Prateek Thakur [Fri, 18 Nov 2016 06:16:31 +0000 (11:46 +0530)]
eet: add NULL check to eet_connection_empty API to avoid segfault
Summary: porting of patch from upstream
commit
0b6e04d94fc3ff6447b6fbf0fcdbdd8b6958e621
Author: Artem Popov <artem.popov@samsung.com>
Date: Wed Nov 16 15:30:29 2016 -0800
eet: add NULL-check to eet_connection_empty API to avoid segfault
@fix
Summary:
Only eet_connection_empty hasn't checking on NULL and try to read from field of object.
Add NULL-check for connection and return result of empty connection.
Reviewers: jpeg, cedric, myoungwoon
Subscribers: t.naumenko, NikaWhite
Differential Revision: https://phab.enlightenment.org/D4399
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Change-Id: I90113cfa41635f35a6c6ef6d1e1497841f484ce7
Signed-off-by: Prateek Thakur <prateek.th@samsung.com>
Jean Guyomarc'h [Wed, 5 Oct 2016 10:00:38 +0000 (12:00 +0200)]
epp: fix memory corruption when using #warning and #error
The epp instructions #warning and #error would led to a segmentation
fault (invalid free) because the malloced buffer's base pointer was
moved.
@fix
Signed-off-by: Pankaj Mittal <m.pankaj@samsung.com>
Change-Id: I694688fedd6249c0ae34df31ebb8f9670aed847e
Awadhesh Singh [Fri, 6 Jan 2017 09:29:09 +0000 (14:59 +0530)]
Evas Text :Safety check to avoid crash
In many case eo_data_scope_get() will return NULL
Such as object is already free , invalid obj_id reference.
Added safety check to avoid crash
Change-Id: Ic509b63df19e3f1d5d8a6ed7a1985821fe835488
Signed-off-by: Awadhesh Singh <awadhesh1.s@samsung.com>
(cherry picked from commit
cd1adb59dad9df70f6e57858f8b750b5d87336a2)
Youngbok Shin [Wed, 18 Jan 2017 09:35:08 +0000 (18:35 +0900)]
evas: don't reinit fontconfig to destroy _fcConfig
FcInitReinitialize() function will destroy previous default(or current) _fcConfig.
If fontconfig is used outside of Evas, it can cause some crash issues.
@tizen_fix
Change-Id: Ied340277155832c802e5ce1f8f66a6c8be5ae115
(cherry picked from commit
0bc545d5555d30cfcb024ac52ad2e1f29457d3b7)
MinJeong Kim [Wed, 18 Jan 2017 07:01:34 +0000 (16:01 +0900)]
ecore_evas_wayland_common: do not check visible state when it's raised/lowered
Change-Id: Ib7260ac65b1420ffb609587f5c10f447b121d8fd
Signed-off-by: MinJeong Kim <minjjj.kim@samsung.com>
Thiep Ha [Wed, 18 Jan 2017 23:50:44 +0000 (08:50 +0900)]
ecore_evas engine: correct the min, max value when rotate device
When rotate device, the min, max is incorrectly set:
- It is set after the size calculation, so the calculation uses
previous min value.
- The w and h are switched in function parameters, but they are switched
again before calling engine function. This makes wrong w, h value.
This patch correct it.
@tizen_fix
Change-Id: I1307358480e707efdf82a68fc4cdd4b9fbd98c8b
sunghyun kim [Mon, 16 Jan 2017 06:57:01 +0000 (15:57 +0900)]
[Extn_GL] Add changed_buffer for tizen remote surface
buffer_update callback will be deprecated next version
so add changed buffer and upgrade version 2 -> 3
Change-Id: Ia89723f823df91ad0ec7e3879ce82270d11c2c54
Youngbok Shin [Tue, 17 Jan 2017 07:52:37 +0000 (16:52 +0900)]
evas text: add "width" which includes advance
The advance value didn't consider bigger width cases.
@tizen_fix
Change-Id: Id1edc93efc303e60bf85bed6847cb9c4c0ec1725
(cherry picked from commit
fffe0924acb4ac03269e399ee66d853010084588)
Youngbok Shin [Tue, 17 Jan 2017 07:18:33 +0000 (16:18 +0900)]
evas text: remove duplicated code which can cause smaller ellipsis position
@tizen_fix
Change-Id: I8a1fa5389a87e454b8c7c5fb5d8948c70fcde4b4
(cherry picked from commit
e5b24963f20cf6594202080d9a8ffead1cca53cb)
Youngbok Shin [Tue, 17 Jan 2017 06:43:54 +0000 (15:43 +0900)]
evas textblock: +1 to formatted width from Textblock
Because, formatted width is always smaller than the given size,
Edje can't use formatted size as a Textblock part's size directly.
If formatted size is used directly, you will see narrower/taller Textblock part.
It replace the following patch.
commit
bbf5d78d385784f2e2dfdc0830808967a919d054
"edje_calc: give +1 width from calculated width of formatted_get"
@tizen_fix
Change-Id: I16532d6fdf94277b54b69629cca0f365cdc85538
(cherry picked from commit
b609c40cbbfc5b9e864fb43875c691fa65eaa445)
Youngbok Shin [Tue, 17 Jan 2017 05:55:24 +0000 (14:55 +0900)]
Revert "edje_calc: give +1 width from calculated width of formatted_get"
This reverts commit
bbf5d78d385784f2e2dfdc0830808967a919d054.
There was a side effect when "text.min: 0 1" is used.
Change-Id: I5edede90aabe2f9346829008f6189d7b5c038ea7
(cherry picked from commit
f54f1bbf4a719fec7704137a388ce58a5a41020b)
Jaehwan Kim [Tue, 17 Jan 2017 06:36:49 +0000 (15:36 +0900)]
Revert "Revert "edje_calc: round the scaled value when it is casted to int type.""
