platform/upstream/efl.git
9 years agoevas-gl-x11: Don't call gl_common_error_set if MakeContextCurrent fails
Chris Michael [Wed, 29 Oct 2014 16:07:48 +0000 (12:07 -0400)]
evas-gl-x11: Don't call gl_common_error_set if MakeContextCurrent fails

Summary: glsym_evas_gl_common_error_set used here leads to an implicit
declaration in compiler warnings. Remove the call to that function and
just print out an error message. This is a cleanup function anyway.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoRevert "evas-gl-x11: Fix copy/paste error from Coverity fix"
Chris Michael [Wed, 29 Oct 2014 16:07:03 +0000 (12:07 -0400)]
Revert "evas-gl-x11: Fix copy/paste error from Coverity fix"

This reverts commit fb4ad043f642e7e1ea6db637750bfe5e11891b77.

Revert this. This also leads to an implicit declaration of the
error_set function.

9 years agoevas-gl-x11: Fix copy/paste error from Coverity fix
Chris Michael [Wed, 29 Oct 2014 16:02:26 +0000 (12:02 -0400)]
evas-gl-x11: Fix copy/paste error from Coverity fix

Summary: In fixing a Coverity issue, I copy/pasted a call to
evas_gl_common_error_set without compiling :( Bad me !! This commit
fixes the issue (data was undefined). Since evas_gl_common_error_set can
take NULL as the data parameter, lets use that.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoevas/gl_drm: Add eeze dependency to fix build race condition
Stefan Schmidt [Wed, 29 Oct 2014 15:30:23 +0000 (16:30 +0100)]
evas/gl_drm: Add eeze dependency to fix build race condition

Like the evas drm engine the evas gl_drm engine now depends on eeze for
backlight. Make sure it is setup to find eeze like the drm engine already
does.

9 years agoedje_cc_out: Do not shadow another local variable.
Stefan Schmidt [Wed, 29 Oct 2014 15:07:39 +0000 (16:07 +0100)]
edje_cc_out: Do not shadow another local variable.

9 years agoevas/gl_x11: Do not shadow another local variable.
Stefan Schmidt [Wed, 29 Oct 2014 15:05:42 +0000 (16:05 +0100)]
evas/gl_x11: Do not shadow another local variable.

9 years agoecore_drm: Silence warnings about unused parameters.
Stefan Schmidt [Wed, 29 Oct 2014 15:00:28 +0000 (16:00 +0100)]
ecore_drm: Silence warnings about unused parameters.

9 years agoeina_file: Give TMPDIR presedence over XDG_RUNTIME_DIR when defined
Stefan Schmidt [Wed, 29 Oct 2014 14:51:11 +0000 (15:51 +0100)]
eina_file: Give TMPDIR presedence over XDG_RUNTIME_DIR when defined

XDG_RUNTIME_DIR gives us a nice securty benefit by only allowing the
same user to read wand write files.

In some configuration this is problematic though. If one looks at the
bug report this fixes  for example you can see that there are build
scripts that use a special build user.

The way this has always worked on unix is that you can define your
own tempdir with TMPDIR. When I was making the original change towards
XDG_RUNTIME_DIR I expected some trouble with it but it worked quite
well so far.

To avoid breaking scripts out there and maybe configurations we
haven't tested yet give TMPDIR precedence over XDG_RUNTIME_DIR.

Fixes T1766

9 years agotests/eina: Enable eina_condition_timedwait() now that it should work.
Stefan Schmidt [Mon, 27 Oct 2014 14:48:24 +0000 (15:48 +0100)]
tests/eina: Enable eina_condition_timedwait() now that it should work.

Commit again now that compile problems are solved.

9 years agoeina: Fix _timedwait to handle the given timeout on top of the absolute time
Stefan Schmidt [Mon, 27 Oct 2014 14:47:51 +0000 (15:47 +0100)]
eina: Fix _timedwait to handle the given timeout on top of the absolute time

The pthread man page clearly states that pthread_cond_timedwait() takes an
absolute time parameter. So far we always passed it epoch plus timeout in
seconds. This would never trigger the timeout.

Making sure we fill out timespec struct with the current time before adding
the timeout as offset now. Also handling the t < 0 error case.

Various version worked up together with Jean-Philippe Andre <jp.andre@samsung.com>

This version does not use _eina_time_get directly as this is currently not a
public API of eina. If we decide to make _eina_time_get public we can remove the
extra code here.

Fixes T1701

9 years agoevas-gl-x11: Check return value of glXMakeContextCurrent
Chris Michael [Wed, 29 Oct 2014 14:28:50 +0000 (10:28 -0400)]
evas-gl-x11: Check return value of glXMakeContextCurrent

Summary: Fix Coverity CID1249651 (unchecked return value)

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-tests: Add missing separator
Chris Michael [Wed, 29 Oct 2014 13:22:25 +0000 (09:22 -0400)]
ecore-tests: Add missing separator

Summary: Previous patch for this was causing build issues due to
missing separator.

