Open
Conversation
This doesn't change any behavior, just fixes a mismatch between function prototype and implementation. Issue detected here: https://skia-review.googlesource.com/c/skia/+/793458/comment/b098fd75_4bd1191d/ Change-Id: I4ae48ed0d22dbb220c9310d91f8db2eee34ab43e Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796396 Reviewed-by: Nolan Scobie <nscobie@google.com> Commit-Queue: Nolan Scobie <nscobie@google.com> Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
Consolidates the variations across Layouts into policies described by a new LayoutRules namespace holding rule functions. Moves allocation logic into an append(alignment, size) function in the header so it can be inlined. Consolidates the write logic into a write<N,Half> and writeArray<N,Half> template. Branching logic is based on both the compile-time dimensions/Half-ness and the LayoutRules. In many cases these LayoutRule checks are skipped due to the compile-time checks. Shared logic in write() and writeArray() is extracted into a LayoutTraits struct to define shared constants and conversion functions. Lastly, there are helpers to write[Array]<SkSLType> used by the public type-safe functions. Looking at `objdump`, it appears most functions are successfully inlined with mainly the more complex array-writing and the general write(Uniform) variant existing as its own function. The layout alignment and offset are now calculated directly from the UniformManager's storage instead of doing a dry-run through UniformOffsetCalculator that doesn't write, and then repeating the logic but with an actual dst pointer. UniformOffsetCalculator has been updated to apply the LayoutRules based on the various SkSLTypeFoo helper functions. Simplifying the logic and allowing the writes to appears to lead to some improvements in overall frametime in MotionMark scenes with high draw counts: On my M1 mini, when running the Paths SKP in viewer before this change, uniform extraction took 18% of the total time, split 10% in ExtractPaintData and 8% in ExtractRenderStepData. With this change, uniform extraction shrunk to about 9% of the frame time. ExtractPaintData went to 7.5% and ExtractRenderStepData went to 1.5%. There's additional streamlining we can do in KeyHelpers to make ExtractPaintData() less of a hot spot too. Running nanobench has the Paths SKP drop from 2.55ms to 2.31ms, which is commensurate with the Instruments hotspot reduction above. Other improvements in MotionMark: Paths: 2.5 -> 2.3 in grmtl Arcs: 4.5 -> 4.2 "" Suits: 6.1 -> 5.8 "" Leaves: .6 -> .55 "" Bug: b/317233287 Bug: b/317233303 Change-Id: Ie2659c4766ae4fa043b85961e2efe8147e460cee Reviewed-on: https://skia-review.googlesource.com/c/skia/+/793356 Commit-Queue: Michael Ludwig <michaelludwig@google.com> Reviewed-by: John Stiles <johnstiles@google.com>
Land after [1]. [1] https://chromium-review.googlesource.com/c/chromium/src/+/5165370 Bug: b/317330945 Change-Id: I7089f8abe6c46aaf20765559ca9007e45cbae008 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796196 Reviewed-by: Ben Wagner <bungeman@google.com> Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Dominik Röttsches <drott@google.com>
Enables the fStorageBufferSupport and fStorageBufferPreferred fields in DawnCaps which will have the side-effect of enabling paint param and render step uniforms to use SSBOs instead of uniform buffers. Local testing has shown this has a huge impact on motionmark scores. Bug: b/309013541 Change-Id: I67df8494fb528bc1a340c260e6845764e424da6f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/778436 Reviewed-by: Michael Ludwig <michaelludwig@google.com> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
We've updated the bots' NDK, so we can start enforcing this rule. Bug: b/318414038 Change-Id: I71102045b05df9df561c201f8f4aae8a919b4a59 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796556 Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: Brian Osman <brianosman@google.com> Reviewed-by: John Stiles <johnstiles@google.com>
https://chromium.googlesource.com/vulkan-deps.git/+log/7b559e9f56e3..c8ff88bfeb04 Also rolling transitive DEPS: https://skia.googlesource.com/external/github.com/KhronosGroup/SPIRV-Tools/+log/0a9f3d1f2a8dd660bec005206a1b86136357d41c..c7affa1707b9c517ea028bf9070c97e6842a6749 If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/vulkan-deps-skia-autoroll Please CC egdaniel@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE Bug: None Tbr: egdaniel@google.com Change-Id: I6be9e4385815f4f975f6be71d20b9e4a9ebbae9f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796222 Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
