4aa07089ce57c4e66e6f4f4cacc915118ea39429: Bug 1267438 - For fixed backgrounds, use the scrolled clip rather than the isClipFixed=false annotation. r=mstange
Botond Ballo <botond@mozilla.com> - Fri, 06 May 2016 19:02:26 -0400 - rev 365139
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - For fixed backgrounds, use the scrolled clip rather than the isClipFixed=false annotation. r=mstange MozReview-Commit-ID: 9STe7ypglpP
1c8e4840f8b40b7e02b9274e446e278e8c907053: Bug 1267438 - Support adding ancestor mask layers from places other than SetupScrollingMetadata. r=mstange
Botond Ballo <botond@mozilla.com> - Fri, 06 May 2016 19:34:16 -0400 - rev 365138
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Support adding ancestor mask layers from places other than SetupScrollingMetadata. r=mstange MozReview-Commit-ID: DwdbSRdEMEc
641eef89309781af7a63df344d0e1f495ba5efd8: Bug 1267438 - Use the scrolled clip in AsyncCompositionManager. r=mstange
Botond Ballo <botond@mozilla.com> - Wed, 04 May 2016 18:52:56 -0400 - rev 365137
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Use the scrolled clip in AsyncCompositionManager. r=mstange MozReview-Commit-ID: LjV8bEhCexE
14ee768229fa8e4e346ff0b62a59c3dff2bc164a: Bug 1267438 - During AlignFixedAndStickyLayers, only un-adjust the fixed portion of a layer's clip rect. r=mstange
Botond Ballo <botond@mozilla.com> - Wed, 04 May 2016 21:05:27 -0400 - rev 365136
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - During AlignFixedAndStickyLayers, only un-adjust the fixed portion of a layer's clip rect. r=mstange MozReview-Commit-ID: 62JVHL5AVhP
68b48bf17f7baa4ac96bd4099c4a4089456eb77c: Bug 1267438 - Factor out a helper function to check if a layer is fixed or sticky. r=mstange
Botond Ballo <botond@mozilla.com> - Wed, 04 May 2016 17:32:12 -0400 - rev 365135
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Factor out a helper function to check if a layer is fixed or sticky. r=mstange MozReview-Commit-ID: A1qrxW3KVVE
6618e805fce8e1b45241783d8ba3b5bec5ad182a: Bug 1267438 - Use the layer's scrolled clip during compositor hit testing. r=mstange
Botond Ballo <botond@mozilla.com> - Wed, 04 May 2016 17:30:26 -0400 - rev 365134
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Use the layer's scrolled clip during compositor hit testing. r=mstange MozReview-Commit-ID: 2nodwaZHeWZ
54cb97cee6810ec56308d1f2f66e5b8fda02da3e: Bug 1267438 - Give layers an optional scrolled clip that is scrolled by all scroll frames associated with the layer. r=mstange
Botond Ballo <botond@mozilla.com> - Fri, 29 Apr 2016 17:32:53 -0400 - rev 365133
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Give layers an optional scrolled clip that is scrolled by all scroll frames associated with the layer. r=mstange MozReview-Commit-ID: EGqVZuISSZQ
5bf9a7962ad4c5d75d6a57a7fc4289f0b6fd680d: Bug 1267438 - Do not propagate the scroll clip to APZC's copy of ScrollMetadata. r=kats
Botond Ballo <botond@mozilla.com> - Fri, 29 Apr 2016 16:16:29 -0400 - rev 365132
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Do not propagate the scroll clip to APZC's copy of ScrollMetadata. r=kats MozReview-Commit-ID: 9E1QGnoxUjN
b802e4ef209948f307565f2ba750b2fbd340bcb6: Bug 1267438 - Group ScrollMetadata's optional clip rect and mask layer index into a LayerClip structure. r=mstange
Botond Ballo <botond@mozilla.com> - Fri, 29 Apr 2016 16:14:21 -0400 - rev 365131
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Group ScrollMetadata's optional clip rect and mask layer index into a LayerClip structure. r=mstange MozReview-Commit-ID: 6W0GZYkioov
1065434be718ec94d30480100f1c409cc446c6ac: Bug 1267438 - Use IntersectMaybeRects() in Layer::GetCombinedClipRect(). r=mstange
Botond Ballo <botond@mozilla.com> - Fri, 29 Apr 2016 15:44:08 -0400 - rev 365130
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1267438 - Use IntersectMaybeRects() in Layer::GetCombinedClipRect(). r=mstange MozReview-Commit-ID: 5DkpzyFrDQ0
f4b3c6c334e9bacbd278dfb4cd5aeee59098ef9d: Bug 1136226 - Fix jit-test shell for SIMD. r=bbouvier
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:31 -0700 - rev 365129
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Fix jit-test shell for SIMD. r=bbouvier Support tests using all SIMD types.
