258117b770a56d136c08bcf3ac9a1fbdf2c085ce: Bug 1624810 - Assert that only JS holders with the approprate flag set can contain pointers to GC things in more than one zone r=mccr8
Jon Coppeard <jcoppeard@mozilla.com> - Thu, 26 Mar 2020 10:49:20 +0000 - rev 520490
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624810 - Assert that only JS holders with the approprate flag set can contain pointers to GC things in more than one zone r=mccr8 Differential Revision: https://phabricator.services.mozilla.com/D68197
f4d11e6a6e3f74cd0f43c2a657b596b1eb4a1e55: Bug 1624810 - Set the multi-zone JS holder flag on appropriate cycle collected classes r=mccr8
Jon Coppeard <jcoppeard@mozilla.com> - Thu, 26 Mar 2020 10:49:06 +0000 - rev 520489
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624810 - Set the multi-zone JS holder flag on appropriate cycle collected classes r=mccr8 Set the flag on the affected classes, which are: - CallbackTimeoutHandler - nsJSArgArray - CallbackObject - Console - MessageEvent - IDBIndexCursor - ExtendableMessageEvent - JSPurpleBuffer Differential Revision: https://phabricator.services.mozilla.com/D68196
4dcdeed2265fe0028fe816097f6cddc62d326977: Bug 1624810 - Add a nsCycleCollectionParticipant flag for JS holders that can have GC things in multiple zones r=mccr8
Jon Coppeard <jcoppeard@mozilla.com> - Thu, 26 Mar 2020 10:48:30 +0000 - rev 520488
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624810 - Add a nsCycleCollectionParticipant flag for JS holders that can have GC things in multiple zones r=mccr8 This adds the flag itself and a means of setting it for a cycle collected class. The proper way to do this would be to add versions of all the NS_DECL_CYCLE_COLLECTION_CLASS macros that also set this flag but we end up needing to create separate versions for five of these and it's a lot of macro code to add. Here I added a version of NS_IMPL_CYCLE_COLLECTION_CLASS that sets the flag. This has the disadvantage that it wouldn't work well if we needed to set the flag on a base class as we'd have to use this macro for every derived class. However that situation doesn't actually arise (and ideally this flag will bet set on the fewest number of classes possible). Differential Revision: https://phabricator.services.mozilla.com/D68195
c6da88c617db64498e4bb6aa2ccd75a14d31b865: Bug 1624810 - Replace individual nsCycleCollectionParticipant flags with a bit field to make it easier pass multiple flags through derived class oonstructors r=mccr8
Jon Coppeard <jcoppeard@mozilla.com> - Thu, 26 Mar 2020 10:48:17 +0000 - rev 520487
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624810 - Replace individual nsCycleCollectionParticipant flags with a bit field to make it easier pass multiple flags through derived class oonstructors r=mccr8 Currently classes derived from nsCycleCollectionParicipant have one flag to pass and adding more makes the class delcaration macros get messy really quickly. This patch replaces boolean flags with a bitfield which makes adding new flags easier. Differential Revision: https://phabricator.services.mozilla.com/D68194
02c284defc4d5f8d9ed1c7c8363becdd93f4b291: Bug 1611848 - Properly suppress shadow dom / display: contents inside svg text. r=heycam
Emilio Cobos Álvarez <emilio@crisal.io> - Thu, 26 Mar 2020 00:06:53 +0000 - rev 520486
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1611848 - Properly suppress shadow dom / display: contents inside svg text. r=heycam Returning null from FindSVGData just means "fall back to whatever display specifies", and that's not great. Differential Revision: https://phabricator.services.mozilla.com/D67706
e3fe601c9f615fadf19f416f84c16f417c07ccb7: Bug 1625061 - [MIPS] Remove unnecessary MOZ_ASSERT in ma_cmp. r=lth
Zhao Jiazhong <zhaojiazhong-hf@loongson.cn> - Thu, 26 Mar 2020 10:26:07 +0000 - rev 520485
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1625061 - [MIPS] Remove unnecessary MOZ_ASSERT in ma_cmp. r=lth Differential Revision: https://phabricator.services.mozilla.com/D68310
ee0d41c18c40e8364705ff6609442c7895a4ec17: Bug 1624436 - Don't include GTK widget padding in outline: auto size. r=karlt,stransky
Emilio Cobos Álvarez <emilio@crisal.io> - Thu, 26 Mar 2020 10:23:39 +0000 - rev 520484
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624436 - Don't include GTK widget padding in outline: auto size. r=karlt,stransky To paint outline: auto, we paint the focused border of a GTK_ENTRY_PAINT. We're also adding the padding of the entry, and that's wrong and causes undesirable padding that looks bogus. Differential Revision: https://phabricator.services.mozilla.com/D68191
