OSDN Git Service
Ben Murdoch [Thu, 15 Sep 2011 11:01:40 +0000 (12:01 +0100)]
Cherry-pick fix for rect sizes from master - do not merge
Fix invalidation rect sizes
When splitting the invalidation rect over the tiles that it
invalidates, we need to use the tile's offset to calculate the
top/left co-ordinate in the case that the inval rect only covers
part of the tile.
(See change Ie3b4be68 for the regression)
Bug:
4965594
Cherry-pick of:
1672c0ff79526ae74ab745724fa57c877811d071
Change-Id: I6a18f1d3d223f5389b0f4f35fab9e579f499e9da
Teng-Hui Zhu [Mon, 19 Sep 2011 22:56:58 +0000 (15:56 -0700)]
Cherry-pick fix for renamed asset from master - do not merge
Fix the missing controller for inline video.
Bug:
5340451
Cherry pick of:
9062292ef8421fc64437c478064f2ba21515ddc8
Change-Id: Ifd706f61f824d1019f998569774b600ae9a9f5b9
Chris Craik [Mon, 19 Sep 2011 22:27:48 +0000 (15:27 -0700)]
Cherry-pick "Fix missing tiles" from master - do not merge
bug:
5288530
Fix issue where stealing a back texture from a tile (while the tile was ready to
swap) wouldn't mark tile as dirty. Now, mark tile dirty whenever texture is
stolen from it, and avoid trying to steal busy textures.
Cherry-pick of CL:
6aa089bded6a1abbf8d61264588e1e3a32be31b3
Change-Id: Iacbda17ba4b4c6382ceb1c84a6e5f02e609a76a5
Kristian Monsen [Thu, 15 Sep 2011 12:56:57 +0000 (13:56 +0100)]
Fix for bug
5226268 [Browser] http keep-alive packets DO NOT MERGE
Disabling SPDY as part of not reusing connections
Also needs the following CL in external/chromium:
https://android-git.corp.google.com/g/#/c/135472/
Change-Id: I246096834c5ca601435df0089fb7f59001506e54
Chris Craik [Wed, 14 Sep 2011 23:56:00 +0000 (16:56 -0700)]
Merge "Improve tile painting prioritization to minimize stalled, blank tiles"
Chris Craik [Wed, 14 Sep 2011 23:06:57 +0000 (16:06 -0700)]
Improve tile painting prioritization to minimize stalled, blank tiles
bug:
5320450
* Greedily deallocate textures from stolen tiles
* Force all prioritization values to be positive
* Create simply described prioritization ordering:
* Draw count
* Has front texture?
* position (if base tile)
Change-Id: Icd5dab931f735c14c18feef5186c8245c5956cfd
Teng-Hui Zhu [Wed, 14 Sep 2011 16:20:56 +0000 (09:20 -0700)]
Merge "A better WAR of driver issue by using GPU fence"
Chris Craik [Tue, 13 Sep 2011 21:54:08 +0000 (14:54 -0700)]
Merge "Allocate textures based purely on viewport size"
Chris Craik [Tue, 13 Sep 2011 18:46:17 +0000 (11:46 -0700)]
Allocate textures based purely on viewport size
bug:
5211417
Allocate textures based upon viewport size, instead of visible nr of tiles. This
will help prevent rare OOM issues triggering from a certain position in the
page.
Change-Id: Ib43fef0e0b9b8e4e26df448d67bb159a105e300c
Teng-Hui Zhu [Tue, 13 Sep 2011 20:24:19 +0000 (13:24 -0700)]
A better WAR of driver issue by using GPU fence
bug:
5270526
Change-Id: I790a3fc995edc73192b2bcb24eaa29e99434f2fb
Chris Craik [Tue, 13 Sep 2011 17:03:55 +0000 (10:03 -0700)]
Merge "Fix layer tile allocation to be more consistent, less disco"
Chris Craik [Mon, 12 Sep 2011 22:49:36 +0000 (15:49 -0700)]
Fix layer tile allocation to be more consistent, less disco
bug:
5290559
Two major things:
* Avoid swapping front/back textures before the textures have been blitted to
avoid race condition where blit fails because it doesn't see a back texture
* Don't allow a tile to acquire its front texture to paint into, as the
acquisition logic doesn't handle this.
