Karthick M [Wed, 30 Mar 2016 12:37:37 +0000 (18:07 +0530)]
fixup! [MM] Webaudio with CAPI port.
Chromium upstream removed audio_decoder_android.h [1], which is however
needed in audio_decoder_efl.h. This CL addresses build issue due to the
missing header file.
[1] https://codereview.chromium.org/
1565623002
Bug: http://suprem.sec.samsung.net/jira/browse/TWF-610
Reviewed by: k2.nagaraju, msu.koo, sm.venugopal, sns.park
Change-Id: I916e7589f1dccec08b0590a9ff12b70c7c209dae
Signed-off-by: Karthick M <karthick.m1@samsung.com>
Gajendra N [Fri, 22 Apr 2016 06:20:24 +0000 (11:50 +0530)]
[M50_2661] Chromium upversion to m50_2661 branch.
This patch deals with the following changes:
1. Removing libvpx.diff patch from build/patches folder, as it is no longer needed.
2. context_factory_efl.cc file will be removed in another patch. To avoid build
break, part of the code is commented under EWK_BRINGUP flag (TWF-893).
3. gfx::Screen::GetNativeScreen() is renamed to GetScreen().
4. RenderWidgetHostImpl::ImeSetComposition() expects additional parameter gfx::Range.
5. delegated_frame_{provider, resource_collection}.h headers are removed.
6. FrameTree::ForEach() is removed, replace with a loop iterating over
FrameTree::Nodes().
7. RenderProcessHostImpl::ResumeRequestsForView() is replaced with
ResourceDispatcherHostImpl::ResumeBlockedRequestsForRouteFromUI().
8. ViewMsg_Resize is renamed to ResizeParams.
9. Set proper dependency path for 'mojo_cpp_bindings'.
10. gpu_messages.h is renamed to gpu_host_messages.h.
11. content::GpuProcess() constructor takes a parameter base::ThreadPriority.
12. |rtc_use_h264| is set to 0, without which link fails due to missing symbol.
13. rjsmin.py is located under third_party/WebKit/Source/devtools/scripts
directory, hence set proper import path in make-file-arrays.py.
14. LayoutTheme::theme is already defined in Blink, which calls nativeTheme()
on a condition, hence define nativeTheme() in RenderThemeChromiumTizen.cpp.
15. ResourceBundle::GetNativeImageNamed() takes only one parameter.
16. ui/gfx/screen_type_delegate.h is removed and SetScreenInstance() takes
only one parameter of type ui::Screen.
17. Add 'ozone_common' target dependency required by ozone_efl.
18. GetEGLSurfaceProperties() is removed and GLSurfaceEGL uses default configuration
attributes, which are same as it was returned by SurfaceFactoryEfl.
19. LoadAccessTokensCallbackType & AccessTokenSet are renamed to
LoadAccessTokensCallback & AccessTokenMap respectively, in AccessTokenStore.
20. "PERMISSION_STATUS_" prefix is trimmed on members of PermissionStatenum members.
21. Changes in prototype of DisplayNotification() & DisplayPersistentNotification()
functions of NotificationControllerEfl.
22. user_gesture parameter of PermissionManager::RequestPermission{s}() is removed.
23. base/prefs/ is moved to components/prefs/ folder.
24. net/base/net_util.h is removed.
25. GetCookieMonster() is removed, replace with GetCookieStore() at necessary
places.
26. MemoryPressureController::GetInstance() is moved to MemoryPressureControllerImpl.
27. Refactor EWebView::Suspend() & Resume() based on changes with respect to
ResourceDispatcherHost.
28. WebContents::ExitFullScreen() takes a bool parameter 'will_cause_resize'.
29. Change in first parameter type from void* to size* for HttpResponseHeaders::
EnumerateHeaderLines()
30 Use blink::WebStringToGURL() to instantiate GURL object from WebString.
31. typedef PlatformCanvas is removed from skia, hence use SkCanvas directly.
32. WebFrame::contentHasText() is removed and reimplemented as WebFrameContentDumper::
dumpFrameTreeAsText().
33. Cast CookieStore to CookieMonster to call SetPersistentCookies() and
SetCookieableSchemes() APIs.
34. Version change from 49.2623 to 50.2661 in build and packaging related files.
Debug build is not working because libchromium-ewk.so size is exceeding 2 GB during linking
time. Temporarily disabling debug build with 'nodebug 1' to avoid the error.
JIRA issue raised at http://suprem.sec.samsung.net/jira/browse/TWF-891
With M50 rebase, we face black screen issue due to some changes in Delegated Rendering,
which will be fixed separately.
JIRA issue raised at http://suprem.sec.samsung.net/jira/browse/TWF-893
'blink_platform' requires utrie2.h which is not supported by system icu yet. To avoid build
break, temporarily using Chromium's third_party/icu by setting 'use_system_icu=0'.
JIRA issue raised at http://suprem.sec.samsung.net/jira/browse/TWF-967
Bug : http://suprem.sec.samsung.net/jira/browse/TWF-728
Change-Id: I8994f376a599ede9045c730cb1c1ce340311deb1
Signed-off-by: Gajendra N <gajendra.n@samsung.com>
Youngcheol Kang [Mon, 22 Feb 2016 11:05:07 +0000 (20:05 +0900)]
Support the TM1 target in mobile profile
This patch cherry-picked from beta/m47 branch.
(http://165.213.202.130/gerrit/#/c/96218/)
We decided to support only TM1, not emulator profile in dev branch,
so the patch which related to emulator was removed in this patch.
[Build Command]
$ ./tizen_src/build/build_mobile.sh -P tzmb_v3.0_TM1 -A armv7l
Bug: http://165.213.149.170/jira/browse/TWF-170
Reviewed by: g.czajkowski, jh11.kwon, sns.park
Change-Id: Ie9b679155555c686ee0c943dc882841b9573ba67
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
Gajendra N [Thu, 25 Feb 2016 13:02:15 +0000 (18:32 +0530)]
[M49_2623] Chromium upversion to m49_2623 branch.
This patch deals with following changes:
1. Set gcc_version gyp variable.
2. base/basictypes.h is removed in upstream.
Replace (u)int{8,16,32,64} with (u)int{8,16,32,64}_t.
3. SingleThreadTaskGraphRunner is now part of upstream available under
cc namespace.
4. scoped_ptr .Pass() is deprecated, use std::move instead.
5. Use GenerateSyncToken() to generate a new SyncToken.
6. ScopedVector<> to std::vector<scoped_ptr<>> conversion is not possible.
Replace when there is mismatch in parameter passing or assignment.
7. COMPILE_ASSERT macro is removed from base/macros.h.
Use static_assert instead.
8. WebPageSerializer has been removed from blink.
Commented part of code that calls it.
9. Changes in number and type of arguments for many API's.
10. Indentation fixes by 'git cl format' for the files that are modified in
this commit only.
Bug: http://165.213.149.170/jira/browse/TWF-345
Change-Id: Ice87e1b73430ba3d269f5632b73c9f14e8fef60b
Signed-off-by: Gajendra N <gajendra.n@samsung.com>
Youngsoo Choi [Mon, 22 Feb 2016 04:09:59 +0000 (13:09 +0900)]
[tizen] Disable a cflag '-Wno-format' for global option change
This patch (http://165.213.202.130/gerrit/#/c/109103/) was cherry-picked
from the m47/beta.
The build option '-Wall', '-Wformat, and '-Wformat-security' have been used as
tizen global options. Recently, tizen platform team decided to remove a option
'-Wformat' which is already included in '-Wall'.
Due to the chnage, following build break happens from chromium-efl build.
>> cc1: error: -Wformat-security ignored without -Wformat [-Werror=format-security]
The build break is caused by a cflag '-Wno-format', defined in 'build/common.gypi'.
So, this CL tries to ignore the cflag for tizen port.
Bug: http://10.113.136.204/jira/i#browse/TWF-338
Change-Id: I2e381eef38083bd50f67a24fcd3439650b95fccd
Signed-off-by: Youngsoo Choi <kenshin.choi@samsung.com>
Grzegorz Czajkowski [Thu, 17 Dec 2015 09:49:02 +0000 (10:49 +0100)]
fixup! Implement ewk_settings_auto_fitting_set
Auto fitting (aka Page Overview) changes the scale factor to adjust the content
width to be equal the device width. This feature is by default turned on for
mobile builds and what's more both Reference Browser and ubrowser enable it
in runtime, optionally adding the user possibility to switch it off via UI.
Disabling this feature in runtime results in not reseting the scale factor,
usually from ~0.4f to 1.0f. This causes bad user experience as the scale is
small but user is able to scroll the (empty) content.
Android exposes WebSettings::setLoadWithOverviewMode(boolean overview) [1] which
in Blink resets the scale when auto fit gets disabled. See
adjustForAndroidWebViewQuirks() for details.
This commit sets setLoadWithOverviewMode along with
setShrinksViewportContentToFit when ewk_settings_auto_fitting_set is called.
Not all utc_blink_ewk_settings_scripts_* tests has been fixed. This patch
fixes them, change comments and test names accordingly to their behaviour.
This is squash of two commits from beta/m47 branch:
- http://web.sec.samsung.net/gerrit/#/c/99946/
Reviewed by: sns.park
Change-Id: I00745a5b1de2252c9259fe9d5c21062a40f62c6d
Author: Grzegorz Czajkowski<g.czajkowski@samsung.com>
- http://web.sec.samsung.net/gerrit/#/c/102256/
Reviewed by: sns.park
Change-Id: I576a465d8ac5702e446df95d210707158c5af36c
Author: Marcin Niesluchowski<m.niesluchow@samsung.com>
[1] http://developer.android.com/reference/android/webkit/WebSettings.html
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15298
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15197
Reviewed by: sns.park
Change-Id: I00745a5b1de2252c9259fe9d5c21062a40f62c6d
Signed-off-by: Grzegorz Czajkowski <g.czajkowski@samsung.com>
Chandan Padhi [Thu, 7 Jan 2016 10:39:54 +0000 (16:09 +0530)]
[M48] Enable sync point related code for delegated renderer.
In M48, sync_point has been replaced by gpu::SyncToken in
GLFrameData class.
