so i was just about to add buffer age debugging evlogs to everywhere
doing buffer age and i found... drm gl and wayland gl engines DONT
HANDLE age change like gl_x11! they dont reset to a "full render" for
that frame. well well. this explains bugs i am seeing for sure. very
very bag! i thought this was handled properly. this does lend some
credence to my thoughts about somehow having a single universal buffer
swapping/update calculating and "applying" api inside efl somewhere...
anyway - this fixes this issue for these 2 engines which is a real
necessary fix to be correct.
@fix
Animations are not supported by Exactness. The test screenshots were not
giving any kind of information as they were taken only when the front
was displayed on the screen.
With this change, animations set on the back of the flip can be replaced by
a background, meaning that flip switches can be checked.
so i have been doing some profiling on my rpi3 ... and it seems
memcmp() is like the number one top used function - especially running
e in wayland compositor mode. it uses accoring to perf top about 9-15%
of samples (samples are not adding up to 100%). no - i cant seem to
get a call graph because all that happens is the whole kernel locks up
solid if i try, so i can only get the leaf node call stats. what
function was currently active at the sample time. memcmp is the
biggest by far. 2-3 times anything else.
13.47% libarmmem.so [.] memcmp
6.43% libevas.so.1.18.99 [.] _evas_render_phase1_object_pro
4.74% libevas.so.1.18.99 [.] evas_render_updates_internal.c
2.84% libeo.so.1.18.99 [.] _eo_obj_pointer_get
2.49% libevas.so.1.18.99 [.] evas_render_updates_internal_l
2.03% libpthread-2.24.so [.] pthread_getspecific
1.61% libeo.so.1.18.99 [.] efl_data_scope_get
1.60% libevas.so.1.18.99 [.] _evas_event_object_list_raw_in
1.54% libevas.so.1.18.99 [.] evas_object_smart_changed_get
1.32% libgcc_s.so.1 [.] __udivsi3
1.21% libevas.so.1.18.99 [.] evas_object_is_active
1.14% libc-2.24.so [.] malloc
0.96% libevas.so.1.18.99 [.] evas_render_mapped
0.85% libeo.so.1.18.99 [.] efl_isa
yeah. it's perf. it's sampling so not 100% accurate, but close to
"good enough" for the bigger stuff. so interestingly memcmp() is
actually in a special library/module (libarmmem.so) and is a REAL
function call. so doing memcmp's for small bits of memory ESPECIALLY
when we know their size in advance is not great. i am not sure our own
use of memcmp() is the actual culprit because even with this patch
memcmp still is right up there. we use it for stringshare which is
harder to remove as stringshare has variable sized memory blobs to
compare.
but the point remains - memcmp() is an ACTUAL function call. even on
x86 (i checked the assembly). and replacing it with a static inline
custom comparer is better. in fact i did that and benchmarked it as a
sample case for eina_tiler which has 4 ints (16 bytes) to compare
every time. i also compiled to assembly on x86 to inspect and make sure
things made sense.
the text color compare was just comparing 4 bytes as a color (an int
worth) which was silly to use memcmp on as it could just cast to an
int and do a == b. the map was a little more evil as it was 2 ptrs
plus 2 bitfields, but the way bitfields work means i can assume the
last byte is both bitfields combined. i can be a little more evil for
the rect tests as 4 ints compared is the same as comparing 2 long
longs (64bit types). yes. don't get pedantic. all platforms efl works
on work this way and this is a base assumption in efl and it's true
everywhere worth talking about.
yes - i tried __int128 too. it was not faster on x86 anyway and can't
compile on armv7. in my speed tests on x86-64, comparing 2 rects by
casting to a struct of 2 long long's and comparing just those is 70%
faster than comapring 4 ints. and the 2 long longs is 360% faster than
a memcmp. on arm (my rpi3) the long long is 12% faster than the 4 ints,
and it is 226% faster than a memcmp().
it'd be best if we didnt even have to compare at all, but with these
algorithms we do, so doing it faster is better.
we probably should nuke all the memcmp's we have that are not of large
bits of memory or variable sized bits of memory.