Change-Id: I84b59396ca9faaf3ddc7c75b6f66e4736bf4c3cf
Teng-Hui Zhu [Mon, 12 Sep 2011 22:57:44 +0000 (15:57 -0700)]
Merge "Fix the poster issue and full screen behavior html5 video"
Teng-Hui Zhu [Mon, 12 Sep 2011 18:20:44 +0000 (11:20 -0700)]
Fix the poster issue and full screen behavior html5 video
This is related to the change in https://bugs.webkit.org/show_bug.cgi?id=49481,
and it seems there is some following up discussion in this area.
We will just follow what we did from Honeycomb and work around issues for now.
bug:
4523481
Change-Id: I3c2225afb078c24db423119a235266f222055431
Chris Craik [Fri, 9 Sep 2011 18:56:55 +0000 (11:56 -0700)]
Fudge layer translation to always line up with view pixels
bug:
3367478
Change-Id: I37632de8503b0f3eba0a183a156d71ea48bdba98
Chris Craik [Fri, 9 Sep 2011 18:04:00 +0000 (11:04 -0700)]
Merge "If unlock base layer marks tiles dirty, repaint"
Derek Sollenberger [Fri, 9 Sep 2011 17:32:39 +0000 (10:32 -0700)]
Merge "Hookup internal plumbing needed to expose new plugin API."
Steve Block [Fri, 9 Sep 2011 16:18:25 +0000 (09:18 -0700)]
Merge "Cherry-pick WebKit change 93521 to fix FrameLoader"
Derek Sollenberger [Fri, 9 Sep 2011 14:03:09 +0000 (10:03 -0400)]
Hookup internal plumbing needed to expose new plugin API.
bug:
5239378
Change-Id: I2078db559315f93e8c43052bf9718566157f971b
Steve Block [Thu, 8 Sep 2011 10:31:09 +0000 (11:31 +0100)]
Cherry-pick WebKit change 93521 to fix FrameLoader
See http://trac.webkit.org/changeset/93521
Note that the change did not apply cleanly to
FrameLoader::stopLoading() because we don't have
http://trac.webkit.org/changeset/90164. See bug for details.
Bug:
5264509
Change-Id: I824284b9daaa8bc033aebca558d5362ebfb617d8
Chris Craik [Thu, 8 Sep 2011 23:54:13 +0000 (16:54 -0700)]
If unlock base layer marks tiles dirty, repaint
bug:
5278818
Change-Id: I03206bdc5e95743054c361f235e2d94b835a95d1
Chris Craik [Thu, 8 Sep 2011 21:53:37 +0000 (14:53 -0700)]
Merge "Stop layer tile flickering"
Chris Craik [Thu, 8 Sep 2011 21:09:43 +0000 (14:09 -0700)]
Stop layer tile flickering
bug:
5265207
Don't steal textures with a draw count from last frame to prevent flickering
Change-Id: I0a37b2417f0721b150b47920feea9ddb1666ec28
Chris Craik [Thu, 8 Sep 2011 20:40:18 +0000 (13:40 -0700)]
Merge "Deallocate graphics memory with onTrimMemory signals"
Derek Sollenberger [Thu, 8 Sep 2011 19:49:37 +0000 (12:49 -0700)]
Merge "Add Plugin API for a video framerate callback."
Chris Craik [Thu, 8 Sep 2011 03:04:24 +0000 (20:04 -0700)]
Deallocate graphics memory with onTrimMemory signals
bug:
5269460
Deallocate the graphics memory backing a BaseTileTexture on onTrimMemory
signals, and accordingly allocate it lazily, as needed.