This commit adapts the same on chromium-efl side as well.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15389
Reviewed by: dhyuna.ko, g.czajkowski, sns.park, venu.musham
Change-Id: Id7b86e28e88fe17794dbcb195827624f4da8216d
Signed-off-by: Chandan Padhi <c.padhi@samsung.com>
Youngcheol Kang [Wed, 27 Jan 2016 04:46:28 +0000 (13:46 +0900)]
Enabled the use_system_re2 flag
Currently we use the re2 library on the internal chromium
(third_party/re2). This patch set the 'use_system_re2' flag
to use re2 library on system instead of internal chromium.
+--------------------+----------+----------+
| Size | Before | After |
+--------------------+----------+----------+
| chromium-ewk.so |
28390640 |
28305872 |
+--------------------+----------+----------+
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15517
Reviewed by: sns.park
Change-Id: I7ceb1112732efe5abdca924a95304dd2cc1e3f9d
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
SeungSeop Park [Mon, 1 Feb 2016 10:26:10 +0000 (19:26 +0900)]
Revert "Build with chromium libusb"
This reverts commit
6f9a252072c9303b827a78c8b09c3bb13917e88e.
Now Tizen 3.0 has libusb 1.0.19, which is newer than chromium's 1.0.17.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15683
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=13185
Reviewed by: sns.park
Change-Id: If4bce3695fe8564edebe975cfdd427b672d07322
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
KangYong Park [Mon, 25 Jan 2016 08:07:20 +0000 (17:07 +0900)]
fixup! [M48_2564] Chromium upversion to m48_2564 branch
The arguments of |ContentRendererClient::GetNavigationErrorStrings|
changed in upstream (https://codereview.chromium.org/
1406303002).
This CL adapts the code accordingly.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15386
Reviewed by: a1.gomes, hyuki.kim, sns.park
Change-Id: I5ceee5ef9c9a3dbea8344e6ae097e394768a7474
Signed-off-by: KangYong Park <kaka.park@samsung.com>
y0.kim [Thu, 14 Jan 2016 05:47:15 +0000 (14:47 +0900)]
fixup! Rename AutofillManagerDelegateEfl to AutofillClientEfl.
Fixes warning:
> no return statement in function returning non-void [-Wreturn-type]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: I444e7fab4f2bb419f5195b657efb1423e4444d61
Signed-off-by: y0.kim <y0.kim@samsung.com>
Antonio Gomes [Thu, 28 Jan 2016 17:41:25 +0000 (13:41 -0400)]
Remove two bogus asserts in GpuChildThreadEfl::OnInitialize.
GpuChildThreadEfl's main goal is to override GpuChannelManager's
shared context and mailbox manager by the ones created by Efl
shared evas gl context.
In order to do that, GpuChildThreadEfl overrides GpuChildThread
and reimplements its OnInitialize method.
OnInitialize first calls GpuChilfThread::OnInitialize, where
gpu_channel_manager_ is instantiated, performs the overrides.
Problem is that by the time GpuChildThread::OnInitialize is executed
and GpuChannelManager is instantiated, its ctor also instantiates
share_group_ and mailbox_manager_ class member variables.
Hence, both DCHECKs (being removed by this patch) fail.
Patch removes them because they check for logically invalid
conditions.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/105899/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15731
Reviewed by: g.czajkowski, sns.park
Change-Id: Ib6d8927f189fdb3595d7da528adebf60d0f2d562
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Thu, 28 Jan 2016 16:45:29 +0000 (12:45 -0400)]
fixup! Unify ContentMainDelegateEfl and WebProcessContentMainDelegateEfl
ui::ResourceBundle::InitSharedInstanceWithPakPath is now called by
ContentMainDelegateEfl for both browser and renderer processes.
No need to explicit call it in EwkGlobalData::GetInstance anymore.
It fixes the following ASSERT failure:
#0 0x7fcf2f7795bf [0128/115917:FATAL:resource_bundle.cc(602)]
Check failed: g_shared_instance_ == NULL. ResourceBundle initialized twice
#0 0x7f4d3b6fd5bf base::debug::StackTrace::StackTrace()
#1 0x7f4d3b73eeb9 logging::LogMessage::~LogMessage()
#2 0x7f4d3e3b6697 ui::ResourceBundle::InitSharedInstance()
#3 0x7f4d3e3b4258 ui::ResourceBundle::InitSharedInstanceWithPakPath()
#4 0x7f4d3b12945b EwkGlobalData::GetInstance()
#5 0x7f4d3b0f8d6d EWebContext::EWebContext()
#6 0x7f4d3b0f8be1 EWebContext::EWebContext()
#7 0x7f4d3b19fefb Ewk_Context::Ewk_Context()
#8 0x7f4d3b19fd3d Ewk_Context::Create()
#9 0x7f4d3b19fc02 Ewk_Context::DefaultContext()
#10 0x7f4d3b1a9ebe ewk_context_default_get
#11 0x000000407f74 Browser::Browser()
#12 0x00000040c85f app_create()
#13 0x00000040cb7e main
#14 0x7f4d38a5fec5 __libc_start_main
#15 0x000000407c89 <unknown>
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/105894/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=14350
Reviewed by: g.czajkowski
Change-Id: I7b1c9c1ba509a521c0b7c16ca534bb3579ff7032
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Wed, 27 Jan 2016 05:16:12 +0000 (01:16 -0400)]
Retire EwkViewMsg_DidFailLoadWithError
Chromium has two messages renderer used to notify browser of a load fail:
FrameHostMsg_DidFailLoadWithError and
FrameHostMsg_DidCommitProvisionalLoad.
Although they are both sent to browser, their hooks are not actually
used by chromium-efl. Instead chromium-efl has a custom message for
its own use: EwkViewMsg_DidFailLoadWithError.
Patch removes EwkViewMsg_DidFailLoadWithError and its plumbing logic,
and use the hooks provided by the former two messages.
Previous stack:
- RenderFrameImpl::didFailLoad
- RenderViewObserverEfl::DidFailLoad
- (IPC - EwkViewMsg_DidFailLoadWithError)
- WebViewBrowserMessageFilterPrivate::OnReceivedDidFailLoadWithError
- EWebView::InvokeLoadError
The stack is:
- RenderFrameImpl::didFailLoad
- (IPC - FrameHostMsg_DidFailLoadWithError)
- RenderFrameHostImpl::OnMessageReceived
- RenderFrameHostImpl::OnDidFailLoadWithError
- (..)
- WebContentsImpl::DidFailLoadWithError
- WebContentsDelegateEfl::DidFailLoad
- EWebView::InvokeLoadError
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/105526/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15697
Reviewed by: g.czajkowski
Change-Id: I680fe89af89bf98181cf02a47e585743f185a994
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Mon, 25 Jan 2016 18:52:50 +0000 (14:52 -0400)]
Reimplement EWebContext::NotifyLowMemory method.
MemoryPurger class is a fork of the obsolete files
chrome/browser/memory_purger.cc/h, removed in [1].
In [2], it is advised for callers of this API to use
RenderThreadImpl::OnMemoryPressure instead.
Patch provides an implementation of EWebContext::NotifyLowMemory
based on MemoryPressureController class that calls out to
MemoryPressureListener and down to RenderThreadImpl::OnMemoryPressure.
ewk/efl_integration/memory_purger.cc/h are also removed, as well
as message EwkViewMsg_PurgeMemory and its associated handler
RenderProcessObserverEfl::OnPurgeMemory.
Experiments with this API shows that it frees up to 40Mb of RAM
when browsing facebook.com.
[1] https://codereview.chromium.org/
192573010
[2] https://code.google.com/p/chromium/issues/detail?id=350455#c13
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/105433/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=4822
Reviewed by: g.czajkowski
Change-Id: I437046a02b4486422dfd2913b1c2b51e405e8e2c
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
SeungSeop Park [Thu, 21 Jan 2016 01:14:18 +0000 (10:14 +0900)]
fixup! Enforce "treat warning as error" policy for chromium-ewk
Although officially CT (and buildbot) support only Tizen 3.0,
there are unofficial activities targeted for old Tizen version.
So, let's restrict the warning/error policy only to Tizen 3.0 for now.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15612
Reviewed by: g.czajkowski, sm.venugopal
Change-Id: Ieac5a206a8579be1875df35d4a0dae28f7c1a505
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SangHyup Lee [Tue, 26 Jan 2016 07:04:47 +0000 (16:04 +0900)]
Fix crash in |SelectionControllerEfl::ClearSelectionViaEWebView|.
If webview evas object deleted by application, destructor of RWHVEfl will be
called. And when RWHVEfl deleted, destructor of SelectionControllerEfl also
will be called because RWHVEfl has a scoped_ptr<SelectionControllerEfl>.
SelectionController try to clear all selection by call |ClearSelectionViaEWebView|
in the destructor. In this case, we should check RWHV and WebContents because
these can be deleted already.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/105349
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15677
Reviewed by: a1.gomes, sns.park
Change-Id: I4f0cb35b2d68a62408be81c4300e93741f3eae6c
Signed-off-by: SangHyup Lee <sh53.lee@samsung.com>
SangHyup Lee [Thu, 28 Jan 2016 05:40:13 +0000 (14:40 +0900)]
fixup! [M48_2564] Chromium upversion to m48_2564 branch
RenderViewHost was separated from RenderWidgetHost after [1].
So we need to get RWHI from RVHI by |GetWidget()| to access
RWHI's member functions.
[1] https://codereview.chromium.org/
1413643002
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15384
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15385
Reviewed by: a1.gomes, sns.park
Change-Id: I51b483144524172ffb428db180d94ffc162d5eec
Signed-off-by: SangHyup Lee <sh53.lee@samsung.com>
Grzegorz Ludwikowski [Wed, 27 Jan 2016 10:16:22 +0000 (11:16 +0100)]
Remove warning from |NavigationPolicyHandlerEfl::DownloadNavigation|
|NavigationPolicyHandlerEfl::DownloadNavigation| contains '#pragma
message', which results in a warning during compilation. The message
warns about code commented out because of change in s-chromium around
M37.
The code was commented out so long ago, that 'git blame' wouldn't point
to actual change. Despite this, no bug was filed for broken
functionallity. This suggests that the code wasn't used anyway.
The code is called through ewk_policy_decision_download, when used on
Ewk_Policy_Decision object of policy type POLICY_NAVIGATION (recevied
from "policy,navigation,decide" callback). Assuming that we call
ewk_policy_decision_download only for Ewk_Policy_Decision objects which
return EWK_POLICY_DECISION_DOWNLOAD decision type from
ewk_policy_decision_type_get, the code isn't used, because objects with
this decision type don't appear in the callback.
Both browser and crosswalk do not use ewk_policy_decision_download API
at the moment.
