
--HG--
extra : source : 56eb23aaff3a5a0dbca8e41c534307c30e7351f4
Automatic update from web-platform-tests
Add .mailmap to map people to their current names
Note that this isn't used by `git log` by default (as of 2.20.1): that
depends on log.mailmap being set to true.
--
Add Tess to .mailmap
--
wpt-commits: 39f54a3dd48b560289e0f455308acf7d32fad043, f8a1bfbe5454352d3f5b58845829968ff212519b
wpt-pr: 15353
Automatic update from web-platform-tests
HTML: noreferer was never a property of hyperlinks
This should have tested the rel attribute value, but as it hasn't been doing that and it would be quite a bit of work, let's remove this instead.
--
wpt-commits: 25c665a93231027be5941838cecb3208e360d4b1
wpt-pr: 15357
Automatic update from web-platform-tests
Add screen orientation test: fire event before promise is resolved (#15267)
--
wpt-commits: 8ed64687560332642c7b34ebd16011ae5b01d99a
wpt-pr: 15267
Automatic update from web-platform-tests
XMLSerializer: Fix prefixed attribute serialization
... in a case where its owner element has xmlns:prefix of which prefix
is same as the attribute and namespace is not same as the attribute.
e.g.
el.setAttributeNS("uri1", "p:n", "v");
el.setAttributeNS(XMLNS_URI, "xmlns:p", "uri2");
ShouldAddNamespaceAttribute() checked only existence of xmlns:prefix in
the element, but we should check existence of (prefix, namespace URI)
pair in the scope according to the specification.
So, this CL adds 'recording the namespace information' step [1] defined
by the specification, and ShouldAddNamespaceAttribute() checks all
available prefixes in the scope.
[1] https://w3c.github.io/DOM-Parsing/#recording-the-namespace
Bug: 929035
Change-Id: I575e8f652ae45f7583202443cc72d5afe5faf59d
Reviewed-on: https://chromium-review.googlesource.com/c/1460643
Reviewed-by: Yoshifumi Inoue <yosin@chromium.org>
Commit-Queue: Kent Tamura <tkent@chromium.org>
Cr-Commit-Position: refs/heads/master@{#631058}
--
wpt-commits: 265330b7d26194cb60157a18ac399127d21f6b99
wpt-pr: 15340
Automatic update from web-platform-tests
decimal points in numbers
Tests that decimal points are allowed *before* digits (like `.1`) and *between* digits (like `0.1`), but not *after* digits (like `1.`).
Tests <https://github.com/w3c/csswg-drafts/issues/3599>
--
fix whitespace for lint
--
Name all tests uniquely
--
Merge pull request #15337 from web-platform-tests/tabatkins-patch-2
decimal points in numbers
--
wpt-commits: 5fdd036dbd18fa6d2e7cf235bd9acc625a6df22e, f17e215eaa37316f1489a7b62ab103d2a58a0d89, 31e1a1bb5fadb08749e4b8070ad0f19ec5ea1088, e2eb25aaa6dd3c512b172588f3400f9c25a410c3
wpt-pr: 15337
Automatic update from web-platform-tests
url whitespace consumption
Ensure that whitespace before or after the string in a url() results in a correct parse.
Tests https://github.com/w3c/csswg-drafts/issues/3600
--
Merge pull request #15339 from web-platform-tests/tabatkins-patch-3
url whitespace consumption
--
wpt-commits: e0b3c4cf2f69e816c6b83432968dc7fc83d700f6, 0de6fa5638c7e314a0391fae7f6bf496309fb65d
wpt-pr: 15339
Automatic update from web-platform-tests
[css-box] Inheritance, initial values (#15312)
Test that CSS Box Model properties do not inherit.
https://drafts.csswg.org/css-box-3/#property-index
--
wpt-commits: 35bd69423b10a361a969139ba0271896dced9691
wpt-pr: 15312
Automatic update from web-platform-tests
Set click count correctly to dispatch "dblclick" event
In order to dispatch the "dblclick" event to DOM by sending synthetic
mouse events, we have to check the most recent two mouse press events.
If their dispatched times and locations are close enough, we will set
the mouse event's click count to 2, which will generate the "dblclick"
event.
Bug: 606367
Change-Id: I049d6463c6bf1393c9e8b496662359eeeef7682b
Reviewed-on: https://chromium-review.googlesource.com/c/1413362
Commit-Queue: Lan Wei <lanwei@chromium.org>
Reviewed-by: Navid Zolghadr <nzolghadr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630933}
--
wpt-commits: 1ddffbeeb6faed345ce7f999544ed1d29c7a203f
wpt-pr: 14911
Automatic update from web-platform-tests
Verify that contentDocument is updated during nested browsing context document parsing.