Change-Id: I52039723f47e6470e4fe8dd987d384017005390f
Ben Murdoch [Thu, 8 Sep 2011 17:59:32 +0000 (10:59 -0700)]
Merge "Roll back fix for b/issue?id=
4203823"
Teng-Hui Zhu [Thu, 8 Sep 2011 16:40:28 +0000 (09:40 -0700)]
Merge "The CPU code path for uploading the textures"
Derek Sollenberger [Thu, 8 Sep 2011 15:25:50 +0000 (11:25 -0400)]
Add Plugin API for a video framerate callback.
bug:
5239378
Change-Id: I5f7d33302d5a40f58ec12a3c0be63cb51d4ffc75
Ben Murdoch [Thu, 8 Sep 2011 02:04:22 +0000 (03:04 +0100)]
Roll back fix for http://b/issue?id=
4203823
Allow positioned divs with overflow to scroll.
Keeps tidying up that fix for http://b/issue?id=
4203823 did and rolls
back the functional change.
Bug:
5216561ยง
Change-Id: I6268bbe880c432c87ca04e68697f66b32b7032d0
John Reck [Wed, 7 Sep 2011 23:08:34 +0000 (16:08 -0700)]
Merge "Implement PictureSet(SkPicture*) ctor in FAST_PICTURESET case"
John Reck [Wed, 7 Sep 2011 22:16:17 +0000 (15:16 -0700)]
Implement PictureSet(SkPicture*) ctor in FAST_PICTURESET case
Bug:
5270617
Change-Id: I5bcdf551a028588aa34956f98039d893d646b362
Teng-Hui Zhu [Wed, 7 Sep 2011 21:44:47 +0000 (14:44 -0700)]
Merge "Revert to allocate textures for layers which had overflow children."
John Reck [Wed, 7 Sep 2011 21:26:50 +0000 (14:26 -0700)]
Merge "OnTrimMemory plumbing"
Teng-Hui Zhu [Wed, 7 Sep 2011 21:17:07 +0000 (14:17 -0700)]
Revert to allocate textures for layers which had overflow children.
The original change on minimizing the texture for the overflow layer is too
aggressive. Revert this until we have a better solution.
bug:
5254818
Change-Id: I77b43f151e801dd896baf197b9e523d20f42b469
John Reck [Wed, 7 Sep 2011 18:33:07 +0000 (11:33 -0700)]
OnTrimMemory plumbing
Bug:
5269460
Change-Id: Iad6513787b6737d45f1537973e08469005ed4dac
Steve Block [Wed, 7 Sep 2011 16:45:18 +0000 (09:45 -0700)]
Merge "Cherry-pick WebKit change 92744 to fix RenderObjectChildList"
Steve Block [Wed, 7 Sep 2011 15:45:49 +0000 (16:45 +0100)]
Cherry-pick WebKit change 92744 to fix RenderObjectChildList
See http://trac.webkit.org/changeset/92744
Bug:
5263877
Change-Id: I672c48d3fb6dc90d0721b6c5638070eb944596f4
Derek Sollenberger [Wed, 7 Sep 2011 12:20:26 +0000 (05:20 -0700)]
Merge "Fix for plugin being "centered" to the left corner instead of center"
Teng-Hui Zhu [Fri, 2 Sep 2011 17:54:20 +0000 (10:54 -0700)]
The CPU code path for uploading the textures
This is just for debugging anything related to the driver/HW.
Change-Id: I26757f8259ec6b6c2406f78bf7aa9c5f527e31e7
Steve Block [Tue, 6 Sep 2011 23:52:17 +0000 (16:52 -0700)]
Merge "Cherry-pick WebKit change 85623 to fix RenderInline::culledInlineAbsoluteQuads()"
Ben Murdoch [Tue, 6 Sep 2011 23:38:59 +0000 (16:38 -0700)]
Merge "Fix logic error in Android style version calculation."
Teng-Hui Zhu [Tue, 6 Sep 2011 22:59:48 +0000 (15:59 -0700)]
Merge "Revert "WebView animation support""
Ben Murdoch [Tue, 6 Sep 2011 22:49:12 +0000 (23:49 +0100)]
Fix logic error in Android style version calculation.
We need to update the style version on a width or height change,
and also detect a transition from display:none to display:something
(and vice versa) in which case WebKit will pass the display:none
style as a null pointer.
