119 Commits

Author SHA1 Message Date
Chris Craik
6c15ffa196 Refactoring of Program ownership/lifecycle, and WIP Glop rendering path
Change-Id: I2549032790bddbc048b0bccc224ed8f386b4517c
2015-02-02 14:08:57 -08:00
Chris Craik
44eb2c0086 Refactor blending and texture gl state
Change-Id: Ia6b3c8b2afd3dfcee7f3ce401d846b789612054a
2015-01-30 10:49:05 -08:00
John Reck
49bc4acfad resolved conflicts for merge of fe5ac4fc to master
Change-Id: I6c0cc82db14b56297586469f940e408c0e218b3b
2015-01-29 12:53:38 -08:00
John Reck
57998017ff Allow ~Layer() to happen after onGLContextDestroyed
Bug: 19146354
Change-Id: I9e885936168bd541bfbed4064ad67ab524f58e32
2015-01-29 10:17:57 -08:00
Chris Craik
96a5c4c7ba Move more GL state management to RenderState and its directory
Change-Id: Ic68584e1c08dc64be2ad43450cb6caa1de834fdc
2015-01-28 16:32:39 -08:00
Chris Craik
65fe5eeb19 Move scissor state to RenderState
Change-Id: I1227a3886fb24e4d9fad79fca469794f06cfb15e
2015-01-27 17:12:39 -08:00
Chris Craik
d41c4d8c73 Add overrides and switch to nullptr keyword for all files
Adds remaining missing overrides and nullptr usages, missed due to
an extreme failure in tool usage.

Change-Id: I56abd72975a3999ad13330003c348db40f59aebf
2015-01-05 16:49:13 -08:00
Chris Craik
51d6a3db97 Cleanup various clang warnings, use unique_ptrs in several places
Change-Id: I347904b25e51fcc7de14b1e72f1acd0f6ba26f3f
2014-12-23 16:53:56 -08:00
John Reck
a2732a2bf9 Stop pretending USE_OPENGL_RENDERER is a thing
Change-Id: I732b490431fe90eafa00a00f3b5429f0d3a067e6
2014-12-18 14:12:27 -08:00
Yohann Roussel
59cf734f9e resolved conflicts for merge of d67bb501 to master
Change-Id: I40698ce1e382cb41eec7af5ea49ac0e2f997d555
2014-12-11 11:10:50 +01:00
John Reck
ebd52610cf Don't preload textures for AssetAtlas
Bug: 18317479

RenderNode::prepareSubTree calls prefetchAndMarkInUse
on every bitmapResoruce in the DisplayList. However,
this resulted in textures being uploaded for bitmaps
that would be drawn from the AssetAtlas instead.

To fix this we teach TextureCache about the AssetAtlas
so that calls to TextureCache return the Texture from
AssetAtlas if it exists. Thus usage of AssetAtlas
is now purely to allow for further optimizations via
draw merging instead of a requirement to get
any benefit at all.

Change-Id: I65282fa05bac46f4e93822b3467ffa0261ccf200
2014-12-10 17:07:41 -08:00
John Reck
27eaec2388 am 842697a3: am decc26df: am f0f68117: Merge "Trace some interesting events" into lmp-mr1-dev
* commit '842697a3602204036e991cfea8b74da3df6e7f14':
  Trace some interesting events
2014-11-18 21:37:59 +00:00
John Reck
fbc8df03e4 Trace some interesting events
Bug: 18337099
Change-Id: Ie2e60da2b9f06e0368061c944d8123ab6903355c
2014-11-17 10:18:31 -08:00
John Reck
6398d2d42e am 8a8a4055: am 541641d0: Merge "Fix counting" into lmp-mr1-dev automerge: 5c8d135
* commit '8a8a40556ea3b0fb4cc8cd1ec063fb9988db497f':
  Fix counting
2014-11-03 18:50:10 +00:00
John Reck
88f5fc7c5a Fix counting
Change-Id: I9a3568908a7223d69430c1765fb5a4db96d7ac9c
2014-11-03 10:32:24 -08:00
John Reck
9481684560 am e05575e9: am a8d83d63: Merge "Layer changes" into lmp-mr1-dev automerge: a51fba0
* commit 'e05575e9c36850d8cfe49396ac9a1372511b12bf':
  Layer changes
2014-10-31 23:30:57 +00:00
John Reck
0e89e2b7bc Layer changes
Bug: 17208461