This reverts commit
d941d18076a51ba6a7a4ec45461699dba22bd27b.
Change-Id: Ide96edcd224dd064e1fbb378f0910d4e07f77653
Jee-Yong Um [Tue, 17 Jan 2017 01:57:04 +0000 (10:57 +0900)]
edje: set repeat_events of swallowed object only when the part has value
Signed-Off-By: Jee-Yong Um <jc9.um@samsung.com>
Change-Id: I2b323825ec4d70a7d6209677eae4bb001a6b69d3
Gwanglim Lee [Thu, 12 Jan 2017 13:55:45 +0000 (22:55 +0900)]
ecore_wl: Fixed window move and resize.
The window with property of the focus skip can not receive the
wl_keyboard::enter event, so there is no way to find the wl_seat
resource needed for the window move or resize request.
To resolve this problem it has changed to use the first wl_seat
resource in the Ecore_Wl_Display when requesting the window move
or resize.
This modification may cause unknown problems, so the tizen extension
protocol that doesn't require wl_seat resource should be added.
Change-Id: I26589c75945bc8005441597ba4b192eec4abc55f
Wonsik, Jung [Thu, 5 Jan 2017 04:00:15 +0000 (13:00 +0900)]
[EvasGL] Add GL[u]int64 type in Evas_GL
To support khronos 64 bit type, this patch is made.
Change-Id: Ib72bd90e7606dba1dbc7797bdb67dc2289e28a55
Joogab Yun [Fri, 6 Jan 2017 08:09:19 +0000 (17:09 +0900)]
[evas_gl] Evas GL Render Threading Patch - 3
-- fixed mistake
Change-Id: I4cb665428b44017003710e6a035625eb62bfad46
Gwanglim Lee [Thu, 5 Jan 2017 07:59:17 +0000 (16:59 +0900)]
ecore_wayland: rotation interface versioning based on the version of server implementation.
Change-Id: Icfd6428cee2b7532b74310f730113fc409259519
Hosang Kim [Thu, 5 Jan 2017 03:54:06 +0000 (12:54 +0900)]
evas_events: fix build error.
Change-Id: Ie42f006664c6338405e6a829e0a1332aba3c435c
(cherry picked from commit
8634d16ae2b38a70351c5e8028eb9c18c4fca046)
Joogab Yun [Tue, 3 Jan 2017 04:45:48 +0000 (13:45 +0900)]
[evas_gl] Evas GL Render Threading Patch - 2
Change-Id: I856a0ba3a2ba57c6b09a9bb5803a6e2461f82cf7
Hosang Kim [Tue, 3 Jan 2017 12:30:35 +0000 (21:30 +0900)]
evas_events: fix mouse cancel logic.
add condition when touch point state is EVAS_TOUCH_POINT_STILL.`
Change-Id: I0d1022badfaaba9c988b63e24eef69d33ec5e861
(cherry picked from commit
d900031385eb3b0777decdb22e89c9f8b9e4e618)
Gwanglim Lee [Wed, 28 Dec 2016 11:55:48 +0000 (20:55 +0900)]
ecore_wayland, ecore_evas_wayland: add support for tizen_rotation v2
angle_change_with_resize event which is newly added to tizen_rotation
v2 clarifies the semantics of changing size of window while rotation.
The clients that receive this event can know exactly how much their
window size should be changed while rotating.
Change-Id: I6dcd0c7222b008b129ef174264c755de1dd42d44
Jongmin Lee [Fri, 30 Dec 2016 06:02:22 +0000 (15:02 +0900)]
Add deprecated comments to deprecated APIs
Change-Id: Id1c755182c695cf685fb1eb998bd4420da97756d
Joogab Yun [Fri, 30 Dec 2016 06:34:15 +0000 (15:34 +0900)]
[evas_gl] don't use a clip in case of the tbm surface type.
- problem)
message app -> new message -> attach -> camera -> change to front
camera -> recoding -> flip is operated which each a clip
Change-Id: I9c213e98c4b7c6f762aa5cd0fd0f52f9dd427a6e
Woochan Lee [Fri, 30 Dec 2016 03:42:27 +0000 (12:42 +0900)]
ecore_wl_dnd.c: Incresed the max size that can be copied.
@tizen_fix
Change-Id: I94382bf173acd740f3be084dd017e4ac43993005
Minkyoung Kim [Thu, 29 Dec 2016 05:34:48 +0000 (14:34 +0900)]
Evas GL:Bind texture to correct one.
If user bind textureA and want to use it continuously, do not call glBindTexture(textureA) again.
But expect that textureA will be binding.
So EvasGL sould not change binded texture silently.
Restore texture to previous bound one after allocating new texture.
And when destroy texture, reset texture to 0 if it is current bound texture.
[Model] Tizen3.0
[BinType] AP
[Customer] OPEN
[Issue#] P161220-02830
[Request] PLM
[Occurrence Version] APL7
[Problem] Collage preview is showing as black while changing the ratio.
[Cause & Measure] EvasGL change the binded texture to 0 in make current job. So resture previous binded texture after allocating new texture.
[Checking Method] Reproduce issue.