Bug:
5137070
Change-Id: I82fa3fbe990cfc4c733a2f4fed8be37f11e44ba4
Teng-Hui Zhu [Tue, 6 Sep 2011 22:08:57 +0000 (15:08 -0700)]
Revert "WebView animation support"
This reverts commit
078f4452393311da6165131451fcf86e04e04f25.
bug:
5264188
Change-Id: I3dd6cf46a168893aa3067615f2f0f902c6f66d74
Chris Craik [Tue, 6 Sep 2011 21:44:13 +0000 (14:44 -0700)]
Fix old textures not being repainted
bug:
5261488
Tile textures that were reacquired by the same tiles weren't being marked as
dirty.
Change-Id: Ie5c41ff7a8166e5e48736bdc504b90e8c5512f79
Derek Sollenberger [Tue, 6 Sep 2011 20:03:58 +0000 (16:03 -0400)]
Fix for plugin being "centered" to the left corner instead of center
bug:
5149930
Change-Id: I23aa7f8292806897cbc99f7a36e6aec428daf859
Steve Block [Tue, 6 Sep 2011 17:26:15 +0000 (18:26 +0100)]
Cherry-pick WebKit change 85623 to fix RenderInline::culledInlineAbsoluteQuads()
See http://trac.webkit.org/changeset/85623
Bug:
4986300
Change-Id: If557cbd8c6914dc6875b87cfe04591f9bfa42047
Chris Craik [Mon, 29 Aug 2011 02:15:00 +0000 (19:15 -0700)]
Enable double buffering via base tiles
bug:
2522049
allocate textures and tiles using the gldraw count when they were most recently
prepared
remaining issues:
-layers still flicker (presumably from texture stealing)
-layers aren't double buffered yet
Change-Id: Iccdf68326d7d476269d4e3a13903aaab249ee92d
Steve Block [Fri, 2 Sep 2011 17:06:59 +0000 (10:06 -0700)]
Merge "Fix the Java bridge to correctly handle null strings"
Steve Block [Fri, 2 Sep 2011 16:50:48 +0000 (09:50 -0700)]
Merge changes Iebbac8a9,I854067cb
* changes:
Incorrect handling of 'display:' property within nested <ruby> tags
Remove some unsafe static_cast in rendering/
Steve Block [Thu, 1 Sep 2011 11:08:57 +0000 (12:08 +0100)]
Fix the Java bridge to correctly handle null strings
Passing null to methods expecting java/lang/String ...
- Previously null was converted to the empty string but should be
converted to Java null.
- convertNPVariantToJavaValue() already converts a null NPVariant to a
null WTF::String.
- Fixed javaValueToJvalue() to convert a null WTF::String to a null
jvalue.
Returning null from a method with return type java/lang/String ...
- Previously this would crash when trying to get the characters from
the null jstring, but should return JavaScript undefined.
- Fixed jvalueToJavaValue() to convert a null jvalue to a null
WTF::String.
- Fixed convertJavaValueToNPVariant() to convert a null WTF::String to
a void NPVariant.
Both of these are regressions from HC due to
http://trac.webkit.org/changeset/82194.
Added tests in https://android-git.corp.google.com/g/#/c/132580
Bug:
5222742
Change-Id: Id08bc7b40a51426fa7638ad0ed5260d09f3e28bb
Steve Block [Fri, 2 Sep 2011 09:16:42 +0000 (10:16 +0100)]
Incorrect handling of 'display:' property within nested <ruby> tags
This is a cherry-pick of
http://trac.webkit.org/changeset/89987
Bug:
5244760
Change-Id: Iebbac8a93d6aabd0f8d568c9b1ace6f9316a6b6c
Steve Block [Fri, 2 Sep 2011 09:15:02 +0000 (10:15 +0100)]
Remove some unsafe static_cast in rendering/
This is a partial cherry-pick of
http://trac.webkit.org/changeset/89304
It is required to allow us to cherry-pick
http://trac.webkit.org/changeset/89987
Bug:
5244760
Change-Id: I854067cbc297e667ec432f947a268eef5c4dfa0f
Kristian Monsen [Fri, 2 Sep 2011 03:30:10 +0000 (20:30 -0700)]
Merge "Closing idle connections, including jni interface"
Kristian Monsen [Thu, 25 Aug 2011 19:33:21 +0000 (20:33 +0100)]
Closing idle connections, including jni interface
Part of fix for bug
4487538 Current Drain
Change-Id: Ib08ea52c1c71726b76a5811663b6990aa13146ad
Derek Sollenberger [Thu, 1 Sep 2011 19:53:19 +0000 (12:53 -0700)]
Merge "Reapply CL for: "Fix for antialiased text with shadows.""