* Switch Layer to be VirtualLightRefBase instead of
  Caches' side-channel ref-counting
* Include active layers in gfxinfo dump
* Run gfxinfo dump on the correct thread
* Dump gfxinfo on Layer creation failure

Change-Id: I28d195699e2334518e215ab28c7a17355aee9678
2014-10-31 15:30:10 -07:00
Tom Hudson
2dc236b2ba Clean up physical coupling
Narrow the use of #include directives in hwui, replacing with forward
declarations where straightforward. Speeds compiles; doesn't do any
restructuring of code.

Change-Id: Icac2baffb5896f55d8c6718e9bd9d4bfa02d3ca0
2014-10-15 15:51:39 -04:00
John Reck
4ced2620b7 glFinish to flush deletes
Bug: 17479800
Change-Id: I4d4a3eeda510e037f806052025184f3cde18c847
2014-09-19 10:10:19 -07:00
Chris Craik
bfd1cd6209 Additional layer logging
bug:17208461

Track layer's last known state.

Change-Id: Ic1799191f5839a1d6cc56f598f2ac2671dc27a6f
2014-09-10 16:10:24 -07:00
Fred Fettinger
70735bd5dd hwui: Caches: use mBoundTextures only for GL_TEXTURE_2D
bug:17441218

GLConsumer uses glBindTexture() directly instead of going through
Caches::bindTexture(). This can cause libhwui to draw with the wrong
texture bound in the following case which involves 2 TextureViews:

Frame 1:
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 1)
HWUI renders TextureView A:
    calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 1) and draws

Frame 2:
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 1)
GLConsumer::updateTexImage() calls glBindTexture(GL_TEXTURE_EXTERNAL_OES, 2)
HWUI renders TextureView A:
    calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 1) and draws
HWUI renders TextureView B:
    calls Caches::bindTexture(GL_TEXTURE_EXTERNAL_OES, 2) and draws

In this case, HWUI will incorrectly draw TextureView A using texture 2 on
frame 2, because mBoundTextures[0]=1, even though the texture currently
bound to GL_TEXTURE_EXTERNAL_OES is 2.

Since GLConsumer is always used with a target of GL_TEXTURE_EXTERNAL_OES,
work around this problem by having mBoundTextures[] store only the
textures bound to the target GL_TEXTURE_2D. This is the common case
where the extra performance is needed. Since it's legal to have
different textures bound to GL_TEXTURE_2D and GL_TEXTURE_EXTERNAL_OES
on one texture unit, Caches::bindTexture() does not need to clear
mBoundTextures[mTextureUnit] when target != GL_TEXTURE_2D.

Change-Id: I8bc54ab8adcfacad7f3ed17a31236dc7a86c967a
Signed-off-by: Fred Fettinger <fred.fettinger@motorola.com>
2014-09-09 20:00:52 +00:00
John Reck
17035b0211 Have destroy call freePrefetchedLayers
Bug: 17208461

 There's a potential race condition between HardwareRenderer.destroy()
 being called (which calls destroyCanvasAndSurface()) and the renderer
 being finalized (which is what calls freePrefetchedLayers), during which
 time it's possible we get a TRIM_MEMORY_COMPLETE and destroy the EGL
 context.

 Fix this race condition by moving stopDrawing() and freePrefetchedLayers()
 into destroyCanvasAndSurface() where they should have been in the first
 place.

 Also, if we hit the assertion failure, dump the current state of
 Caches to try and provide more context for the failure.

Change-Id: Ife0ba3562041e8b08e87e3e13640472b3004eed6
2014-09-03 11:13:53 -07:00
ztenghui
14a4e35208 Bring back the shadow strength tweak
bug:16712006

Change-Id: Ifc0ecca139d58140b45d7d227536a53069e1d480
2014-08-13 18:22:42 +00:00
Chris Craik
a736cd9d35 Remove scaling bucket hack
bug:16626221

Fixes glyph cache pressure and double scaling artifacts.

