The memory usage graph was going up and to the right!
I was told this is always a good thing!
... maybe not this time :)
Hopefully I didn't forget a case. An intense session of
genlist scrolling with masks all over the place and masks
of masks didn't show any glitch, crash or memory leak.
The main difference between 1.12 and 1.13 memory foot print is actually
related to this two pointer to mask. I am wondering if there is not an
issue here also has we do have a duplicated pointer. We have prev_mask
and mask in both cur and prev state of an Evas_Object, but only mask
and prev_mask from the cur state seems to be accessed.
If we can remove two pointers from those state, we should have a decent
win in expedite benchmark. Hopefully 300KB to win there (Close to half
the additional cost in memory).
Summary: It turns out we don't need to handle open & setup of the tty
unless we are Not using systemd. This fixes an issue during
enlightenment restart where we are unable to reopen the tty.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: When we shutdown outputs, we should reset the hardware cursor
state and the saved crtc mode in order to reset things back to normal.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: When we are shutting down a tty, restore the old keyboard
mode that we retain from tty_setup.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: If we are using a touchscreen, there will be no pointer and
thus no cursor_name to use. Trap for that case by checking if
input->cursor_name is valid.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
The flag should be set on the mask itself.
Checking for (x,y) being inside the mask can be an expensive operation,
so further optimization will be required.
there are possible security implications by not checking values of
size fields to see if they are within the data range AND are not 0 or
negative. so do this.
@fix
Macros should be surrounded with do {} while() statements,
otherwise there can be issues, like the macro not requiring a semicolon
after the statement or etc.
@fix
Summary: Now that things are shaping up, we can re-enable the K_OFF
tty mode which will help with buffering input, etc.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: When we do an input_ungrab, we should be sending a mouse up
event so that apps/elm/etc know that the mouse has been released. This
fixes an issue in the Elm Features DnD test report by Daniel Zaoui
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: We should not be setting a 'move' cursor unless we actually
make the call to start a drag
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
have a pointer to outputs
Summary: In order to limit pointer motion, we need to assign an Output
to each Input device. In order to accomplish that, we need to
initialize Outputs first
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: This patch adds support to limit the mouse from moving
outside the output dimensions, and fixes coordinate transformation
calls to use the Output from the Input device
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: This just removes the call to the no longer used/needed
function to update the input device's view of the output.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: Since we now have a reference to the actual output inside the
input device, we no longer need this function to update the Input
device's view of the output size.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: This adds an actual Ecore_Drm_Output pointer to the
Ecore_Drm_Evdev structure so we can check the output size to limit
mouse corrdinates.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary:
Nothing was printed to journald before because sd_journal_send_with_location
expects file and line to be of the NAME=VALUE form.
Change-Id: I382b82b665558fddebae61b7d0a8d4de87638511
Signed-off-by: Vasyl Vavrychuk <vasyl.vavrychuk@globallogic.com>
Reviewers: kuri, cedric
Reviewed By: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1865
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
@fix
this fixes the eina mmap safety handler to only handle what comes
from actual i/o errors (not actual corrupt/hw memory faults or invalid
objects), and if it's not a sigbus eina mmap should handle then pass
it onto the previous handler before eina mmap took over.
Without this missing dependency the file would not show up in the
tarball and the build on MAC OSX would fail.
Thanks a lot to Leif Middelschulte for testing and the report!
When running eo_test_suite we get an memory error: "double free or corruption".
That error arises because we try to free an Eina_Value value in eo_test_value
that doesn't need to freed.
I switched the eina_value_free to wina_value_flush, the proper way of
releasing it.
@fix
Summary: Coverity reports idential code for different branches here.
Not sure WHY that was done, but I commented out the existing if
statement for posterity.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
It is not ready as a public API yet. It needs to run on GeoClue1
and GeoClue2 which will definitely bring API changes as some parts
are tied to mcuh to what GeoClue1 offers right now.
configure event handler
Summary: This patch ports the fix for windows without a min/max size
being set over to the configure event handler (which was also not
taking into account the fact that Some windows may Not have a min/max
property set on them.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
If we are going to test for 10000000 messages we should not print out for each
of them to stdout. The log file for ecore_suite will catch them all. It ended
up being 199MB on my machine just filled with these debug statements.
I left it in in case someone needs it for debugging but we should not have it
on by default.
The header.id was masked before using it as index in the _eo_classes
array and was not unmasked when used.
It hasn't caused segfault (by sheer luck) but was wrong.
@fix
Summary:
Whenever creating a window via elm_win_add(), save netwm_sync_counter values to edata->netwm_sync_counter
by ecore_x_sync_counter_new().
However, never free the values when a window has been destroyed.