Teng-Hui Zhu [Thu, 1 Sep 2011 16:30:09 +0000 (09:30 -0700)]
Merge "WebView animation support"
John Reck [Thu, 1 Sep 2011 15:43:25 +0000 (08:43 -0700)]
Merge "Allow native to query selection region"
Derek Sollenberger [Thu, 1 Sep 2011 13:04:10 +0000 (06:04 -0700)]
Reapply CL for: "Fix for antialiased text with shadows."
This reverts commit
e6303b0e3d1473670390053d70a691697ac8b0e6
Steve Block [Thu, 1 Sep 2011 09:00:48 +0000 (02:00 -0700)]
Merge "Fix some leaked JNI local references in the Java Bridge"
Steve Block [Wed, 6 Jul 2011 11:57:47 +0000 (12:57 +0100)]
Fix some leaked JNI local references in the Java Bridge
This includes a cherry-pick of http://trac.webkit.org/changeset/92689
Bug:
5244518
Change-Id: I38704f2307ab667ee6019fb096265609e27618cc
John Reck [Thu, 1 Sep 2011 00:01:41 +0000 (17:01 -0700)]
Allow native to query selection region
Bug:
5224394
Bug:
5208328
Change-Id: I2141ad80a78f717e3005740285661cd1f558dfca
Teng-Hui Zhu [Wed, 31 Aug 2011 21:46:17 +0000 (14:46 -0700)]
WebView animation support
bug:
4982054
Change-Id: I6f4fe313d242f728a515c485a2531611d7166198
Nicolas Roard [Wed, 31 Aug 2011 22:06:47 +0000 (15:06 -0700)]
Disable UI-side animations for now, and let webkit deal with them.
Webkit communication/updates are now fast enough that it's not really an
issue, and this fixes a few annoying bugs.
bug:
4104067
Change-Id: I5994f37de488bb5df1cd06523ed4f0c8cc8a2027
Nicolas Roard [Wed, 31 Aug 2011 20:51:42 +0000 (13:51 -0700)]
Enable the FAST_PICTURESET codepath
bug:
5218173 bug:
5145259
Change-Id: I4518185eb8501dffeef99c08c4ae703c63326e91
Nicolas Roard [Wed, 31 Aug 2011 20:21:42 +0000 (13:21 -0700)]
Merge "Implement a faster pictureset Cluster the invalidations in regular buckets"
Nicolas Roard [Wed, 31 Aug 2011 02:33:56 +0000 (19:33 -0700)]
Implement a faster pictureset
Cluster the invalidations in regular buckets
bug:
5218173 bug:
5145259
Change-Id: Ie3b4be6848b51ca0306bd163803635f9481ace9b
Steve Block [Wed, 31 Aug 2011 13:23:18 +0000 (06:23 -0700)]
Merge "Make sure WebViewClient.onPageStarted() doesn't preceed WebViewClient.shouldOverrideUrlLoading()"
Steve Block [Thu, 25 Aug 2011 11:41:57 +0000 (12:41 +0100)]
Make sure WebViewClient.onPageStarted() doesn't preceed WebViewClient.shouldOverrideUrlLoading()
In the case of a server-side redirect, we were triggering a callback to
WebViewClient.onPageStarted() before we'd called
WebViewClient.shouldOverrideUrlLoading().
This change delays calling WebViewClient.onPageStarted() until after
WebViewClient.shouldOverrideUrlLoading() has returned and has indicated that the
WebView should continue with the load.