https://skia.googlesource.com/buildbot.git/+log/58247e808c13..6de9699bbab6 2024-01-04 yasinsonui@google.com Upgrading Testhaus to React 18 2024-01-04 borenet@google.com [task scheduler] Disable tracing 2024-01-04 yasinsonui@google.com Allow list required packages for upgrading Testhaus to React 18 2024-01-04 jeffyoon@google.com [bisect] Create a client for skia-bridge 2024-01-04 borenet@google.com [task scheduler] Reduce tracing frequency 2024-01-04 borenet@google.com [task scheduler] Reduce tracing sample proportion 2024-01-04 borenet@google.com Revert "[task scheduler] Add logging for non-IN_PROGRESS jobs" 2024-01-04 borenet@google.com [task scheduler] Fix race where we might start a Job multiple times 2024-01-04 jcgregorio@google.com Change dfbuilder to not assume that all commits exist. 2024-01-04 skia-autoroll@skia-public.iam.gserviceaccount.com Roll Skia Infra CIPD packages from b9d08e6f5a57 to 58247e808c13 (8 revisions) 2024-01-03 ashwinpv@google.com Add checkbox support for paramset-sk If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/infra-skia Please CC lovisolo@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Skia Infra: https://bugs.chromium.org/p/skia/issues/entry To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Bug: None Tbr: lovisolo@google.com Change-Id: I9050066be6324fee51e37545230fae270267608e Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796756 Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Rename cargo package name to just fontations, instead of fontations_bridge, which is a more suitable name. Roll some dependencies and propagate a build define when the fontations backend is being built (to match what the GN build does.) Bug: b/301562314 Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian10-Clang-x86_64-Debug-Fontations,Build-Mac-Clang-x86_64-Debug-Fontations,Test-Debian10-Clang-GCE-CPU-AVX2-x86_64-Debug-All-Fontations,Test-Mac12-Clang-MacBookPro16.2-CPU-AppleIntel-x86_64-Debug-All-Fontations Change-Id: I666ce7b4976d81b19307335c55c2b3ba110441c9 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796197 Auto-Submit: Dominik Röttsches <drott@chromium.org> Commit-Queue: Dominik Röttsches <drott@google.com> Reviewed-by: Ben Wagner <bungeman@google.com>
COLRv1 data may contain cycles. If a cycle was detected then drawing was halted and an error returned. Instead, simply don't draw the cycle and return normally without error. Change-Id: I1119b3454f075f371093531a35752cb65c879c03 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796597 Reviewed-by: Dominik Röttsches <drott@google.com> Commit-Queue: Dominik Röttsches <drott@google.com>
https://skia.googlesource.com/buildbot.git/+log/6de9699bbab6..251696c8011d 2024-01-05 skia-autoroll@skia-public.iam.gserviceaccount.com Roll Skia Infra CIPD packages from 58247e808c13 to 6de9699bbab6 (11 revisions) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/sk-tool-skia Please CC lovisolo@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Tbr: lovisolo@google.com Change-Id: I8f0ec71b4697bb6d10f0777a5722433fcf7212ac Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796700 Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
The SPIR-V code generator would attempt to synthesize u_skRTFlip in every interface block that it found, under the (normally reasonable) assumption that we would only have one interface block. We now only attempt to synthesize u_skRTFlip one time. Previously, this error was just reported as "symbol 'u_skRTFlip' declared more than once" but due to http://review.skia.org/793576 we no longer have a ThreadContext set up while the SPIR-V code generator executes. The ThreadContext is used to report this error because `SymbolTable::add` has historically never taken a Context reference. In a followup I will remove the reliance on ThreadContext here. (This is, in fact, one of the last few places where ThreadContext is used at all.) Bug: oss-fuzz:65538 Change-Id: I5b81c00764da8cca0d4edad6da1aa85c953753e2 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796736 Auto-Submit: John Stiles <johnstiles@google.com> Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
Implement the `ColorPainter` trait in the FFI interface and pass draw calls generated by Skrifa to the SkCanvas we're operating on. Modify SkTypeface_fontations.cpp to not unconditionally generate bounds and image from path, but instead, assign track glyph format using ScalerContextBits and call the parent `GenerateImageFromPath()` function for path drawing, as the DirectWrite backend does it. Pass calls back up to the base SkScalerContext for handling monochromatic glyphs from path. Bug: b/301562314 Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian10-Clang-x86_64-Debug-Fontations,Build-Mac-Clang-x86_64-Debug-Fontations,Test-Debian10-Clang-GCE-CPU-AVX2-x86_64-Debug-All-Fontations,Test-Mac12-Clang-MacBookPro16.2-CPU-AppleIntel-x86_64-Debug-All-Fontations Change-Id: Ibea9e29f3f4ae5de8b876d5167a6290cf73fae71 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/774596 Reviewed-by: Ben Wagner <bungeman@google.com> Commit-Queue: Dominik Röttsches <drott@google.com>