--
wpt-commits: ae6d56308684b97f30a01b455ce3eb205cba79aa
wpt-pr: 15268
Automatic update from web-platform-tests
Fix canonical form of rotate property.
Expected behavior is outlined in the spec here:
https://drafts.csswg.org/css-transforms-2/#individual-transform-serialization.
Summary:
* 2d serializes to just an angle.
* Axis must be specified for 3-D rotations. If the axis is parallel to the x, y or z axis, the appropriate axis name must be used.
Bug: 905274
Change-Id: I5d16f3301baf9720099e4b5b42a1b87077b97998
Reviewed-on: https://chromium-review.googlesource.com/c/1447844
Commit-Queue: Kevin Ellis <kevers@chromium.org>
Reviewed-by: Anders Hartvoll Ruud <andruud@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630796}
--
wpt-commits: a01ac89be3f48fa64bd96d36ac1a65ac9074eb4f
wpt-pr: 15205
Automatic update from web-platform-tests
Make stability checks provide an explicit message for duplicate test names
--
wpt-commits: 23008e948bff4c245148cfcda2ebaa72149a2bb1
wpt-pr: 15330
Automatic update from web-platform-tests
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
Update mozprocess from 0.26 to 1.0.0
--
wpt-commits: 9a874520787308eecb14d85399cb5184efaf99ec, 999c546d4958eb649a71f0339563218a22f6d281, e032140d4ba4954c0704356f4848cc8f1d2dd590, ad09e7414f098c7833a65d336f01e76ba8c1dc15, a54f2b6ed9445d74fded895c45b561b1e7fd9314, 47bb668028ce2b7f6e564b47aee7ed12ce720532, f55469ea2fe1ed49f6b3f2d9701b793fc051bb67, 78c2264a4d9b28e89070670d6a73df470ae45aaf, 725876f98dc3fe1dd86c131b73724ac0e0a41147
wpt-pr: 15294
Automatic update from web-platform-tests
[css-grid] Overflow should be computed with the actual logical bottom
In order to compute the 'auto' height of the grid container, we set
temporarily a value based on the tracks' size. We can see this value as
a kind of 'intrinsic height'. Then, we call to the UpdateLogicalHeight
function to finally compute the grid container's actual logical height.
In order to compute the overflow area in the grid container we should
use the result of the ClientLogicalBottom function. However, we were
retrieving this value just after setting the tracks-based temporary
height, and before computing the actual logical height.
The consequence of this wrong logic is that empty grid areas may
affect the result of the ClientLogicalBottom and, as it's described
in the bug, conclude that there is a content overflow in the grid
container, even if such container has no grid items at all.
Since the grid itself is not a box, it can't contribute to the grid
container's content size; instead, its grid items should.
This change ensures that we always use the actual grid container's
logical bottom to properly compute the overflow area.
Bug: 928885
Change-Id: I05f86fbce06a83c0dbc0d5389bf0416763f8588a
Reviewed-on: https://chromium-review.googlesource.com/c/1459620
Commit-Queue: Javier Fernandez <jfernandez@igalia.com>
Reviewed-by: Manuel Rego <rego@igalia.com>
Cr-Commit-Position: refs/heads/master@{#630735}
--
wpt-commits: aa6f52fc7b99559377c760a550e842438efa91a1
wpt-pr: 15284
Automatic update from web-platform-tests
Fix remaining preload tests flaking on wpt.fyi
This is a follow-up to https://github.com/web-platform-tests/wpt/pull/15075.
https://crrev.com/c/1436777 successfully fixed the wpt.fyi flakiness
for link-header-preload-srcset.tentative.html and
link-header-preload-nonce.html. This patch fixes that for the remaining
preload tests.
- Use step_timeout instead of dummy.js?pipe=trickle, because
dummy.js?pipe=trickle may be loaded from cache with no delay.
- Replace verifyNumberOfDownloads() with
- verifyNumberOfResourceTimingEntries() if possible
- A new helper function verifyLoadedAndNoDoubleDownload() that
doesn't fail if the resource was already cached before running
the test.
Bug: 922343
Change-Id: Ia9b7c4bd49dd76463df4607349fe4f935e5410d0
Reviewed-on: https://chromium-review.googlesource.com/c/1442020
Reviewed-by: Kinuko Yasuda <kinuko@chromium.org>
Commit-Queue: Kunihiko Sakamoto <ksakamoto@chromium.org>
Cr-Commit-Position: refs/heads/master@{#629029}
--
wpt-commits: eb5aa0038ee7a938beabbe4c732a685af2581dbc
wpt-pr: 15129
Automatic update from web-platform-tests
Initial support for WorkletAnimation.playbackRate
Changes for scroll-linked worklet animations will come in a separate pull request.