[Team] UI System
[Developer] Minkyoung Kim
[Solution company] Samsung
[Change Type] Self Test
Change-Id: Ib14c2be0d69cc4f1c88ace42fc1fb4518571afbf
Joogab Yun [Thu, 29 Dec 2016 07:16:38 +0000 (16:16 +0900)]
[evas_gl] EvasGL : Disable Partial Rendering On EvasGL
- webkit app -> scroll repeat -> sometimes display blink
- browser -> long press on web page -> Magnifer UI -> move ->
display empty
Change-Id: I78b6d80948c532382861c358e89e27468d582537
Joogab Yun [Thu, 29 Dec 2016 06:30:44 +0000 (15:30 +0900)]
[evas_gl] SVACE issue fixed
- It is void type function but it returned bool type
Change-Id: Ie3f7a90c3d589e6e9169ad58238aca32af122528
Youngbok Shin [Thu, 29 Dec 2016 05:45:56 +0000 (14:45 +0900)]
edje: fix to work color_class del for textblock styles
@tizen_fix
Change-Id: I7406eac6d5e986f442d78ae30a54acd252ccfdae
JEONGHYUN YUN [Thu, 29 Dec 2016 02:05:58 +0000 (11:05 +0900)]
ecore_wayland: insert exception code for SVACE issue
Change-Id: If5554e833ed3e0479885d14fc5acd0197f2acebc
Signed-off-by: JEONGHYUN YUN <jh0506.yun@samsung.com>
Shinwoo Kim [Thu, 7 May 2015 11:24:29 +0000 (20:24 +0900)]
[ecore_input] add ECORE_EVENT_JOYSTICK_BUTTON_PLAY
Change-Id: I946df16ab557c11e97af476557ef6236621b2b1c
DaeKwang Ryu [Wed, 28 Dec 2016 06:21:02 +0000 (15:21 +0900)]
[EvasGL] bugfix framebuffer_create function usage
_framebuffer_create function parameter is EvasGL Version.
Not Eina_Bool.
Change-Id: I2252561ad9d7d4c4d616cf4369af9dbb46607b24
DaeKwang Ryu [Tue, 27 Dec 2016 15:20:34 +0000 (00:20 +0900)]
[EvasGL] fix GLES 1.1 FBO creating
Depth and Stencil buffers of GLES 1.1 FBO were created with
GLES 2.0 capability test results.
But GLES 1.1 doesn't support OES_depth_texture.
Originally we need to test capability test with GLES 1.1 context.
Change-Id: I2e2979d126818f64bfef3e88f3447927e6281c7c
DaeKwang Ryu [Wed, 28 Dec 2016 06:28:31 +0000 (15:28 +0900)]
Revert "[EvasGL] Fix to OpenGLES 1.1 rendering issue."
This reverts commit
6c48fedaf3681a044a2ad993fb389de444da729e.
Change-Id: Ib572a074f0914cf37832bd73c47e18702ae5a560
DaeKwang Ryu [Wed, 28 Dec 2016 06:27:54 +0000 (15:27 +0900)]
Revert "[EvasGL] add to check opengles 1.1"
This reverts commit
3d823b6f5a0771719acd834efeb4bd1aa2d3d3dd.
Change-Id: Ia789c73416a6b972c4a33cf40b46691b8220e531
Youngbok Shin [Tue, 27 Dec 2016 06:21:56 +0000 (15:21 +0900)]
edje/entry: fix to get start position of selection from surrounding_cb
The cursor position when a entry has selection should be same from
_edje_entry_imf_retrieve_surrounding_cb and
_edje_entry_imf_event_delete_surrounding_cb functions.
@tizen_fix
Change-Id: Id3ea75e3ab574e1e99220ed826f3bfb0a02bb5a4
Minkyoung Kim [Tue, 27 Dec 2016 08:15:30 +0000 (17:15 +0900)]
Evas GL: Skip FBO creation when direct_mem_opt is set, unless direct fallback is set.
Summary:
When direct_mem_opt is set, FBO should not be created, unless
the render target for direct rendering is FBO.
There are cases where evas_gl_make_current is called outside
on pixel callback in webkit, and since they use direct_mem_opt,
FBO is repeatedly created/destroyed during switch from direct to indirect.
This should fix it.
[Model] Tizen3.0
[BinType] AP
[Customer] OPEN
[Issue#] N/A
[Request] N/A
[Occurrence Version] N/A
[Problem] Webkit UI has a few second black screen on first time.
[Cause & Measure] Internal make current didn't work properly.
[Checking Method] Run email app and compose new mail.
[Team] UI System
[Developer] Minkyoung Kim
[Solution company] Samsung
[Change Type] Self Test
Change-Id: Ic37772b5ccf05c345d880714604ee21fd890ad4d
wonki kim [Wed, 28 Dec 2016 01:06:20 +0000 (17:06 -0800)]
Revert "evas_events: Fix to update clip area with more patch."
This reverts commit
25cdb5d702976c3c920fc64ac06df9e4a2799191.
Change-Id: I0fe6bd257d5255e806ae55515602b39a8d02bb30
wonki kim [Wed, 28 Dec 2016 01:06:17 +0000 (17:06 -0800)]
Revert "Evas_Render: upstream performance improvements port, so object map works now"
This reverts commit
60b2fe175ede63b944c4540470f657543ef0340c.
Change-Id: I3e7a2ff3c8efbf5cfc60bdf0cab5cbee0c9d239c
Wonki Kim [Tue, 27 Dec 2016 02:56:08 +0000 (11:56 +0900)]
Ecore_Input : Modify public structure member order
If public structure is changed, application which has been alread built
doesn't access to the member of structure properly unless developer
built it again with the new header.
the commit(
1c0707d8f297df00f703e16141de5a30d128ab31) changed Ecore_Iput
header then, I guess that make something wrong.
this commit modify public structure member order to keep ABI compet.
Change-Id: Ic9ddbd8ab9c1af1737e184a3545929870b6c7b46
Signed-off-by: Wonki Kim <wonki_.kim@samsung.com>
Hosang Kim [Tue, 27 Dec 2016 10:23:22 +0000 (19:23 +0900)]
evas_events: Fix to update clip area with more patch.