The majority of changes come from two basic facts: 1) GCC does not allow implicit conversion of scalars to vectors. To deal with this, I added promotion helper functions. These are zero-cost on clang. For a few functions that are called *often* with scalars, I added overloads to limit the noise later in the file. 2) GCC does not allow implicit conversion between integral vector types and the __mXXXi types that appear in intrinsic signatures. (It does allow float vectors to implicitly convert to __mXXX.) Therefore, all of the integer intrinsic arguments and return values need explicit casts. (Now that we've turned off lax-vector-conversions in clang, most of these were already fixed, but a few more are caught by GCC). I've monitored code-gen before and after this change on both compilers. clang had some trivial code motion (reordering of instructions within a few stage functions), but no new overhead at all. GCC only had scalar RP previously, and how has a vectorized implementation that's comparable to clang's (many functions appear slightly worse, but not dramatically). Bug: b/316617002 Bug: b/318414038 Change-Id: I9c88acbf8ecf0adfcb1225c81b9e8ec3644620c2 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/790319 Commit-Queue: Brian Osman <brianosman@google.com> Reviewed-by: John Stiles <johnstiles@google.com>
Bug: b/318817249 Change-Id: Ic5b4ac946287663f9fd6d0017f9ba30d3dde57b8 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796737 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Michael Ludwig <michaelludwig@google.com>
The Parser's symbol-adding code now has two paths: - add/addWithoutOwnership takes a Context& - addOrDie/addWithoutOwnershipOrDie will abort on failure We use the former calls in most places, and the latter calls in cases where we expect failure is impossible (root symbol table creation, adding array dimensions). This should fix for the root-cause of the crash in oss-fuzz:65538. Change-Id: I8a4d3fd46ee443a43af73c303e7d8cd2fa2f33e9 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796557 Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: John Stiles <johnstiles@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
https://chromium.googlesource.com/vulkan-deps.git/+log/c8ff88bfeb04..dd0c87f2c1d5 Also rolling transitive DEPS: https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Cross/+log/f349c91274b91c1a7c173f2df70ec53080076191..cffc08a542b01dfd15f6ed7682251d002dbc4e64 https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Headers/+log/80207f9da86423ce33aff8328a792fd715f3c08f..217e93c664ec6704ec2d8c36fa116c1a4a1e2d40 If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/vulkan-deps-skia-autoroll Please CC egdaniel@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE Bug: None Tbr: egdaniel@google.com Change-Id: Id96645ac560853f8db11f19f7d397da4188d43f6 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796704 Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
This reverts commit c0a39cf. Reason for revert: GCC x86 build failure, Win-MSVC test failure Original change's description: > Add GCC support to Raster Pipeline > > The majority of changes come from two basic facts: > > 1) GCC does not allow implicit conversion of scalars to vectors. To deal > with this, I added promotion helper functions. These are zero-cost on > clang. For a few functions that are called *often* with scalars, I > added overloads to limit the noise later in the file. > > 2) GCC does not allow implicit conversion between integral vector types > and the __mXXXi types that appear in intrinsic signatures. (It does > allow float vectors to implicitly convert to __mXXX.) Therefore, all > of the integer intrinsic arguments and return values need explicit > casts. (Now that we've turned off lax-vector-conversions in clang, > most of these were already fixed, but a few more are caught by GCC). > > I've monitored code-gen before and after this change on both compilers. > clang had some trivial code motion (reordering of instructions within a > few stage functions), but no new overhead at all. GCC only had scalar RP > previously, and how has a vectorized implementation that's comparable to > clang's (many functions appear slightly worse, but not dramatically). > > Bug: b/316617002 > Bug: b/318414038 > Change-Id: I9c88acbf8ecf0adfcb1225c81b9e8ec3644620c2 > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/790319 > Commit-Queue: Brian Osman <brianosman@google.com> > Reviewed-by: John Stiles <johnstiles@google.com> Bug: b/316617002 Bug: b/318414038 Change-Id: I45cfd730da7e70fdeb50d672b1cd7da25823cb4a No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796673 Auto-Submit: Brian Osman <brianosman@google.com> Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Change-Id: I383b15c025979cfa840d8028bc185926394157c8 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796738 Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: Michael Ludwig <michaelludwig@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
