Jon Bauman
619acc91bb
Bug 1729539 - Hit MOZ_CRASH(assertion failed: y2 > 1. / 12. && y2 <= 1.) at gfx/qcms/src/iccread.rs:1392. r=jrmuizel
...
The assertion is due to an inappropriate test of exact floating-point values.
build_trc_table() handles this saturating case, so there's no need to assert.
Add more test coverage to be certain no fuzzing inputs will lead to crashes.
Differential Revision: https://phabricator.services.mozilla.com/D125006
2021-09-17 17:41:26 +00:00
Butkovits Atila
f2a5d438bb
Backed out changeset d629f462a605 (bug 1729539) for causing Wd failures. CLOSED TREE
2021-09-16 02:50:35 +03:00
Jon Bauman
6d6715ba10
Bug 1729539 - Hit MOZ_CRASH(assertion failed: y2 > 1. / 12. && y2 <= 1.) at gfx/qcms/src/iccread.rs:1392. r=jrmuizel
...
The assertion is due to an inappropriate test of exact floating-point values.
build_trc_table() handles this saturating case, so there's no need to assert.
Add more test coverage to be certain no fuzzing inputs will lead to crashes.
Differential Revision: https://phabricator.services.mozilla.com/D125006
2021-09-15 19:57:15 +00:00
Marian-Vasile Laza
699488300d
Backed out changeset bd23fb0c95cc (bug 1729539) for causing web platform failures.
2021-09-15 08:41:00 +03:00
Jon Bauman
f5e56d8781
Bug 1729539 - Hit MOZ_CRASH(assertion failed: y2 > 1. / 12. && y2 <= 1.) at gfx/qcms/src/iccread.rs:1392. r=jrmuizel
...
The assertion is due to an inappropriate test of exact floating-point values.
build_trc_table() handles this saturating case, so there's no need to assert.
Add more test coverage to be certain no fuzzing inputs will lead to crashes.
Differential Revision: https://phabricator.services.mozilla.com/D125006
2021-09-14 23:14:41 +00:00
Cristian Tuns
3a65ad3358
Backed out changeset db3338456ae1 (bug 1729539) for causing multiple failures. CLOSED TREE
2021-09-14 15:02:43 -04:00
Jon Bauman
cf0312ede2
Bug 1729539 - Hit MOZ_CRASH(assertion failed: y2 > 1. / 12. && y2 <= 1.) at gfx/qcms/src/iccread.rs:1392. r=jrmuizel
...
The assertion is due to an inappropriate test of exact floating-point values.
build_trc_table() handles this saturating case, so there's no need to assert.
Add more test coverage to be certain no fuzzing inputs will lead to crashes.
Differential Revision: https://phabricator.services.mozilla.com/D125006
2021-09-14 16:32:37 +00:00
Jon Bauman
6155249a72
Bug 1696045 - Add AVIF telemetry for unimplemented features and performance. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D123378
2021-08-24 22:34:32 +00:00
Sandor Molnar
bb9a848857
Backed out changeset d0f659f7ea9b (bug 1696045) for causing gtest failures in image/decoders/nsAVIFDecoder. CLOSED TREE
2021-08-24 21:27:20 +03:00
Jon Bauman
fc5f21743a
Bug 1696045 - Add AVIF telemetry for unimplemented features and performance. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D123378
2021-08-24 17:20:02 +00:00
Jeff Muizelaar
1916934834
Bug 1725742. Validate tag offsets upfront to avoid weird overflow situations later on. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D122856
2021-08-19 16:33:03 +00:00
Jon Bauman
bacfba2091
Bug 1725056 - qcms: avoid undefined behavior when handling CICP values. r=jrmuizel,tsmith
...
Differential Revision: https://phabricator.services.mozilla.com/D122318
2021-08-16 23:13:36 +00:00
Jon Bauman
1fbcf87b11
Bug 1634741 - AVIF (AV1 Image File Format): proper color space support. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D121764
2021-08-05 15:47:59 +00:00
Jon Bauman
ad9d5759ea
Bug 1723253 - qcms: Add CICP (Coding-independent code points) support. r=jrmuizel
...
Add CICP-specific functionality
Differential Revision: https://phabricator.services.mozilla.com/D121452
2021-08-05 05:10:37 +00:00
Jon Bauman
bbd6119d94
Bug 1723253 - qcms: Clippy/costmetic changes. r=jrmuizel
...
Clippy/costmetic changes
Differential Revision: https://phabricator.services.mozilla.com/D121458
2021-08-05 05:10:37 +00:00
Jeff Muizelaar
ee0926e1db
Bug 1721596 - qcms: Rerun rustfmt. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D120466
2021-07-21 13:03:23 +00:00
Jeff Muizelaar
132a09618b
Bug 1721515 - qcms: Remove some now unneeded unstable features. r=aosmond
...
These were used by our local NEON intrinsics implementation.
We don't need them anymore.