This patch fixes that image, which is clipped by precise image clipper
and covered by other object, does not receive mouse event.
This patch is a part of upstream commit
f6b3c31561276a6c7afc8fb56ae2e5363772782c.
(evas event handling3 - fix yet more corner cases for clipped objects)
addition:
evas event handling2 - fix incorrect object reporting
(
25d77bc1d24d9fd539c681fa58db976c1ca65051)
evas event handling4- fix more corner cases where bounding is bad
(
e57084d5c5b8b55a4823952ba43c61a15457e369)
Change-Id: I517fd3ac78dd08ef664cb0e1f97b83d0a2c6c40d
Prince Kumar Dubey [Mon, 11 Jul 2016 13:29:04 +0000 (22:29 +0900)]
If there is situation when server is closed prior to client, the client pointer pointing to server must be reset and check must be there during client deletion
Summary: During mobile product testing, we got a crash with callstack which suggest server is getting deleted prior to client. On valgrind analysis we found invalid write operation with same callstack. callstack is pasted in comment section.
Test Plan: create a situation where server got deleted prior to client.
Reviewers: raster, cedric
Subscribers: govi, rajeshps, jpeg
Differential Revision: https://phab.enlightenment.org/D4152
Change-Id: I725b299b327311b8bbdaf75f79cef58c484c1d2c
Joogab Yun [Tue, 27 Dec 2016 01:48:57 +0000 (10:48 +0900)]
[evas_gl] GL program is setup without checking the previous one and
without updating current state in efl (i.e. gc->state.current.prog).
So when shader_array_flush is called, there is a state inconsistency
between GL and efl. Eventually it ends up in using wrong GL program
Change-Id: I140f9661e92232425cfb3e71a0e67380a5bef142
Jee-Yong Um [Fri, 23 Dec 2016 11:43:16 +0000 (20:43 +0900)]
Revert "evas_events: Fix to update clip area."
This reverts commit
c76101eef4baa375e0e63e3471aa3615f7e223cd.
Signed-Off-By: Jee-Yong Um <jc9.um@samsung.com>
Change-Id: Ibc329c68c971187887278dfd34407d21ec15985d
Joogab Yun [Tue, 27 Dec 2016 01:43:41 +0000 (10:43 +0900)]
[evas_gl] change merge mode from MERGE_SMART to MERGE_BOUNDING
- partial rendering issue happend
- My Files -> Device storage -> repeat scroll -> sometimes blink
Change-Id: Ib7f62b38caf3e363b5b25453a9bfd03aa83c8a79
Youngbok Shin [Mon, 26 Dec 2016 11:04:23 +0000 (20:04 +0900)]
edje: skip complex calculation for height with zero width
Getting formatted size from Evas Textblock with zero width
to get only its height is very wasteful in performance aspect.
The patch only ignore formatted size in the above cases.
It'll reduce about 100ms when launching a application which is
consisted with elm_toolbar, elm_naviframe and elm_layout.
@tizen_fix
Change-Id: I1c7fb14a7fc7c1fa2093e624b2af8cb1a98dfcdb
Youngbok Shin [Thu, 10 Nov 2016 07:21:05 +0000 (16:21 +0900)]
edje entry: keep cursor position on mouse down and move
When elm_entry takes focus by a mouse click event,
elm_entry needs to get new cursor position which is changed by mouse down/move/up.
But, elm_entry takes focus before updating the cursor of edje_entry.
It causes an issue about showing proper region.
So, the internal API will be used when elm_entry takes focus.
@tizen_fix
Change-Id: I951c79ed96cd96a4e3c71dbeafe5c51ae432d0af
Woochan Lee [Mon, 26 Dec 2016 06:25:25 +0000 (15:25 +0900)]
ecore_wl_dnd: Change buffer size to get big size of text than before.
The value(102441) was requested by web team. the previous max buffer size was (PATH_MAX = 4096).
There was no limitation In X environment, we should changed this to get text without limitation to
maintain backward compatibility.
@tizen_fix
Change-Id: I2f463ddac9d506c0b2046ec446784055d891c822
Hosang Kim [Mon, 26 Dec 2016 06:40:52 +0000 (15:40 +0900)]
evas_events: add more NULL check logic to source mouse event.
@fix
Change-Id: Iaffad086e1817b74a48dcf7fde337fc37dae7c92
Andrii Kroitor [Thu, 22 Dec 2016 11:15:42 +0000 (13:15 +0200)]
ethumb: rewrite slave input/output
Using stdio instead of low-level read and write because low-level API has
different behaviour on Linux and Windows.
Change-Id: Ibc202ffc3005b29efd3e4eb6a3b9d0c68e8501bd
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Andrii Kroitor [Thu, 22 Dec 2016 16:49:27 +0000 (18:49 +0200)]
ecore_exe: do not try to send 0 bytes
Summary: This action is meaningless when communicating with child process.
Reviewers: barbieri
Reviewed By: barbieri
Subscribers: jpeg, vtorri, cedric, raster
Differential Revision: https://phab.enlightenment.org/D4510
Change-Id: Ia78ab9e33081ab766df75212f005b6bff4a0851a
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Andrii Kroitor [Thu, 22 Dec 2016 12:38:26 +0000 (14:38 +0200)]
ecore_exe_win32: fix typo in input poll thread
Change-Id: Iefbb54961e0fa45d2990a743659ffcb0c2fb7cc9
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Andrii Kroitor [Thu, 22 Dec 2016 12:10:10 +0000 (14:10 +0200)]
ecore_exe: fix send on Windows
Check for bytes written only if more than 0 bytes were sent.
I don't know why some efl code is trying to send 0 bytes, but that works on
Linux and therefore should be fixed on Windows.