This change removes the compile time warning, and adds
'NOTIMPLEMENTED()' macro instead, so we will see at runtime that not
implemented code path is used.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15638
Reviewed by: sns.park
Change-Id: Ie07171f41c8c45ba70699ed42b8956b5de0deece
Signed-off-by: Grzegorz Ludwikowski <g.ludwikowsk@samsung.com>
Antonio Gomes [Wed, 20 Jan 2016 23:13:59 +0000 (19:13 -0400)]
Unify ContentMainDelegateEfl and WebProcessContentMainDelegateEfl
Both classes are EWK delegates to the main process kick off routines.
It turns out though that chromium-efl is the only content-based
app that has two different delegates for browser and renderer processes.
Chrome (ChromeMainDelegate) and ContentShell (ShellMainDelegate) for
instance have this class unified.
That said, patch merges WebProcessContentMainDelegateEfl into
ContentMainDelegateEfl, simplifying our logic.
WebProcessContentMainDelegateEfl's main method PreSandboxStartup
was merged into ContentMainDelegateEfl counterpart.
Original beta/m47 patch:
- http://165.213.202.130/gerrit/#/c/105265/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=14350
Reviewed by: djmix.kim, sns.park
Change-Id: Idff02ac417b8247d0fd498db7d99cf3a60b60fb7
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Wed, 13 Jan 2016 20:51:26 +0000 (16:51 -0400)]
RenderProcessHostImpl dtor is not executed.
Since forcibly shutdown our UI message loop, it is possible
that important pending Tasks are simply not executed.
Among these, the destruction of RenderProcessHostImpl class
(scheduled with DeleteSoon) performs various process um-plumbing
stuff.
Particularly, RenderProcessHostImpl holds a strong references of lots
of other objects (e.g. URLRequestContextGetterEfl, RenderWidgetHelper,
ChildProcessLauncher, etc) and not performing the proper destruction
of them might result in crashes at shutdown or other undefined problems.
Patch gives the UI message loop a extra run ('till it gets idle)
to perform such pending tasks.
In order to implement this, it was required to unstub the
implementation of MessagePumpForUIEfl::Run, based on the logic
described in base/message_loop/message_pump.h, as well as in
message_pump_{default,glib}.cc
Callstack looks like:
- EwkGlobalData::~EwkGlobalData
- MessageLoop::RunUntilIdle
- RunLoop::RunUntilIdle
- RunLoop::Run
- MessageLoop::RunHandler
- MessagePumpForUIEfl::Run # idle state will break the loop
Original beta/m47 patches:
- http://165.213.202.130/gerrit/#/c/103446/
- http://165.213.202.130/gerrit/#/c/104287/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15557
Reviewed by: djmix.kim, sns.park
Change-Id: I2957a0a7fb942daf9bcbf74431d91acb8cff7902
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Leszek Syroka [Thu, 21 Jan 2016 15:00:01 +0000 (16:00 +0100)]
Remove warnings from MemoryPurger
Refactoring of MemoryPurger class. Warnings, unused includes and dead
code have been removed.
Erased code contained helper class calling non-existent memory purging
methods of ProxyService. MemomoryPurger was marked in previous commits
as obsolete https://code.google.com/p/chromium/issues/detail?id=350455.
MemoryPurger has not been removed from code base due to existing references
to its interface. Currently only EWebContext::NotifyLowMemory invokes
MemoryPurger methods.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/104842/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=4822
Reviewed by: a1.gomes
Change-Id: Ib77e5d2983c38fa1281f37902370d2d1ad387748
Signed-off-by: Leszek Syroka <l.syroka@partner.samsung.com>
(cherry picked from commit
f776a5291d3af1bea61497155cb3724cf0ef3aa9)
Youngsoo Choi [Mon, 25 Jan 2016 12:42:24 +0000 (21:42 +0900)]
[tizen v3.0] Remove local build.conf files for better mainternance
Disabling rpmlint is for improving build time around 15 minutes.
The previous way to achieve the goal was to keep a custom build.conf inside
source tree, which was made by cloning & modifiying specific platform version's.
Refer to: http://165.213.202.130/gerrit/#/c/75855/
But problem in the approach is that
content of build.conf can be suddenly changed from gbs repo, such as
build macros, options, and package list, which can cause build break.
Whenever build.conf is changed from gbs repo, custom build.conf has to be also
fixed in source tree and uploaded to git repo.
To prevent the burden, build.conf needs to be automatically sync'ed with
the latest one from gbs repo.
So, this CL downloads build.conf from remote repo as part of build procedure,
and optionally modifies the content of the file if user does not wants to
run rpmlint (which is default behaviour).
- Disable rpmlint : build_{target}.sh
- Enable rpmlint : build_{target}.sh --rpmlint
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15671
Reviewed by: sns.park
Change-Id: I6ae5a468748d476db62d877d29080ef4239a7906
Signed-off-by: Youngsoo Choi <kenshin.choi@samsung.com>
SeungSeop Park [Mon, 25 Jan 2016 08:33:20 +0000 (17:33 +0900)]
Enforce "treat warning as error" policy for ubrowser
SSIA.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15612
Reviewed by: sns.park
Change-Id: I4623cd66a8d9213fa3b70c594ee7f5a0eec422b9
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
Leszek Syroka [Tue, 26 Jan 2016 11:40:14 +0000 (12:40 +0100)]
Fix compiler warnings for uBrowser
Warnings caused by calling deprecated API have been erased. Switching
tab to private mode in uBrowser have been removed because it is deprecated
and will not be supported in further versions of Chromium.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15664
Reviewed by: sns.park
Change-Id: Ideb1cb9700e951d2e759494eed99ecdde6c32bd6
Signed-off-by: Leszek Syroka <l.syroka@partner.samsung.com>
SeungSeop Park [Mon, 25 Jan 2016 08:26:11 +0000 (17:26 +0900)]
Enforce "treat warning as error" policy for efl_webview_app and mini_browser
SSIA.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15612
Reviewed by: sns.park
Change-Id: Ic1276552cdae91ce1054b9cb8a03dbc0cfc0a472
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
Youngcheol Kang [Wed, 30 Dec 2015 11:28:40 +0000 (20:28 +0900)]
fixup! Replacing deprecated 'efl-assist' with 'efl-extension'
This patch changes the codes in order to use the 'efl-extension'
in both Mobile and TV profile.
Reviewed by: g.czajkowski, sns.park
Change-Id: I4db84b345ebb7360459695944342412fe2145c1f
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
ws29.jung [Tue, 19 Jan 2016 10:28:50 +0000 (19:28 +0900)]
fixup! [MM] Introduce a switch for zero-copy support from multimedia
Previously, build break happens from missing libyuv headers
when |TIZEN_MULTIMEDIA_ZEROCOPY_SUPPORT| is turned off.
This fix includes changes:
- added missing libyuv headers.
- replaced "base::Process::Current().Handle()" with
"base::GetCurrentProcessHandle()"
Reviewed by: sns.park
Change-Id: I1b8f54297ac52366d8274ccf61fcbf9d1979ff94
Signed-off-by: ws29.jung <ws29.jung@samsung.com>
y0.kim [Thu, 21 Jan 2016 11:45:00 +0000 (20:45 +0900)]
Fix compiler warnings in ewk/efl_webview_app/app.c
Fixes following warnings in app.c:
> 'prompt_entry' defined but not used [-Wunused-variable]
> '__scroll_down_cb' defined but not used [-Wunused-function]
> '__scroll_up_cb' defined but not used [-Wunused-function]
> '__scroll_left_cb' defined but not used [-Wunused-function]
> '__scroll_right_cb' defined but not used [-Wunused-function]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15559
Reviewed by: a1.gomes, sns.park
Change-Id: Ic9c7ae0109e3b2cb75ecdab0b05f644590ce28d3
Signed-off-by: y0.kim <y0.kim@samsung.com>
y0.kim [Thu, 21 Jan 2016 09:37:45 +0000 (18:37 +0900)]
Fix warnings in ewk/efl_webview_app/mini_browser.c
This patch fix 9 warnings in mini_browser.c
> passing argument 1 of 'ewk_context_vibration_client_callbacks_set'
from incompatible pointer type
> passing argument 3 of 'evas_object_smart_callback_add' from
incompatible pointer type
> initialization from incompatible pointer type
> (near initialization for 'values[0].strp')
> assignment from incompatible pointer typen
> unused variable 'permissionRequest' [-Wunused-variable]
> implicit declaration of function 'ewk_start_tracing'
[-Wimplicit-function-declaration]
> implicit declaration of function 'ewk_stop_tracing'
[-Wimplicit-function-declaration]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15558
Reviewed by: sns.park
Change-Id: Iac5fe427fbde3b2093cce95c4651ac6c8261ee31
Signed-off-by: y0.kim <y0.kim@samsung.com>
SeungSeop Park [Fri, 22 Jan 2016 01:56:40 +0000 (10:56 +0900)]
fixup! [M48_2564] Chromium upversion to m48_2564 branch
Fix warning(error):
> .../ewk/efl_integration/renderer/render_view_observer_efl.cc:357:24:\
> error: unused variable 'failed_request' [-Werror=unused-variable]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: msu.koo, sm.venugopal, sns.park
Change-Id: Ia94f8d5ca56b52ed08749de9808ef7159306d87f
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
Suneel Kota [Wed, 20 Jan 2016 13:49:29 +0000 (19:19 +0530)]
fixup! [wayland][tizen v3.0] Initial support for TV and mobile
This CL fixes following warning(error) in building for tizen 3.0 hawkp:
> error: unused variable 'DownloadPath' [-Werror=unused-variable]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: Ib4e45b2c847c0062f55c4445ec4caf87c34a225b
Signed-off-by: Suneel Kota <suneel.kota@samsung.com>
samyukta.c [Tue, 12 Jan 2016 04:10:55 +0000 (09:40 +0530)]
[MM] Remove |format| as argument
Removed the unused argument |format| and changed the order of resolution
parameters |width|, |height|.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15536
Reviewed by: k2.nagaraju, msu.koo, sm.venugopal
Change-Id: Ic7a9723e39d724f6a78cd777565b9e851cb5a411
Signed-off-by: samyukta.c <samyukta.c@samsung.com>
y0.kim [Thu, 7 Jan 2016 07:13:37 +0000 (16:13 +0900)]
fixup! Merge efl-port POC code into tizen_src
Below log is emitted during the build, which clutters the build log:
> #pramga message "TODO: Add proper delegate class for ..."