i set breakpoints for memcmp and found at least a chunk in efl. but
also it seems the vc4 driver was also doing it too. i have no idea how
much memory it was doing this to and it may ultimately be the biggest
culprit here, BUT we may as well reduce our overhead since i've found
this anyway. less "false positives" when hunting problems.
why am i doing this? i'm setting framerate hiccups. eg like we drop 3,
5 or 10 frames, then drop another bunch, then go back to smooth, then
this hiccup again. finding out WHAT is causing that hiccup is hard. i
can only SEE the hiccups on my rpi3 - not on x86. i am not so sure
it's cpufreq bouncing about as i've locked cpu to 600mhz and it still
happens. it's something else. maybe something we are polling? maybe
it's something in our drm/kms backend? maybe its in the vc4 drivers or
kernel parts? i have no idea. trying to hunt this is hard, but this is
important as this is something that possibly is affecting everyone but
other hw is fast enough to hide it...
in the meantime find and optimize what i find along the way.
@optimize
this adds eina_freeq api's for c land for deferring freeing of
pointers and can be used a s a simple copy & paste drop-in for free()
just to "do this later". the pointer will eveentually be freed as
eina_shutdown will free the main free queue and this will in turn free
everything in it. as long as the main lo0op keeps pumping things will
og on the queue and then be freed from it. free queues have limits so
if they get full they will clear out old pointers and free them so it
won't grow without bound. the default max is 1mb of data or 16384
items whichever limit is hit first and at that point the oldest item
will be freed to make room for the newest. the mainloop whenever it
finishes idle enterers will add an idler to spin and free while idle.
the sizes can be tuned and aruged about as to what defaults should be.
this also allows for better memory debugging too by being able to fill
freed memory with patterns if its small enough etc. etc.
@feature
test_efl_ui_text was creating a very small window,
with cropped components, etc.
So let's create it with a mininum size to be able to
properly see this test.
it was using old API, updated, but still doesn't work as expected,
lots of warnings from children being left alive, all proxies are
reporting no properties...
when model dies, all children proxies should die as well, otherwise we
get on console:
```
CRI:eldbus lib/eldbus/eldbus_core.c:215 eldbus_shutdown() Alive TYPE_SYSTEM connection
ERR:eldbus lib/eldbus/eldbus_core.c:175 print_live_connection() conn=0x8219230 alive object=0x8276d50 net.connman of bus=net.connman
...
```
Also, all proxies are reporting no properties "(no properties yet)",
likely they are missing to fetch such... even if "--wait" to let it
run, no asynchronous properties are delivered, at least not triggering
EFL_MODEL_EVENT_PROPERTIES_CHANGED.
remove setters that do not make sense, they are set in the constructor
functions and cannot be changed.
define the Eldbus_Connection_Type in .eo, properly type functions
using it.
Transit zoom effect didn't care image fill area in case of manual filling.
This additional logic computes map uvs with regards to the current image fill area.
@fix
Evil implementation of pipe() function uses sockets. Windows functions
"write", "read" and "close" doesn't works with sockets. In this commit
added macros, that replace "read" with "recv", "write" with "send" and
"close" with "closesocket".
@fix
The first patch did not work for maps. This explains why the
original code was so weird. But it actually made sense.
After struggling a bit I realized that we really just need
to shuffle around the pixel position on the window to map that
of the position in the canvas (unrotate it).
Note that compatibility with GLSL-ES (for OpenGL ES) implies
we can not use an array initializer like:
vec2 pos[4] = vec2[4](a,b,c,d);
So the code could probably be optimized. But at least this works.
This patch also avoids calling glGetUniformLocation again and
again.
post event callbacks must return 0 to stop processing when an event is
consumed, and 1 when processing should continue. this is the only place in
all of efl which used this functionality, and it did so incorrectly.
@fix
ref 248b6beeee
ref D2393
so bu5hman pointed out a compile warning from clang that
{ 0x20000, 42711, EVAS_SCRIPT_HAN },
has 42711 exceeding a signed short. true. so this should be an
unsigned short. but this drew me to the fact the whole array could be
shorter by packing this short with the style memeber after it making
them pack into a nicely aligned 4 byte chunk next to the start unicode
value before it, thus chopping 1324 bytes off this table. even worse
the 8192 entry fast table above is using a full 32bits per entry where
they data they store is not even exceeding 7bits, so move this to an
unsigned char saving another 24k. this should reduce cache misses and
memory footprint and binary footprint of the evas .so files etc.