NB: Totally my fault for not testing the patch !!

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoEcore File : Added test suite for Ecore File module
kabeer khan [Wed, 29 Oct 2014 13:01:59 +0000 (09:01 -0400)]
Ecore File : Added test suite for Ecore File module

Summary:
Added 2 test cases. First to test initialise, shutdown Ecore_File module and second to test all file operation functions defined in the module.

Signed-off-by: kabeer khan <kabeer.khan@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1600

9 years agoecore_input: Added test suite for ecore_input.
Srivardhan Hebbar [Wed, 29 Oct 2014 12:39:56 +0000 (08:39 -0400)]
ecore_input: Added test suite for ecore_input.

Summary:
Added test suite for ecore_input with one test case.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1602

9 years agoecore_con: Added test case to check if _ecore_con_init_count goes below zero.
Srivardhan Hebbar [Wed, 29 Oct 2014 12:24:53 +0000 (08:24 -0400)]
ecore_con: Added test case to check if _ecore_con_init_count goes below zero.

Summary:
_ecore_con_init_count should not go below zero. I've added a test case to test this and also the code to fix this.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1603

9 years agoecore-fb: Adding test suite for ecore_fb
vivek [Wed, 29 Oct 2014 12:14:26 +0000 (08:14 -0400)]
ecore-fb: Adding test suite for ecore_fb

Summary:
Added test suite for ecore_fb with test for ecore_fb_init
w
Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns

Reviewed By: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1594

9 years agoedje_cc_out: update the id of set images.
Jaehwan Kim [Wed, 29 Oct 2014 12:04:37 +0000 (21:04 +0900)]
edje_cc_out: update the id of set images.

Before it removes unused images, it updates ids.
For the set images, it should be done.

@fix

9 years agoecore_evas - ecore_evas's should start withdrawn then normal on show
Carsten Haitzler (Rasterman) [Wed, 29 Oct 2014 10:33:16 +0000 (19:33 +0900)]
ecore_evas - ecore_evas's should start withdrawn then normal on show

this fixes initial state of an ee where you couldn't detect when the
window had been accepted by the wm (goes to normal state from
withdrawn).

@fix

9 years agoEvas filters: Fix COW usage.
Jean-Philippe Andre [Wed, 29 Oct 2014 06:10:03 +0000 (15:10 +0900)]
Evas filters: Fix COW usage.

Thanks cedric for the report.
We'll need a new function (after 1.12) to detect if the current
cow pointer is the default data or not. Maybe eina_cow_default_get()
or maybe something just returning a bool.

9 years agoEvas GL: Fix potential crash
Jean-Philippe Andre [Wed, 29 Oct 2014 03:56:00 +0000 (12:56 +0900)]
Evas GL: Fix potential crash

Welp, it looks like this crash actually happened once even though
it should not. glGetString() seems to have returned NULL somehow.

9 years agoevas: fix increased memory usage in Evas.
Cedric BAIL [Wed, 29 Oct 2014 05:15:11 +0000 (06:15 +0100)]
evas: fix increased memory usage in Evas.

Eina_Cow does memory comparison during standby time. So in expedite as we don't
have any standby time, we end up allocating some vast amount of memory that
will never be compared. It is way simpler to compare that the data are not
going to change before hand. It should also reduce the CPU consumed during
idle time.

This patch save about 1MB of data at peak time in expedite.

9 years agoautotools: fix build with NEON support.
Cedric BAIL [Tue, 28 Oct 2014 23:12:06 +0000 (00:12 +0100)]
autotools: fix build with NEON support.

9 years agoeina: fix overlapped doxygen string output
Tae-Hwan Kim [Tue, 28 Oct 2014 20:22:06 +0000 (21:22 +0100)]
eina: fix overlapped doxygen string output

Summary:
Add top padding to prevent overlapping between "Value:" and below
strings when long macro is used.

Reviewers: raster, cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1582

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: enhance doxygen in eina_error.h
Tae-Hwan Kim [Tue, 28 Oct 2014 20:20:33 +0000 (21:20 +0100)]
eina: enhance doxygen in eina_error.h

Summary:
Arrange grouping
Add @brief for brief description
Add @details for detailed description
Add @note for noted description
Add [in] & [out] for parameters
Fix indentation & Fix typeof

Reviewers: raster, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1588

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: remove reference to API that never made it to a stable release.
Cedric BAIL [Tue, 28 Oct 2014 20:19:10 +0000 (21:19 +0100)]
eina: remove reference to API that never made it to a stable release.