Patch demotes it to plain code comment for clean build log.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: g.czajkowski, sns.park
Change-Id: Icdf3a4ea1b148dfdc133a79310738898afce6449
Signed-off-by: y0.kim <y0.kim@samsung.com>
Chandan Padhi [Tue, 19 Jan 2016 13:28:12 +0000 (18:58 +0530)]
White screen at content shell launch.
White screen is caused due to failure of gl draw calls.
GPU was blacklisted for content_shell due to which context
provider creation failed. This resulted in creation of
software resources instead of GL resources, for which GL
draw calls stand invalid.
This commit ignores gpu blacklist for content_shell similar
to efl_webview_app and ubrowser on desktop.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15626
Reviewed by: a1.gomes, g.czajkowski
Change-Id: I53444ce02e6005a7a5021dbd5e540dde9b0a382b
Signed-off-by: Chandan Padhi <c.padhi@samsung.com>
SeungSeop Park [Fri, 15 Jan 2016 00:25:46 +0000 (09:25 +0900)]
Enforce "treat warning as error" policy for chromium-ewk
1. Remove top-level "-Dwerror=" gyp define from gyp_chromiumefl.sh.
This will let chromium targets use their own policy regarding whether to treat
warnings as error or ignore them.
2. Add '-Werror' cflag to chromium-ewk target.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15612
Reviewed by: a1.gomes, g.czajkowski
Change-Id: Id2634bcb1958c53ddc53a4259404d3cf54047a66
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SeungSeop Park [Tue, 19 Jan 2016 02:47:04 +0000 (11:47 +0900)]
fixup! Implement device display info and get device scale factor.
Fix following warning in tv build:
> 'int gfx::{anonymous}::GetDensityRange(int)' defined but not used \
> [-Werror=unused-function]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: a1.gomes, msu.koo, sns.park
Change-Id: I4f01bd642ec2117015bc502b136df9bb979b63f9
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SeungSeop Park [Tue, 19 Jan 2016 09:10:17 +0000 (18:10 +0900)]
fixup! [M48_2564] Chromium upversion to m48_2564 branch
Fix warnings of '-Wunused-variable'.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: a1.gomes, msu.koo, sns.park
Change-Id: I6063531c4a9e67277765b3e2475156e8dec1f9a2
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
y0.kim [Thu, 14 Jan 2016 03:12:50 +0000 (12:12 +0900)]
fixup! [MM] Webaudio with CAPI port.
Fix compilation warnings in audio_decoder_capi.cc
>comparison between signed and unsigned integer expressions [-Wsign-compare]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15560
Reviewed by: a1.gomes, msu.koo, sns.park
Change-Id: Ie22b4f062e13474e428c20b7ce713d006e1a7bcc
Signed-off-by: y0.kim <y0.kim@samsung.com>
Youngsoo Choi [Tue, 19 Jan 2016 04:11:25 +0000 (13:11 +0900)]
[tizen v3.0] Add api-version to meet syntax
The definition of property api-version is essential on tizen v3.0.
It can be used for API 'privileg_manager_get_mapped_privilege_list'.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15627
Reviewed by: a1.gomes, sns.park
Change-Id: Icf2f52495eb42ba1c0398e8eebfba070e27a951c
Signed-off-by: Youngsoo Choi <kenshin.choi@samsung.com>
SeungSeop Park [Tue, 12 Jan 2016 11:41:32 +0000 (20:41 +0900)]
fixup! [M40 Merge][Tizen3.0][WRT] Refactor ewk_send_widget_info API
Patch fixes warning:
> 'EWebContext::inspector_server_' will be initialized after...
> 'double EWebContext::widget_scale_' [-Wreorder]
> 'EWebContext::m_pixmap' will be initialized after...
> 'double EWebContext::widget_scale_' [-Wreorder]
Original beta/m47 patches:
- http://165.213.202.130/gerrit/#/c/103997/
- http://165.213.202.130/gerrit/#/c/103199/
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: msu.koo
Change-Id: I087826db0dab99611f3b77646f559f8e912a128d
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
Tomasz Czekala [Thu, 14 Jan 2016 10:25:29 +0000 (11:25 +0100)]
Add return value for SSLHostStateDelegateEfl::HasAllowException
bool SSLHostStateDelegateEfl::HasAllowException was not returning anything
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: Ifb8ac5d0c3b1d6e5559b1fc4dd2549bd91cb33aa
Signed-off-by: Tomasz Czekala <t.czekala@samsung.com>
y0.kim [Thu, 7 Jan 2016 06:35:58 +0000 (15:35 +0900)]
fixup! [wayland] Webcursor for wayland
Fix warning: deprecated conversion from string constant to 'char*' [-Wwrite-strings]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: I7fc56ea91c46ff39d7259f51d230d99e5dcf0e93
Signed-off-by: y0.kim <y0.kim@samsung.com>
SeungSeop Park [Mon, 18 Jan 2016 09:14:04 +0000 (18:14 +0900)]
Fix compilation warning in permission_manager_efl.cc
Patch returns a dummy value to suppress warning.
> warning: no return statement in function returning non-void [-Wreturn-type]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15541
Reviewed by: msu.koo
Change-Id: I8f093960c853b9b1bc13573b175c81b27241337d
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
msu.koo [Mon, 11 Jan 2016 08:20:09 +0000 (17:20 +0900)]
[MM] Consolidate error handling routines of MediaPlayerEfl.
This patch is to consolidate the error handling routines
for ME/MSE implementations.
Bug: http://10.113.136.204/jira/browse/TSAM-806
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15564
Reviewed by: k2.nagaraju, msu.koo
Change-Id: If0b2fe034b5a0055233d938421ddb0626715aa9a
Signed-off-by: msu.koo <msu.koo@samsung.com>
Antonio Gomes [Fri, 15 Jan 2016 15:41:49 +0000 (11:41 -0400)]
Hide disambiguation popup on HW BACK keypress.
SSIA.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/103852/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15609
Reviewed by: sns.park
Change-Id: I3aa284fc1d400c8d6906794a00214ad5fe3c8605
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
SeungSeop Park [Tue, 19 Jan 2016 02:13:32 +0000 (11:13 +0900)]
Fix "libpng warning: iCCP: known incorrect sRGB profile"
There are 32 warnings emitted during the build process.
The cause of warning is that "libpng-1.6 is more stringent about checking
ICC profiles than previos versions"[1].
All png files in chromium-efl are converted by running following command:
$ find . -type f -name "*.png" -exec convert {} -strip {} \;
[1] http://stackoverflow.com/questions/
22745076/libpng-warning-iccp-known-incorrect-srgb-profile
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15534
Reviewed by: sns.park
Change-Id: Ic17441d15648d8fa776646e70d934b8bde0fd430
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
msu.koo [Mon, 21 Dec 2015 11:01:41 +0000 (20:01 +0900)]
[MM] Fix random crashes and flickering while playing video.
Random crash and flickering happens because
- Lifecycle between BrowserMediaPlayerManagerEfl(BMPMEfl) and
MediaPlayerEfls(MPEfl) is not properly managed.
- Lifecycle of delivered media_packet is not properly managed so that
Renderer occasionally accesses media_packet which was already destroyed.
Bundle of patches fix theses issues by
- Introducing weak pointer when posting tasks not to make task outlive
than tasker.
- Making MPEfl's lifecycle dependent with BMPMEfl.
- Introducing IPC for destroying media_packet explicitly after Renderer
uses it.
- Introducing scoped_ptr for media_packet not to make leaks even task
is cancelled.
In detail, this patch consists of following beta/m47 patches:
1. http://165.213.202.130/gerrit/#/c/100528/ by msu.koo
Previously, crash happens on refresh while playing videos using HTML's
iframe tag, because |OnNewTbmBufferAvailable| is queued after RenderFrame
is deleted, so it refers already-deleted RenderFrame.
This patch includes 2 major changes.
- On Renderer-side,
This separates OnNewTbmBufferAvailable() from RenderFrame's lifecycle.
- On Browser-side,
This introduces BrowserMediaPacketManager which is Control IPC handler
outlives than BrowsermediaPlayerManager. This class can handle releasing
|media_packet| without affected by Player's lifecycle.
2. http://165.213.202.130/gerrit/#/c/100838/ by sangdeug.kim
This patch fixes crash during run |MediaPlayerEfl::DeliverMediaPacket| by
checking the destructing state.
3. http://165.213.202.130/gerrit/#/c/100839/ by msu.koo
Previously, random crash happens because player accesses manager
without considering the manager's lifecycle.
Currently, BrowserMediaPlayerManager request to delete Players
when it's deconstructed, which means that Player can outlive
than Manager.
This fix introduces WeakPtr and let Player access Manager
via WeakPtr for lifetime handling between Player and Manager.
4. http://165.213.202.130/gerrit/#/c/102313/ by msu.koo
"Improve MediaPlayerEfl lifecycle management."
Previously, MediaPlayerEfl outlives than BrowserMediaPlayerManagerEfl
so additional checking if Manager is alive before callback.
This patch make MediaPlayerEfl's lifetime dependant with BMPMEfl
and remove above checking routines.
Also remove IsPlayerDestructing and relates, which is no more required
after introducing WeakPtr for PostTask.
5. http://165.213.202.130/gerrit/#/c/102463/ by msu.koo
"Fix possible |media_packet| leaks when DeliverMediaPacket is cancelled."
Previously, DeliverMediaPacket task can be cancelled if queued after
MediaPlayerEfl is destroyed. But when DeliverMediaPacket is cancelled,
there is no handling to destroy the passed |media_packet| on it, so it
causes |media_packet| leaks.
This patch introduces MediaPacketProxy and scoped_ptr to fix it.
6. http://165.213.202.130/gerrit/#/c/103502/ by sangdeug.kim
After making MediaPlayerEfl's lifetime dependent with
BrowserMediaPlayerManager(BMPM), crash happens while destructing player.
Root cause is
- WeakPtr of BMPM is invalidated before Players are destroyed
by the destruction sequence of member variables.
- While destructing Players, Player accesses BMPM
via already-invalidated WeakPtr -> Crash happens.
This patch removes redundant WeakPtr of BMPM and related codes,
and let Player access Manager with raw pointer.
This patch also ensure destroy all players before BMPM is destroyed.
7. http://165.213.202.130/gerrit/#/c/103039/ by msu.koo
Previous implementation implements Proxy Class for media_packet,
but it can be replaced with Chromium's scoped_ptr.