Conflicts:
src/lib/ecore/ecore_exe_win32.c
Change-Id: I80903a5ad5be4e6fd3314ed90e81a9b5ca6bcde3
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Andrii Kroitor [Wed, 21 Dec 2016 13:08:58 +0000 (15:08 +0200)]
ecore: fix wait for stdin on Windows
We can't directly wait on stdin handle because it is signaled imediatly even
if there is no new data on input stream.
Change-Id: Iec44c2944b2f638f4dd7b2fb184861f3913910fb
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Andrii Kroitor [Thu, 22 Dec 2016 02:10:42 +0000 (11:10 +0900)]
ecore_exe is broken on Windows
Summary:
T4938
diff from @raster
Aaaargh! There is no other way to get code from diff on phab..
Reviewers: vtorri
Subscribers: vtorri, i.furs, cedric, jpeg, raster
Differential Revision: https://phab.enlightenment.org/D4448
Conflicts:
src/lib/ecore/ecore_exe_private.h
src/lib/ecore/ecore_exe_win32.c
Change-Id: I922742bde863b06660f3c62660aa8bb53672901b
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Vitalii Vorobiov [Mon, 13 Jun 2016 12:31:48 +0000 (15:31 +0300)]
edje_calc: fix interpolate calculation of map colors
There were a problem when while swithcing between states
(because of program running in transition),
there is SIGSEV appearing when first state has
only one point color, and next state more than one.
Basically it looks like this:
....
part {
name: "rectangle";
type: RECT;
description { state: "default" 0.0;
map {
on: 1;
color: 3 255 90 0 255;
}
}
description { state: "moved" 0.0;
map {
on: 1;
color: 0 0 0 255 255;
color: 1 255 0 0 255;
color: 2 255 90 0 255;
color: 3 41 68 59 255;
}
}
}
<and program that change 'rectangle' state with transition>
....
@fix
Change-Id: I078f692cf1e8f959fed68dbf79b38357f6882764
Signed-off-by: Vitalii Vorobiov <vi.vorobiov@samsung.com>
Vyacheslav Reutskiy [Fri, 2 Dec 2016 09:48:14 +0000 (11:48 +0200)]
ecore_exe_win32: fix ecore_exe_send
Write to the child pipe corect data size.
@fix
Change-Id: I96dad8066e5d63a30eaeb466f4799af568458859
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Vitalii Vorobiov [Thu, 22 Dec 2016 10:01:38 +0000 (12:01 +0200)]
Evas_Render: upstream performance improvements port, so object map works now
Based on commits that improve render performance from upstream EFL.
This commit is important for map on Windows, because if there are
objects with map.on: 1, it caused app window to be re-rendered over and
over, so it leads to high CPU load and lots of events didn't worked (so
combobox, scroller, etc turns out to be broken). This is very important
commit that won't recalculate entire canvas every time if mapped object
is already calculated.
Source code was taken (fully or just partially) from lots of render-related
commits of upstream efl:
> Carsten Haitzler <raster@rasterman.com>
Commits are:
ac10a00a,
606e8658,
69cb85aa,
496f2ed4,
47047803,
8dd20ad8,
26189dc9,
7bdcb0b8,
a4aac60a,
d7c6fca6,
ecc556c7,
c844d9ae,
6a462f92,
5fa3815e,
4dee873a,
e3489d5f
> Jean-Philippe ANDRÉ <jp.andre@samsung.com>
Commits are:
f00f4662,
f50b0fed,
bff8dcfe,
4d1c53d9,
81ed0de0,
f00f4662,
ac5ca928
> Hermet Park <hermet@hermet.pe.kr>
Commit:
8233770e
> Cedric BAIL <cedric@osg.samsung.com>
Commit:
ee16c647
> Jean Guyomarc'h <jean@guyomarch.bzh>
Commit:
5b0ddfec
Change-Id: Ifd632d05900ab7766dd66bd5148b5eb469131ec0
Signed-off-by: Vitalii Vorobiov <vi.vorobiov@samsung.com>
Joogab Yun [Fri, 23 Dec 2016 12:29:37 +0000 (21:29 +0900)]
[evas_gl] fix reference counting
- if using elm_image_orient_set in genlist then 3D memory is increased
so, crash happend
sample code:
{
Elm_Genlist_Item_Class *itc = elm_genlist_item_class_new();
itc->item_style = strdup("full");
itc->func.content_get = _content_get_cb;
}
static Evas_Object *_content_get_cb(void *data, Evas_Object *obj, const
char *part)
{
if (strcmp(part, "elm.swallow.content"))
return NULL;
int i = (int)(intptr_t)data;
Evas_Object *img = elm_image_add(obj);
elm_image_file_set(img, img_path[i%6], NULL);
elm_image_preload_disabled_set(img, EINA_TRUE);
elm_image_orient_set(img, ELM_IMAGE_ORIENT_90);
evas_object_size_hint_min_set(img, 500, 500);
return img;
}
Change-Id: I12d0b734650066e851ea612ed68971e08b83fdd4
Viacheslav Reutskyi [Fri, 23 Dec 2016 12:23:19 +0000 (04:23 -0800)]
Revert "edje_cc: add svg file name '.png' if it use as image"
This reverts commit
ff56fa1c2ae214c86822bfc32e3ec56d73e95a54.
Change-Id: I91a219b2ebfa81c6f318da03db8e665e4666632d
Vyacheslav Reutskiy [Tue, 9 Aug 2016 11:22:39 +0000 (14:22 +0300)]
edje_pick: set the correct file name for copied sample
@fix
Change-Id: Iffbae1949e1838d69ec7e8b6b0e1d2787dc2532e
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Vyacheslav Reutskiy [Thu, 22 Dec 2016 16:08:27 +0000 (18:08 +0200)]
edje_cc: add svg file name '.png' if it use as image
This path add notify for user in case when svg use in image section. It
need for clear understanding why vertor image scale like bitmap. And
add to file name new extension '.png'
Conflicts:
src/bin/edje/edje_cc_out.c
Change-Id: Ic7074990ee29bc5fdc249c98b2457839a2af0fde
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Jaehwan Kim [Fri, 23 Dec 2016 04:11:14 +0000 (13:11 +0900)]
edje_data: add the missed vector data.