The Parser is the only class which actually builds the list of ProgramElements (now that DSL has been removed). It can own the list of ProgramElements directly, rather than indirecting through ThreadContext. Change-Id: I38ff4bc8ee46f375a81bd0f96ac1806b4d70de21 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/795736 Reviewed-by: Jorge Betancourt <jmbetancourt@google.com> Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
At some point, this vector became dead code. Nothing is adding into it, so it's always empty. The program's shared-elements list is built up after the Program is created, by the FindAndDeclare transformation passes. Change-Id: I7bc99189ff9bf14dfc380dc3e06b3079262b3235 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796856 Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: John Stiles <johnstiles@google.com> Reviewed-by: Greg Daniel <egdaniel@google.com>
When removing SK_RESOLVE_FILTERS_BEFORE_RESTORE, there were two failures in Flutters tests. One was a rounding issue because a color filter now was operating on the original values instead of the 8-bit values. The other involved applying a "group opacity" optimization to a drawImage call that also had an image filter and expecting that the alpha-modulated output match the manually modulated original image. This boils down to adding an alpha value on the SkPaint. The Flutter test in question used a non-AA paint with nearest neighbor filtering and a dst rect aligned on 1/2px boundaries. With the legacy build flag enabled, both the opaque and the "group opacity" case would draw the non-AA image to a new layer (possibly with alpha), and then pass that to an erode/dilate filter. The non-AA rendering meant that there were no pixels with partial coverage. With the build flag removed, both cases skipped the auto layer within SkCanvas. The opaque case deferred the src->dst transform and nearest neighbor filtering into the erode/dilate evaluation. The use of nearest neighbor filtering produced an equivalent result to the original's non-AA snapping in the layer itself. However, for the "group opacity" case, the alpha is now applied as a color filter to the FilterResult before going to the erode/dilate filter. This is deferred but because erode/dilate pass in the kSampledRepeatedly flag, asShader() resolves the image. FilterResult uses AA by default because it gives more stable behavior when images are being processed by later filters, but in this case we now introduce partial coverage on the border pixels because of the half-pixel aligned dst rect in the deferred transform. While this actually leads to more stable behavior in the erode/dilate evaluation when dragging the GM around in viewer, it is definitely a different output and triggering the assert failures in Flutter. When the kSampledRepeatedly behavior was added, there were some regressions in performance on some platforms (higher end where math is probably cheaper than another image). The current check for "any" deferred color filter is pretty aggressive. Something like an alpha multiply is probably always worth deferring. It also happens to avoid the aliasing issue that tickles the Flutter test. Bug: b/40042615 Bug: b/40040587 Change-Id: I7f89bc47ac8c3c0ea4b53b8f4036f6a166e58fa7 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796876 Commit-Queue: Michael Ludwig <michaelludwig@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
As it turned out, this code only existed to be tested. We do clone Expressions extensively, but there really isn't a pressing need to clone a Statement or ProgramElement in its entirety. Change-Id: I3f2d618de60f74785a4ecd7d7a424d319e1b0016 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796857 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: John Stiles <johnstiles@google.com> Commit-Queue: Brian Osman <brianosman@google.com> Auto-Submit: John Stiles <johnstiles@google.com>
Much like http://review.skia.org/796856, this member was never modified, so it always remained in its default-constructed state and didn't do anything meaningful. The program's Interface is actually configured by the FindAndDeclare transform pass. Change-Id: I0917f19e2c7130d2ab0d0bb1630a4371e640234e Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796859 Commit-Queue: John Stiles <johnstiles@google.com> Auto-Submit: John Stiles <johnstiles@google.com> Reviewed-by: Greg Daniel <egdaniel@google.com>
