There were 2 issues remaining after a recent change to support
glyph caching from multiple textures:
- memory in the GPU for all textures was being allocated automatically.
This is now lazy, being allocated only when those textures are first
needed.
- filtering (applied when a rendered object is transformed) was ignoring
the new multiple-texture structure. Filtering should be applied correctly
whenever we change textures.
Change-Id: I5c8eb8d46c73cd01782a353fc79b11cacc2146ab
Some GPU architectures could not handle the previous implementation
of our glyph cache. Frequent uploads would cause memory problems in the GPU
and eventually a crash due to these memory issues. The solution is to move to
a system of several, smaller caches instead of one monolythic cache for all
glyphs.
Change-Id: I0fc7a323360940d16d5a33eeb33abfab194c5920
This optimization along with the previous one lets us render an
application like Gmail using only 30% of the number of GL commands
previously required
Change-Id: Ifee63edaf495e04490b5abd5433bb9a07bc327a8
I don't know who's to blame, SGX or Tegra2 but one of those two GPUs is not
following the OpenGL ES 2.0 spec.
Change-Id: I2624e0efbc9c57d571c55c8b440a5e43f08a54f2
Bug #5706056
A newly introduced optimization relied on the display list renderer
to properly measure text to perform fast clipping. The paint used
to measure text needs to have AA and glyph id encoding set to return
the correct results. Unfortunately these properties were set by
the GL renderer and not by the display list renderer. This change
simply sets the properties in the display list renderer instead.
This change also improves the error message printed out when the
application attempts to use a bitmap larger than the max texture
size.
Change-Id: I4d84e1c7d194aed9ad476f69434eaa2c8f3836a8
There was an error in some of the OpenGL layer logic such that we would
occasionally set up a layer for rendering and then not clean up when it was
done. This caused future OpenGL rendering to go into that layer instead of
to the buffers being displayed on the screen, resulting in artifacts including
flashes and displaying of stale content. This happened specifically when
using the wifi settings dialog with the InputMethod keyboard displayed,
but it was probably visible in other situations as well.
Issue #5628248: Flickering/flashing after entering password for WiFi
Change-Id: I38139f620b310f4309570fa7224552d2ee633999
Bug #5659476
The FontRenderer was not cleaning up its temporary state, leading
to crashes when invoking renderDropShadow.
Change-Id: I43b24820dd5625af8c080bbe11b64de2f74164b2
Bug #5650514
After invoking a GL functor, libhwui restores a few OpenGL states
including the current FBO. The renderer was however making the
wrong assumption that the FBO to restore to was the base layer
instead of the FBO associated with the current canvas state.
Change-Id: Ie565500832ebffd673f6a43b83422d6cc05470a0
Use const references to keys and values where appropriate to avoid
copying them unnecessarily.
Deleted some dead code.
Simplified a few pieces that were doing unnecessary redundant work.
Change-Id: Ib2145b7094a40db2d679e05dafe050fe1e87b846
* commit 'cb0db0306b5849a35d3d99eea1b34ce019c6f0d8':
Make the overridden ImageView#setVisibility remotable
Clamp non-monotonic stats instead of dropping.
DO NOT MERGE. Fix leak in LayoutTransition
Fix lastVisible/global rects
Fix Wimax-less build.
Fix leak in LayoutTransition
Deferring wallpaper update to improve workspace scrolling (issue 5506959)
Terminate EGL when an app goes in the background
boot animation is dithered and scaled
Fix NdefRecord byte-stream constructor.
PopupWindow dismiss() can get into a recursive loop.
Fold WiMAX state into the mobile RSSI.
Remove dedicated wimax icon to fix RSSI layout.
This does not happen on high end gfx devices. This happens
only if only one EGL context is initialized in the current
process.
Change-Id: Ibd1737efdf84eef8a84108b05795440d1ae9964e
Bug #5566149
Lazily initialize font renderers
Keep 60% of the texture cache when an app goes to the background
Delete least used font renderer when going to the background
Delete all font renderers on full memory trim
Change-Id: I3c2454d46dc1107ec0f0f72a9ce69cbbcc8825e7
IMPORTANT: this change needs two patches for Harfbuzz:
- one concerning hb_utf16_script_run_prev() which was not returning the correct "previous" script
- one for the "script_properties" table that was missing Arabic code point ranges and declaring
HB_Script_Inherited instead of HB_Script_Arabic
The current change is doing the following:
- pass the correct typeface for Harbuzz shaping (depending on the script of the run)
- offset correctly the glyphIDs returned by Harfbuzz
We need to offset the glyphsID as Harfbuzz will return local glyphIDs (meaning in the
local range of the font used for shapping).
We then cannot use those glyphIDs when we are using a fallback Font (Arabic, Hebrews...)
because the FontRenderer needs glyphIDs in the range of all the Fonts (including the fallbacks)
Change-Id: I494897435bbc59293b02392ee2059cebcdf0e571
This change removes unnessary symbols. All symbols are hidden by
default, public APIs with exported symbols are explicitly marked
with ANDROID_API.
Change-Id: I692fde432a86c12108de1cfd1f6504919a7d5f3f
This bug was affecting several third party applications. The bug was
introduced by an optimization that prevented the local clip to be
recomputed when needed.
Change-Id: I7b5c464db21b5ff1a7274af75a3a64d0e9d84772
- we want functions like isEmpty() to return true if NANs are
involved in the Rect
- also clean-up the intersect familly of calls
- minor cleanup in the int32_t Rect as well
These played a role in http://b/5331198.
Bug: 5331198
Change-Id: I5369725ab482e4b83da9f1bd4cee5256e5de75b2
Bug: 5327776
Currently LTE_FLOAT isn't even a valid partial order, let alone
a strict weak order as needed by binary search. Consequently,
KeyedVectors with keys whose types have operator< using LTE_FLOAT
sometimes find the wrong values!
This problem affects several graphics caches including the
TextLayoutCache.
(This is still not a total order because NaNs are not comparable
but hopefully we don't ever search for NaNs this way!)
Change-Id: Ic447a34c7d87f681ee81538f22a0937a039072dd