Change-Id: Iba4f9222d2943d28860535dc8087a25b693d3075
Jaehwan Kim [Fri, 23 Dec 2016 04:10:14 +0000 (13:10 +0900)]
edje_pick: add vector resources into the output file.
Change-Id: I585e0773efd8f82c18a2ad25fd63408b0d274173
Jaehwan Kim [Thu, 22 Dec 2016 01:16:32 +0000 (10:16 +0900)]
edje_pick: include the resources that is only used.
When it pick any groups, it just includes the resources that is used
in those groups.
Change-Id: I981063926ee93563825b49c288b03139b9e6b1e1
Vyacheslav Reutskiy [Wed, 21 Dec 2016 11:57:41 +0000 (13:57 +0200)]
edje_cc: calculate and save the parts count in "edje/file"
Conflicts:
src/bin/edje/edje_cc_handlers.c
Change-Id: I31fe1e7ad32c2dffb58b9959f621d20e5558bd08
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Vyacheslav Reutskiy [Wed, 21 Dec 2016 09:05:22 +0000 (11:05 +0200)]
edje_pick: fix calculation of index for parent and depended groups
Conflicts:
src/bin/edje/edje_pick.c
Change-Id: Ie1ad3db12535a50f0d84ec12bd1b59ae60bc5766
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
sunghyun kim [Thu, 22 Dec 2016 06:38:15 +0000 (15:38 +0900)]
[Extn_GL] Fix build error in other OS
fix build error in other OS ( such as MacOS, Window ..)
Extn_GL can't use without wayland_egl and tizen-remote-surface-client.
so add build define and check this in extn_gl feature.
Change-Id: I28b4b53e91b9f20f1e6f722900749427ade4f28e
Mykyta Biliavskyi [Thu, 22 Dec 2016 09:18:19 +0000 (11:18 +0200)]
Edje_Edit: replace eet_read_direct by eet_read function.
eet_read_direct function doesn't works with ciphered data, but it is
possible that edje file was ciphered. In this case data, that returned
by eet_read_direct always will be NULL.
Change-Id: I0efee8ce1fefe12efa2ac91a71cf2b59018ef5f4
Signed-off-by: Mykyta Biliavskyi <m.biliavskyi@samsung.com>
Joogab Yun [Thu, 22 Dec 2016 06:30:41 +0000 (15:30 +0900)]
1. Evas GL: Direct rendering render to map fix
if target buffer is not window surface then do FBO rendering
2. Evas GL: Disable direct rendering if there is a map
The engine itself (gl_generic) will detect whether the target
surface (canvas where to draw the gl stuff) is the window
backbuffer or not.
Direct rendering should also be disabled even if direct override
flag is set.
Change-Id: Iae31395f3247c9c9fa8a4359594585af95829777
Hosang Kim [Thu, 22 Dec 2016 05:10:04 +0000 (14:10 +0900)]
evas_events: add NULL check logic to source mouse event.
When the proxy's source is updated, the proxy's event list will be
updated, too. But there is no way now. So add defensive code for
protecting segment fault.
@fix
Change-Id: Iaf0b6d8c519c4c478c2b16b923066c59ba69f7e2
JEONGHYUN YUN [Thu, 22 Dec 2016 02:17:01 +0000 (11:17 +0900)]
ecore_wayland: insert ecore_wl_display output sync in ecore_wl_dpi_get()
insert _ecore_wl_display_output_sync logic in ecore_wl_dpi_get()
for recieving correct dpi value in some case
@tizen_fix
Change-Id: I7885551cb356b283d04e47d0ac84bde4d0fde51a
Signed-off-by: JEONGHYUN YUN <jh0506.yun@samsung.com>
Mykyta Biliavskyi [Tue, 20 Dec 2016 06:35:32 +0000 (08:35 +0200)]
Ecore_win32: do not handle WM_SYSKEY* events.
Summary:
In case if WM_SYSKEYDOWN or WM_SYSKEYUP events are handled in
DefWindowProc - system doesn't init event for a system shortcuts.
For example any EFL application on Windows couldn't be closed by
Alt+F4 combination.
@fix
Test Plan:
Launch Elemenatry_test app.
Try to close by Alt+F4 combination.
Reviewers: vtorri, raster
Reviewed By: raster
Subscribers: an.kroitor, bowonryu, cedric, jpeg, #eflete
Differential Revision: https://phab.enlightenment.org/D4477
Change-Id: If56057c1e2f37e7ac6a4d2db016e40762a8f8f34
Signed-off-by: Mykyta Biliavskyi <m.biliavskyi@samsung.com>
Wonsik, Jung [Wed, 21 Dec 2016 02:15:22 +0000 (11:15 +0900)]
[EvasGL] Add env to enable PreRotation
For DDK does not support PreRotation, add ENV to enable Prerotation.
Change-Id: Ibcd15977ed7bb486b164039b83d1a16da6918b00
Shinwoo Kim [Wed, 21 Dec 2016 00:54:18 +0000 (09:54 +0900)]
Revert "eldbus: change to idle_enterer from idler"
This reverts commit
eee1a1d4a0ed83f43508dbe0e23fa2f5cafa0a33.