a8a0b4464969d0b76cb23556e6bc940d74e481db: Bug 1136226 - Rename MSimdSplatX4 to MSimdSplat. r=bbouvier
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:31 -0700 - rev 365128
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Rename MSimdSplatX4 to MSimdSplat. r=bbouvier This MIR instruction will be used for splatting all vector shapes.
117800523a0368e2a98d90befbc9c15854c7ce6c: Bug 1136226 - Make MSimdSwizzle and MSimdShuffle length-agnostic. r=bbouvier
Jakob Stoklund Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:31 -0700 - rev 365127
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Make MSimdSwizzle and MSimdShuffle length-agnostic. r=bbouvier Constructors and factories take lane lists as arrays instead of four separate lane arguments.
bb54ae23b22a7434ffef46e9678731338971d1bd: Bug 1136226 - Add masm.simd128Constant(). r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365126
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Add masm.simd128Constant(). r=sunfish Replace the existing int32x4Constant and float32x4Constant masm methods with a generic simd128Constant.
f63aa4a372a4af819f067d0aad23cf60a14619f6: Bug 1136226 - Rename 32x4 SIMD masm methods to "Simd128". r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365125
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Rename 32x4 SIMD masm methods to "Simd128". r=sunfish Load, store, and move methods are generic for all 128-bit SIMD types. Keep the difference between float and int vectors since x86 sometimes cares.
c515aeae0c850113c1127a0db4f3a72e8822f71e: Bug 1136226 - Update LDefinition and MoveOp for 8x16 and 16x8. r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365124
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Update LDefinition and MoveOp for 8x16 and 16x8. r=sunfish In both classes, rename enums: INT32X4 -> SIMD128INT FLOAT32X4 -> SIMD128FLOAT Catch the new MIRTypes to produce SIMD128INT.
2d6da6b2c98ab18cbb2f071c7e623180d48bd218: Bug 1136226 - Materialize 8x16 and 16x8 SIMD constants. r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365123
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Materialize 8x16 and 16x8 SIMD constants. r=sunfish Rename LIR instructions: LInt32x4 -> LSimd128Int LFloat32x4 -> LSimd128Float. These two LIR instructions can be used to materialize 128-bit SIMD vectors of other geometries too. Also rename the masm.loadConstant{Int,Float}32x4() functions to indicate that they can be used for other geometries.
5de8e799eae3901a46aba497de1601be02c61a1a: Bug 1136226 - Add 16x8 and 8x16 MIRTypes. r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365122
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Add 16x8 and 8x16 MIRTypes. r=sunfish This adds 8x16 and 16x8 types to MIRType, but it does not enable inlining of the corresponding SIMD operations yet. Explicitly disable the inlining of non-4-lane constructor calls until we have support for them.
55710af96f41086584b3c9065e552ec129f23a96: Bug 1136226 - Support 8x16 and 16x8 types in SimdConstant. r=sunfish
Jakob Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365121
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Support 8x16 and 16x8 types in SimdConstant. r=sunfish Extend the SimdConstant class to handle other 128-bit SIMD vector types. Remove some unused methods.
48e12dc46e554777b2048cbf515cd4f808364fb7: Bug 1136226 - Remove SimdLane enumeration. r=sunfish
Jakob Stoklund Olesen <jolesen@mozilla.com> - Mon, 09 May 2016 16:48:30 -0700 - rev 365120
Push 17650 by martin.thomson@gmail.com at Tue, 10 May 2016 05:06:10 +0000
Bug 1136226 - Remove SimdLane enumeration. r=sunfish We're about to add 16-lane and 8-lane SIMD types to IonMonkey, so we don't want an enum to identify lanes. Just use an unsigned instead. Also note that SIMD.js no longer has the .x .y .z .w properties. Change a few lane < 4 checks to use the number of lanes in the Simd type.
(0) -300000 -100000 -30000 -10000 -3000 -1000 -300 -100 -50 -20 +20 +50 +100 +300 +1000 +3000 +10000 +30000 +100000 +300000 tip