Automatic update from web-platform-tests
[TablesNG] clusterfuzz multicol causes illegal table layout trees
The cause of the crash is legacy LayoutTableColumn inside
LayoutNGTable. This triggers bad typecast, wrong methods get called...
The fix is to propagate SetShouldForceLegacyLayoutForChild
correctly. It was only propagating until it found parent
with legacy layout.
Note: SetShouldForceLegacyLayoutForChild flag never gets cleared.
Also added DCHECKs to detect mixed table trees earlier.
Bug: 1178263
Change-Id: Ifebaaccbad4006819ca66fed8310ddeb59e1aa64
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2692390
Commit-Queue: Aleks Totic <atotic@chromium.org>
Reviewed-by: Morten Stenshorne <mstensho@chromium.org>
Cr-Commit-Position: refs/heads/master@{#854106}
--
wpt-commits: 4e40ddcf74f404619016b333cd47c1ac58c6bc95
wpt-pr: 27631
Automatic update from web-platform-tests
Consider transform change countering layout shift
This is to ignore layout shift when transform and location change at the
same time, and visual representation is kept unchanged. This happens in
many websites containing carousel UI.
Summary of layout_shift.cluster_telemetry
https://ct.skia.org/results/cluster-telemetry/tasks/chromium_perf_runs/wangxianzhu-ChromiumPerf-5629/html/index.html:
mainFrameCumulativeLayoutShift -4.5%
overallCumulativeLayoutShift -2.5%
Bug: 1169326
Change-Id: Icb8f5748e927753ef244429699cc1396d8cb7f71
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2673965
Commit-Queue: Xianzhu Wang <wangxianzhu@chromium.org>
Reviewed-by: Chris Harrelson <chrishtr@chromium.org>
Reviewed-by: Steve Kobes <skobes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#854104}
--
wpt-commits: 363d949874454dc5a93c817b7eaa8c0427f9c0b5
wpt-pr: 27467
Automatic update from web-platform-tests
Adds a test for use of datachannel after handover.
This particular pattern works. There are others that do not.
Bug: chromium:1165692
Change-Id: Ifef02d6b5e41f6b3179aed5a0af3258be394bcfd
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2695350
Commit-Queue: Harald Alvestrand <hta@chromium.org>
Reviewed-by: Henrik Boström <hbos@chromium.org>
Cr-Commit-Position: refs/heads/master@{#854052}
--
wpt-commits: da00061247e7d6d63e0f5d1802e45001c3bd2c98
wpt-pr: 27633
Automatic update from web-platform-tests
[TablesNG] clusterfuzz empty cells crash
If all cells in a table column were empty, collapsed border
dimensions were computed incorrectly.
Fix: do not skip empty cells when dimensioning borders.
Bug: 1178110
Change-Id: I55b099c095b163d198a8dac5cfc7f172e6046a20
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2694726
Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org>
Commit-Queue: Aleks Totic <atotic@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853950}
--
wpt-commits: a8c6d1fe99eb0ab6ea4e4c19d7a918616e22bb82
wpt-pr: 27624
Automatic update from web-platform-tests
Support a crossorigin= attribute for webundle subresource loading
There are several changes in this CL. Let me summarize at first.
The current behavior is:
- No support crossorigin= attribute
- request's mode is set to "cors" in default.
- request's creditial-mode is "omit"; UA never sends a credential.
This CL changes the behavior as follows:
- If crossorigin= attribute is not specified,
- request's mode is set to "no-cors"
- request's credential-mode is set to "include"
- If crossorigin= attribute is "anonymous" or empty (See [1] for details)
- request's mode is set to "cors"
- request's credential-mode is set to "same-origin"
- If crossorigin= attribute is "use-credential",
- request's mode is set to "cors"
- request's credential-mode is set to "include"
Note: In the current implementation, subresources can be loaded from a
cross-origin bundle which the server returns without a valid
"Access-Control-Allow-Origin" response header. That should be considered
as a bug. This Cl changes request's mode to no-cors correctly in
default.
Regarding tests,
A general behavior of a crossorigin attribute is well tested in existing
tests, such as fast/dom/HTMLLinkElement/link-crossOrigin.htm.