Bug: 852475
Change-Id: Ie6dd14242797cbc14a7e1e377f23c3dda174fe15
Reviewed-on: https://chromium-review.googlesource.com/c/1423537
Reviewed-by: Stephen McGruer <smcgruer@chromium.org>
Reviewed-by: Majid Valipour <majidvp@chromium.org>
Commit-Queue: Olga Gerchikov <gerchiko@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#628800}
--
wpt-commits: 9d1de21ddfa4338cfac14a887622b98207efe247
wpt-pr: 14946
Automatic update from web-platform-tests
[animation-worklet] Basic pause implementation (reland)
Pausing worklet animation now holds the time. This works as expected for main
thread animations. Implementing this for composited worklet animations will be
done in a follow up patch.
Major changes:
- Add and expose pause() method pausing the animation.
- Introduce hold_time that is used when animation is paused.
- Rework how current time is computed, it is now closer to
regular animations i.e., we either compute it based on "start time and
timeline.currentTime" or use "hold time".
- Instead of setting start time we now set the current time
which then works backward to compute either the start time or the hold time
based on the animation state.
- When transitioning animation play state, we now always set
the current time. Previously this was adhoc and inconsistent.
- Introduce has_started_ to differentiate when playing an
animation for the first time vs playing it from pause.
- Update playback_rate related calculation to use new logic.
Relanding: Original CL was reverted here [1]. Took the following steps to
address this:
- Address flakiness in unit test by using more accurate error
value matching other tests.
- Fix flakiness in layout test (worklet-animation-pause.https.html)
by using startTime and waiting for time to actually advance working around
pre-existing animation clock issue [2]. Also got rid of an invalid state
transition from pause->idle.
[1] https://chromium-review.googlesource.com/c/chromium/src/+/1434815
[2] http://crbug.com/785940
TEST:
- wpt/animation-worklet/worklet-animation-pause.https.html: js test for basic
current time calculations
- wpt/animation-worklet/worklet-animation-pause-immediately.https.html:
reftest for basic pause
- wpt/animation-worklet/worklet-animation-pause-result.https.html: reftest for
pause/resume.
- WorkletAnimationTest.PausePlay: unit test for basic state
transition and time calc
Bug: 821910,
Change-Id: I11fd2960443081be81055904d6d56a2abc3282f5
Reviewed-on: https://chromium-review.googlesource.com/c/1456640
Reviewed-by: Majid Valipour <majidvp@chromium.org>
Reviewed-by: Yi Gu <yigu@chromium.org>
Commit-Queue: Majid Valipour <majidvp@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630548}
--
wpt-commits: 345300fad3945a5f1441fb2b2001109ca48f36e8
wpt-pr: 15279
Automatic update from web-platform-tests
[Payment Request] Add Web Platform Tests for hasEnrolledInstrument().
The failing expectations are required because hasEnrolledInstrument()
is not yet turned on by default in Blink. These tests pass when running
Chrome with --enable-features=PaymentRequestHasEnrolledInstrument.
Bug: 915907
Change-Id: I2189bfedad813e7942c7ddbd44aac78d1b895a30
Reviewed-on: https://chromium-review.googlesource.com/c/1461313
Commit-Queue: Danyao Wang <danyao@chromium.org>
Reviewed-by: Rouslan Solomakhin <rouslan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630578}
--
wpt-commits: a7498ddcfeeca02e7b62ab9e9b359fcaa5764c62
wpt-pr: 15306
Automatic update from web-platform-tests
Deflake RTCPeerConnection-track-stats.https.html.
This CL does the following:
1. Add a test that verifies we have a succeeded candidate pair in
getStats() as soon as iceConnectionState is "connected". This is
currently flaky (https://crbug.com/926170) and marked as such.
2. Add workaround to RTCPeerConnection-track-stats.https.html so that
it does not flake due to 1), and mark it not flaky.
3. Move shared helper functions to RTCPeerConnection-helper.js.
4. Add test coverage for making sure we can become connected in the
recvonly use case. This requires changing the prefix to
".https.html" of "RTCPeerConnection-iceConnectionState".
// Already reviewed by jonasolsson@chromium.orgTBR=guidou@chromium.org
Bug: 922955, 926170
Change-Id: I828e46273a84447c817595a466a5e143bde30eca
Reviewed-on: https://chromium-review.googlesource.com/c/1442201
Reviewed-by: Henrik Boström <hbos@chromium.org>
Commit-Queue: Henrik Boström <hbos@chromium.org>
Cr-Commit-Position: refs/heads/master@{#627421}
--
wpt-commits: 7179024d5dfb231073c7319948499d7e30f671cb
wpt-pr: 15133
Automatic update from web-platform-tests
Update dynamic import tests for HTML event handler changes
Fixes https://github.com/web-platform-tests/wpt/issues/15183. Follows https://github.com/whatwg/html/pull/4302.