Together with: Iecc266fee83ae009285554dce4581d6c7e9bf545
Bug: http://10.113.136.204/jira/browse/TSAM-751
Bug: http://10.113.136.204/jira/browse/TSAM-836
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15296
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15410
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15425
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15439
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15445
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15458
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15502
Reviewed by: msu.koo, sm.venugopal, sns.park
Change-Id: I59747d19e18cdcb8708d6ba91c49bc2eb7fad66b
Signed-off-by: msu.koo <msu.koo@samsung.com>
Signed-off-by: sangdeug.kim <sangdeug.kim@samsung.com>
y0.kim [Thu, 14 Jan 2016 04:35:35 +0000 (13:35 +0900)]
fixup! [M44_2403] Chromium upversion to m44_2403 branch
Fix compilation warning in ewk_geolocation_private.cc
> '_Ewk_Geolocation_Permission_Request::callback_' will be initialized after [-Wreorder]
> '_Ewk_Security_Origin* _Ewk_Geolocation_Permission_Request::origin_' [-Wreorder]
> when initialized here [-Wreorder]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: Ied47e7e95a1aa13e9926bd6f907ee9c762dea685
Signed-off-by: y0.kim <y0.kim@samsung.com>
sm.venugopal [Wed, 13 Jan 2016 09:31:15 +0000 (15:01 +0530)]
[MM] Fix compiler warnings in tizen_video_decode_accelerator
|release| was used to giveup the object held by scoped pointer. Unused
return from |release| is passed to |ignore_result| to remove warning.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15533
Reviewed by: msu.koo, sns.park
Change-Id: I155c25784511d5ad77a72fc1966bc68c1d4ecf63
Signed-off-by: sm.venugopal <sm.venugopal@samsung.com>
SeungSeop Park [Mon, 18 Jan 2016 08:53:37 +0000 (17:53 +0900)]
Replace '#warning' directive with '#pragma message'
If turn on '-Werror' option, '#warning' will be treated as an error,
which however cannot be suppressed even with "-Wno-error=cpp" option.
As a workaround, patch instead uses "#pragma message" which is information
only and is neither a warning nor an error.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15612
Reviewed by: sns.park
Change-Id: Icfbe90bcd9d9817307704189755d413a808fcc22
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SeungSeop Park [Fri, 15 Jan 2016 04:13:49 +0000 (13:13 +0900)]
Ignore warnings in clipboard_helper_efl.cc
There are several warnings of -Wunused-variable and -Wunused-function.
These are originated from temporary code under defined(WAYLAND_BRINGUP) guard.
Rather than fixing the temporary code, patch just ignores the warnings.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15535
Reviewed by: sns.park
Change-Id: Ief83a2628d119154cd4f2b81adadc00636811987
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SeungSeop Park [Mon, 18 Jan 2016 10:15:23 +0000 (19:15 +0900)]
Make PermissionPopup's destructor as virtual.
There are warnings in permission_popup_manager.cc:
> ... deleting object of abstract class type 'PermissionPopup' which has \
> non-virtual destructor will cause undefined behaviour \
> [-Werror=delete-non-virtual-dtor]
Since PermissionPopup class is used as base class of several classes, e.g,
- GeolocationPermissionPopup,
- NotificationPermissionPopup,
- NotificationPermissionPopup,
its destructor needs to be virtual.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15621
Reviewed by: sns.park
Change-Id: I634825ac2a8b2c3b5a9164569242fb09d67d96b7
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
j-hyun.lee [Fri, 15 Jan 2016 05:25:00 +0000 (14:25 +0900)]
fixup! [M47_2526] Chromium upversion to m47_2526 branch
Fix compiler warning in plugin_placeholder_efl.cc
blink::WebPlugin* PluginPlaceholderEfl::CreatePlugin() was not returning
anything. Set return value as nullptr.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15544
Reviewed by: sns.park
Change-Id: Ic8c9e5570686a3f8f6549fdbd3e07b24460beb93
Signed-off-by: j-hyun.lee <j-hyun.lee@samsung.com>
Antonio Gomes [Wed, 16 Dec 2015 19:35:02 +0000 (15:35 -0400)]
Disambiguation popup should send tap gestures not mouse events
When disambiguation popup is shown, and user clicks on its
content, RWHVEfl synthesizes and sends mouse down/up events
to the renderer.
Mouse events are different from a regular touch input, and various
touch-oriented features are not triggered:
- zoom to focused editable node;
- insert large handle on tap.
Patch switches from the disambiguation popup handling logic
from synthesized mouse events to tap gestures.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15359
Reviewed by: a1.gomes, g.czajkowski
Change-Id: I291e107634c4cc56c8512705bcf05ecbcfd81f19
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Fri, 8 Jan 2016 17:19:20 +0000 (13:19 -0400)]
Fix ewk_cookie_manager_persistent_storage_set when called before a WebView
Patch indirectly creates requestcontext, resourceContextEfl
and cookiemanager instances as part of EWebContext ctor.
These objects end up getting created as part of the
EWebView::InitializeContents routine anyways. The reason on why it
is manually triggered here is because there are ewk_cookie_manager
APIs that can be called before a webview instance is created. In such
circumstances, APIs fail to execute, including
ewk_cookie_manager_persistent_storage_set.
Also, patch sets file:// as a cookieable protocol
for WRT purposes.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/102798/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15514
Reviewed by: djmix.kim
Change-Id: Ieffcbf1058237e032486aa3c715dbee1282304c9
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Mon, 4 Jan 2016 20:31:54 +0000 (16:31 -0400)]
Templatize EWebView callback classes.
Patch templatizes the following classes in order to remove duplication:
- EwkViewPlainTextGetCallback
- MHTMLCallbackDetails
- QuotaPermissionRequestCallback
- WebViewGeolocationPermissionCallback
- WebViewUserMediaPermissionCallback
- NotificationPermissionCallback
- WebViewUnfocusAllowCallback
They all share the same logic.
Original beta/m47 patches:
- http://165.213.202.130/gerrit/#/c/102267/
- http://165.213.202.130/gerrit/#/c/102906/ (by Grzegorz Czajkowski)
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15494
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I6b8e97b05bed2fe30c4bfe236e9a48748108945c
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Tue, 15 Dec 2015 18:59:25 +0000 (14:59 -0400)]
Decouple context menu from selection controller class.
Chromium-efl enters selection mode by a long press gesture.
Such event is only triggered when touch events are enabled.
The upstream counterpart of our SelectionControllerEfl
class is TouchSelectionController (see
ui/touch_selection/touch_selection_controller.cc).
As the name implies, its logic is also tied to touch events.
Patch decouples our context menu logic from the
SelectionControllerEfl class, tying it up to touch
event status. This allows context menus to be shown on
TV and desktop builds.
Original beta/m47 patches:
- http://165.213.202.130/gerrit/#/c/99552/
- http://165.213.202.130/gerrit/#/c/100348/
- http://165.213.202.130/gerrit/#/c/100324/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15220
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15393
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15414
Reviewed by: g.czajkowski
Change-Id: I949afb612d3fb0806d95c865d34d51101014734b
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Thu, 7 Jan 2016 19:42:35 +0000 (15:42 -0400)]
Improve ewk_cookie_manager_persistent_storage_set signature
Patch replaces 'bool' parameters part of the call chain of
ewk_cookie_manager_persistent_storage_set.
There is an existing EWK enum we can use instead.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/102797/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15514
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I5320a523231392338bf894554ded4a240c7ee3fa
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Wed, 30 Dec 2015 17:45:54 +0000 (13:45 -0400)]
Enforce proper ewk_view creation in response to "create,window".
When a new webview is created in response to a request from the
engine, the BrowserContext instance of the originating WebContents
is shared with the newly created WebContents object.
In other words, both the WebContents instance that triggered
the new window creation, and the WebContents instance derived
from that request, will share the same BrowserContext instance.
This is an implementation detail that can be seen at
WebContentsImplEfl::HandleNewWebContentsCreate.
Hence, if as part of the WebView creation, the embedding APP
passes in a Ewk_Context instance that wraps a different BrowserContext
instance than the one the originating WebContents holds,
undefined behavior can be seen, including hard crashes.
Below is a snippet code that illustrates the scenario:
(..)
evas_object_smart_callback_add(web_view_, "create,window",
&OnNewWindowRequest, this);
(..)
void OnNewWindowRequest(void *data, Evas_Object*, void* out_view) {
(..)
EvasObject* new_web_view = ewk_view_add_with_context(GetEvas(), ewk_context_new());
*static_cast<Evas_Object**>(out_view) = new_web_view;
(..)
}
The new Ewk_Context object created and passed in as parameter to
ewk_view_add_with_context wraps a different instance of BrowserContext
than the one the new WebContents object will actually hold.
In order to prevent that, patch documents the proper usage of this API,
and adds a CHECK clause aiming at catching any misuse of it.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/101739/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15459
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I51260370eef965c7f1c1d63cde82c82fadc8fbe1
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Wed, 30 Dec 2015 14:28:05 +0000 (10:28 -0400)]
Remove leftover code from EWebView
Patch removes internal classes/methods no longer needed,
namely:
- EWebView::SelectClosestWord
- AsyncRequestHitTestPrivate (declaration only)
- class WebViewAsyncRequestHitTestDataInternalCallback
Original beta/m47 patches:
- http://165.213.202.130/gerrit/#/c/102076/
- http://165.213.202.130/gerrit/#/c/102075/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15459
Reviewed by: djmix.kim, g.czajkowski
Change-Id: Icf84b4f921ecd943f01e747b0de9e3e05c7ca5b0
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Tue, 29 Dec 2015 15:21:33 +0000 (11:21 -0400)]
Defer implementation of HTTPCustomHeaderXXX methods to EWebContext.
Ewk_Context struct should be a proxy to EWebContext.
It works this way for most of its methods but not all.
Patch fixes this.
Additionally, some char* signatures are converted to string&.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/101544/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15459
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I9719cd5ffccb8659b779fd4a99f41d33c32c0f1f
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Mon, 28 Dec 2015 22:30:24 +0000 (18:30 -0400)]
Crash @CookieManager::DeleteSessionCookiesOnIOThread
The way Tizen browser is creating WebView instances
is through ewk_view_add_with_context(bleh, ewk_context_new()).
In other words, it passes in a new context instance for
every view created.
However, given the way chromium-efl (specially EWK layer)
is implemented, it is possible that the instance of EWebContext
passed in is not going to be used by the WebContents object
associated to the receiver WebView.