This CL tests only the effects of crossorign= attributes for a bundle
as follows:
1. Test for credentials:
subresource-loading-credential.tentative.sub.html
2. Test for CORS behavior:
subresource-loading-cors.tentative.html
3. Test for CORS behavior (error case):
subresource-loading-cors-error.tentative.html
The existing test, subresource-loading-cross-origin.tentative.html is
rewritten and split into 2 and 3.
Due to https://crbug.com/1168449, the test 3 is marked as TIMEOUT.
This bug will be fixed as another CL.
[1]: https://html.spec.whatwg.org/multipage/#cors-settings-attribute
BUG=1149816, 1168449
Change-Id: I354e203a3a536bc829ef7ba6a97d4edd6a6340a0
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2670486
Reviewed-by: Tsuyoshi Horo <horo@chromium.org>
Reviewed-by: Kinuko Yasuda <kinuko@chromium.org>
Reviewed-by: Kunihiko Sakamoto <ksakamoto@chromium.org>
Commit-Queue: Hayato Ito <hayato@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853942}
--
wpt-commits: cf9a213635193035a3409436a594995e84e5346d
wpt-pr: 27452
Automatic update from web-platform-tests
Let OverflowClip instead of ScrollTranslation create layout shift root
This avoids layout shift root status change (causing layout shifts in
descendants ignored) when the existence of ScrollTranslation changes,
e.g. when overflowing status changes or scroll offset changes from 0 to
non-zero.
Summary of layout_shift.cluster_telemetry
https://ct.skia.org/results/cluster-telemetry/tasks/chromium_perf_runs/wangxianzhu-ChromiumPerf-5625/html/index.html:
mainFrameCumulativeLayoutShift 1.441%
overallCumulativeLayoutShift 0.519%
Bug: 1165353
Change-Id: I61cb5994e9d70c751a2767de417b8e70aab79def
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2690998
Reviewed-by: Chris Harrelson <chrishtr@chromium.org>
Commit-Queue: Xianzhu Wang <wangxianzhu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853843}
--
wpt-commits: 9b0d1ce0412c990e8d1e715c88ad37a6d88ebf0b
wpt-pr: 27615
Automatic update from web-platform-tests
Stricter mixed content check for blob and filesystem URLs
The mixed content spec relies on the concept of potentially trustworthy
URLs which checks trustworthiness of blob and filesystem by checking the
origin of the context in which they were created [1] [2]. Concretely,
"blob:https://" is trustworthy while "blob:http://" is not. Currently
Chromium treats any blob: or filesystem: URLs as trustworthy. This CL
updates the behavior to match the stricter rule from the specification.
[1] https://w3c.github.io/webappsec-mixed-content/
[2] https://w3c.github.io/webappsec-secure-contexts/#is-url-trustworthy
[3] https://groups.google.com/a/chromium.org/g/blink-dev/c/nrpl_ewkmaQ
Bug: 1172988
Change-Id: I6fe6d47d0436d405294194e0fcdd4d48461ad7b5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2662116
Commit-Queue: Frédéric Wang <fwang@igalia.com>
Reviewed-by: Matt Falkenhagen <falken@chromium.org>
Reviewed-by: Mike West <mkwst@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853814}
--
wpt-commits: 06098389364eb8c5046ea0f62db58fd6b05754cc
wpt-pr: 27604
Automatic update from web-platform-tests
Portals: focus() inside portal should change active element
Allows a portal document's activeElement to update when focus is called.
Note that focus events won't be dispatched as the portal doesn't get
page focus, and the portal WebContents isn't set as the focused
WebContents (with the exception of orphaned portals). Updating the
activeElement also means an autofocused element inside a portal will
receive focus when the portal is activated.
More discussion here: https://github.com/WICG/portals/issues/257
Change-Id: If67be1f424114653983b473a80c515337d49596a
Bug: 1059404
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2566072
Commit-Queue: Adithya Srinivasan <adithyas@chromium.org>
Reviewed-by: Lucas Gadani <lfg@chromium.org>
Reviewed-by: Jeremy Roman <jbroman@chromium.org>
Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Reviewed-by: David Trainor <dtrainor@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853647}
--
wpt-commits: a507854e826d126010758ba44847d1b67fb29c01
wpt-pr: 27461
Automatic update from web-platform-tests
[GridNG] Resolve both min-length & max-length for auto repetitions.