57e989c8a37cbe84810e37dce41331a77847bd66: Bug 1624842 - Update base shapes on the main thread after compacting GC r=sfink
Jon Coppeard <jcoppeard@mozilla.com> - Wed, 25 Mar 2020 22:36:29 +0000 - rev 520483
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624842 - Update base shapes on the main thread after compacting GC r=sfink Previously we updated base shapes in parallel on helper threads, but this can cause a race with updating shapes. The race occurs between getting a base shape's unowned base shape in fixupShapeTreeAfterMovingGC and updating that field of the base shape. I don't think this race is currently causing problems. The fix is to update base shapes on the main thread along with shapes (which are already updated on the main thread). It's unlikely that this will make much difference to performance due to the relatively small number of base shapes. I took the opportunity to refactor Shape::fixupShapeTreeAfterMovingGC slightly. Differential Revision: https://phabricator.services.mozilla.com/D68206
f72280a54c6e1889d33de7a4e79d4fb0b3927c11: Bug 1624793 part 3 - Handle JSOp::ThrowSetConst in BytecodeFallsThrough. r=iain
Jan de Mooij <jdemooij@mozilla.com> - Wed, 25 Mar 2020 16:39:23 +0000 - rev 520482
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624793 part 3 - Handle JSOp::ThrowSetConst in BytecodeFallsThrough. r=iain Just a drive-by change to be more consistent with JSOp::Throw and JSOp::ThrowMsg. It also results in slightly more compact code in Baseline Interpreter and JIT. Differential Revision: https://phabricator.services.mozilla.com/D68166
635cc9ec177ccc0d0f6dbd3c60a87318b7e6f3d4: Bug 1624793 part 2 - Do the same thing for MThrowRuntimeLexicalError. r=iain
Jan de Mooij <jdemooij@mozilla.com> - Thu, 26 Mar 2020 08:30:13 +0000 - rev 520481
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624793 part 2 - Do the same thing for MThrowRuntimeLexicalError. r=iain We can now easily terminate the block for JSOp::ThrowSetConst too. In theory we could also terminate the block after the MThrowRuntimeLexicalError added in IonBuilder::addLexicalCheck, but because that depends on type information instead of being a static property of the bytecode I'm not completely sure it's always safe (considering arguments/definite-properties analysis, for example). Differential Revision: https://phabricator.services.mozilla.com/D68165
f0a1f1eade8b279ed64881620b509aaf7fc5b366: Bug 1624793 part 1 - Simplify MThrow and IonBuilder::visitThrow. r=iain
Jan de Mooij <jdemooij@mozilla.com> - Thu, 26 Mar 2020 08:29:50 +0000 - rev 520480
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624793 part 1 - Simplify MThrow and IonBuilder::visitThrow. r=iain The changes: * Make MThrow a non-control-instruction (followed by the MUnreachable control instruction). * This allows us to remove the separate MNop and attach the resume point to MThrow itself. * WarpBuilder only allows resumeAfter() on effectful instructions, so make sure MThrow isEffectful() by giving it a store AliasSet. Differential Revision: https://phabricator.services.mozilla.com/D68164
fff410e32a31f0849fc3d65048ed07d32fa12a08: Bug 1600193 - Reduce page load timeouts for some raptor benchmark tests. r=whimboo,perftest-reviewers,AlexandruIonescu
Gregory Mierzwinski <gmierz2@outlook.com> - Thu, 26 Mar 2020 06:47:56 +0000 - rev 520479
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1600193 - Reduce page load timeouts for some raptor benchmark tests. r=whimboo,perftest-reviewers,AlexandruIonescu This patch reduces the page load timeout for speedometer and unity-webgl tests to 7 minutes. Differential Revision: https://phabricator.services.mozilla.com/D68059
46bf06a42ba623133cfdaceac9d3f9f778975b79: Bug 1589497 - Make PendingAction of TSFStore copy of key message r=m_kato
Masayuki Nakano <masayuki@d-toybox.com> - Thu, 26 Mar 2020 07:33:37 +0000 - rev 520478
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1589497 - Make PendingAction of TSFStore copy of key message r=m_kato `TSFTextStore::sHandlingKeyMsg` refers pointer of struct, but referred via `TSFTextStore::PendingAction` so that we should make it has a copy of `sHandlingKeyMsg` because of for async handling. Differential Revision: https://phabricator.services.mozilla.com/D68049
1fe283e3bb59d87eb597ee4a0287fbafe264bc34: Bug 1624333 - Don't close autocomplete popup when pressing Delete key. r=Honza.