9 years agoeina: enhance doxygen in eina_binshare.h
Tae-Hwan Kim [Tue, 28 Oct 2014 20:18:28 +0000 (21:18 +0100)]
eina: enhance doxygen in eina_binshare.h

Summary:
Add @brief for brief description
Add @details for detailed description
Add @warning for important description
Add [in] & [out] for parameters
Fix indentation & Fix typeof

Reviewers: raster, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1586

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: enhance doxygen in eina_inarray.h
Tae-Hwan Kim [Tue, 28 Oct 2014 20:16:38 +0000 (21:16 +0100)]
eina: enhance doxygen in eina_inarray.h

Summary:
Arrange grouping
Add @brief for brief description
Add @details for detailed description
Add @note for noted description
Add [in] & [out] for parameters
Fix indentation & Fix typeof

Reviewers: raster, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1592

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoEo composite objs: Add a test for the recent comp obj fix.
Tom Hacohen [Tue, 28 Oct 2014 16:19:30 +0000 (16:19 +0000)]
Eo composite objs: Add a test for the recent comp obj fix.

This adds a fragile test for cedric's fix from commit
3550c3808085175cf322670b290bf0c4a70a2fa7.

It assumes that all the added 64 functions are in order and that each
eo_op chain is 32 in length. As long as those don't change, this should
be a decent test.

9 years agoEo comp: Remove dead code following the composite fix.
Tom Hacohen [Tue, 28 Oct 2014 15:19:41 +0000 (15:19 +0000)]
Eo comp: Remove dead code following the composite fix.

This removes dead code resulting from Cedric's fix:
3550c3808085175cf322670b290bf0c4a70a2fa7.

9 years agoecore_wayland: Checking if default wayland socket exists before executing test case.
Srivardhan Hebbar [Tue, 28 Oct 2014 14:27:31 +0000 (10:27 -0400)]
ecore_wayland: Checking if default wayland socket exists before executing test case.

Summary:
If WAYLAND_DISPLAY is not set or if wayland-0 fd doesn't exist in the XDG_RUNTIME_DIR path, then this test case fails. That should not be the case. So made changes in the test case.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1590

9 years agoecore_drm: Added test case calling ecore_drm_shutdown before calling ecore_drm_init.
Srivardhan Hebbar [Tue, 28 Oct 2014 14:26:39 +0000 (10:26 -0400)]
ecore_drm: Added test case calling ecore_drm_shutdown before calling ecore_drm_init.

Summary:
_ecore_drm_init_count should not go below zero. So added test case to verify that.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1589

9 years agoevas - fix new issue with segv when switching to alpha win with gl
Carsten Haitzler (Rasterman) [Tue, 28 Oct 2014 11:09:33 +0000 (20:09 +0900)]
evas - fix new issue with segv when switching to alpha win with gl

evas gl window was freed before new one was created getting shared ctx
to 0 refcount thus freeing everything there - BAD. fix. new issue in
git, not from release

9 years agoEdje_calc: set the image set border again after resize.
Jaehwan Kim [Tue, 28 Oct 2014 06:50:38 +0000 (15:50 +0900)]
Edje_calc: set the image set border again after resize.

_edje_real_part_image_set can change the image of part,
if the part use the image that is set by image set.
If the image is changed, the border should be changed.

@fix

9 years agoEvas: Fix lock usage around async_current (gl preload)
Jean-Philippe Andre [Tue, 28 Oct 2014 02:19:24 +0000 (11:19 +0900)]
Evas: Fix lock usage around async_current (gl preload)

Fixes Coverity:
 - CID 1039361

9 years agoEvas convert: Remove dead code
Jean-Philippe Andre [Mon, 27 Oct 2014 14:09:18 +0000 (23:09 +0900)]
Evas convert: Remove dead code

Woah coverity...
Fixes CID 1039448

9 years agoinotify: Fix safety after read in 3 places
Jean-Philippe Andre [Mon, 27 Oct 2014 13:57:53 +0000 (22:57 +0900)]
inotify: Fix safety after read in 3 places

Fixes Coverity issues:
 - CID 1039565
 - CID 1039566

9 years agoEvas cserve2: Check message size
Jean-Philippe Andre [Mon, 27 Oct 2014 13:16:53 +0000 (22:16 +0900)]
Evas cserve2: Check message size

Fixes CID 1039570

9 years agoEmbryo: Use strncpy instead of strcpy in a few places
Jean-Philippe Andre [Mon, 27 Oct 2014 13:07:36 +0000 (22:07 +0900)]
Embryo: Use strncpy instead of strcpy in a few places

There are still lots of unsafe uses of strcpy, strcat, etc...
Dunno if they are relevant though. Also, assert() may be
compiled out at build time (release mode) so don't rely on it
for safety.

Fixes CID 1039593

9 years agoEvas GL: Set umask before calling mkstemp
Jean-Philippe Andre [Mon, 27 Oct 2014 12:48:37 +0000 (21:48 +0900)]
Evas GL: Set umask before calling mkstemp

Fixes Coverity issues:
 - CID 1039609
 - CID 1039608

9 years agoEvas filters: Prevent division by 0
Jean-Philippe Andre [Mon, 27 Oct 2014 12:37:39 +0000 (21:37 +0900)]
Evas filters: Prevent division by 0

While it really shouldn't happen, let's just add a quick if()
and make Coverity shut up.

