NOTE: Linear blending is currently disabled in this CL as the
feature is still a work in progress
Android currently performs all blending (any kind of linear math
on colors really) on gamma-encoded colors. Since Android assumes
that the default color space is sRGB, all bitmaps and colors
are encoded with the sRGB Opto-Electronic Conversion Function
(OECF, which can be approximated with a power function). Since
the power curve is not linear, our linear math is incorrect.
The result is that we generate colors that tend to be too dark;
this affects blending but also anti-aliasing, gradients, blurs,
etc.
The solution is to convert gamma-encoded colors back to linear
space before doing any math on them, using the sRGB Electo-Optical
Conversion Function (EOCF). This is achieved in different
ways in different parts of the pipeline:
- Using hardware conversions when sampling from OpenGL textures
or writing into OpenGL frame buffers
- Using software conversion functions, to translate app-supplied
colors to and from sRGB
- Using Skia's color spaces
Any type of processing on colors must roughly ollow these steps:
[sRGB input]->EOCF->[linear data]->[processing]->OECF->[sRGB output]
For the sRGB color space, the conversion functions are defined as
follows:
OECF(linear) :=
linear <= 0.0031308 ? linear * 12.92 : (pow(linear, 1/2.4) * 1.055) - 0.055
EOCF(srgb) :=
srgb <= 0.04045 ? srgb / 12.92 : pow((srgb + 0.055) / 1.055, 2.4)
The EOCF is simply the reciprocal of the OECF.
While it is highly recommended to use the exact sRGB conversion
functions everywhere possible, it is sometimes useful or beneficial
to rely on approximations:
- pow(x,2.2) and pow(x,1/2.2)
- x^2 and sqrt(x)
The latter is particularly useful in fragment shaders (for instance
to apply dithering in sRGB space), especially if the sqrt() can be
replaced with an inversesqrt().
Here is a fairly exhaustive list of modifications implemented
in this CL:
- Set TARGET_ENABLE_LINEAR_BLENDING := false in BoardConfig.mk
to disable linear blending. This is only for GLES 2.0 GPUs
with no hardware sRGB support. This flag is currently assumed
to be false (see note above)
- sRGB writes are disabled when entering a functor (WebView).
This will need to be fixed at some point
- Skia bitmaps are created with the sRGB color space
- Bitmaps using a 565 config are expanded to 888
- Linear blending is disabled when entering a functor
- External textures are not properly sampled (see below)
- Gradients are interpolated in linear space
- Texture-based dithering was replaced with analytical dithering
- Dithering is done in the quantization color space, which is
why we must do EOCF(OECF(color)+dither)
- Text is now gamma corrected differently depending on the luminance
of the source pixel. The asumption is that a bright pixel will be
blended on a dark background and the other way around. The source
alpha is gamma corrected to thicken dark on bright and thin
bright on dark to match the intended design of fonts. This also
matches the behavior of popular design/drawing applications
- Removed the asset atlas. It did not contain anything useful and
could not be sampled in sRGB without a yet-to-be-defined GL
extension
- The last column of color matrices is converted to linear space
because its value are added to linear colors
Missing features:
- Resource qualifier?
- Regeneration of goldeng images for automated tests
- Handle alpha8/grey8 properly
- Disable sRGB write for layers with external textures
Test: Manual testing while work in progress
Bug: 29940137
Change-Id: I6a07b15ab49b554377cd33a36b6d9971a15e9a0b
SystemProperties.get() is not particularly fast,
especially if a string is returned. Since ro.* values
are unable to be changed, there's no need to
continously re-query them. Cache the value at
static init time to trivially fix this.
Test: refactoring CL.
Change-Id: Iccb021d3cb2ba3a4a1d0048ddec6811bb7409eec
Let RuntimeInit use an UncaughtExceptionPreHandler to log an exception
rather than relying on UncaughtHandler, which apps can replace. This
makes it easier to diagnose application death, especially during app
compatibility testing for a new version of Android.
Test: Verified manually, with the help of a small sample app (not
checked in), that stacktraces for RuntimeExceptions thrown on main
or background threads are logged even when the app set a default
UncaughtExceptionHandler that swallows the exception with no action.
Note that such an inappropriate UncaughtExceptionHandler will still
cause threads to die without the app being killed, which it should be.
In an exception then happens on the main thread, the app will freeze
until the ANR dialog kicks in after a few seconds. I have manually
verified that this behavior is unchanged from before this CL.
No new integration tests are included because the default system
behavior has not changed.