Nicolas Chevobbe <nchevobbe@mozilla.com> - Thu, 26 Mar 2020 07:18:23 +0000 - rev 520477
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1624333 - Don't close autocomplete popup when pressing Delete key. r=Honza. If the delete key is pressed, even if the input isn't modified, codeMirror will still emit _onEditorBeforeChange, which would close the popup in the end. That's not something we should do, so in this patch, we bail out if we detect that there was no changes in the input. A test is added to ensure this works as expected and we avoid regression. Differential Revision: https://phabricator.services.mozilla.com/D67903
3beaf308af00777f66e62a4da013668777e68ca5: Bug 1623919 - Add jest tests for toolbox redux store. r=jlast.
Nicolas Chevobbe <nchevobbe@mozilla.com> - Wed, 25 Mar 2020 20:59:56 +0000 - rev 520476
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1623919 - Add jest tests for toolbox redux store. r=jlast. This patch adds some unit tests for the toolbox redux store, especially for the threads reducers and actions. In order for jest to work, we need to add babel as well as some mocks. Differential Revision: https://phabricator.services.mozilla.com/D68157
f2f4628ef59132f35a4571876952f32f09175847: Bug 1623913 - Make `WSRunObject::PrepareToDeleteRangePriv()` invalidate child of `mScanStartPoint` before reusing it r=m_kato
Masayuki Nakano <masayuki@d-toybox.com> - Thu, 26 Mar 2020 05:53:22 +0000 - rev 520475
Push 37252 by malexandru@mozilla.com at Thu, 26 Mar 2020 15:34:27 +0000
Bug 1623913 - Make `WSRunObject::PrepareToDeleteRangePriv()` invalidate child of `mScanStartPoint` before reusing it r=m_kato I give up to write a clean patch for this bug with current design. The trigger is indeed bug 1618089, but this is a hidden regression of bug 1530649. Starting from bug 1530649, `WSRunObject` started to use `EditorDOMPoint` for storing the specified point. And it may store (or only store) child node. Therefore, if it points a text node and it's removed by `WSRunObject::DeleteRange()`, the point becomes invalid even if its offset is still valid. Therefore, we should make `mStartScanPoint` and `mEndScanPoint` forget their child before DOM tree change ideally, but it means that we need to compute offset of the child every time before changing the DOM tree. We cannot accept this safest approach due to performance reason. Therefore, this patch just invalidates `mStartScanPoint`'s child node only when it's reused after the DOM tree is modified. Differential Revision: https://phabricator.services.mozilla.com/D68048
3e5a7430c8d7d87bfeff471e62643185393a34c6: Bug 1622599 - preserves columns state in clearFrames Action. r=Honza
Pranav Pandey <pranavpandey1998official@gmail.com> - Thu, 26 Mar 2020 07:00:04 +0000 - rev 520474
Push 37251 by malexandru@mozilla.com at Thu, 26 Mar 2020 09:33:08 +0000
Bug 1622599 - preserves columns state in clearFrames Action. r=Honza Differential Revision: https://phabricator.services.mozilla.com/D68321
3aecd3848e20c483ea3f7fcbf85bfc92baa0019c: Bug 1623544 - Install python3-pip on Debian during bootstrap; r=firefox-build-system-reviewers,rstewart
byron jones <glob@mozilla.com> - Wed, 25 Mar 2020 15:31:09 +0000 - rev 520473
Push 37251 by malexandru@mozilla.com at Thu, 26 Mar 2020 09:33:08 +0000
Bug 1623544 - Install python3-pip on Debian during bootstrap; r=firefox-build-system-reviewers,rstewart Differential Revision: https://phabricator.services.mozilla.com/D68160
e7c568e7ecddc180841cb267d4f9e079b515f277: Bug 1625040 - Fix clang-5 toolchain task symbol. r=dmajor
Mike Hommey <mh+mozilla@glandium.org> - Thu, 26 Mar 2020 03:18:35 +0000 - rev 520472
Push 37251 by malexandru@mozilla.com at Thu, 26 Mar 2020 09:33:08 +0000
Bug 1625040 - Fix clang-5 toolchain task symbol. r=dmajor Differential Revision: https://phabricator.services.mozilla.com/D68292
08ad6f3a8a55d4a3a53c70c662516b146296cd1e: Bug 1624216 - Dirty rect is not valid when Tile's local valid rect is changed r=gw
sotaro <sotaro.ikeda.g@gmail.com> - Thu, 26 Mar 2020 02:37:57 +0000 - rev 520471
Push 37251 by malexandru@mozilla.com at Thu, 26 Mar 2020 09:33:08 +0000
Bug 1624216 - Dirty rect is not valid when Tile's local valid rect is changed r=gw Differential Revision: https://phabricator.services.mozilla.com/D68298
(0) -300000 -100000 -30000 -10000 -3000 -1000 -300 -100 -50 -20 +20 +50 +100 +300 +1000 +3000 +10000 tip