So this patch is for free the values in _ecore_evas_x_free().
@fix
Reviewers: cedric, raster, Hermet, devilhorns
Reviewed By: devilhorns
Subscribers: cedric, seoz
Differential Revision: https://phab.enlightenment.org/D1855
Previously Eina_Stringshare printf variant where returning NULL with ("%s", ""). This
is highly disturbing as you would have gotten "(null)", if you asked for ("s", NULL).
To clarify and make things logical, we now return "".
Summary:
There was no conversion to the double quotation mark in the evas_textblock_text_utf8_to_markup function.
The price of the text coming out to API and text coming out to Textblock was different as a result.
As a result, Two text lengths came out differently.
So, I added the exceptional treatment part in the evas_textblock_text_utf8_to_markup function.
@fix
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1853
Summary:
There will be several methods to set orientation in edc, so we have decided to make one big vector,
the main reason is that we use quaternion by default, but look_at, for example, is given as 6 coordinates.
Reviewers: Hermet, cedric, raster
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1693
Summary: This fixes an issue for windows which do not set a min or max
size in the properties. This was discovered when running Enlightenment
in a Wayland-Only scenario, and trying to bring up the settings panel
which would cause an endless loop in calculating the proper window
size due to min/max not being set.
@fix
NB: Thanks to Mike for the help in tracing this ! :)
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: We can make opaque and input region handling simplier if we
just use one opaque & input region per window. Previous code would
always create a new region, set it to the surface, then destroy it.
This code works much nicer in that it hammers the protocol with less
region create/destroy calls.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This should avoid potential crash during shutdown while some Eio thread
were still running. We are still not blocking for more than 30s, so if
an IO is blocked on a dead device, you should be fine.
This enable the possibility to block the main loop until a
specific thread is done. It may trigger still process ending
of other thread during that function call, but not any other
type of event (timer, animator, idler, ... are all ignored).
Well... actually this is not exactly a fix.
It just restores the previous behaviour, and allows AA to
work. As in, it won't draw ugly black lines but properly
blend to transparent.
But there is still a problem:
The image map render function changes the alpha flag on the source
image if AA is enabled or if the map has an alpha color. This is
actually wrong as images forcefully set to not have any alpha
(with evas_object_image_alpha_set(0)) will then not be opaque
anymore.
Right now I can't think of a solution (also I don't quite follow
the entire pipeline in evas map...). Changing the flag will
make some opaque areas transparent. Not changing the flag will
produce ugly artifacts where AA blending should happen. Fix one
bug and the other appears, and vice versa.
This can be tested with the example evas-map-aa and adding an
alpha channel to cube1.png (with gimp for instance) but manually
setting alpha to 0 in the code. Weird stuff will happen (try
playing with the map and pressing I to switch to/from image mode).
Here are only 3 very basic test cases.
One is a dumb set/get to check that image objects can
be passed as clippers.
The other one is a pixel verification test with extremely
basic data (NEAREST scaling and just rectangles). It also
compares text clipping and masking.
The last one performs a very basic verification that masks
of masks work.
The selected op func was not performing the correct operation,
thus producing rendering artifacts. These functions should not
be used anywhere except in case of masking... which was not an
available option earlier.
It was doing (wrong):
dst = interp(mask, src, dst)
Instead of (correct):
dst = dst + (1 - mask) * src
NOTE:
This commit also disables MMX, SSE3 & NEON implementations of
pixel_mask blend operations, since they are also broken.
In case the clipper is a mask object, we should use precise
event masking. By default precise_is_inside is not enabled
because it is expensive, but it should probably be set by
the application when they use masks as clippers.
Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
This implements supports for masking inside evas_render, which
means:
- Render the mask itself into a surface (ALPHA if possible)
- Pass this mask surface to the draw context
- Apply mask recursively in case a masked object is contained
by another masked object.
@feature
Work done by Jaeun Choi, rebased & squashed by jpeg.
This commit introduces changes to the low-level draw functions
of the SW engine considering the existence of an alpha mask image.
Features:
- Font masking (TEXT, TEXTBLOCK),
- Rectangle masking,
- Image masking (all image scaling functions should be handled).
The mask image itself is not yet set in the draw context (see
following commits).
@feature
Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
Summary: The dbus calls to TakeControl and ReleaseControl of a session
are actual Methods that need to be setup and called in order to
operate properly. As such, this commit fixes that issue by using the
proper eldbus method calls, and fixes an issue where shutting down
Enlightenment would lead to "cannot release control" error messages.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: We need to clean tty handlers and close the tty Before we can
call logind_disconnect as that function ends up shuttting down our
dbus connection.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: This fixes an incorrect order with shutdown of drm library.