Bug:
5191031
Change-Id: I43d12c4df46e5d6e2defad10e80efbf0d5c30aed
Nicolas Roard [Wed, 31 Aug 2011 00:21:01 +0000 (17:21 -0700)]
Reduce the number of dirty buffers (not needed anymore)
bug:
5145259
Change-Id: I05ab5682a1a95b5d0f14c6ff7609b8f6141a2fc5
Nicolas Roard [Wed, 31 Aug 2011 00:13:28 +0000 (17:13 -0700)]
Fix repaint issues with BaseTile
bug:
5225313
Change-Id: I1cf4f60f005f00f8b1cc8454fce189b6a91d2dac
Teng-Hui Zhu [Tue, 30 Aug 2011 16:22:26 +0000 (09:22 -0700)]
Merge "Fix the null pointer crash when playing full screen video."
Teng-Hui Zhu [Mon, 29 Aug 2011 22:11:12 +0000 (15:11 -0700)]
Fix the null pointer crash when playing full screen video.
bug:
5220693
Change-Id: I3696a8523c3328363ec9d162ef32525c7fbab930
Russell Brenner [Mon, 29 Aug 2011 18:38:24 +0000 (11:38 -0700)]
Merge "Add Harfbuzz support for fallback fonts (redux)"
Derek Sollenberger [Mon, 29 Aug 2011 17:51:19 +0000 (13:51 -0400)]
Fix plugin crashes due to bad ref counting.
bug:
5197948
Change-Id: Iece573a141feda56af2123b5fc9ace4cefa5139d
Ben Murdoch [Fri, 26 Aug 2011 15:14:44 +0000 (16:14 +0100)]
Fix PictureSet invlidation rectangle bug
Set the height of the invalidation rectangle correctly. The last
parameter to SkRect::set should be the bottom of the rectangle,
not the desited height.
Also, clear the content when we navigate between pages to avoid
showing any stale content.
Bug:
5213749
Change-Id: Ic75b9e9d8632750a15a03ebfbd4291fad03fca6a
Ben Murdoch [Fri, 26 Aug 2011 10:35:50 +0000 (11:35 +0100)]
Fix crash when scrolling non-HW accelerated WebView
Need to null check m_glWebViewState. It should also have
a compile time guard.
Bug:
5219373
Change-Id: I36fc559fd20e54dc34912f7843a9aa6d2316765b
Chris Craik [Thu, 25 Aug 2011 22:13:37 +0000 (15:13 -0700)]
Merge "better locking while scrolling, but disables double buffering"
Chris Craik [Tue, 23 Aug 2011 20:36:18 +0000 (13:36 -0700)]
better locking while scrolling, but disables double buffering
don't detect scrolling, query from java
removes double buffering
depends on CL in frameworks/base: https://android-git.corp.google.com/g/#/c/130213/
bug:
5106313
Change-Id: Ibfa7f305874f4209fed001c030bb94e752d479b9
Teng-Hui Zhu [Thu, 25 Aug 2011 21:21:38 +0000 (14:21 -0700)]
Merge "Fix the duration value for the html5 inline video"
Teng-Hui Zhu [Thu, 25 Aug 2011 18:51:51 +0000 (11:51 -0700)]
Fix the duration value for the html5 inline video
bug:
4600281
Change-Id: Ied1cd64744b59b9b272547656bc250c9eea5ed9e
Teng-Hui Zhu [Wed, 24 Aug 2011 20:23:03 +0000 (13:23 -0700)]
Merge "Prevent preparing a page with an invalid size."
Teng-Hui Zhu [Tue, 23 Aug 2011 23:18:08 +0000 (16:18 -0700)]
Prevent preparing a page with an invalid size.
bug:
5185840
Change-Id: I9ac0ad0fcee76be8864494fd55398edf1e68506f
Teng-Hui Zhu [Wed, 24 Aug 2011 20:07:49 +0000 (13:07 -0700)]
Merge "The real fix for the video forward/back button."
Teng-Hui Zhu [Wed, 24 Aug 2011 18:16:07 +0000 (11:16 -0700)]
The real fix for the video forward/back button.