Fixes:
 - CID 1191912
 - CID 1191911
 - CID 1191910
 - CID 1191909

9 years agoeo: fix composite to actually work.
Cedric BAIL [Mon, 27 Oct 2014 22:23:45 +0000 (23:23 +0100)]
eo: fix composite to actually work.

So I don't really understand why the code was not there before, but it resulted
in my experiment of making a combobox for elementary just impossible. Now it
work at least.

9 years agoevas/3d: Fix resource leaks found by coverity upstream/1.12.0+beta1+3+g0e736ee
Bogdan Devichev [Mon, 27 Oct 2014 13:57:22 +0000 (14:57 +0100)]
evas/3d: Fix resource leaks found by coverity

CID: 1244183 - 1244187

Reviewers: cedric, artem.popov, Oleksander

Subscribers: stefan_schmidt, cedric

Differential Revision: https://phab.enlightenment.org/D1585

9 years agoecore_wayland: Added test case calling ecore_wl_shutdown before calling ecore_wl_init.
Srivardhan Hebbar [Mon, 27 Oct 2014 12:09:43 +0000 (08:09 -0400)]
ecore_wayland: Added test case calling ecore_wl_shutdown before calling ecore_wl_init.

Summary:
Added a test case to calling ecore_wl_shutdown before ecore_wl_init. The return value of ecore_wl_shutdown should not go below 0.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1584

9 years agoecore_drm: Adding test-suite to ecore_drm.
Srivardhan Hebbar [Mon, 27 Oct 2014 12:00:38 +0000 (08:00 -0400)]
ecore_drm: Adding test-suite to ecore_drm.

Summary:
Adding test-suite to ecore_drm with one test case.

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1583

9 years agorelease: Update NEWS and bump version for 1.12.0-beta1 release v1.12.0-beta1
Stefan Schmidt [Mon, 27 Oct 2014 11:32:01 +0000 (12:32 +0100)]
release: Update NEWS and bump version for 1.12.0-beta1 release

9 years agoevas - sw generic - fix osmesa to not use rgb buffers as evas doesnt know
Carsten Haitzler (Rasterman) [Mon, 27 Oct 2014 10:51:10 +0000 (19:51 +0900)]
evas - sw generic - fix osmesa to not use rgb buffers as evas doesnt know

this fixes incorrect sw osmesa rendering when no alpha cannel is used
on evas gl targets.

9 years agoedje: edje_edit - fix posible resource leak
Vyacheslav Reutskiy [Mon, 27 Oct 2014 10:28:25 +0000 (10:28 +0000)]
edje: edje_edit - fix posible resource leak

Fixes: https://phab.enlightenment.org/T1760

@fix

9 years agoedje: edje_edit - don't unlick(NULL)
Vyacheslav Reutskiy [Mon, 27 Oct 2014 09:21:45 +0000 (09:21 +0000)]
edje: edje_edit - don't unlick(NULL)

Summery: fix, posible to delete file by 'NULL' path (T1760)

@fix

9 years agoeina_semaphore: OSX support
Jean Guyomarc'h [Mon, 27 Oct 2014 08:11:19 +0000 (09:11 +0100)]
eina_semaphore: OSX support

Summary:
OSX only support named semaphores. Eina_Semaphore was actually broken on OSX.
Since OSX 10.10 sem_init() and sem_destroy() (were not implemented) are also marked as
"deprecated", which adds huge pollution to the output when compiling.

Reviewers: cedric, raster, stefan_schmidt

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1576

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: enhance doxygen in eina_accessor.h
Tae-Hwan Kim [Mon, 27 Oct 2014 08:10:16 +0000 (09:10 +0100)]
eina: enhance doxygen in eina_accessor.h

Summary:
Add @brief for brief description
Add @details for detailed description
Add [in] & [out] for parameters
Fix indentation & Fix typeof

Reviewers: raster, cedric

Reviewed By: cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1580

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: documentation enhance - add @brief tag
Tae-Hwan Kim [Mon, 27 Oct 2014 08:08:42 +0000 (09:08 +0100)]
eina: documentation enhance - add @brief tag

Summary: Add @brief for brief description

Reviewers: raster, cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1577

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoeina: enhance doxygen for eina_counter.h
Tae-Hwan Kim [Mon, 27 Oct 2014 08:06:16 +0000 (09:06 +0100)]
eina: enhance doxygen for eina_counter.h

Summary:
Arrange grouping
Add @brief for brief description
Add @details for detailed description
Add [in] & [out] for parameters
Fix indentation & Fix typeof

Reviewers: raster, cedric

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1579

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoEvas GL: string_query returns a const char *
Jean-Philippe Andre [Fri, 24 Oct 2014 11:52:09 +0000 (20:52 +0900)]
Evas GL: string_query returns a const char *

Why cast to/from void* in the first place?