Sprites are created (during init) before inputs, so they should be
shutdown After inputs are.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
So I've discovered some weird output values after drawing
some text. The destination alpha would become 0xFE even
when the back buffer had a background with 0xFF alpha.
Example:
Dest is 0xff00ff00 (green).
Color is 0xffffffff (white).
Current font alpha is 170 (0xaa).
--> Output was 0xFEaaFEaa instead of 0xFFaaFFaa.
This is because of some slightly invalid calculation
when doing the font masking (mtab[v] = 0x55 above).
Indeed, MUL_256 takes alpha values in the range [1-256]
and not [0-256] as was assumed.
Updated the code for the "inheritance helper" functions and classes
to be in conformance with the new wrapper architecture.
Member variable "parents" of efl::eo::eo_class struct was split in two
member variables, "ancestors" and "parents", the former containing all
Eo class ancestors and the later only the direct bases.
Changed all required files accordingly.
Check to avoid using biding string when the parameter is @out now matches
any variation of "char".
Add default constructor to efl::eo::concrete in order to allow it to be
NULL initialized in the "inheritance helper" classes.
Removed conflicting parent_set member function in the efl::eo::inherit
class.
Removed the "inheritance_extension_function" generator since it is no
longer used.
Overloaded address-of operator for C++ Eolian wrappers for implicit converting
to any ancestor pointer type when getting the object memory address.
Added new grammars to aid the creation of the pointer wrappers
responsible of doing the implicit conversions.
Added a unit test which checks the address-of overload for arbitrary classes.
Added new .eo files to be used in it.
Solved diamond inheritance problem by completely removing inheritance in
the abstract class.
All ancestors are inherited directly in the concrete class.
The algorithm that list the ancestors also avoid repetition.
Now concrete classes define methods too. This helps referring the correct
method directly by the object type (when there are methods with the same
name).
Moved the declaration and definition of constructor methods to the concrete
class, since they should not be used in derived classes.
Updated example that call "color_set".
With this model, if two ancestor classes have a method with the same name,
to call one of them from a derived class you must write the scoped name of
the member function in the wrapper.
In this case, either Evas.Object and Evas.SmartObject have a property
named "color".
Added "from_global" option to the full_name grammar too.
Using a new architecture in the generated files that simplify multiple
inheritance and allows the use of interface types as parameters.
No longer using a hand-crafted C++ header for eo_base.eo.
This file was added to the generation process.
Updated all files that are dependent in the hand-crafted eo_base
C++ header.
Now there is a class that contains the essentials functions of the former
eo::base wrapper and that is used to create (through inheritance) the
"concrete" classes for all Eo generated wrappers.
No longer binding any function or property that are protected, private or
legacy for now.
eolian_type_instance is now a struct with general information for the
whole type.
Added the new header file namespace_generator.hh to hold namespace
generation grammars.
Separated declaration and definition of Eo wrappers methods.
Referring for most objects by its full name (starting at the global
namespace ::) in the generated files.
Created additional helper grammars to avoid code replication.
Removed a TODO comment referring to a doubt about inheritance of
constructor methods.
Added a TODO comment regarding memory allocation for callbacks in static
member functions.
No longer reverting to the native type when the parameter has "@out"
direction.
Added "is_out" member variable to eolian_type class.
With that, generators can keep track of the direction of the parameters.
Also added helper functions "type_is_out" and "type_is_complex".
Created "to_native" functions in eo_cxx_interop.hh to convert binding types
from C++ arguments to the actual C function arguments.
Added static assertions in these functions to enforce compatibility
between the binding and the native type (Required by @out parameters).
Reworked the overload of the "to_c" function for eo::base derivated
objects. Now there is a overload that rely in the compatibility
between the native type and the wrapper, enabling a wrapper to be used as
an output parameter.
Added callbacks to the event EO_EV_DEL for deleting heap-allocated
function objects. This will enforce that the allocated memory,
necessary for maintaining these callable objects alive, is freed at
least when the underlaying Eo object is destroyed.
Functions and constructor methods are now able to have multiple
callback types.
Removed some unused generators, since they become inconsistent now that
functions are required to handle multiple callback types.
Allocating callback objects in the constructor methods instead of
delaying it until the final constructor is called.
Created some generators to avoid code repetition.
Now the generator parameters_forward_to_c replicate the behavior of the
generator parameters_list. The generator parameters_list was, then,
removed since it have a less intuitive name.
Added a TODO comment regarding the behaviour of simple not translating
callbacks that are not followed by a user data pointer.
The generator parameter_type was moved from "parameters_generator.hh" to
"type_generator.hh" for more consistency and convenience.
In convert.cc: Reading functions instead of implements to convert the
Eolian_Class. It avoids creation of methods that do not belong to the
class, in special it avoids calling the default constructor twice in the
generated code.