@fix + @optimize
as per mailing list discussion about dropping xcb support now. it
hasn't been complete for a long time, thus not recommented for being
turned on. as we are moving to a wayland world xcbmakes even less
sense. as agreed, time to clean up a bit and remove a distraction as
well as not well tested code. this also updates po's too.
@feature
Clang 3.9.0 told me:
warning: passing an object that undergoes default argument
promotion to 'va_start' has undefined behavior [-Wvarargs]
So I told it to shut up and changed Eina_Bool to int.
Note that edje_edit_state_external_param_set has the same issue.
I'm trying to fix a crash that seems to happens in some very odd
circumstances under stress testing. I have absolutely no idea
what is going wrong... So let's just add some extra safety.
This fixes a a crash on NULL and ensures the EAPI call is
done on an elm_image. Just checking NULL is not good enough
as _efl_ui_image_sizing_eval() doesn't check the type first.
The original solution was really complex and relied on
transforming the current gl_Position into the screen
coordinate in pixels, and map that to the pixel position
in the mask.
This new solution simply pushes the required vertices for
the mask, based on its geometry. This fixes masks when used
in a rotated window.
Why was it so hard to get right? :(
@fix
draw_frame is a legacy feature that draws a very ugly window border
with a white rect and a black text as title bar. This could be
used in wayland when using only the ecore_evas APIs, rather than
elm_win.
Note that the API ecore_evas_draw_frame_set() can not possibly work
as the flag is checked when the ecore_evas is created, so changing
the flag has no effect on existing windows.
Summary:
Datetime widget is module based, so datetime widget is used as base for efl_ui_clock and merged dayselector/clock features into efl_ui_clock.
Added day selection and seconds support in efl_ui_clock.
Added clock features like auto updation of time, stop timer etc in efl_ui_clock.
Added API to enable/disable edit_mode. efl_ui_clock can be configurable to display either only day/date/time or display any two of them or display all three.
Added efl_ui_clock.c and test_ui_clock.c. Theme and Module is added in another patch by Amitesh.
Original author is Yeshwanth <r.yeshwanth@samsung.com>. I have polished this patch a bit and make it compatible with current EFL code.
Test Plan: test_ui_clock
Reviewers: bu5hm4n, tasn, yashu21985, jpeg, cedric, raster
Subscribers: CHAN, woohyun
Differential Revision: https://phab.enlightenment.org/D3938
Finally bring these together as much as possible to avoid future
diversions when bugs are only fixed in one or the other.
There are functional changes - state tracking for client side effects is
now added to the shm engine, some bug fixes for the egl engine have been
brought to the shm engine.
Brings resize code into the common implementation - there is a functional
change. There appears to have been a bug in the egl resize where it
used the same w, h order for portrait evases as for landscape. This was
fixed in shm. I've used the shm variant for the common code.
Moves transparent set into the common implementation - there is a
functional change here - the egl engine now calls transparent_set in
render_updates like the shm engine.
It is probable the this was the intended behaviour all along.
Moves alpha set into the common implementation - there is a functional
change here - the egl engine now calls alpha_do in render_updates like
the shm engine.
It is probable that this was the intended behaviour all along.
These engines are incredibly similar - by sharing the same engine info
structure we'll be able to simplify the wayland ecore_evas bits and
make them much more maintainable.
If a NULL surface gets passed into ecore_wl2_window_pointer_set that
would mean we are unsetting the cursor surface, so we should still be
calling wl_pointer_set_cursor even with a NULL surface.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This patch adds an API that can be used to set a pointer acceleration
profile. This API should be used when addressing T4736
ref T4736
@feature
Signed-off-by: Chris Michael <cp.michael@samsung.com>
not the ideal solution, but we need a decision if we're going to copy
the long code from OpenSSL into our library just to support legacy
users, given that Efl.Net targeted at the future.
Summary:
The style user should be kept when entry's mode is changed.