Change-Id: I06cde720db646531b45943283cb4ecb48af8ff5b
2014-08-04 13:18:38 -07:00
Chris Craik
58ddced277 Initializing variables - generally a good practice
Change-Id: Ib29a69f741eda869701493bb04e155d786e6e402
2014-07-16 19:11:46 -07:00
Chris Craik
59744b79ec Add hack for custom re-rasterization buckets
bug:14083128

Moves all of the font transform management into
OpenGLRenderer::findBestFontTransform(), and now simply passes down
final rasterization transforms into the FontRenderer.

Change-Id: Ie02752e6af863347b142367c7d628db5f9fc2998
2014-07-07 19:14:55 +00:00
Chris Craik
0732ba5756 am 88b1c9ad: am e709d53f: am 0d28bb47: Merge "Fix the texture ID reuse issue in HWUI."
* commit '88b1c9ad87ca6d6faaf824bff4ee9d3c799362e5':
  Fix the texture ID reuse issue in HWUI.
2014-06-17 01:31:21 +00:00
jiayuanr
4a473c7d94 Fix the texture ID reuse issue in HWUI.
Issue: When the layer of previous frame is destroyed, it doesn't clear the
texture id in mBoundTextures[mTextureUnit], so in the next frame, if
glGenTexture returns same texture ID of the previous frame,
the new texture is not bound.

CRs-fixed: 671736

Change-Id: Ifc5fd2115fb1863b3c006ab14b0faabeaeb4eab4
2014-06-16 09:29:22 +05:30
Chris Craik
05f3d6e511 Tessellate on worker threads
Tessellate and cache (where possible) shadow and round rect
tessellation tasks.

Change-Id: I2cfda8e11d83d51ea74af871235cf26e8f831d40
2014-06-10 01:53:17 +00:00
Chris Craik
797b95b26b Define light position (using new lighting spec) in Java
Also updates the relative shadow strengths.

Change-Id: I6cac7275d38df98aea9f0dda463cd7207102986a
2014-05-21 13:50:29 -07:00
Chris Craik
f5be3ca5cc Add more shadow control knobs
Change-Id: I1ff500bf429029a97b681ced9df652f4ee9f1332
2014-04-30 18:20:30 -07:00
John Reck
087bc0c14b Refcount RenderNode
Change-Id: I7a86db8acc2b78ef33d987a43a119f5933d7d752
2014-04-07 14:53:08 -07:00
ztenghui
50ecf849cb Create one hole inside the umbra area to avoid overdraw.
bug:13439450

Change-Id: I859575196bd5a3029f447883025a6ec3a1f1face
2014-03-17 10:10:02 -07:00
John Reck
e18264b079 Rename DisplayList->RenderNode
Change-Id: Id42e23c9a1a6eb6eaeafef707ced7fa6887b03d0
2014-03-12 13:56:30 -07:00
Chris Craik
e361ad7ab1 Remove experimental perspective projection
Change-Id: Iad688c8395ccd4673cc129973802b5f01dfb45ba
2014-03-11 15:41:40 -07:00
Chris Craik
726118b352 Improve shadow tessellation performance
- Tune and simplify shadow parameters
- Remove additional inner rings
- Improve polygon ray casting algorithm

Change-Id: If0f28b2d66ae0480b675942bb65e8fcd2864425d
2014-03-11 15:23:36 -07:00
ztenghui
63d41abb40 Use pre-computed index to draw the shadow.
Also draw the umbra part as triangle fans instead of zig zag fashion.

b/12840179

Change-Id: Iaa5d15e77351acdd71f076bd8f9bb2d4d2b92faf
2014-02-25 14:18:25 -08:00
ztenghui
ef94c6f88f Separate spot and ambient shadow strength setting
Change-Id: I4530e618b09a7f44b5382f8a40646c0ebf5f214c
2014-02-13 17:09:45 -08:00
ztenghui
cc3c25622c Property support for light positioning.
Tune up the light size to make it look better.

Change-Id: I139a05f3dd53dacbe55759b91188f0e1cc2c7f80
2014-01-17 13:42:09 -08:00
Chris Craik
ba9b613437 Create private properties on GLCanvas for experimentation with 3d
Change-Id: I17772f61efce727cb4c1111f4d97f58c741786b8
2013-12-16 14:57:00 -08:00
ztenghui
55bfb4e728 Calculate and show Ambient shadow.
Basically we compute the shadow as a strip of triangles, whose alpha value
is the strength of the shadow.
We use the normal to extend the geometry.
And we use static function and try to avoid new/malloc in the computation.