9 years agoEvas GL: Fix typo in an extension's name
Jean-Philippe Andre [Fri, 24 Oct 2014 10:52:21 +0000 (19:52 +0900)]
Evas GL: Fix typo in an extension's name

This could potentially break applications relying on this
typo to check for the extension. The chances are very slim.

9 years agodnd/x: fix type set for case there is no type list exists
Daniel Zaoui [Mon, 27 Oct 2014 06:10:11 +0000 (08:10 +0200)]
dnd/x: fix type set for case there is no type list exists

Summary:
DND does not work in X11 because we cannot set type.
The f8e036d5af3c99d74c3b0f4dcc612c4e0a949e00 causes this.
Since the xdnd type list does not exists at the beginning,
if we always return without setting new property, we cannot set dnd type.
This patch brings dnd work again by correcting the type set operation.

@fix

Test Plan: Try dnd tests in elementary_test.

Reviewers: raster, woohyun, JackDanielZ

Reviewed By: JackDanielZ

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1578

9 years agoevas gl rgba visual/rendering fix try ... again
Carsten Haitzler (Rasterman) [Sat, 25 Oct 2014 11:04:52 +0000 (20:04 +0900)]
evas gl rgba visual/rendering fix try ... again

so this is a re-try at the evas gl destination alpha fix. this is what
cedric tried, but done RIGHT. it required adding an ecore_x call to
create a window with correct visual/colormap. it requires doing
visuals totally correctly all the way from ecore_evas to the evas
gl_x11 core. nvidia drivers are very picky about visuals. i also had
to vid the egl/gles code too to do the same thing. nvidia gles/egl
drivers are also picky, mesa is not. this all requires a lot of code
changes. it's far from trivial

this isn't backported for a few reasons:

1. verify this fix doesn't break for anyone.

i tested:

nvidia glx + egl/gles
intel glx + egl/gles
radeon glx

it needs wider testing. nouveau, fglrx for starters and maybe
some other gles/egl drivers.

2. have some review time
3. time to settle before blasting to stable branches

@fix

9 years agoecore_exe_win32: CreateProcess was called with random flags.
Adrien Nader [Sat, 25 Oct 2014 14:35:06 +0000 (16:35 +0200)]
ecore_exe_win32: CreateProcess was called with random flags.

CreateProcess() has a flags parameter which is being passed
"run_pri | CREATE_SUSPENDED".

The issue lies in the value of run_pri. It is best explained by the
following code somewhere else in the file:

   switch (run_pri)
     {
      case IDLE_PRIORITY_CLASS:
        return ECORE_EXE_WIN32_PRIORITY_IDLE;

The run_pri variable is supposed to store a value from the win32 API while
it was used to store one from the ecore API.

If I recall correctly, the windows one is equal to 32 and the ecore one to
9999. Meaning 9999 ended up used as flags so let's have a look at what that
actually enabled; the reference is "Process Creation Flags" from MSDN
http://msdn.microsoft.com/en-us/library/ms684863%28v=vs.85%29.aspx .

9999 gives 0x0000270F and this matches
  DEBUG_PROCESS | DETACHED_PROCESS | DEBUG_ONLY_THIS_PROCESS
  | CREATE_SUSPENDED | CREATE_NEW_PROCESS_GROUP | CREATE_SEPARATE_WOW_VDM
  | CREATE_UNICODE_ENVIRONMENT | <0x00002000 matches nothing>

Matches nothing? Weird. Well, maybe. Except that I stumbled upon this define
in the mingw-w64 headers:
  #define CREATE_FORCEDOS 0x2000

Mingw-w64 only has a #define, Wine has nothing (they don't do DOS anyway),
but ReactOS has some code about it:
  https://git.reactos.org/?p=reactos.git;a=blob;f=reactos/dll/win32/kernel32/client/proc.c;hb=f60941f8dc775427af04eb0a3c3e4d38160c7641#l3007

Overall the actual set of flags probably made very little sense and wasn't
working very well. :)

I also noticed the following in the mingw-w64 headers:
  #define INHERIT_CALLER_PRIORITY 0x20000

This should be a better match for what seemed to be the original intent of
inheriting the priority. I haven't tested it and it's only documented on
MSDN for Windows CE and similar so I'm really not sure about what it does.

MSDN however mentions that the child processes will have at most the
"normal" priority by default (same as its parent if the parent has less
than the default one) but I'm under the impression a process can raise its
own priority level... Anyway, "NORMAL_PRIORITY_CLASS" will do for now.

With this change and a couple others, elementary's theme builds properly
on Windows (_on_ Windows). I'll assess the usefulness of the other changes
in my tree over the next few days.