As above. This also considers transferred sizes from the aspect-ratio.
The new tests are for dynamic changes which were previously untested
(and for some cases not working), and for aspect-ratio.
Bug: 1045599
Change-Id: I061e97a15190225616ceb18c031029f8c0791641
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2664997
Commit-Queue: Ian Kilpatrick <ikilpatrick@chromium.org>
Reviewed-by: Ethan Jimenez <ethavar@microsoft.com>
Reviewed-by: Kurt Catti-Schmidt <kschmi@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#853613}
--
wpt-commits: 333c487de3e5912deb80dc55c31bf467f97df6d2
wpt-pr: 27450
Automatic update from web-platform-tests
Fix paint2d-gradient.https.html
The test has a linear gradient, together with OOP-R enabled
everyone, it seems we have very subtle pixel difference where
human eyes cannot see. Here is a picture shows what the diff
looks like:
https://drive.google.com/file/d/15dU0JLc2lj00J_eoN6PQQ3lZMpymQrMm/view?usp=sharing
I changed the test to be fuzzy matching, this should fix the
test. The fix is this line:
<meta name=fuzzy content="0-5;3000-4000">
It means that when there is a pixel diff, the diff for each
color channel is between 0-5. And there are 3000-4000 pixels
diff.
Bug: 1034807
Change-Id: Iae83f00815c4231c1073a812e7f62f3abe3e4c5e
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2692506
Reviewed-by: Kevin Ellis <kevers@chromium.org>
Commit-Queue: Xida Chen <xidachen@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853600}
--
wpt-commits: 2d2308f7f690c310e542dcac941eb2c86a64c7c0
wpt-pr: 27611
Automatic update from web-platform-tests
[python3] Fix regen_certs.yml for the python3 migration (#27612)
This was trying to call 'wpt' with python 2, which is now rejected.
--
wpt-commits: 94e17303bfa24d017d68eb4e81984bd145fb0a73
wpt-pr: 27612
Automatic update from web-platform-tests
Cleanup unnecessary `__init__.py` files
Now that we are Py3-only and have PEP 420, we should not need paths of
__init__.py files to file handlers.
Fixes#26552
--
wpt-commits: 3586ff740b00aa1fa82ab00cccbc36cca0bb8ccb
wpt-pr: 27529
Automatic update from web-platform-tests
[LayoutNGTable] Cells and captions are never suitable engine boundaries.
We need to check for this also when marking the ancestry, and not just
at the target element.
Bug: 1177684
Change-Id: I0543d2a8a1791282aeb81757a1e82f8f6e918c52
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2690854
Commit-Queue: Aleks Totic <atotic@chromium.org>
Reviewed-by: Aleks Totic <atotic@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853547}
--
wpt-commits: 13610f951cb8d2d1750d1feaf98c89b486b0c6ba
wpt-pr: 27608
Automatic update from web-platform-tests
Clear reports in WPT collector when queried.
This fixes a bug in the WPT report collector where reports are not
removed from the stash when queried, as they were intended to be.
There was one test in the Network Error Logging suite which relied on
this bug, and tested reports in two passes, which could now fail if all
reports are received before the first query is performed, so this CL
also fixes that test by adding an optional 'retain' query parameter to
the report collector.
Bug: 1177757
Change-Id: I945325f15a2ce633cfa32bcebeedb5be5afaa860
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2689369
Reviewed-by: Stephen McGruer <smcgruer@chromium.org>
Commit-Queue: Ian Clelland <iclelland@chromium.org>
Cr-Commit-Position: refs/heads/master@{#853528}
--
wpt-commits: be055c46bb610b18e2c3f5612aff388f9003750f
wpt-pr: 27607
Automatic update from web-platform-tests
Fix (#27610)
In PR #27520 we landed a new test for buttons with children with percentage heights called percentage-descendant-of-anonymous-flex-item.html
In the review process the height: 100% was replaced by height: 100px. However that was a bad decision because every UA styles buttons in a different way (even the same UA in different platforms) meaning that due to margins and paddings that height could be smaller than 100px. Switching back to the percentage height in the expectations.