@fix
Test Plan:
1. Run "elementary_test -to "entry style user"
2. Click "Singleline Mode" toggle
3. See the result
Reviewers: raster, tasn, herdsman, cedric
Reviewed By: cedric
Subscribers: jpeg
Differential Revision: https://phab.enlightenment.org/D4366
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
It was kind of bug state before.
There was no any action when the user gives a focus on an item. then press the enter key.
The item should be set as selected state after that should call clicked signal.
@fix
Test Plan: elementary test MBE sample.
Reviewers: woohyun, bu5hm4n, Hermet
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4368
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
So when mouse / keyboard are present or not it will
generate events.
ecore_evas/wayland will handle that creating or
deleting evas devices for each one (seat device
will be used as parent).
Create or delete evas_devices with class EVAS_DEVICE_CLASS_SEAT
for seats on each ecore_evas created so far.
Initially it's named considering its Wayland id.
Summary:
When Evas is deleted the function _evas_device_cleanup() goes thru all
devices and unref them. Since Evas_Devices are Efl_Input_Device, the user
may still hold a reference to the device (efl_ref()),
thus causing the device to do not be deleted *yet*.
This causes a problem, because when the user calls efl_unref()
and the device itself is deleted the Evas _del_cb
callback will be called and will try to access the Evas_Public_Data from
a deleted object.
In order to avoid this problem all devices will be kept in the devices
list and Evas will unregister the EFL_EVENT_DEL from those devices that
were not deleted.
Reviewers: jpeg, bdilly, barbieri, cedric
Reviewed By: bdilly, cedric
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4369
There seem to be an issue with the ecore_evas_wayland_egl engine when
using them for cursors. The issue is that a black square shows up
behind the mouse pointer. This does not happen with the wayland_shm
engine so use wayland_shm engine for mouse pointers (for now) until
this can be sorted out.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
After the commit 97c9fa64a4 (Remove group_show and group_hide),
some Efl.Ui.Image objects would not render properly. The reason
being that the object call to show() was aborted too early when
the image is still preloading.
This made for really random results as an image preload could
take more or less time, depending on chance.
in the previous commit we're manually upgrading an existing TCP socket
to SSL. It is desired since some protocols need to negotiate, like
STARTTLS and the likes
Now we offer 2 classes that does autostart SSL once the socket is
ready.
As done by write, if we try to read and we can't, then don't give
up. This happens with streams that wraps another, like SSL, may report
there are data to read, but once you try it may not result in enough
data to upper layers.
As we call the same code during ecore_wl2_window_surface_get and
ecore_wl2_window_show functions that basically create the wl_surface
for a given window we can unify that code into one function that can
be called from various places. This also fixes an issue inside
ecore_wl2_window_show where the window surface_id may not have been
getting filled properly.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
the buttons should scroll up and down by 1 px, but nothing move here
That test was still broken in other ways, for example:
play a bit with the "to X Y" buttons and see if it always do the correct thing
there are spu apis to turn subtitles on and off and this should be off
until turned on by api. you really have to be able to choose the
subtitles to display - eg language etc. to use them effectively.
this fixes T4795
@fix
pulse insists on connecting to the xserver on init/setup context if:
1. DISPLAY is set
AND
2. DISPLAY is not empty
so to do a pretty horrible worka-round, empty off the display if its
set so pa doesnt go connect to x and do this if WAYLAND_DISPLAy is set
assuming we'll use wayland then. this is far better than a solid
rock-hard hang. :)
@fix
eolian_cxx tests failed to link because of massively inexistant symbols.
I assume eolian_cxx tests have been working at some point. Maybe they
were gcc-only? I don't get what's going on with gcc and non-existant
symbols. Is there a sugar-coating of some kind? Because when a symbol
does not exist, clang throws you away. Is it because we are only
referring to the eo implementation functions via function pointers?
That's the second time I'm doing a fix like this. Maybe we should change
a bit our linking flags (see --unresolved-symbols in ld)??
Anyway, now we have our symbols. Clang is happy, make check can go on...
Clang raised a massive amount of warnings due to the struct sockaddr_un
not being declared before using it. So, include the header that declares
this structure first.
The ascii circumflex (^) can be typed by pressing twice the ^ key on a
mac keyboard. A single press allows composition (e.g. ^+e = ê).
Pressing ^ twice though, led to a segmentation fault in elementary,
because the result character of the operation (^) appeared in the raw
characters stack, and not in the filtered one.