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
9 years agoecore_evas: Add since tag to new ecore_evas_gl_drm_new() function
Stefan Schmidt [Fri, 24 Oct 2014 12:29:35 +0000 (14:29 +0200)]
ecore_evas: Add since tag to new ecore_evas_gl_drm_new() function

9 years agoecore-wayland: Add since tag to new ecore_wl_window_xdg_surface_get() function
Stefan Schmidt [Fri, 24 Oct 2014 12:28:02 +0000 (14:28 +0200)]
ecore-wayland: Add since tag to new ecore_wl_window_xdg_surface_get() function

9 years agoevas-gl-drm: Fix initialization from incompatible pointer type
Chris Michael [Fri, 24 Oct 2014 12:37:51 +0000 (08:37 -0400)]
evas-gl-drm: Fix initialization from incompatible pointer type

Summary: The function prototype for eng_context_create has recently
changed in gl_common, however nobody thought it wise to update all
engines using it, so this commit fixes the function for the gl_drm
engine.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoeina_cxx: Include newly added header to unbreak make distcheck
Stefan Schmidt [Fri, 24 Oct 2014 11:02:07 +0000 (13:02 +0200)]
eina_cxx: Include newly added header to unbreak make distcheck

9 years ago[eina-cxx] Fixed EFL_CXX_THROW not defined
Felipe Magno de Almeida [Thu, 23 Oct 2014 20:06:13 +0000 (18:06 -0200)]
[eina-cxx] Fixed EFL_CXX_THROW not defined

Fixed errors of EFL_CXX_THROW not defined when #include'ing specific
eina C++ headers.

9 years ago[cxx] Fixed eina::value get for string
Felipe Magno de Almeida [Thu, 23 Oct 2014 19:38:15 +0000 (17:38 -0200)]
[cxx] Fixed eina::value get for string

The get function for string in eina::value wronfully was freeing the
string returned from eina_value_get. Which crashed applications that
used it.

9 years agoecore_wayland: Making _ecore_wl_shutdown return int instead of EINA_BOOL.
Srivardhan Hebbar [Thu, 23 Oct 2014 14:40:39 +0000 (10:40 -0400)]
ecore_wayland: Making _ecore_wl_shutdown return int instead of EINA_BOOL.

Summary:
_ecore_wl_shutdown should return int instead of EINA_BOOL. So changing the function prototype.

@fix

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1574

9 years agoecore-wayland: Add error message for calling shutdown without calling
Chris Michael [Thu, 23 Oct 2014 14:33:40 +0000 (10:33 -0400)]
ecore-wayland: Add error message for calling shutdown without calling
init

Summary: If someone calls ecore_wayland_shutdown without first calling
ecore_wl_init, then the init count is wrong. Warn the caller.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Add error message for calling shutdown without calling init
Chris Michael [Thu, 23 Oct 2014 14:32:59 +0000 (10:32 -0400)]
ecore-drm: Add error message for calling shutdown without calling init
first.

Summary: If someone calls ecore_drm_shutdown without first calling
ecore_drm_init, then the init count is wrong. Warn the caller.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoRevert "tests/eina: Enable eina_condition_timedwait() now that it should work."
Stefan Schmidt [Thu, 23 Oct 2014 13:13:20 +0000 (15:13 +0200)]
Revert "tests/eina: Enable eina_condition_timedwait() now that it should work."

This has to work until we really fixed up eina_condition_timedwait()

This reverts commit 5c9024424f7102abd6c7d858628b52da6065cf9a.

9 years agoRevert "eina: Fix _timedwait to handle the given timeout on top of the absolute time"
Stefan Schmidt [Thu, 23 Oct 2014 13:10:39 +0000 (15:10 +0200)]
Revert "eina: Fix _timedwait to handle the given timeout on top of the absolute time"

Not ready yet as it uses _eina_time_get which is internal only right now. Compiling
works fine for efl alone as the private header is in teh include search part but it
blows up when compiling elementary.

Need to think a bit more about this. Maybe exposing _eina_time_get as API but that
should wait until after the release.

This reverts commit f0a02a92beff49879eba84b4d4eee265fe156511.

9 years agoecore_drm: Making _ecore_drm_init_count not to go below zero in _ecore_drm_shutdown.
Srivardhan Hebbar [Thu, 23 Oct 2014 13:04:29 +0000 (09:04 -0400)]
ecore_drm: Making _ecore_drm_init_count not to go below zero in _ecore_drm_shutdown.

Summary:
If _ecore_drm_init_count goes below zero, then when next time ecore_drm_init is called, it won't do the initializations which it is supposed to do. So preventing this scenario by not making it go
below zero in _ecore_wl_shutdown function.

@fix

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Reviewed By: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1573

9 years agoecore_wayland: Making _ecore_wl_init_count not to go below zero in shutdown.
Srivardhan Hebbar [Thu, 23 Oct 2014 13:02:11 +0000 (09:02 -0400)]
ecore_wayland: Making _ecore_wl_init_count not to go below zero in shutdown.

Summary:
If _ecore_wl_init_count goes below zero, then there would be problem if someone calls ecore_wl_shutdown 1st and then ecore_wl_init later. So fixing this issue in ecore_wl_shutdown.