Change-Id: I382286f1cad351bd5ff983f76f446c075819dcaf
2013-12-13 17:25:47 -08:00
Romain Guy
627c6fd913 Add overdraw debugging that accounts for Deuteranomaly
Change-Id: I31f68a07aa7cf0490d2572e24e4c5ac2066a1151
2013-08-21 11:53:18 -07:00
Romain Guy
46bfc48110 Fix hardware layers lifecycle
Bug #10075732

Hardware layers could survive across EGL terminate events.

Change-Id: Ie8565d55cb29fe6625fa1584d695edfecd37ab5e
2013-08-16 18:38:29 -07:00
Romain Guy
b0a41ed3dc Prevent ANR in apps using drawPath()
Bug #10347089

If an app clears its path cache before stopping background tasks, it could
get into an infinite loop in PathCache::trim().

Change-Id: Ieb865b762e7b00aebaba0c023769c2db286a94f5
2013-08-16 14:44:38 -07:00
Romain Guy
b746371de7 Clear FBO cache on full memory flush
Change-Id: I44e06d5d15cd899a0522c62d7c0d042170665abb
2013-08-16 13:55:29 -07:00
Victoria Lease
1e546815bb Support RGBA fonts and bitmap fonts (and RGBA bitmap fonts)
Quite a few things going on in this commit:

- Enable bitmap strikes by default in Paint objects.

The SkPaint parameter that enables bitmap strikes was not previously
included in DEFAULT_PAINT_FLAGS. This effectively disabled bitmap
fonts. Oops! It's for the best, though, as additional work was needed
in Skia to make bitmap fonts work anyway.

- Complain if TEXTURE_BORDER_SIZE is not 1.

Our glyph cache code does not currently handle any value other than 1
here, including zero. I've added a little C preprocessor check to
prevent future engineers (including especially future-me) from
thinking that they can change this value without updating the related
code.

- Add GL_RGBA support to hwui's FontRenderer and friends

This also happened to involve some refactoring for convenience and
cleanliness.

Bug: 9577689
Change-Id: I0abd1e5a0d6623106247fb6421787e2c2f2ea19c
2013-07-31 15:50:47 -07:00
Chris Craik
9ab2d18475 Ensure glActiveTexture is cleaned up correctly on functor resume
Change-Id: I103d7d63b17289d599c2c08dcc442cfba9b8e51d
2013-07-22 16:21:55 -07:00
Romain Guy
e3b0a0117a Refcount 9-patches and properly handle GC events
This change adds refcounting of Res_png_9patch instances, the native
data structure used to represent 9-patches. The Dalvik NinePatch class
now holds a native pointer instead of a Dalvik byte[]. This pointer
is used whenever we need to draw the 9-patch (software or hardware.)

Since we are now tracking garbage collection of NinePatch objects
libhwui's PatchCache must keep a list of free blocks in the VBO
used to store the meshes.

This change also removes unnecessary instances tracking from
GLES20DisplayList. Bitmaps and 9-patches are refcounted at the
native level and do not need to be tracked by the Dalvik layer.

Change-Id: Ib8682d573a538aaf1945f8ec5a9bd5da5d16f74b
2013-06-26 17:15:08 -07:00
Romain Guy
31e08e953f Share Caches' index buffer with FontRenderer
This reduces state changes when we draw 9patches and text together,
which happens *a lot*. Also disable the NV profiling extension by
default since it doesn't play nice with display lists deferrals.
To enable it set debug.hwui.nv_profiling to true.

Change-Id: I518b44b7d294e5def10c78911ceb9f01ae401609
2013-06-18 15:53:53 -07:00
Romain Guy
7f4307668b Add new Query class for debugging
This class can be used to perform occlusion queries. An occlusion query
can be used to test whether an object is entirely hidden or not.

Change-Id: Ida456df81dbe008a64d3ff4cb7879340785c6abf
2013-06-13 14:58:32 -07:00