--
wpt-commits: f3a7512a062994d1d2f4f0fbc7fe89c011e6ee4a
wpt-pr: 27610
Roughly:
- Don't reset on load
- Don't reset on URL parsing error
- Reset when playing a MediaStream, a MediaSource
- Otherwise follow the resource selection algorithm, and set it to what has been selected
Differential Revision: https://phabricator.services.mozilla.com/D101937
This test highlights the issue reported in Bug 1673829, where the location links
in the rule view are incorrect after navigating, when `devtools.testing.enableServerWatcherSupport`
is true.
At the moment, this test fails when we try to enable the stylesheet resource
watcher support (Bug 1685268).
Differential Revision: https://phabricator.services.mozilla.com/D105286
This adds methods to verify the chunk metadata very GC in debug builds (held relocated arenas mess things up because they're mprotected so it's checked after we free those).
The code in question was hard to exercise and I had to update reportLargeAllocationFailure to take a byte count so we could avoid calling the shell's large allocation failure callback, which causes the shell to exit (see JSRuntime::onOutOfMemoryCanGC).
It's debateable whether it's worth handling OOM from MarkPagesUnusedSoft. If we assume pages have been been successfully marked unused regardless of failure I think the worst that happens is that we don't reuse them immediately... I ended up handling this anyway though, and rebuilding the list rather than clearing it entirely in decommitFreeArenasWithoutUnlocking.
Differential Revision: https://phabricator.services.mozilla.com/D104837
```
2021-02-05 Danh <congdanhqx@gmail.com>
* gtests/manifest.mn:
Bug 1688374 - Fix parallel build NSS-3.61 with make. r=kjacobs
[a5c857139b37] [NSS_3_62_BETA1]
2021-02-05 Robert Relyea <rrelyea@redhat.com>
* lib/libpkix/pkix/util/pkix_tools.c:
Bug 1682044 pkix_Build_GatherCerts() + pkix_CacheCert_Add() can
corrupt "cachedCertTable"
Patch by Andrew Cagney Preliminary Review by Ryan Sleevie Tested
against all.sh rrelyea. r=kjacobs
(this bug is old)
pkix_Build_GatherCerts() has two code paths for creating the list
"certsFound":
pkix_CacheCert_Lookup() this sets "certsFound" to a new list
"certsFound" and "cachedCertTable" share items but not the list
pkix_CacheCert_Add(pkix_pl_Pk11CertStore_CertQuery()) this sets
"certsFound" to a new list; and then adds the list to
"cachedCertTable" "certsFound" and "cachedCertTable" share a linked
list
Because the latter doesn't create a separate list, deleting list
elements from "certsFound" can also delete list elements from within
"cacheCertTable". And if this happens while pkix_CacheCert_Lookup()
is trying to update the same element's reference, a core dump can
result.
In detail (note that reference counts may occasionally seem off by
1, its because data is being captured before function local
variables release their reference):
pkix_Build_GatherCerts() calls pkix_pl_Pk11CertStore_CertQuery()
(via a pointer) to sets "certsFound":
PKIX_CHECK(getCerts (certStore, state->certSel, state->verifyNode,
&nbioContext, &certsFound, plContext), PKIX_GETCERTSFAILED);
it then calls:
PKIX_CHECK(pkix_CacheCert_Add (certStore, certSelParams,
certsFound, plContext), PKIX_CACHECERTADDFAILED);
[dafda4eee75c]
```
Differential Revision: https://phabricator.services.mozilla.com/D105209
I thought we didn't support outline-offset on auto-style outline.
The rect we get is already inflated, so we just got to compute the
radius using that.
Differential Revision: https://phabricator.services.mozilla.com/D105125
From GTK 3.24.25 on we have a new API that allows us to savely apply
opaque regions to our own surfaces without risking to freeze GDK.
Differential Revision: https://phabricator.services.mozilla.com/D102835
Expectation to timeout got set in latest synchronization with upstream
repository of web-platform-tests (bug 1692513).
Depends on D105275
Differential Revision: https://phabricator.services.mozilla.com/D105276
Expectation as timeout got set in latest synchronization with upstream
web-platform-tests repository (bug 1692513).
Depends on D105274
Differential Revision: https://phabricator.services.mozilla.com/D105275