Manual roll requested by kainino@google.com https://dawn.googlesource.com/dawn.git/+log/7330dd2dfe64..9fbef511d79e 2023-12-26 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll ANGLE from 3f9f34b48c80 to 3dfb2158b912 (2 revisions) 2023-12-26 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll vulkan-deps from a901b66f77bf to ddef466fc7d4 (1 revision) 2023-12-26 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll ANGLE from c8af75dde7f0 to 3f9f34b48c80 (1 revision) 2023-12-25 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll vulkan-deps from 75058a70503d to a901b66f77bf (1 revision) 2023-12-25 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll ANGLE from 3f5f4aa476b9 to c8af75dde7f0 (1 revision) 2023-12-24 dawn-autoroll@skia-public.iam.gserviceaccount.com Roll vulkan-deps from c3b062a5b7ab to 75058a70503d (5 revisions) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/dawn-skia-autoroll Please CC cwallez@google.com,kainino@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Dawn: https://bugs.chromium.org/p/dawn/issues/entry To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Debug-Graphite_Dawn_Vulkan;skia/skia.primary:Test-Mac11-Clang-MacMini9.1-GPU-AppleM1-arm64-Debug-All-Graphite_Dawn_Metal;skia/skia.primary:Test-Win10-MSVC-Golo-GPU-QuadroP400-x86_64-Release-All-Graphite_Dawn_D3D12;skia/skia.primary:Test-Win10-MSVC-Golo-GPU-QuadroP400-x86_64-Release-All-Graphite_Dawn_D3D11 Bug: None Tbr: kainino@google.com Change-Id: I5b6d28141fb6b572ea926b26d888524c398eb450 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796220 Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: Kai Ninomiya <kainino@chromium.org>
https://chromium.googlesource.com/vulkan-deps.git/+log/dd0c87f2c1d5..4441d7f010ce Also rolling transitive DEPS: https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Tools/+log/e21dc3deb042f6354320222903214f491b7b6ad0..2c0a644db855f40f100f9f39e5a8a8dfa2b0014d https://chromium.googlesource.com/external/github.com/KhronosGroup/Vulkan-Utility-Libraries/+log/fce11d52fee0344bb10a098b0a398dff42cb5d51..4cfc176e3242b4dbdfd3f6c5680c5d8f2cb7db45 If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/vulkan-deps-skia-autoroll Please CC egdaniel@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE Bug: None Tbr: egdaniel@google.com Change-Id: I7982416882e09590853aff6b7aa9377afc3db40d Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796639 Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/skottie-base-skia Please CC lovisolo@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Tbr: lovisolo@google.com Change-Id: I7b3c276a7086a49c76138fa4fe003141fd5570f1 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796641 Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/debugger-app-base-skia Please CC lovisolo@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Tbr: lovisolo@google.com Change-Id: Iabdce3e52e39fb0d0c9482518f4865d4e0bb5a1b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796710 Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/shaders-base-skia Please CC lovisolo@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Tbr: lovisolo@google.com Change-Id: I9496cb922421481f38dddd9b4adfa5aa04903f16 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796642 Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
…end" This reverts commit 9612d6c. Reason for revert: speculative revert for chrome and g3 failures Original change's description: > [Fontations-backend] Add COLRv0/v1 support to Fontations backend > > Implement the `ColorPainter` trait in the FFI interface and pass draw > calls generated by Skrifa to the SkCanvas we're operating on. > > Modify SkTypeface_fontations.cpp to not unconditionally generate bounds > and image from path, but instead, assign track glyph format using > ScalerContextBits and call the parent `GenerateImageFromPath()` function > for path drawing, as the DirectWrite backend does it. > > Pass calls back up to the base SkScalerContext for handling > monochromatic glyphs from path. > > Bug: b/301562314 > Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian10-Clang-x86_64-Debug-Fontations,Build-Mac-Clang-x86_64-Debug-Fontations,Test-Debian10-Clang-GCE-CPU-AVX2-x86_64-Debug-All-Fontations,Test-Mac12-Clang-MacBookPro16.2-CPU-AppleIntel-x86_64-Debug-All-Fontations > Change-Id: Ibea9e29f3f4ae5de8b876d5167a6290cf73fae71 > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/774596 > Reviewed-by: Ben Wagner <bungeman@google.com> > Commit-Queue: Dominik Röttsches <drott@google.com> Bug: b/301562314 Change-Id: I2d3ba6cbf98da3dfad92d557dfc50eb758e79d8d Cq-Include-Trybots: luci.skia.skia.primary:Build-Debian10-Clang-x86_64-Debug-Fontations,Build-Mac-Clang-x86_64-Debug-Fontations,Test-Debian10-Clang-GCE-CPU-AVX2-x86_64-Debug-All-Fontations,Test-Mac12-Clang-MacBookPro16.2-CPU-AppleIntel-x86_64-Debug-All-Fontations Reviewed-on: https://skia-review.googlesource.com/c/skia/+/797237 Commit-Queue: Greg Daniel <egdaniel@google.com> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