Differential Revision: https://phabricator.services.mozilla.com/D120420
2021-07-21 01:42:38 +00:00
Jeff Muizelaar
6edd219c43
Bug 1721484
- qcms: Fix extra space. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D120395
2021-07-20 21:36:45 +00:00
Jeff Muizelaar
ad5985aca6
Bug 1721483 - qcms: Remove unneeded vim modelines. r=aosmond
...
These were inherited from the C code and are wrong and unneeded in
Rust.
Differential Revision: https://phabricator.services.mozilla.com/D120394
2021-07-20 21:36:18 +00:00
Jeff Muizelaar
2ed71226c3
Bug 1697818 - Drop handrolled NEON SIMD wrappers. r=aosmond
...
We require Rust 1.50 now which is new enough to not need these
wrappers anymore.
Differential Revision: https://phabricator.services.mozilla.com/D108337
2021-07-20 21:29:01 +00:00
Jeff Muizelaar
62908eaa10
Bug 1713012. Only set RUSTC_BOOTSTRAP if it's not already set. r=aosmond
...
In bug 1670538 we started setting RUSTC_BOOTSTRAP for Rust 1.50+
Rust 1.50+ gives a warning if we set in build.rs so don't.
Differential Revision: https://phabricator.services.mozilla.com/D116131
2021-05-28 20:18:23 +00:00
Jeff Muizelaar
54a52e060a
Bug 1710897 - Handle matrix inversion failing more gracefully. r=aosmond
...
Prior to making invert() return an Option we just gleefully
assumed that it succeeded here. Instead of panicing let's
fail to create the transform.
The problem was found by oss-fuzz:
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=34216
Differential Revision: https://phabricator.services.mozilla.com/D114999
2021-05-12 20:52:53 +00:00
Jeff Muizelaar
c92ee2a045
Bug 1710542 - Remove 'invalid' field from Matrix. r=aosmond
...
This changes to using Option<Matrix> instead of
having an 'invalid' field. This makes sure we check
when we need to and don't have to check otherwise.
Differential Revision: https://phabricator.services.mozilla.com/D114808
2021-05-11 13:58:20 +00:00
Jeff Muizelaar
9b9ddc9cc2
Bug 1710450 - Pass a reference to read_tag_s15Fixed16ArrayType r=aosmond
...
This lets us avoid having to look up the tag again.
Differential Revision: https://phabricator.services.mozilla.com/D114751
2021-05-11 13:57:43 +00:00
Jeff Muizelaar
7c5d88731c
Bug 1708805 - Reject output luts with channels != 3. r=aosmond
...
This problem was introduced by the CMYK support
patch not rejecting 4 channel LUTs at parse time.
The problem was caught by oss-fuzz: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=33647
Differential Revision: https://phabricator.services.mozilla.com/D114014
2021-05-01 02:39:06 +00:00
Jeff Muizelaar
db8307e53c
Bug 1707794 - Add support for CMYK to the fuzzing target. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D113460
2021-04-27 14:38:51 +00:00
Jeff Muizelaar
cdd4899132
Bug 1555331. Add a test that parametric functions are properly inverted. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D113227
2021-04-26 12:48:33 +00:00
Jeff Muizelaar
0fe884da2d
Bug 1555331. Use parametric inversion when building an output_lut. r=aosmond
...
Previously we didn't invert at all which was causing over saturation.
Differential Revision: https://phabricator.services.mozilla.com/D113226
2021-04-26 12:48:33 +00:00
Jeff Muizelaar
bf77763c5d
Bug 1555331. Implement analytic inversion for parametric transfer functions. r=aosmond
...
This also adjusts the initialization of Param to make inversion better behaved.
Differential Revision: https://phabricator.services.mozilla.com/D113225
2021-04-26 12:48:32 +00:00
Jeff Muizelaar
e98ff078aa
Bug 1555331. Add a Param struct. r=aosmond
...
This makes our parsing of parametric transfer functions
more clear because you can see exactly what output variables
the input variables map to.
Differential Revision: https://phabricator.services.mozilla.com/D113224
2021-04-26 12:48:32 +00:00
Jeff Muizelaar
404d7d8cfe
Bug 1706144 - Add support for CMYK. r=aosmond
...
This is not enabled in Gecko yet.
Differential Revision: https://phabricator.services.mozilla.com/D112643
2021-04-23 02:42:53 +00:00
Butkovits Atila
0a6b07942c
Backed out changeset d87602a10873 (bug 1706144) for causing build bustages. CLOSED TREE
2021-04-23 04:59:43 +03:00
Jeff Muizelaar
33c441281c
Bug 1706144 - Add support for CMYK. r=aosmond
...
This is not enabled in Gecko yet.
Differential Revision: https://phabricator.services.mozilla.com/D112643
2021-04-23 01:36:00 +00:00
Jeff Muizelaar
84dc71e689
Bug 1705950 - Add safe API for transforming with different formats. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D112489
2021-04-19 18:15:05 +00:00
Jeff Muizelaar
d0d3ba387d
Bug 1705847 - Use the input_gamma_table_gray as an array. r=aosmond
...