Fixing this might require a larger rework of some of these assumptions.
Patch emergency basis fixes a related crash due to a null pointer
being deferenced. The same pointer is null-checked in all other
occurrences of this class.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/101375/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15459
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I72ad30ff660eddbb698339f496142553b5538d8a
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Antonio Gomes [Fri, 18 Dec 2015 19:49:56 +0000 (15:49 -0400)]
Long press on links does not trigger context menu in some circumstances
In the current SelectionControllEfl, In order to confirm if a long
press gesture is ongoing, one just needed to check "long_mouse_press_".
However, when long press happens on a link or image, we set
long_mouse_press_ to false in ::HandleLongPressEvent, although
finger is still down. This is important in the current selection
controller logic.
To compensate this, patch adds an extra check asking from the
engine if a finger is still touching down.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/100309/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15407
Reviewed by: djmix.kim, g.czajkowski
Change-Id: I19b664907d79ec4d18e782396e22b3c6c335d2b5
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
msu.koo [Mon, 11 Jan 2016 05:34:02 +0000 (14:34 +0900)]
[MM] Introduce a switch for zero-copy support from multimedia
This patch is to introduce a switch to support zero-copy
when delivering video frames. This switch can be used to test
multimedia features without affected by renderer.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=14015
Reviewed by: sns.park
Change-Id: I3d7232c6338b2fd789eba29765ddd3285bd89fe7
Signed-off-by: msu.koo <msu.koo@samsung.com>
SeungSeop Park [Mon, 4 Jan 2016 05:46:28 +0000 (14:46 +0900)]
[MM] Remove dependency to ffmpeg library in mobile/tv build
Patch set |media_use_ffmpeg| to 0 if |building_for_tizen| is 1.
If media_use_ffmpeg=0, third_party/ffmpeg would be excluded from the build.
Hence, the ffmpeg patch under build/patches/ can be also removed.
Together with: I107068ac5ebfbf2c4da4c7e923747a8db9f0e57b
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=14620
Reviewed by: a1.gomes, sm.venugopal, sns.park
Change-Id: Iecbb6086a3b1b7de869ac8c7c26b4344b2b82f1f
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
SeungSeop Park [Mon, 4 Jan 2016 04:36:55 +0000 (13:36 +0900)]
[MM] Disable speech recognition and remove dependency to flac library
Tizen does not require speech recognition feature.
And it's desirable to minimize third-party library dependency to avoid
unnecessary maintenance overhead.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15477
Reviewed by: a1.gomes, sm.venugopal, sns.park
Change-Id: I6b867dac31734eb92425fada12cf99d445028dc1
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
sm.venugopal [Tue, 29 Dec 2015 05:37:23 +0000 (11:07 +0530)]
[MM] Remove unused gstreamer port.
1) CAPI support is added for ME / MSE / WebAudio. So, removed
gstreamer port. On desktop, chromium's pipeline will be used.
2) Tizen MM flag is enabled only for Tizen3.0.
3) Tizen MM files are compiled based on "tizen_multimedia_support"
flag.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15468
Reviewed by: a1.gomes, sns.park
Change-Id: I1a397265f920c74c953b83c46bfe8f89eeae6dfc
Signed-off-by: sm.venugopal <sm.venugopal@samsung.com>
sangdeug.kim [Tue, 22 Dec 2015 06:52:36 +0000 (15:52 +0900)]
fixup! [MM] Handling multiple instances of media players.
Patch removes workaround code which was added to avoid |player_unprepare|
error due to platform issue.
* beta/m47 cherry-pick: http://165.213.202.130/gerrit/#/c/100693
On SetVolume call, after setting the player volume, the volume was not getting
stored in MediaPlayerBridgeCAPI. Because of this the initial volume value (0,
which gets stored during CreatePlayer), was being set and audio playback was
not happening unless we set the volume again. Updated the code to store volume
in MediaPlayerBridgeCAPI whenever SetVolume call is made.
* beta/m47 cherry-pick: http://165.213.202.130/gerrit/#/c/102232
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=14833
Reviewed by: k2.nagaraju, sm.venugopal, sns.park
Change-Id: Ifc322aa1ad19080d1cd62b9c061e8324df7f7b76
Signed-off-by: sangdeug.kim <sangdeug.kim@samsung.com>
Signed-off-by: mck.giri <mck.giri@samsung.com>
y0.kim [Tue, 5 Jan 2016 08:10:56 +0000 (17:10 +0900)]
fixup! [MM] Webaudio with CAPI port.
Fix warning 'media::AudioDecoderCapi::pcm_output_' will be initialized after
'player_s* media::AudioDecoderCapi::player_' [-Wreorder]
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: If7fc3cd01e6374f8d65546d5039fe2b316dea3c5
Signed-off-by: y0.kim <y0.kim@samsung.com>
msu.koo [Wed, 16 Dec 2015 05:43:15 +0000 (14:43 +0900)]
fixup! [MM] Fix flickering when playing video.
Previously, indexing is required to find the proper |media_packet|
from |tbm_surface| by manager.
But from the previous patch, it has been guaranteed that every |tbm_surface|
will be released, so managing |tbm_surface| itself is just a overhead.
This fixup removes manager,
and let BrowserMediaPlayer directly release the matched |media_packet|
by also passing |media_packet_h| via TbmBufferHandle structure.
Reviewed by: sm.venugopal, sns.park
Change-Id: I0cd888b45ed9a4a64c4ba901c5d2e008de6d4f9a
Signed-off-by: msu.koo <msu.koo@samsung.com>
SeungSeop Park [Fri, 8 Jan 2016 06:35:59 +0000 (15:35 +0900)]
Fix compilation warnings in libvpx (-Wunused-function)
Currently, there are 27 warnings saying:
> ../../third_party/libvpx_new/source/libvpx/vp9/encoder/vp9_denoiser.h:63:12:\
> warning: 'total_adj_strong_thresh' defined but not used [-Wunused-function]
It's because |total_adj_strong_thresh| is defined as static function in .h file.
So warning will appear in any file that includes the header but does not use
the function.
The suggested solution is to declare the static function as inline.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15496
Reviewed by: sns.park
Change-Id: Ic2173c0c3560256df1319d4344988618c71ec9a2
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
msu.koo [Tue, 22 Dec 2015 04:57:39 +0000 (13:57 +0900)]
fixup! [MM] Consolidated routines for video frame delivering.
This fixup removes redundant member variables from previous patch.
Reviewed by: sm.venugopal, sns.park
Change-Id: I1eacf74c73444e46fda16b62adf462aa83d9e952
Signed-off-by: msu.koo <msu.koo@samsung.com>
Youngcheol Kang [Tue, 5 Jan 2016 07:20:48 +0000 (16:20 +0900)]
Removed the use_system_minizip flag
The gyp flag |use_system_minizip| no longer exists in upstream
chromium, hence can be safely removed.
As result, we don't need |use_system_minizip| flag any more.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15486
Reviewed by: a1.gomes, sns.park
Change-Id: Iaffefe5006d883d523c0b682db20f3a646f0b093
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
Youngcheol Kang [Mon, 4 Jan 2016 07:12:07 +0000 (16:12 +0900)]
Removed the speex dependency
The gyp flag |use_system_speex| no longer exists in upstream chromium,
hence can be safely removed.
As result, we don't need speex package any more.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15479
Reviewed by: a1.gomes, sns.park
Change-Id: I2f5ae8528a1bd263dbc23f25ad2fbfd7ae89fd2b
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
Antonio Gomes [Thu, 17 Dec 2015 15:30:12 +0000 (07:30 -0800)]
Use system icu in Tizen builds.
After coordinating with Platform team, icu version was
bumped from 51 to 54, which is the base version of icu
bundled with chromium (third_party/icu).
It turns out that icu-54 is not enough to build chromium
because it lacks TimeZone::detectHostTimeZone API used in
content/browser/time_zone_monitor.cc.
After some more investigation, it was figured out that icu bundled
with chromium is manually patched (see third_party/icu/patches).
It was request to the platform team to apply two patches then
as part of package build:
- third_party/icu/patches/tzdetect{2}.patch
It is now possible to build chromium m47 with system icu.
This saves about 1.5Mb in libchromium-efl.so size (stripped).
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=12893
Reviewed by: sns.park
Change-Id: I90ce970b444c7865d7769a219949a699deb9b893
Signed-off-by: Antonio Gomes <a1.gomes@samsung.com>
Tomasz Czekala [Mon, 28 Dec 2015 11:00:23 +0000 (12:00 +0100)]
Revert "Filter out events jittery for single-touch gestures."
This reverts commit
f7f8b960f56cbd66560b7590e24983beb9e63509.
The reverted commit was removing the events which are required for
smoothness. The gesture_detector.cc module handles extra events
received. We just need to pass them to it.
This is a port of the following patch:
http://165.213.202.130/gerrit/#/c/100452/
Reviewed by: a1.gomes, djmix.kim, prashant.n
Change-Id: Iadc6aa67c7178f3b600c2fcf0aa6c7b0907b7a33
Signed-off-by: Tomasz Czekala <t.czekala@samsung.com>
Tomasz Czekala [Mon, 28 Dec 2015 08:50:31 +0000 (09:50 +0100)]
Fix wrong computation of scroll velocity.
Currently timestamp for touch event (ui::TouchEvent) is using
function ecore_time_get() which gives time in seconds. This makes
difficult to distinguish touch events occurring in sequence as time
comes almost same in seconds. This further breaks gesture event
handing as timestamp resolution is too low.
This patch adds current time in microseconds which is high enough
to distinguish two consecutive touch events distinctly.
This improves touch smoothness in all gestures quite a lot.
This patch sends the timestamp of event actually occurred in platform.
When browser process is busy, system may not be able to deliver the
events in time, so we may get events delayed. Actual events occurred
in platform are equi-spaced in time.
This solves the incorrect fling occurring in small scrolls due to
wrong velocity computation. This also solves now the smoothness with
all gestures and removes jitters from fling events.
This is a port of the following patches:
http://165.213.202.130/gerrit/#/c/100329/
http://165.213.202.130/gerrit/#/c/100453/
Reviewed by: a1.gomes, djmix.kim, prashant.n
Change-Id: I6669f167deb61dc004856985785daf62b39b8f61
Signed-off-by: Tomasz Czekala <t.czekala@samsung.com>
Tomasz Czekala [Thu, 24 Dec 2015 11:17:12 +0000 (12:17 +0100)]
Revert "Change velocity tracker strategy to INT1"
This reverts commit
603fa30d109aa0a36dcf8bfb334e44c69f545dc8.