This is a bit weird, as backtick (`) appears in the filtered keys stack.
@fix
Commit e44c48b904 failed to translate the
deprecated API into the Sierra API... replacing the Command key flags
by the Option key flags. This resulted of Opt+q quitting the program.
@fix
CoreAudio support was initially introduced by commit
62e29b39f4 as an experimental feature.
It played basic sounds, but suffered from drawbacks: it was controlling
the master channel, and therefore any sound played by ecore_audio would
shut down a previous sound (e.g. background music) for the time of the
sound being played. So that wasn't exactly great... Also, after some
time, some hangs have been reported when playing a sound on input. Most
of the time, it translated as a pause in the main loop (see T3797).
More recently (several months ago), ecore_audio with CoreAudio stopped
working during 1.19 development...
So... CoreAudio support on macOS has never been great. And now it's fully
broken. Instead of trying to revive the thing, let just use PulseAudio.
PulseAudio can be installed without any trouble on macOS thanks to
package managers such as Homebrew. Actually, the efl package provided by
Homebrew already provides PulseAudio as a dependency. And it actually
just works very fine. Dropping CoreAudio seems therefore a nice option:
removes unmaintained code, fixes bugs, and add features.
Assigning id->edje to EINA_TRUE in a code path triggered when id->edje
is EINA_FALSE instead of unconditionnaly setting it to EINA_TRUE avoids
to assign id->edje to itself.
This patch changes els_cursor.c to work with the pointer object
created for EFL Wayland Application windows. This allows any EFL cursors
specified by an application to work in Wayland (ex: elm_test Cursor
tests)
NB: While the code is working, there will still be missing cursors due
to those not being included (yet) in the elm theme. Will continue to
add those as time goes.
@feature
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Add an internal elm function we can call from withint els_cursor.c so
that we can set window cursor/pointer images based on what is supplied
by els_cursor.
@feature
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary:
This change removes the necessity to link EFL against the libvncserver
Please ignore the first three commits, they're being reviewed here:
https://phab.enlightenment.org/D4323
Reviewers: bdilly, cedric
Reviewed By: cedric
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4338
Signed-off-by: Cedric Bail <cedric@osg.samsung.com>
This fixes an issue where build would be broken if not building with
Wayland support.
Fixes T4778
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
As we already call _ecore_wl2_input_cursor_update_stop above, we
actually don't need to call it a second time.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
prctl allows us on some platforms to request a thread be woken up more
agressively e.g. due to a timeout bu setting timerslack. since we use
a dedicated thread just for vsync events, this is a very good idea to
ask the kernel to be as exact as possible for this thread as it only
wakes up once per frame (or should only) and accuracy is important. so
use this.
also improve prctl checks to be more explicit in configure.ac and use
these ifdefs in ecore exe too where prctl is used as well.
@feature
As most (if not all) compositors will draw their own "move" cursor
when moving windows around, there's no real need to set this one
client side.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
As we only want to show our custom EFL cursors when the mouse is
actually inside the window, use the _elm_win_mouse_in/out functions to
control pointer visibility.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
In the event that an ecore_evas (using wayland_shm) gets hidden then
the corresponding wl_surface gets destroyed. If evas_norender is
called after that, the outbuf_redraws_clear function follows.
Outbuf_redraw_clear function ends up trying to post updates to the
wl_surface however if that wl_surface is gone, then we end up crashing.
This patch addresses that issue by checking for a valid wl_surface
inside the engine before trying to post updates to that wl_surface.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
If we have no cursor surface, then we don't need the pointer frame
callback anymore so call cursor_update_stop which will delete the
pointer frame callback
Signed-off-by: Chris Michael <cp.michael@samsung.com>
surface
In the event that we have no cursor surface, then we should not be
creating a cursor frame callback.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
As we don't need a pointer or pointer canvas for
ELM_WIN_INLINED_IMAGE, add missing type check and skip creating
pointer canvas there.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
As under the drm engine a softcursor will be created, but not a
pointer canvas, we need to use separate if checks in certain places
like showing, hiding the mouse pointer/canvas.
Signed-off-by: Chris Michael <cp.michael@samsung.com>