https://chromium.googlesource.com/vulkan-deps.git/+log/4441d7f010ce..784f7328cede If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/vulkan-deps-skia-autoroll Please CC egdaniel@google.com,skiabot@google.com on the revert to ensure that a human is aware of the problem. To file a bug in skia: https://bugs.chromium.org/p/skia/issues/entry To report a problem with the AutoRoller itself, please file a bug: https://issues.skia.org/issues/new?component=1389291&template=1850622 Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE Bug: None Tbr: egdaniel@google.com Change-Id: Ib2221ca92d3d1e29b90c0cc77a4e389bf1db2814 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/796714 Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com> Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
https://skia-review.googlesource.com/c/skia/+/761237 introduced an explicit FontMgrRunIterator, which can trigger font fallback (previously impossible with SkShaper_primitive). This causes issues for clients which don't have a fallback-capable fontmgr. Allow clients to opt into using a TrivialFontRunIterator. Bug: chromium:1520148 Change-Id: Ie8530d01f0c61daecb3c888c51e2c8a4c414738d Reviewed-on: https://skia-review.googlesource.com/c/skia/+/805536 Reviewed-by: Ben Wagner <bungeman@google.com> Reviewed-by: Florin Malita <fmalita@google.com> Commit-Queue: Florin Malita <fmalita@google.com> (cherry picked from commit 29b545e) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/806936 Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Auto-Submit: Florin Malita <fmalita@google.com> Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Ben Wagner <bungeman@google.com>
Adobe Reader takes issue with the color profiles produced from SkColorSpace when applied to single channel images. While this is investigated, disable embedding such color profiles on single channel images. Bug: chromium:1523360 Change-Id: I7039e3ebf0644679d9b024a84b6f4e72b009987b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/807597 Commit-Queue: Ben Wagner <bungeman@google.com> Reviewed-by: Jim Van Verth <jvanverth@google.com> (cherry picked from commit 76da7ec) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/807716 Reviewed-by: Brian Osman <brianosman@google.com>
Initially motivated by the drawing of Type3 glyphs into the PDF document, which normally happens after the last page is ended and the document is being finished. In order to properly target bookmarks it is necessary to keep track of where marks (drawings) are made, which is a position on a page. However, there may be marks (like those of a Type3 font) which are not bound with or drawn to any particular page. When these are emitted there is no current page and attempting to access the current page attributes currently results in a crash. This can also happen if a user draws an attribute or tags a structure element node while there is no currently active page. The SkPDFDocument already abstracts the currently active page and any marks made to it. Update this abstraction to handle the cases where there is no currently active page. Also add a portable test by making the test Jupiter glyph somewhat transparent to force a layer. Bug: chromium:1523260 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/807817 Reverted-on: https://skia-review.googlesource.com/c/skia/+/808498 Change-Id: I4bb74ea3a023ea57f82ec6e1115ae33e1085a9f0 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/808660 Commit-Queue: Ben Wagner <bungeman@google.com> Reviewed-by: Brian Osman <brianosman@google.com> (cherry picked from commit 909cf06) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/809603
Some printers have been found to have issues with ICC profiles that have an empty profile description string. Ensure that generated ICC profiles have non-empty descriptions. Bug: chromium:329032158 Change-Id: I024e70313b9919a90baca2a04ddd0a189da59e7c Reviewed-on: https://skia-review.googlesource.com/c/skia/+/825723 Commit-Queue: Ben Wagner <bungeman@google.com> Reviewed-by: Brian Osman <brianosman@google.com> (cherry picked from commit 18f41d3) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/826439
Change-Id: I70ba2400ca607f70fe7d3daadef40d131dbc9b32 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/827560 Reviewed-by: Heather Miller <hcm@google.com>
Change-Id: I2e01c4bd906b715abe577da2e74467f712660ea7 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/841202 Reviewed-by: Heather Miller <hcm@google.com>
Change-Id: Ideb02cf83177568ea614d7c5559adf8c215ba1c4 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/852349 Reviewed-by: Heather Miller <hcm@google.com>
M120 merge issues: Conflicting includes Bug: b/360758697 No-Try: true No-Presubmit: true No-Tree-Checks: true Change-Id: I46eb92ac6ed71646fb05a910f8d577ec851e3b3f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/891636 Commit-Queue: James Godfrey-Kittle <jamesgk@google.com> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/894936 Reviewed-by: James Godfrey-Kittle <jamesgk@google.com>