Reviewed by: a1.gomes, djmix.kim, g.czajkowski, prashant.n
Change-Id: If90df856273a089466b0ff16329e512734c40361
Signed-off-by: Tomasz Czekala <t.czekala@samsung.com>
Youngcheol Kang [Mon, 28 Dec 2015 11:33:21 +0000 (20:33 +0900)]
Updated the gbs.conf by seperating base and main repositories
The snapshot was divided into base and main repositories.
By the request of platform team, the gbs.conf should be updated.
(http://165.213.149.170/confluence/pages/viewpage.action?pageId=
45394582)
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15463
Change-Id: I290f707942de2348001959be8ea0f84726da6f24
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
msu.koo [Fri, 18 Dec 2015 01:38:28 +0000 (10:38 +0900)]
[MM] Remove dead codes: TIZEN_VIDEO_HOLE
TIZEN_VIDEO_HOLE was used on Tizen 2.4 for HAWK-P,
but it's not used anymore on Tizen 3.0
Bug: http://165.213.149.170/jira/browse/TNEF-6526
Reviewed by: sns.park
Change-Id: Ie14b530195f58b590da6c9a5f2f0db1aad0fc73e
Signed-off-by: msu.koo <msu.koo@samsung.com>
msu.koo [Fri, 18 Dec 2015 01:29:23 +0000 (10:29 +0900)]
[MM] Remove dead codes: USE_LEGACY_PLAYER_CAPI
USE_LEGACY_PLAYER_CAPI was used for tizen 2.4 on HAWK-P
but it's not used anymore on Tizen 3.0
Bug: http://165.213.149.170/jira/browse/TNEF-6526
This patch includes patchs from beta branch:
8afad0c [MM] Remove dead codes: USE_LEGACY_PLAYER_CAPI
3b13a6e fixup! [MM] Remove dead codes: USE_LEGACY_PLAYER_CAPI
Reviewed by: sns.park
Change-Id: I1b30dadfc428051e6816beb3c250e6af16db86bd
Signed-off-by: msu.koo <msu.koo@samsung.com>
karthick.m1 [Wed, 16 Dec 2015 16:23:49 +0000 (21:53 +0530)]
[MM][MSE] Fix crash when clicking another clip while video playing in youtube
Crash happens, because media packet comes after player destroyed.
This patch introduces WeaKPointer not to make task outlive player.
Reference: http://165.213.202.130/gerrit/94565/
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=14678
Reviewed by: sm.venugopal, sns.park
Change-Id: I9c666dcf2bc2bb44a601e15a5395548b695701cf
Signed-off-by: karthick.m1 <karthick.m1@samsung.com>
Aron Kim [Wed, 16 Dec 2015 02:40:14 +0000 (11:40 +0900)]
Add gles20 package dependency
After DDK team's asked to apply this change in order to avoid breakage
due to replacing mesa with coregl.
Reviewed by: Antonio Gomes, DaeHyun Ko, venu musham
Change-Id: Ia1e2ee62251bcaaee5a908ed32114bf725af82e1
Signed-off-by: Aron Kim <aron.kim@samsung.com>
puru [Wed, 16 Dec 2015 11:53:07 +0000 (17:23 +0530)]
[MM][MSE] EOS Handling and multiple clip playback is not working
When Video underflow occurs Pause() is called and meanwhile if EOS for Video
is pushed, player could not be put to play state without manual intervention.
So added check for audio and video EOS.
Maxium player instances for TV is 1 and mobile is 5. In case of TV when you
play one MSE clip and press second clips to play, black screen comes and
browser hangs. Same behaviour is noticed in mobile after playing 5 clips.
So after destroying the player |OnSuspendCompleted| should be called,
otherwise browser media player manager considers player still exists,
so the playback does not starts.
Bug: http://165.213.149.170/jira/browse/TSAM-548
Reviewed by: sm.venugopal, sns.park
Change-Id: I971f2463b3e0d3add79db86f64c4c0304f27b56a
Signed-off-by: puru <puru.dharma@samsung.com>
SeungSeop Park [Mon, 21 Dec 2015 05:54:29 +0000 (14:54 +0900)]
[MM] Up-version to M48 and re-enable Tizen MM support
Update MM code as per upstream changes.
- VideoCaptureDevice::Client::OnError() has an additional argument.
Upstream CL: https://codereview.chromium.org/
1418263006
- Include path changed for skcanvas_video_renderer.h
- Method renamed: media::SetDecryptorReadyCB => media::SetCdmReadyCB
- Header renamed: media/base/decryptor.h => media/base/cdm_context.h
- Method removed: media::{Audio|Video}DecoderConfig::extra_data_size().
=> Cherry-picked beta/m47 patch: http://165.213.202.130/gerrit/#/c/96258
- Argument changes for WebmediaPlayer::setSinkId()
- Replaced scoped_refptr<media::GpuVideoAcceleratorFactories>&
-> media::GpuVideoAcceleratorFactories*
as per Upstream CL: https://codereview.chromium.org/
1369843002
- Removed duplicated definition IPC::ParamTraits<media::SubsampleEntry>
from chromium-efl. It's moved to content/common/gpu/gpu_messages.h (M48)
from content/common/media/media_player_messages_android.h (M47).
Finally, patch reverts the temporary codes made by M48 bringup patch
(http://165.213.202.130/gerrit/#/c/99857/):
- Turn on |tizen_multimedia_support| flag again
- Remove a temp patch file.
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15381
Reviewed by: msu.koo, sm.venugopal, sns.park
Change-Id: I78be30dc896724ca1c45d5fb7122f3ef850b25be
Signed-off-by: SeungSeop Park <sns.park@samsung.com>
Sanghyup Lee [Mon, 7 Dec 2015 22:28:29 +0000 (07:28 +0900)]
Default behavior for HW back key should be implemented in web engine.
Currently browser and WRT gets HWBackKey and use it to:
1. If it is full screen mode, exit it.
2. If there is text selection, clear it.
3. WRT: execute script to send HWBackKey to web app.
Browser: Back navigation.
Currently it is implemented in applications.
This patch moves common implements from application to web engine.
|WebViewEvasEventHandler::HandleEvent_HWBack| is added to handle HWBack event.
On mobile this function is called by eext package, on the other profile it is
called by EVAS_CALLBACK_KEY_UP.
"hwback,unhandled" signal is added which is called if event isn't handled in engine.
The next patch will fix ubrowser back navigation with this signal.
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15225
Reviewed by: a1.gomes, djmix.kim, g.czajkowski
Change-Id: I87beaf8c3c8a79b883c0de658077df98504c8b38
Signed-off-by: Sanghyup Lee <sh53.lee@samsung.com>
Grzegorz Czajkowski [Mon, 14 Dec 2015 15:29:57 +0000 (16:29 +0100)]
Do not show selection ctx when selection range is offscreen
Currently, regardless of selection visibilty (start, end) we show selection ctx.
This patch changes ShowHandleAndContextMenuIfRequired() method in a way that
at least one handler visibilty (selection start or end) is required to show
ctx menu.
This fixes WCS TC 10.
Inspired by http://165.213.202.130/gerrit/#/c/98864/.
Additionally, remove ::IsAnyHandleOnScreen because with composited updates the
text selection contains such information.
Original beta/m47 patch:
- http://165.213.202.130/gerrit/#/c/99275/
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=12055
Reviewed by: a1.gomes, djmix.kim, g.czajkowski
Change-Id: I56d2ce545aeecec8f4e4bf9a5be22180eb98596c
Signed-off-by: Grzegorz Czajkowski <g.czajkowski@samsung.com>
Sanghyup Lee [Wed, 16 Dec 2015 04:47:23 +0000 (13:47 +0900)]
Fix popup size for input pickers.
Current popup covers only webview, so browser buttons are available
even if input picker is shown.
This patch changes popup size from webview size to top widget size to
prevent browser buttons clicked if input picker is shown.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/99608/
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15150
Reviewed by: a1.gomes, djmix.kim, g.czajkowski
Change-Id: I6ce011a400fe7248d1e2ec2d5e829606448b9c9f
Signed-off-by: Sanghyup Lee <sh53.lee@samsung.com>
Sanghyup Lee [Wed, 16 Dec 2015 01:46:59 +0000 (10:46 +0900)]
Fix page visibility when user changes browser tab.
When we change browser tab, page visibility should be sent to renderer.
EWebView::HandleShow|Hide is called by Evas_Smart_Class callbacks.
This patch add calling EWebView::Show|Hide in these functions.
Original beta/m47 patch:
- http://165.213.202.130/gerrit/#/c/99575/
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15074
Reviewed by: a1.gomes, djmix.kim
Change-Id: Iab5fe40dc67551323a8e182f5606cd5baf731455
Signed-off-by: Sanghyup Lee <sh53.lee@samsung.com>
Youngcheol Kang [Thu, 17 Dec 2015 00:40:12 +0000 (09:40 +0900)]
[M48_2564] Chromium upversion to m48_2564 branch
This patch deals with following upstream changes:
1. Folder and class name changes
- Changed the name of the inertial_sensor_consts.h to the
device_sensors_consts.h in content/browser/device_sensors/
- Changed the namespace of the GLImageEflPixmap to the gl from the gfx
- Changed the path of the strong_binding.h to the mojo/public/cpp
/bindings from third_party/mojo/src/mojo/public/cpp/bindings
- Removed the third_party/WebKit/public/web/WebNodeList.h
- Removed the wtf/FastMalloc.h
2. API changes
- Added the ConfirmSaveCreditCardLocally, ConfirmSaveCreditCardToCloud,
LoadRiskData API in AutofillClient class
- Added the GetBackgroundSyncController API in BrowserContext class
- Added the LockCompositingSurface, UnlockCompositingSurface API in
RenderWidgetHostViewBase class
- Added the RequestPermissions API in PermissionManager class
- Changed the argument of CancelPermissionRequest, RequestPermission API
in PermissionManager class
- Changed the argument of CopyFromCompositingSurfaceHasResultForVideo,
CopyFromCompositingSurfaceFinishedForVideo, CopyFromCompositingSurface
FinishedForVideo, CopyFromCompositingSurfaceToVideoFrame API
in DelegatedFrameHostClient class
- Changed the argument of CreateNewWindow API in RenderViewHostImpl
class
- Changed the argument of GetNavigationErrorStrings API in
ContentRendererClient class
- Changed the argument of IsFullscreenForCurrentTab API in
RenderWidgetHostDelegate class
- Changed the argument of Init API in PersonalDataManager class
- Changed the argument of net::HttpCache class
- Changed the argument of OnUnmaskVerificationResult API in
AutofillClient class
- Changed the argument of ReturnSubscriberTexture API in
DelegatedFrameHostClient class
- Changed the argument of Run API in Callback class
- Changed the argument of setPageScaleFactor API in WebView class
- Changed the argument of ScalePoint API in gfx::PointF class
- Changed the argument of ShouldCreateWebContents API in
WebContentsDelegate class
- Changed the argument of ViewHostMsg_StartContentIntent IPC message in
view_messages.h
- Changed the name of EnableRecording API to the StartTracing in
TracingController class
- Changed the name of DisableRecording API to the StopTracing in
TracingController class
- Changed the type of GetPasswordManager API in PasswordManagerClient
class
- Changed the sync_point(uint32) variable to the
sync_token(gpu::SyncToken) in gpu namespace
- Removed the ConfirmSaveCreditCard API in AutofillClient class
- Removed the WillUseTexImage, DidUseTexImage, WillModifyTexImage,
DidModifyTexImage API in GLImage class
3. Need to change API (TODO issues)
- The multimedia function has been changed. Temporary disabling the
'tizen_multimedia_support' flag
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15381)
- The autofill support function has been changed. Temporary disabling
the 'tizen_autofill_support' flag
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15382)
- The gold of system had a build problem. Temporary using the gold of
prebuilt in chromium-efl instead of the system
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15383)
- The CreateViewForWidget API in WebContentsView class has been changed.