No longer generating one constructor in the C++ wrapper for each eolian
class constructor, since the correct behavior demands that all constructor
should be called. Now the wrappers have "constructor methods" that
must be called when creating a new object.
Updated test cases and examples to match the new interface. Some class
constructors and some test cases have to be removed since they were based
on the wrong assumption that constructors are mutually exclusive.
Created new generators for forwarding parameters and for looping over
the relevant parameters to the C++ wrapper executing a generic lambda.
Added a TODO comment regarding the call of constructor methods of all
base classes. Currently there is no base type with constructors, so this
situation should be discussed more.
Added a TODO comment regarding the way callback parameters are being
processed.
Added static assertion in the generated header to ensure that the wrapper
have the same size of Eo*, thus grating compatibility between these types.
Added static assertion in the generated header to ensure that the wrapper
have standard layout. This should ensure correct type sizes when dealing
with inheritance.
Created a test to ensure that eo::base and the eolian wrappers have the
same size of a Eo*.
Added eolian_cxx_test_wrapper.cc to the list of test source files in
Makefile_Eolian_Cxx.am.
Removed the eina_init call from the test suite main function to enforce
that each test case creates an instance of the eina::eina_init object.
Added an eina::eina_init object instanciation for each test case that
were lacking this code.
Created a specialization of the eina::accessor for C++ Eo wrappers.
Created an override of the "accessor" methods in the eina::list specialization
for Eo wrappers (in eina_list.hh). It is necessary to create accessors with the
correct type.
Created specializations of std::is_base_of to avoid compilation errors related
with _Eo_Opaque type.
Added new test cases in "eina_cxx_test_accessor.cc" to test the accessor
specialization.
Added efl::eina::eina_init to the existing test cases to correctly do the
on demand initialization required by the tests.
@fix
Argument count limit is now supported. The source has been cleaned up and
unified so that less code repeating is done. The module is now environment
safe, not depending on specific metatable being set on strings. More callbacks
have been added and error messages are now more descriptive.
Summary: Add new function for replacing string with using eina_stringshare_replace() and change _edje_if_string_free() to setup passed string to NULL.
@fix
Reviewers: raster, Hermet, seoz, cedric
Reviewed By: cedric
Subscribers: reutskiy.v.v
Projects: #efl
Differential Revision: https://phab.enlightenment.org/D1835
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: Refine odd lines to delete _edje_timer
Test Plan: Run example using edc programs to change state
Reviewers: raster, cedric, woohyun, jaehwan, Hermet
Reviewed By: Hermet
Subscribers: cedric, seoz
Differential Revision: https://phab.enlightenment.org/D1838
"Eina_Stringshare *" type is interchangeable with "const char *".
The "stealing" constructor from efl::eina::stringshare was incorrectly
using "char *" instead "const char *", preventing it from being used.
curl is dumb. it needs to poll its own fd for data, it gets confused with its own timeouts, and sometimes it forgets that it's supposed to be doing anything.
this fixes:
* connection timeout processing
* connection data processing order
also curl_multi_timeout calls are now done from a single function to handle all of this stupidness in one place
maybe backport after more testing...
Summary:
Without EGL_PLATFORM environment variable, eglInitialize() can be
failed because egl tried to load DRM platform instead of X11 platform and it
tried to handle XDisplay pointer as a gbm_device pointer as well.
The failure seems to be occured especially if the egl was built
with DRM platform as native platform.
This revision can prevent the failure by indicating proper egl platform using
EGL_PLATFORM environment variable.
@fix
Reviewers: gwanglim, jaehwan, seoz
Reviewed By: seoz
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1828
Summary:
Include sound samples, that uses in programs of this group in collections block.
@fix
Reviewers: cedric, Hermet, raster, reutskiy.v.v
Projects: #efl
Differential Revision: https://phab.enlightenment.org/D1824
Summary:
Memory allocated to dir was not freed. so added code to free dir struct
@fix
Signed-off-by: vivek <vivek.ellur@samsung.com>
Reviewers: devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1814
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
eina_inarray_free was not called. So calling it to fix the memory leak.
@fix
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1808
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
eina_inarray_new returns NULL if malloc fails. This check was not performed. So have added the check for the error case and returning NULL if any of it fails.
@fix
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1807
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
While going through eina for understanding, wrote a program to understand
he differences between different eina inarray functions. Thought, this might
be useful for others too, so adding the same.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: devilhorns
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D1803
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
If MSAA was requested, it is very likely that no config was
found (depending on the driver), so we'll try again without
MSAA. Yeah, this might not look very smooth but it should be
better that failing at eglMakeCurrent.