The callers of PatternHelper which are not updated here pass in a TArray size as repeatCount, which already prevents overflow: https://crsrc.org/c/third_party/skia/include/private/base/SkTArray.h?q=kMaxCapacity Bug: b/361461526 Change-Id: I86c494cb00223f0bb8d68540d33d7230b60c9486 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/893916 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: James Godfrey-Kittle <jamesgk@google.com> (cherry picked from commit 07fcb9a) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/898036 Reviewed-by: James Godfrey-Kittle <jamesgk@google.com>
Check total index count in onCombineIfPossible. Bug: b/360265320 Change-Id: I02f04593b60dcd2470580110d0a555ed4bf47280 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/890322 Commit-Queue: Michael Ludwig <michaelludwig@google.com> (cherry picked from commit fdc8c2d) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/894465 Reviewed-by: Michael Ludwig <michaelludwig@google.com>
…aces Bug: 355465305 Change-Id: Ifb87db5e8d0d0c29449e6a3e82254189e3f2d33b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/886696 Commit-Queue: Brian Osman <brianosman@google.com> (cherry picked from commit 05097fb) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/892616 Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Bug: 355465305 Change-Id: Ife25289f7b3489701c67b7dc5d30e473019a1193 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/888376 Commit-Queue: Brian Osman <brianosman@google.com> (cherry picked from commit d1b243b) Reviewed-on: https://skia-review.googlesource.com/c/skia/+/892676 Reviewed-by: Michael Ludwig <michaelludwig@google.com>
…cation size" This reverts commit a59291e. Reason for revert: unit test uses code not in m120. Original change's description: > [M120-LTS][SkSL:RP] Prevent overflow when computing slot allocation size > > Bug: 355465305 > Change-Id: Ife25289f7b3489701c67b7dc5d30e473019a1193 > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/888376 > Commit-Queue: Brian Osman <brianosman@google.com> > (cherry picked from commit d1b243b) > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/892676 > Reviewed-by: Michael Ludwig <michaelludwig@google.com> Bug: 355465305 Change-Id: I27fe9fa6d769c84955bbbc2ca01c10305d4349b2 No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://skia-review.googlesource.com/c/skia/+/899717 Auto-Submit: Michael Ludwig <michaelludwig@google.com> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
…cation size" This reverts commit 7ba220b. Reason for revert: removing unit test that fails to compile in m120 since all that's needed is the actual SkSL::RP updates for the LTS chrome release. Original change's description: > Revert "[M120-LTS][SkSL:RP] Prevent overflow when computing slot allocation size" > > This reverts commit a59291e. > > Reason for revert: unit test uses code not in m120. > > Original change's description: > > [M120-LTS][SkSL:RP] Prevent overflow when computing slot allocation size > > > > Bug: 355465305 > > Change-Id: Ife25289f7b3489701c67b7dc5d30e473019a1193 > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/888376 > > Commit-Queue: Brian Osman <brianosman@google.com> > > (cherry picked from commit d1b243b) > > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/892676 > > Reviewed-by: Michael Ludwig <michaelludwig@google.com> > > Bug: 355465305 > Change-Id: I27fe9fa6d769c84955bbbc2ca01c10305d4349b2 > No-Presubmit: true > No-Tree-Checks: true > No-Try: true > Reviewed-on: https://skia-review.googlesource.com/c/skia/+/899717 > Auto-Submit: Michael Ludwig <michaelludwig@google.com> > Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Bug: 355465305 Change-Id: Iea2884487313206150f98becbbf22a5a286ff512 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/901140 Reviewed-by: Robert Phillips <robertphillips@google.com>
Merge google/skia chrome/m120 branch. Conflicts resolved: - DEPS: took upstream versions - SkTypeface.cpp: kept custom RefDefault(), removed UniqueID(), added MakeEmpty() - DrawMeshOp.cpp: took upstream overflow fix - RELEASE_NOTES.md, infra/bots/jobs.json: took upstream
Merge google/skia chrome/m121 branch. Conflicts resolved: - GrDirectContext.cpp: took upstream (deprecated Metal void* overloads) - DrawMeshOp.cpp: took upstream - SkFontMgr_fontconfig.cpp: took upstream - RELEASE_NOTES.md, infra/bots/jobs.json, tools/git-sync-deps: took upstream
Merge google/skia chrome/m122 branch. All conflicts resolved by taking upstream: - SkTypeface.h/cpp: removed legacy MakeDefault, RefDefault, MakeFromName/File/Stream/Data, UniqueID - SkFontMgr.h: removed RefDefault, MakeDefault, Factory - Various gn/build/infrastructure files: took upstream Custom font manager and typeface defaults will be implemented in C API layer.