@fix

Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns

Reviewed By: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1571

9 years agoecore-drm: Added documentation for various APIs in ecore_drm_output module
vivek [Thu, 23 Oct 2014 12:52:02 +0000 (08:52 -0400)]
ecore-drm: Added documentation for various APIs in ecore_drm_output module

Summary:
Added documentation for ecore_drm_outputs_create, ecore_drm_output_free,
ecore_drm_output_cursor_size_set, ecore_drm_output_size_get functions

@fix

Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns

Reviewed By: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1572

9 years agoevas/cserve2_slave: Fix build race with linking to eo but not depending on it.
Stefan Schmidt [Thu, 23 Oct 2014 11:53:40 +0000 (13:53 +0200)]
evas/cserve2_slave: Fix build race with linking to eo but not depending on it.

I have not been able to reproduce this myself but I have seen a build log where
the binary tries to link to libeo and fails due to the missing file.
A similar problem was "fixed" in 0e4b847deb79c957dbc042ea0c6a98282aead6e0, but
this really makes me wonder where the linking against eo comes from for cserve2
which is not using eo as far as I can see.

9 years agoEina Test : Added test case for convert float in eina_test_value
kabeer khan [Thu, 23 Oct 2014 12:34:14 +0000 (08:34 -0400)]
Eina Test : Added test case for convert float in eina_test_value

Summary: Signed-off-by: kabeer khan <kabeer.khan@samsung.com>

Reviewers: cedric, devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1567

9 years agoecore-drm: Added documentation for various APIs in ecore_drm_device module
vivek [Thu, 23 Oct 2014 12:31:20 +0000 (08:31 -0400)]
ecore-drm: Added documentation for various APIs in ecore_drm_device module

Summary:
Added documentation for ecore_drm_device_fd_get, ecore_drm_device_window_set
and ecore_drm_device_name_get APIs in ecore_drm_device module.

@fix

Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1568

9 years agoecore-drm: Fix formatting
Chris Michael [Thu, 23 Oct 2014 12:28:33 +0000 (08:28 -0400)]
ecore-drm: Fix formatting

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Cleanup backlight and drm device in output_free function code
vivek [Thu, 23 Oct 2014 12:28:14 +0000 (08:28 -0400)]
ecore-drm: Cleanup backlight and drm device in output_free function code

Summary:
Added code to cleanup backlight structure and to close the drm device and
delete it.

@fix

Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1566

9 years agoeina: Add a note about potential use of CLOCK_MONOTONIC
Stefan Schmidt [Thu, 23 Oct 2014 09:12:00 +0000 (11:12 +0200)]
eina: Add a note about potential use of CLOCK_MONOTONIC

Note it down so we can think about this at some later point.

9 years agotests/eina: Enable eina_condition_timedwait() now that it should work.
Stefan Schmidt [Thu, 23 Oct 2014 09:09:55 +0000 (11:09 +0200)]
tests/eina: Enable eina_condition_timedwait() now that it should work.

9 years agoeina: Fix _timedwait to handle the given timeout on top of the absolute time
Stefan Schmidt [Wed, 22 Oct 2014 13:36:57 +0000 (15:36 +0200)]
eina: Fix _timedwait to handle the given timeout on top of the absolute time

The pthread man page clearly states that pthread_cond_timedwait() takes an
absolute time parameter. So far we always passed it epoch plus timeout in
seconds. This would never trigger the timeout.

Making sure we fill out timespec struct with the current time before adding
the timeout as offset now. Also handling the t < 0 error case.

Various version worked up together with Jean-Philippe Andre <jp.andre@samsung.com>

Fixes T1701

9 years agoeet: shutup stupid compiler warning.
Cedric BAIL [Wed, 22 Oct 2014 20:30:29 +0000 (22:30 +0200)]
eet: shutup stupid compiler warning.

9 years agoevas-gl-drm: Cleanup formatting
Chris Michael [Wed, 22 Oct 2014 13:18:05 +0000 (09:18 -0400)]
evas-gl-drm: Cleanup formatting

NB: No functional changes

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Check for valid crtc before calling free
Chris Michael [Wed, 22 Oct 2014 13:13:44 +0000 (09:13 -0400)]
ecore-drm: Check for valid crtc before calling free

Summary: We cannot call drmModeFreeCrtc with an invalid crtc, so check
that it is set inside the output structure before trying to make this
call

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoecore-drm: Cleanup/Unify output_free function code
Chris Michael [Wed, 22 Oct 2014 13:12:29 +0000 (09:12 -0400)]
ecore-drm: Cleanup/Unify output_free function code

Summary: There is no point in having different code in each output
free function (internal one and API exposed one), so let's unify the
code here.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
9 years agoemotion: Add wma as possible playable extension
Stefan Schmidt [Wed, 22 Oct 2014 13:11:25 +0000 (15:11 +0200)]
emotion: Add wma as possible playable extension

This allows to preview (listen) a wma file in efm for example.