Cleaner, safer and just as fast.
Differential Revision: https://phabricator.services.mozilla.com/D112446
2021-04-19 18:14:47 +00:00
Jeff Muizelaar
1a31db3a4d
Bug 1704909 - Use a Box<[f32; 256]> instead of a Vec. r=aosmond
...
This gives a known length at compile time at the cost of slightly
worse ergonomics. Knowing the length should let the compiler drop
some bounds checks and will make it easier to make some of the
transform functions safe.
Differential Revision: https://phabricator.services.mozilla.com/D111917
2021-04-16 22:34:13 +00:00
Jan-Erik Rediger
598f4a226a
Bug 1705693 - Pass verifier callback as an option. r=aosmond
...
`ProduceVerifyOutput` takes an `transform_fn_t`, which is a type
definition for `Option<unsafe fn(...)>`.
So we need to pass `Some`. This code can never have worked correctly.
Differential Revision: https://phabricator.services.mozilla.com/D112341
2021-04-16 12:40:33 +00:00
Jeff Muizelaar
5f53667340
Bug 1704407 - Use a CLU helper like the original C++ code. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D111572
2021-04-14 01:09:37 +00:00
Jeff Muizelaar
ad361a869e
Bug 1703101 - Bump the version of libfuzzer-sys.
...
This fixes a build error on oss-fuzz
Differential Revision: https://phabricator.services.mozilla.com/D110862
2021-04-05 20:59:58 +00:00
Butkovits Atila
45016e9a4a
Backed out changeset 1bba2e7410e1 (bug 1697818) as requested by Jeff. CLOSED TREE
2021-03-31 17:55:28 +03:00
Jeff Muizelaar
7bd7df3eeb
Bug 1702080 - Make RUSTC_BOOTSTRAP usage conditional on 'neon' feature. r=aosmond
...
This makes it so our subversion to get NEON doesn't happen by default.
Differential Revision: https://phabricator.services.mozilla.com/D110348
2021-03-31 14:31:23 +00:00
Jeff Muizelaar
266a347d35
Bug 1702073 - Enable ICCv4 by default in the crate but disable it in Gecko. r=aosmond
...
This gives other users of the crate ICCv4 by default while
still giving Gecko the ability to enable it dynamically.
Differential Revision: https://phabricator.services.mozilla.com/D110344
2021-03-31 14:29:35 +00:00
Jeff Muizelaar
23af50ef79
Bug 1701758. Avoid identity sRGB transforms. r=aosmond
...
This avoids the performance cost and ensures the data remains unchanged.
Differential Revision: https://phabricator.services.mozilla.com/D110289
2021-03-31 13:15:26 +00:00
Jeff Muizelaar
ce00bef848
Bug 1701457 - Rewrite compute_curve_gamma_table_type_parametric. r=aosmond
...
This rewites the function to use slice patterns and make the computation
more closely follow the ICC spec. It also fixes a bug where we forgot to
divide X by 255 before applying the threshold.
Differential Revision: https://phabricator.services.mozilla.com/D110037
2021-03-29 16:41:09 +00:00
Jeff Muizelaar
4a6a749d05
Bug 1701125 - Remove the Option<> from build_lut_matrix. r=aosmond
...
We always pass Some(lut) to build_lut_matrix so we
don't need take an Option.
Differential Revision: https://phabricator.services.mozilla.com/D109839
2021-03-26 14:51:44 +00:00
Jeff Muizelaar
720be19554
Bug 1698363
- Use trait objects instead of home grown polymorphism. r=aosmond
...
This lets each ModularTransform contain only the data that needs
instead of all of the possible data and will allow for removing
some of the Option wrappers.
Differential Revision: https://phabricator.services.mozilla.com/D108357
2021-03-15 18:44:17 +00:00
Jeff Muizelaar
83e8eeb430
Bug 1698363
- Use a Vec instead of a list to store ModularTransforms. r=aosmond
...
This makes things a little simpler at the cost of the Vec<> allocation.
However, it lets us switch to using trait objects because we don't
need a common next_transform field.
Differential Revision: https://phabricator.services.mozilla.com/D108356
2021-03-15 18:44:17 +00:00
Jeff Muizelaar
4408cb9135
Bug 1698325 - Remove dead field. r=aosmond
...
Differential Revision: https://phabricator.services.mozilla.com/D108341
2021-03-15 17:58:12 +00:00
Jeff Muizelaar
c2592d4467
Bug 1698324 - Store an array of curves instead of explicit rgb ones. r=aosmond
...
This will make supporting CMYK easier.
Differential Revision: https://phabricator.services.mozilla.com/D108340
2021-03-15 17:05:26 +00:00
Jeff Muizelaar
812f96aeba
Bug 1697818 - Drop handrolled NEON SIMD wrappers. r=aosmond
...
We require Rust 1.50 now which is new enough to not need these
wrappers anymore.
Differential Revision: https://phabricator.services.mozilla.com/D108337
2021-03-14 19:44:57 +00:00