- sk_typeface: stub out removed factory methods (moved to fontmgr) - sk_fontmgr: implement create/ref_default using platform-specific factories - sk_font: translate empty typeface ↔ null for backwards compatibility - gr_context: GrDirectContext::MakeVulkan → GrDirectContexts::MakeVulkan - gr_context/sk_image: GrMipMapped → skgpu::Mipmapped - Add required includes for GpuTypes.h and GrVkDirectContext.h Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- sk_typeface: stub out removed factory methods (moved to fontmgr) - sk_fontmgr: implement create/ref_default using platform-specific factories - sk_font: translate empty typeface ↔ null for backwards compatibility - gr_context: GrDirectContext::MakeVulkan → GrDirectContexts::MakeVulkan - gr_context/sk_image: GrMipMapped → skgpu::Mipmapped - skresources: bool predecode → ImageDecodeStrategy enum - Add required includes for GpuTypes.h and GrVkDirectContext.h Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- GrVkYcbcrConversionInfo: use DEF_MAP instead of DEF_STRUCT_MAP (now a type alias) - GrDirectContext::MakeGL → GrDirectContexts::MakeGL - Add include for GrGLDirectContext.h Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…aint In Skia m122+, SkFont(nullptr) creates an empty typeface that draws nothing. SkiaSharp preserves the legacy behavior where null means 'platform default' by routing through the shared font manager singleton. Affected functions: - sk_font_new() - now creates font with default typeface - sk_font_new_with_values(nullptr, ...) - uses default typeface - sk_font_set_typeface(font, nullptr) - sets default typeface - SkCompatPaint() - initializes internal font with default typeface Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Author
|
Companion PR: mono/SkiaSharp#3559 |
These 6 functions were stubbed to return nullptr after the m122 upstream removal. Since the C API has no ABI compatibility requirement, remove them entirely rather than keeping useless stubs. Removed: - sk_typeface_create_default - sk_typeface_ref_default - sk_typeface_create_from_name - sk_typeface_create_from_file - sk_typeface_create_from_stream - sk_typeface_create_from_data All callers have been migrated to sk_fontmgr_* equivalents. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Expose the upstream skresources::ImageDecodeStrategy enum as skresources_image_decode_strategy_t with values LazyDecode and PreDecode. New functions: - skresources_file_resource_provider_make2(base_dir, strategy) - skresources_data_uri_resource_provider_proxy_make2(rp, strategy) Existing bool overloads kept for compatibility, delegating to the new enum-based functions internally. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
SkFontMgr_New_Custom_Empty() requires skia_enable_fontmgr_custom_empty which is disabled on WASM builds. Use SkFontMgr::RefEmpty() instead, which is always available from core Skia and doesn't require any port library to be linked. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
On WASM (and platforms with SK_FONTMGR_FREETYPE_EMBEDDED_AVAILABLE), use SkFontMgr_New_Custom_Data instead of SkFontMgr::RefEmpty() as the default platform font manager. The empty font manager returns nullptr from onMakeFromData, making it impossible to create typefaces from font data at runtime. The custom data font manager uses FreeType and fully supports creating typefaces from data/streams. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Switch from SkFontMgr_New_Custom_Data (empty, no fonts) to SkFontMgr_New_Custom_Embedded(&SK_EMBEDDED_FONTS) for platforms with SK_FONTMGR_FREETYPE_EMBEDDED_AVAILABLE. The WASM build compiles NotoMono-Regular.ttf into the binary via embed_resources.py as the SK_EMBEDDED_FONTS symbol. Using the embedded font manager ensures text rendering works out of the box without requiring explicit font registration — matching the previous behavior where WASM had a monospace fallback font. The embedded font manager also supports creating additional typefaces from data/streams via FreeType at runtime. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
PR that fix LoongArch64 LSX error: #169 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Update Skia to Chrome Milestone 122
Sequential merge of upstream chrome/m120 → m121 → m122 into the skiasharp branch.
Breaking changes addressed in C API shim:
SkTypeface::MakeDefault/RefDefault/MakeFromName/File/Stream/Data— C API stubs return nullptr; behavior moved to font managerSkFontMgr::RefDefault/MakeDefault/Factory— implemented platform-specific singleton (SkFontMgr_New_CoreText,SkFontMgr_New_DirectWrite, etc.)SkFont(nullptr)now creates empty typeface — C API intercepts null to provide platform default via shared font manager singletonGrDirectContext::MakeGL/MakeVulkan→GrDirectContexts::MakeGL/MakeVulkanGrMipmapped→skgpu::MipmappedGrVkYcbcrConversionInfofrom struct to type alias —DEF_STRUCT_MAP→DEF_MAPwithSK_VULKANguardskresourcesbool →ImageDecodeStrategyenumCustom dependency pins restored:
Upstream m122 DEPS reverted our security bumps. Restored: libwebp 1.6.0, brotli 1.2.0, zlib, libpng, freetype, harfbuzz, libjpeg-turbo, expat.
Companion PR:
mono/SkiaSharp#3559