Change-Id: I35f1fc1056ac78de23b8537688d30fd2cdcb2f0e
Reason: request of profile team
Shinwoo Kim [Wed, 21 Dec 2016 00:52:34 +0000 (09:52 +0900)]
Revert "eldbus: quick fix to solve GEM usage issue"
This reverts commit
eeccf061bb9627f21d75bd7a340c3319bec259f0.
Change-Id: If247254e489b4ae88ddd5de252db803b89fe69ef
Reason: request of profile team
Youngbok Shin [Tue, 20 Dec 2016 11:37:01 +0000 (20:37 +0900)]
edje: fix text marquee issues
Fixed
- RTL text/textblock marquee start position was wrong.
- Internal clipper for marquee was created in a job callback.
It could cause text flickering issue at the outside of part size.
Change-Id: I481be30e3ff5b98893996cc3afb0b98ecba258d6
Jaehyun Cho [Tue, 20 Dec 2016 12:21:05 +0000 (21:21 +0900)]
evas_events: Fix to update clip area.
This patch fixes that image, which is clipped by precise image clipper
and covered by other object, does not receive mouse event.
This patch is a part of upstream commit
f6b3c31561276a6c7afc8fb56ae2e5363772782c.
(evas event handling3 - fix yet more corner cases for clipped objects)
Change-Id: I231fde137a2aae3331747b1bbe661a75a65716e2
sunghyun kim [Tue, 20 Dec 2016 11:55:15 +0000 (20:55 +0900)]
[evas gl] fix bug when application use evas_object_image_save() in gl backend
Usally, evas_object_image_save() in gl backend need to rotate image_data for save.
but if application call evas_object_image_save() after elm_image_orient_set(),
it can make strange calculation by img_object for save().
so it need to be deleted after save image file.
Change-Id: I1ddf7bf3e7a49888f2ca48e499305f808cafe297
DaeKwang Ryu [Tue, 20 Dec 2016 11:04:31 +0000 (20:04 +0900)]
[EvasGL] wrap glGetString. support up to GLES 3.1
EvasGL supports up to GLES 3.1
Change-Id: I518aa0373bfa00427c447e5db12bc02743d4c759
DaeKwang Ryu [Tue, 20 Dec 2016 10:56:27 +0000 (19:56 +0900)]
[EvasGL] add some GLES extensions.
verified with CTS on TM2.
Change-Id: I605d2b52b1708cb5423e3129eeadacd51d95d942
DaeKwang Ryu [Tue, 20 Dec 2016 10:53:54 +0000 (19:53 +0900)]
[EvasGL] update evas_gl.h with latest GLES3 header
Khronos GLES 3 header was updated.
GLhalf is added.
Change-Id: I03c408f52b6e30be4fe577ac4a13054cfa458a72
DaeKwang Ryu [Tue, 20 Dec 2016 10:52:04 +0000 (19:52 +0900)]
[EvasGL] add GLES 3.1 macros which were missed
GLES 3.1 macros were missed.
Change-Id: I9eafa54f3aee02e0d51ba3f86f30bd34a76a0536
Jee-Yong Um [Tue, 20 Dec 2016 09:52:04 +0000 (18:52 +0900)]
edje: convert data structure for text/size class to public
Signed-Off-By: Jee-Yong Um <jc9.um@samsung.com>
Change-Id: I5291d2e25efc286932e942b5542feb9ffce14f85
Carsten Haitzler (Rasterman) [Mon, 11 Jul 2016 13:16:06 +0000 (22:16 +0900)]
evas render2 region - handle if src region is empty correctly on add
fix coverity CID 1313542
Change-Id: I6726d1695e905d74457fab1b59399ebda8245bf6
Signed-off-by: Mykyta Biliavskyi <m.biliavskyi@samsung.com>
Tom Hacohen [Tue, 8 Dec 2015 11:35:25 +0000 (11:35 +0000)]
Evas render2: Fix always false comparisons.
This looks like a classic copy and paste error. It didn't make any sense
before, and it seems like no one was going to take a look at it.
This looks correct. If this breaks something (is render2 even used),
someone should remove those lines. I wasn't entirely sure if I should
just remove them, or correct them, as some of the code looked redundant
anyway.
Change-Id: I805473d988bd2b76b0f77e2f6247fb39a0eae879
Signed-off-by: Mykyta Biliavskyi <m.biliavskyi@samsung.com>
sunghyun kim [Tue, 20 Dec 2016 03:57:36 +0000 (12:57 +0900)]
[Extn_GL] add release request in update callback
Change-Id: Ic18a9c588dbd6dac0826a324c4f8146335b90c76
Wonsik, Jung [Tue, 20 Dec 2016 05:18:59 +0000 (14:18 +0900)]
[EvasGL] Fix build error for PreRotation
Fix build error for PreRotation
Change-Id: I41d848a730621ec2e67d33185b3d4cf6a56ba993
huiyu.eun [Tue, 20 Dec 2016 02:40:11 +0000 (11:40 +0900)]
[EvasGL] Add Pre-Rotation enum type
[Problem] EFL Integration FAILED in VD.
[Solve] Add Pre-Rotation enum types.
Change-Id: If0feb8a55b3f84d5edd6c465db7ec8393cefe7df
Signed-off-by: huiyu.eun <huiyu.eun@samsung.com>
SubodhKumar [Mon, 19 Dec 2016 10:14:30 +0000 (15:44 +0530)]
edje_entry: Show item object when it is inside viewport.
@tizen_fix
Change-Id: I5f55b74204ac5fbeb1b5b0b243103ffe19406e3a
Cedric BAIL [Fri, 16 Dec 2016 23:15:17 +0000 (15:15 -0800)]
edje: handle case when strdup would be NULL.
alternate solution to D4488. Would be better to have a buffer and recycle it.