Bug: 29624607
Change-Id: Ie87377b0bcadc3ba4083a8ab1bedb8f3dd95a4bd
By default, AVD animates on RT thread. But since in some cases there's a
need for a finer control on when the frame update for the animation should
happen, such as coordiating the AVD animation with other animations that
run on UI thread, we are providing a way to force the AVD to run on UI
thread.
Bug: 27278616
Change-Id: I372ecd3dc52e3fa0bdce3a1e9c19443f9b199027
The provider is installed early in ActivityThread to ensure that no
TLS objects are created in the application before the provider is
installed.
Change-Id: I5f77addfa75a4ee19301de54e01507d8dca33657
This moves the non-public API classes backing Android Keystore from
android.security to android.security.keystore, a package specially
created for Android Keystore.
Bug: 18088752
Change-Id: Ibf04d6a26c54d310b0501fc5e34f37b1176324ad
Bouncy Castle JCA provider incorrectly declares that its Cipher, Mac,
Signature, and KeyAgreement implementations accept arbitrary keys (
including AndroidKeyStore keys). As a result, when a Cipher, Mac,
Signature, or KeyAgreement instance is requested from JCA without
explicitly specifying the provider (which follows best practices)
and then initialied with an AndroidKeyStore key, JCA chooses the
BouncyCastle's implementation, which in turn blows up because it
can't handle such keys.
The workaround is to install Cipher, Mac, Signature, and
KeyAgreement implementations backed by AndroidKeyStore as a
higher-priority JCA provider than the Bouncy Castle one. This is
achieved by splitting out the above implementations from
AndroidKeyStoreProvider into AndroidKeyStoreBCWorkaroundProvider
and installing the AndroidKeyStoreProvider at the usual priority
(below Bouncy Castle) and the AndroidKeyStoreBCWorkaroundProvider
at above Bouncy Castle priority.
Bug: 20691708
Change-Id: I336464f4a49bc30c6845ddc4e84b07f4105424dd
Previous ActionBar animations didn't handle configuration changes
or other situations in which the view would get detached. listeners
would stay on the view and would attempt to do something nonsensical
in the new window. This new approach removes listeners on window
detach to avoid this problem.
Issue #20125407 Settings Crashes when changing orientation of device
Change-Id: I0b3bbd0f6fc23cdb4cbd646b0d2772d72d3d795d
This avoids loading all AndroidKeyStore crypto SPI classes when the
AndroidKeyStore provider is instantiated and installed. This provider
is installed early on during the initialization of each app. Most apps
don't need these classes loaded.
Bug: 18088752
Change-Id: Ib43c9dd6a7d434b128916e0f9c8652ba61df0d47
We don't want to preload the WebView library in the zygote process any
more, as loading an updatable WebView in the zygote is a stability risk.
The memory benefits of preloading will be obtained in other ways.
Bug: 13005501
Change-Id: Ic89f2a1d057dc92b01fb775bf326b47ac2d4caa2
Some cipher modes were removed from OpenSSLCipher. This change removes
those classes from the preloaded classes list.
Change-Id: Ib4450c392513973836684b4d5df7fffb200b6260
For image creating, art needs this in the preloaded-classes list.
Otherwise it complains loudly about not being able to load it from a
class initializer in Provider.java.
Change-Id: Ie7ed5655cba038b504767c3de9f70b47a8965cce
http://ag/352924 renamed INdefPushCallback
to IAppCallback. Update preloaded-classes
to reflect this.
This is a cherry-pick of Id552e94e3a01cdd61593480f3c0aedb32d185f80
Change-Id: I0901f14fb41f94e3c7572c09e090ec5903a878d1
Bug: 10427705
- layoutlib has references to classes that no longer build into the host
core JAR when WebViewClassic is removed.
- preloaded-classes should not reference WebViewClassic classes.
Change-Id: I4d6773a88ea2932982278127a3c96d38be54ddf5
Allow the appropriate WebView to be preloaded in the zygote by
constructing the currently selected WebViewFactoryProvider when the
WebViewFactory is preloaded. At runtime, if the preloaded provider is
still the current selection, the preloaded instance is used, otherwise
the provider is loaded at that time.
This change also removes "graceful" fallback from the experimental
WebView to the classic implementation: if the option to use the
experimental WebView is selected and it could not be loaded
successfully at the time a WebView is created, an exception will be
thrown, rather than allowing execution to continue with the classic
implementation, as the fallback may mislead developers who do not
examine logcat output in detail.
Change-Id: I0cd01c784d7048abeac55ab5863ca16b8fd9ecf2