Fixes T1568

9 years agoevas: tests only for extension compiled in.
Cedric BAIL [Wed, 22 Oct 2014 12:14:26 +0000 (14:14 +0200)]
evas: tests only for extension compiled in.

9 years agoecore_drm: Added internal function to check if drm device is hotplug device
vivek [Wed, 22 Oct 2014 12:14:26 +0000 (08:14 -0400)]
ecore_drm: Added internal function to check if drm device is hotplug device

Summary:
Added _ecore_drm_output_device_is_hotplug API to check if the
drm device is hotplug device. It returns EINA_TRUE if device is
hotplug else returns EINA_FALSE

@feature

Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns

Subscribers: cedric

Differential Revision: https://phab.enlightenment.org/D1559

9 years agoEo id: Add regression tests to eo id sanity checks
Tom Hacohen [Wed, 22 Oct 2014 10:43:45 +0000 (11:43 +0100)]
Eo id: Add regression tests to eo id sanity checks

Relevant commit: 0b86334a858d9ba88a1642ec5671185292e75ce5

9 years agoEo id: Fix id security checks for invalid objects.
Tom Hacohen [Wed, 22 Oct 2014 10:31:06 +0000 (11:31 +0100)]
Eo id: Fix id security checks for invalid objects.

In some cases, invalid object ids (e.g 0x1) would pass validation and
represent completely different objects (0x80...01). This happened because
we weren't properly checking a given object id is actually an object id.

@fix.

9 years agoEina: Check for errors after eina_tls_new
Jean-Philippe Andre [Wed, 22 Oct 2014 06:48:11 +0000 (15:48 +0900)]
Eina: Check for errors after eina_tls_new

Fix CID 1248913.

9 years agoEvas filters: Remove useless source_count and cow checks
Jean-Philippe Andre [Wed, 22 Oct 2014 05:14:58 +0000 (14:14 +0900)]
Evas filters: Remove useless source_count and cow checks

9 years agoEvas filters: Feed some cow to the image objects
Jean-Philippe Andre [Wed, 22 Oct 2014 02:27:11 +0000 (11:27 +0900)]
Evas filters: Feed some cow to the image objects

Same as for the text objects.
Fixes T1725.

Cedric, please run again your memory test, please.

9 years agoEvas filters: Move filter COW to evas global stuff
Jean-Philippe Andre [Wed, 22 Oct 2014 02:12:33 +0000 (11:12 +0900)]
Evas filters: Move filter COW to evas global stuff

It will be shared between TEXT and IMAGE objects.

9 years agoEvas filters: Use a cow to reduce memory usage of text objects
Jean-Philippe Andre [Wed, 22 Oct 2014 01:53:45 +0000 (10:53 +0900)]
Evas filters: Use a cow to reduce memory usage of text objects

As pointed out by Cedric, the memory usage of basic evas objects
has increased a lot in recent versions of EFL, in part due
to this excessive use of filters data.

This is a partial fix for ticket 1725.

9 years agoEvas GL: Don't pass shared context for GLX with 1.1 API
Jean-Philippe Andre [Wed, 22 Oct 2014 01:44:06 +0000 (10:44 +0900)]
Evas GL: Don't pass shared context for GLX with 1.1 API

Not sure if this is very relevant, since GLX does not support
GL-ES as such, anyways... We should be using the extension
GLX_EXT_create_context_es_profile to create proper contexts.

Note: GLX + OpenGL-ES 1.1 crashes at any function call on my
machine (binary bloc driver), while EGL + GLES1.1 is fine.

9 years agoEvas GL: Remove useless #ifdef/#endif
Jean-Philippe Andre [Tue, 21 Oct 2014 23:43:28 +0000 (08:43 +0900)]
Evas GL: Remove useless #ifdef/#endif

9 years agoEvas GL: Fix unused variable
Jean-Philippe Andre [Tue, 21 Oct 2014 23:20:13 +0000 (08:20 +0900)]
Evas GL: Fix unused variable

CID 1248912

9 years agoeldbus: Fix wrong comments.
Pierre Le Magourou [Tue, 21 Oct 2014 13:53:49 +0000 (15:53 +0200)]
eldbus: Fix wrong comments.

9 years agoevas: let's handle dds also during tests properly.
Cedric BAIL [Tue, 21 Oct 2014 14:47:05 +0000 (16:47 +0200)]
evas: let's handle dds also during tests properly.

9 years agoevas: tests more file format loader.
Cedric BAIL [Tue, 21 Oct 2014 14:46:12 +0000 (16:46 +0200)]
evas: tests more file format loader.

NOTE: It seems that the xpm one is broken. If somebody could give it
a better look and check if it work for him or not, before opening a bug
in phab.

9 years agoevas: don't redefine macro in tgv saver.
Cedric BAIL [Tue, 21 Oct 2014 14:18:55 +0000 (16:18 +0200)]
evas: don't redefine macro in tgv saver.