Change-Id: I599239762d49cd0f459610891f60f4eb3ba2cbe6
Vyacheslav Reutskiy [Fri, 16 Dec 2016 13:30:20 +0000 (15:30 +0200)]
edje_pick: copy images from image sets
@fix
Change-Id: I903b52e3153f2504e16831aecc857b965140d6fc
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Vyacheslav Reutskiy [Fri, 16 Dec 2016 13:27:53 +0000 (15:27 +0200)]
edje_pick: use separate Eina_List vars for different EINA_LIST_FOREACH
Change-Id: Ia897aa7a282cfd922bdd3893e4a31e8f175a3825
Signed-off-by: Vyacheslav Reutskiy <v.reutskiy@samsung.com>
Wonsik, Jung [Fri, 16 Dec 2016 11:23:57 +0000 (20:23 +0900)]
[EvasGL] add to check opengles 1.1
When evas gl runs with Opengles 1.1,
depth and stencil buffer should be created, not depth_stencil buffer.
Change-Id: I5f8ac7c9cac68c1d044cf72e1028154e1be0cd77
Youngbok Shin [Fri, 16 Dec 2016 04:12:59 +0000 (13:12 +0900)]
edje_calc: give +1 width from calculated width of formatted_get
When Edje resizes Textblock object with the width from formatted_get(),
Textblock object will be shrunken on its width and expanded on its height.
It can cause text cutoff issues in multiline Textblock.
@tizen_fix
Change-Id: Id45e9f43c6e1bea488f9b39b1e28377f660d72d9
Jean-Philippe Andre [Thu, 15 Dec 2016 03:09:44 +0000 (12:09 +0900)]
evas: Try to fix masks of masks inside proxies
Yeah it's another fun bit of evas_render.
Issue #1:
Here's the situation:
1. A container (genlist) has a mask, M0.
2. An item I0 inside this container uses a proxy P0 as render object
rather than the item directly (eg. for zooming in/out).
3. An element E0 inside this item has another mask, M1.
Theory:
1. The proxy surface for P0 is rendered, and M1 is applied to
the element E0.
2. The proxy P0 is rendered on the canvas, with M0 applied.
Practice:
1. The element E0 is prepared for rendering, this triggers
a mask subrender for M1.
2. M1 is rendered with M0 as a prev mask, then kept in cache and
not redrawn (no geometry change, etc...)
3. When P0's surface is rendered, M1's surface is the result of M1+M0.
4. When P0 is drawn on screen, we can see the effect of M1+M0 as
P0's geometry might be different from the item's I0.
Solution:
Discard prev masks and force a mask redraw when we're inside a
proxy. Ideally we should detect if the prev mask belongs to the
insides of the proxy or not.
Issue #2:
Somehow evas_render_mapped() messed up the rendering of a simple
mask in some rare situation. As a workaround, we can directly call
the object render function (masks are image objects anyway). This
issue could cause the disappearance of some text in a genlist.
Extras:
Remove use_mapped_ctx flag. Always assume it as true, always pass
a proper context to mapped_render. This simplifies the logic a bit.
Also avoid context free/new when we can just change two values.
This is a backport of (first tested on this tizen branch):
-
3e0f3822f144b254c9d541a32c0d789036453da4
-
1c21cdfdbb782d389818871d1697ece1098e85c2
-
2bed30b3984b4e73d64b99fe05d6574001173497
-
489815457b9e7baeadb0d8050dc25434c77d0fe9
Change-Id: I9ee857f0625a43d281eb945c2f354dd63cc0b9cd
Shinwoo Kim [Fri, 16 Dec 2016 01:10:33 +0000 (10:10 +0900)]
eldbus: quick fix to solve GEM usage issue
Change-Id: Ib4e62b0a5d384f187fab04c85282f46cb620570b
Joogab Yun [Fri, 16 Dec 2016 02:05:31 +0000 (11:05 +0900)]
[ecore_evas_wayland_shm] bug fixed - wayland_tbm_client_create_surface_queue /
destroy_queue
Change-Id: I26ea5b3e7affe6ec3a899b68ce5516f782723dfc
Wonsik, Jung [Thu, 15 Dec 2016 12:10:22 +0000 (21:10 +0900)]
[EvasGL] Fix to OpenGLES 1.1 rendering issue.
If the DDK driver supports depth_stencil buffer,
evas gl can not use GL_UNSIGNED_INT_24_8_OES.
Because OES_depth_texture is supported since OpenGLES 2.0.
So, if application want to use OpenGLES 1.1,
Evas GL should use depth and stencil buffer
Change-Id: I5fe7c01ee5a37fd0f670192e187e5b233f8aa5d6
Mykyta Biliavskyi [Wed, 14 Dec 2016 13:40:49 +0000 (15:40 +0200)]
Fix generate ecore key events on windows.
If Ctrl+number combination pressed/unpressed function
_ecore_win32_event_keystroke_get return NULL. It happens because
ToUnicode WinAPI func fails to prepare unicode string for given scancode
and the keyboard state.
This commit add exception for the case with digits keys. In case if
there no translate
string, but the digit key processed - will be created a normal event as
usual.
Change-Id: Ia99feffdbc2263cc7922b9e79d836704209cfbd8
Signed-off-by: Mykyta Biliavskyi <m.biliavskyi@samsung.com>
Vitalii Vorobiov [Fri, 9 Dec 2016 09:52:33 +0000 (11:52 +0200)]
edje_pick: export aliases and entire dependency list
A lot of cases were missed, cases that are making resulting edj file
useless.
> Now export all aliases
> If group "1" source of group "2" which is source of group "3", it will
> be imported as well!
@fix
Change-Id: I37015a463f8bf787d8e056e9febdeea51a8e8489
Signed-off-by: Vitalii Vorobiov <vi.vorobiov@samsung.com>