Temporary disabling the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15384)
- The ExecuteEditCommand API in RenderWidgetHostImpl class has been
changed. Temporary disabling the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15385)
- The GetNavigationErrorStrings API in ContentRendererClient class has
been changed. Temporary disabling the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15386)
- The RequestPermission API in PermissionManager class has been changed.
Temporary disabling the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15387)
- The argument of the ViewHostMsg_StartContentIntent IPC message has
been changed. Need to implement the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15388)
- The sync_point variable was removed in GLFrameData class. Temporary
disabling the codes
(http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15389)
4. Verification
- This patch supports to the desktop and mobile(Tizen 3.0 32/64bit) and
tv(Tizen 3.0 32bit) profiles
- There is no build problem in all target, also the browser operates in
all target
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15310
Change-Id: I0de1cea6b793c8e542fa8359f642b21d7fcb6db5
Signed-off-by: Youngcheol Kang <ychul.kang@samsung.com>
msu.koo [Fri, 11 Dec 2015 07:03:28 +0000 (16:03 +0900)]
[MM] Consolidated routines for video frame delivering.
Previously, both MSE and ME take the same approach
for video frame delivering routines, but codes are duplicated and diverged.
This patch is to consolidate the routines into MediaPlayerEfl,
then let derives(MediaPlayerBridgeCapi, MediaSourcePlayerCapi) use it.
Bug: http://165.213.149.170/jira/browse/TNEF-6573
Reviewed by: sm.venugopal, sns.park
Change-Id: Ie7c18a38daa7d2e32f1ab1361c11286c04bd0faa
Signed-off-by: msu.koo <msu.koo@samsung.com>
Pawel Niemirski [Mon, 14 Dec 2015 16:41:11 +0000 (17:41 +0100)]
Consider webview position while deciding on context menu visibility
Currently, the position of webview is not considered while
deciding on context menu visibility.
This fixes it, and indirectly fixes WCS TC 21.
Original beta/m47 patch:
- http://165.213.202.130/gerrit/#/c/99334/
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=15137
Reviewed by: a1.gomes
Change-Id: I70b0e911bcdf70ef92a16d62f2ab9fc54cbd0977
Signed-off-by: Pawel Niemirski <p.niemirski@samsung.com>
Grzegorz Czajkowski [Fri, 11 Dec 2015 08:44:36 +0000 (09:44 +0100)]
fixup! Reimplement range and page scale functionality
Since chromium-efl sets scale asynchronously after [1] ewk_view_scale_get is
likely to fail after immediately call to ewk_view_scale_set.
This patch adds this note to the doxygen documentation for ewk_view_scale_set.
Additionally, fix ewk_view_scale_set coding style.
[1] http://165.213.202.130/gerrit/95891
Bug: http://web.sec.samsung.net/bugzilla/show_bug.cgi?id=12080
Reviewed by: a1.gomes, sns.park
Change-Id: Ie00a64dc619d047f6c3f94c4afae24931ed82bae
Signed-off-by: Grzegorz Czajkowski <g.czajkowski@samsung.com>
Tomasz Czekala [Fri, 11 Dec 2015 13:51:42 +0000 (14:51 +0100)]
Fix for black screen in new window for web apps
Currently, we see black screen for the new window that is opened
using window.open() in web apps. physical_backing_size(view size
in pixels) was not sent in the resize message to the renderer.
The same is now sent in the resize message.
Original beta/m47 patch: http://165.213.202.130/gerrit/#/c/98852
Bug: http://165.213.149.170/jira/browse/TSAM-613
Reviewed by: a.renevier, djmix.kim
Change-Id: Iafcd2a64eed7e29af2e4b212b8601b5df4bfa442
Signed-off-by: Tomasz Czekala <t.czekala@samsung.com>
Chandan Padhi [Mon, 7 Dec 2015 13:01:32 +0000 (18:31 +0530)]
Fix for crash on webview resize for a new page opened in the same tab.
Currently, segmentation fault occurs on webview resize when we load a
new page on an already existing tab. In this scenario, a new RWHVEfl
is created first and then the previous RWHVEfl is destroyed. However,
the same WebContentsViewEfl(therefore the same native view) is used
for the new RWHVEfl as well.
Two resize callbacks(one for each RWHVEfl) are added to the same
native view in order of their creation.
In destructor of the previous RWHVEfl, evas_object_event_callback_del()
deletes the most recently added callback(the one for the new RWHVEfl).
When resize occurs for the new RWHVEfl, the callback for the previous
RWHVEfl(already deleted) is invoked that results in crash.
To fix this issue, we now use evas_object_event_callback_del_full()
instead of evas_object_event_callback_del().
evas_object_event_callback_del_full() deletes the callback for the
corresponding RWHVEfl only.
Bug: http://suprem.sec.samsung.net/jira/browse/CBEFL-806
Reviewed by: sm.venugopal, sns.park
Change-Id: I36e4538c12b1847bedd8263fe1d18e3815ee0e3d
Signed-off-by: Chandan Padhi <c.padhi@samsung.com>
msu.koo [Tue, 15 Dec 2015 04:27:29 +0000 (13:27 +0900)]
fixup! [MM] Fix flickering when playing video.
Previous patch relied mostly on CleanUp() for releasing
used tbm_surface. But it has issues like
- It causes that many number of tbm_surfaces are pended for releasing,
because compositor sometimes does not signal to release.
This causes locking on capi-player because capi-player has constraints
on output tbm_surface counts.
- CleanUp() is cost operation.
This fixup introduces Closure to bound the lifecycle of
tbm_surface with VideoFrame by following the Chromium's VideoFrame approach.
(So no more hack is required on GLImage for releasing tbm_surface.)
When Chromium invalidates the VideoFrame, tbm_surface will also be released.
For this approach, this fixup introduces IPC message for releasing tbm_surface,
so any process/thread can request to release tbm_surface.
(On previous approach, Browser-Process can only request releasing tbm_surface).
By taking this approach, it's guaranteed that every tbm_handles are
requested to be released, so CleanUp is removed.
Bug: http://165.213.149.170/jira/browse/TSAM-676
Bug: http://165.213.149.170/jira/browse/TSAM-662
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15255
Bug: http://suprem.sec.samsung.net/jira/browse/CBEFL-835
Reviewed by: msu.koo, sm.venugopal, sns.park
Change-Id: I9c622396402ec34e0aa77e28bff7bf3e937f83eb
Signed-off-by: msu.koo <msu.koo@samsung.com>
msu.koo [Wed, 9 Dec 2015 10:18:47 +0000 (19:18 +0900)]
[MM] Fix flickering when playing video.
Previously, Player deletes |tbm_surface| when next frame is delivered.
But occationally, it causes frame drops and crashs
because |tbm_surface| is released before the gpu does not finish to handle it.
This patch introduces |MediaPacketManager|
to hold media_packets which wraps tbm_surface and CleanUp regularly.
Also supplying APIs to let Renderer explicitly release the |tbm_surface|s
by indexing the |media_packet|s using |tbm_surface_h| as the key.
(Renderer only knows tbm_surface.)
CleanUp is introduced to prevent media_packet(tbm_surface) leaks
even Renderer does not request to release.
Theoritically CleanUp should do nothing because Renderer invoke
Release() for all |tbm_surface|s afer rendering so that
implementation is focused on optimizing Release() rather than CleanUp().
Bug: http://165.213.149.170/jira/browse/TSAM-653
Bug: http://107.108.218.239/bugzilla/show_bug.cgi?id=15255
Reviewed by: sns.park
Change-Id: I9a1e3b0710da0072b3459818690d16e5c43d6fd3
Signed-off-by: msu.koo <msu.koo@samsung.com>
sm.venugopal [Wed, 9 Dec 2015 07:50:09 +0000 (13:20 +0530)]
[MM][MSE]CAPI code refactor
Removed unused code. Re-arranged code to improve readablity.
Bug: http://165.213.149.170/jira/browse/TSAM-663
Reviewed by: msu.koo, sm.venugopal
Change-Id: I5d481488d5f9ba1398d2d5058b2866cdd5b80f1c
Signed-off-by: sm.venugopal <sm.venugopal@samsung.com>
Piotr Ganicz [Thu, 10 Dec 2015 14:44:03 +0000 (15:44 +0100)]
Refactoring code with creating OK button on popup.
The issue is that we has CreateOKButton() function in Layout class,
but in the code there is not any usage. In every function with
name started with the "show" word (e.g. showDatePopup) there is
created "OK" button the same way, without using special function
for it. It creates a redundancy of code.
Reviewed by: a1.gomes, g.czajkowski
Change-Id: Ic7a98c43de3dcdc4106187b9e2624d12d9f42cd9
Signed-off-by: Piotr Ganicz <p.ganicz@samsung.com>
projects / platform / framework / web / chromium-efl.git / log