This updates paths/expected nonce behavior now some methods of loading are not based off the current script.
--
wpt-commits: 8ecb503bc748ded75e7f5c224b95344a2966eeb9
wpt-pr: 15251
Automatic update from web-platform-tests
Remove OriginalTimingAllowOrigin from ResourceTimingInfo
The OriginalTimingAllowOrigin attribute was added in
https://codereview.chromium.org/271083002 to ensure TAO headers are
preserved on cached resources (304 responses). This CL removes it
because it seems that it is no longer needed. It also moves the
corresponding test to WPT.
Bug: 929453
Change-Id: I41e0c84b04a94acc27c22e375f55398d9fc411e5
Reviewed-on: https://chromium-review.googlesource.com/c/1459333
Commit-Queue: Nicolás Peña Moreno <npm@chromium.org>
Reviewed-by: Yoav Weiss <yoavweiss@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630441}
--
wpt-commits: f84a067164651cc9069c36d6098d6a6b03d63816
wpt-pr: 15300
Automatic update from web-platform-tests
[css-overscroll-behavior] Computed values (#15287)
The computed values of the overscroll-behavior-x and overscroll-behavior-y
properties are the keywords as specified.
https://drafts.csswg.org/css-overscroll-behavior/#property-index
--
wpt-commits: 374bcf2e5aa678ce021ea8e8d0872114b9fd1acf
wpt-pr: 15287
Automatic update from web-platform-tests
[Payment Request][WPT] Fix canMakePayment() tests.
Before this patch, a couple of WPT test cases for canMakePayment() would
fail when attempting to issue an automated click with a user gesture to
to show the payment sheet.
This patch splits the test cases that need user gestures into a manual
test that solicits a real user gesture click to show the payment sheet.
After this patch, the canMakePayment() tests are passing.
Bug: 929773
Change-Id: Ide9a310f39c69957b2b182f00e6db37ea6391e12
Reviewed-on: https://chromium-review.googlesource.com/c/1459061
Reviewed-by: Danyao Wang <danyao@chromium.org>
Commit-Queue: Rouslan Solomakhin <rouslan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630358}
--
wpt-commits: 7d5d8b89aba590b018d7ac402c233df89b42dcbd
wpt-pr: 15281
Automatic update from web-platform-tests
Round up to the next render quantum for suspend
OfflineAudioContext.suspend rounds the time up to the next render
quantum instead of down. It is sometimes confusing when suspend
occurs before the given time, which is different from how everything
else works in WebAudio.
A few tests need to be fixed because they computed the boundary by
rounding down instead of up. One test needed to be adjusted because
the times were rounded to different boundaries instead of the same
boundary.
See also https://github.com/WebAudio/web-audio-api/issues/1822
Bug: 927895
Change-Id: Ie6685c620dd38eb2a059901d13b907764ba1a2db
Reviewed-on: https://chromium-review.googlesource.com/c/1450536
Reviewed-by: Kent Tamura <tkent@chromium.org>
Reviewed-by: Hongchan Choi <hongchan@chromium.org>
Commit-Queue: Raymond Toy <rtoy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#630346}
--
wpt-commits: 1b1a627e597f15b60327191799b8081beac2a3cd
wpt-pr: 15214
Automatic update from web-platform-tests
Add an id="log" container at the top of the page for MathML testharness tests. (#15298)
Currently the test summary is appended at the end of the page, sometimes after
a lot of MathML formulae. Adding an id="log" container allows to display the
summary at the top of the page.
--
wpt-commits: e2dd57c3f273468b3c6a087978c531dcd00933de
wpt-pr: 15298
Automatic update from web-platform-tests
Remove two uses of dedicated-workers.idl (#14952)
Prerequisite for https://github.com/web-platform-tests/wpt/pull/12803.
--
wpt-commits: 45b6a5be22e5ea7596c55f5d1e65bae5e62e1656
wpt-pr: 14952
James Graham
Julian Seward
moz-wptsync-bot
Masatoshi Kimura
Sam Sneddon
Anne van Kesteren
pyup.io bot
Johanna
autofoolip
Kent Tamura
Tab Atkins Jr
Xianzhu Wang
Eric Willigers
Lan Wei
Josh Matthews
Kevin Ellis
Javier Fernandez
Jordan Taylor
Kunihiko Sakamoto
Olga Gerchikov
Majid Valipour
Danyao Wang
Henrik Boström
Jon Coppeard
Nicolas Pena
Rouslan Solomakhin
Raymond Toy
Frédéric Wang
Philip Jägenstedt