Since the code move to the /shadow part, revert the previous change c/130242
and put in this new change.
bug:
4598827
Change-Id: I6592050e198d2801d87ccc1558ceb2bf6dfc25d5
Teng-Hui Zhu [Wed, 24 Aug 2011 17:38:00 +0000 (10:38 -0700)]
Merge "video UI fix, hide the forward/back button"
Teng-Hui Zhu [Mon, 22 Aug 2011 23:09:12 +0000 (16:09 -0700)]
video UI fix, hide the forward/back button
bug:
4598827
Change-Id: I7de60b79e79bbb0de8c3524764ccdbd6ff582268
Nicolas Roard [Tue, 23 Aug 2011 22:45:06 +0000 (15:45 -0700)]
Merge "Limit the number of inval rects we send to the UI"
Nicolas Roard [Tue, 23 Aug 2011 22:31:32 +0000 (15:31 -0700)]
Limit the number of inval rects we send to the UI
bug:
5204192
Change-Id: I5c29e9b2411ac2e05274d9a89c472285ed1aeb67
Russell Brenner [Thu, 18 Aug 2011 20:39:43 +0000 (13:39 -0700)]
Add Harfbuzz support for fallback fonts (redux)
When Skia resorts to fallback fonts for non-latin characters, it
retains the font ID of originally requested font, concealing the
fact that a fallback was invoked. Harfbuzz needs to know the fallback
font ID to read the correct GSUB/GPOS tables from the fallback font
file so that these complex languages can be properly rendered.
This change uses the script recognized by Harfbuzz as a means to
surmise the fallback font that would be used by Skia and then injects
that font file as a replacement for the originally requested font.
Fallback fonts are loaded once and reused multiple times. An earlier
implementation did not cache fonts in this way, causing large pages
to run out of files for mmap streaming.
Bug:
5087744
Change-Id: I61222829209d9bd6df465d37923ccff3681c2c54
Chris Craik [Tue, 23 Aug 2011 17:02:28 +0000 (10:02 -0700)]
Merge "Re-enable tile prefetching, correct and adjust max tile allocation logic"
Teng-Hui Zhu [Mon, 22 Aug 2011 23:46:13 +0000 (16:46 -0700)]
Merge "Avoid the lock when dealing with Surface Texture"
Chris Craik [Mon, 22 Aug 2011 22:49:32 +0000 (15:49 -0700)]
Re-enable tile prefetching, correct and adjust max tile allocation logic
bug:
5168261
Change-Id: I37ec223da4ac1555cd925b08f105f9ed571cb2d9
Teng-Hui Zhu [Mon, 22 Aug 2011 21:40:07 +0000 (14:40 -0700)]
Merge "Merge the line by line copy into one when appropiate"
John Reck [Mon, 22 Aug 2011 21:37:54 +0000 (14:37 -0700)]
Merge "Press state colors update"
John Reck [Mon, 22 Aug 2011 21:31:24 +0000 (14:31 -0700)]
Press state colors update
Bug:
5173644
Change-Id: Ic5a30410e7805e44880773420a3a20b587f80a28
Teng-Hui Zhu [Mon, 22 Aug 2011 20:37:08 +0000 (13:37 -0700)]
Avoid the lock when dealing with Surface Texture
The lock over the surface texture operation is unnecessary.
They just need to be around the transfer queue operation.
bug:
5185840
Change-Id: Ia09ec5f6049421047aac1944efbf601cdc179a52
Teng-Hui Zhu [Mon, 22 Aug 2011 20:20:28 +0000 (13:20 -0700)]
Merge the line by line copy into one when appropiate
When the stride is the same as the bitmap width, we can just use single memcpy
to copy the whole content.
bug:
5196685
Change-Id: I01c9a673c8e95e14eb23f3d7df4002c886b7554d
Teng-Hui Zhu [Mon, 22 Aug 2011 18:27:10 +0000 (11:27 -0700)]
Fix the mouse handling for the video control
bug:
5137664
Change-Id: I0f2b1a84beebe03d4282e728bb26ec70d05bf813