as per other recent benchmarking, moving rearely run code (in this
case code to init the op etc.) out of the l1 cacheline prefetch inot a
blob of code at the end of the function where we goto and goto back
again should provide decent-ish speedups for the resolv cache in
avoding this code. yes it makes the code less pretty to read but at
this really low level hot path ... evil things must happen to get the
speed we want/need.
As we always set this flag in the drm2_fb_flip function, having this
check here is now pointless.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
before the format was not passed correctly, now the format is passed
correctly to the callback.
Also if a dnd operation was started while a cnp receive call was
going on (this happend because weston-editor failed to close the fd, so
EOF was never sent), then elm_cnp would behave wrong, since the
requestwidget, action, format would be different.
This commits adds api to deal with wayland offers.
It also ports elm_cnp to use the new api.
The selection_get and dnd_drag_get calls are replaced by simply receive
data from the offer.
The Offer object is now also emitted in every Enter,Motion,Drop and
Leave event, so a potential user can prefetch data and display it.
To finish a dnd operation positiv, the user has to call the finish call
before the offer is destroyed
the drag.source is only available if the drag is happing inside the same
client.
So for the case there are two efl apps (A,B) , where A started the dnd and
the B is currently entered by the dnd, then A has the data source, and B
has the offer, if B sets a action on the offer A tried to set it on a
field which does not exist there, this fixes that. It saves the action
to a undepended field.
If buf->priv.fb.fb was NULL the function would have crashed. So
buf->priv.fb.fb can't be NULL. I'm keeping the if(buf->priv.fb.fb)
anyway, but not sure the else case is valid.
Thanks @jiin.moon for the report.
Summary:
Fix parsing error which occurs when lazEDC keyword is combined
with other keyword by period. (like parts.rect)
@fix
Test Plan: Download attached file and run "make"
Reviewers: Hermet, cedric, jpeg
Subscribers: taxi2se
Differential Revision: https://phab.enlightenment.org/D4309
SEGV would happen if the cache was NULL, as the error pointer
was also NULL in some cases (root cause for cache == NULL not
quite known, happens in elm_suite with CK_FORK=no).
Note: simply adding evas_common_init() to evas_init() leads to
a whole new set of issues with CK_FORK=no elm_suite - not good.
This adds a few callbacks to inform applications about
async open and preload:
- load,open
- load,ready
- load,error
- load,cancel
This patch adds a new EAPI as well: elm_image_async_open_set.
This is the only way to use async file open with the legacy APIs
and should be easily matched with whatever EO API we end up using
(be it based on promises or events).
Alter the test cases for make check as they were using the
unstable EO API which I just removed.
Thanks @arosis for the original patch. And sorry for the huge
delay in merging this.
See also: https://phab.enlightenment.org/D4215
@feature
In efl_ui_win, _evas_event_key_cb gives efl_event_callback_call,
though the focused object is not the window itself.
Becuase of this, _propagate_event was called twice for event single
key down.
So, now, _propagate_event returns itself whenever focused object is
not the win but it's called from efl_ui_win's event call.
The problem can be checked with following stpes.
1. elementary_test -> focus
2. Input any directions (ex: Down)
3. Focus movement happens twice for every single input.
of mode
This fixes an issue where gl_drm engine would end up flickering
everytime a frame was being set.
Thanks derek ;)
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This patch modifies our ecore_drm2_fb_flip code to use Atomic/Nuclear
pageflips.
NB: Works perfectly under software drm engine .. some flickering with the
gl_drm engine that needs investigating.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This patch adds code to enable Atomic Modesetting support (via ioctl)
and to fill in Atomic Crtc state during startup.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This code will detect the drm driver name and check that the kernel
itself is new enough to use Atomic Modesetting. This is needed as some
drivers (i915) do not handle Atomic Modesetting propertly without a
new enough kernel.
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: ELM_POPUP_ACTION_BUTTON_MAX is already defined in the included header file, elm_widget_popup.h.
Reviewers: cedric, jpeg, minkyu, Hermet
Reviewed By: Hermet
Subscribers: seoz, jehun.lim
Differential Revision: https://phab.enlightenment.org/D4312
This has been in the making for a very long time. Thanks to Marcel for
reminding me to do it.
What is it?
This is a tool to help application developers debug their apps with
everything Eo. Eo is strict, but not as strict as it can be. Many strict
tests and debug are very expensive to implement, and we have so many
hot-paths that even basic "ifs" to check if debugging is enabled will
add significant overhead to normal running applications. This is why I
created this library. All the expensive tests and bookkeeping should be
wrapped around with "#ifdef EO_DEBUG". With this change, libeo.so is
compiled twice, once normally, and once with this define set (as
libeo_dbg.so). This means that normal eo code will not be affected, but
if you decide to debug your application, all you need to do is:
LD_PRELOAD=/path/to/libeo_dbg.so ./app
Or use the convenient wrapper:
eo_debug ./app
Which will load the debug heavy version.
What's currently there: at the moment, EO_DEBUG enables xref and
data_xref and stricter tests when fetching object data.
In the future, I also plan introducing "zombie objects", which
essentially mean that objects are never really deleted, so you can query
them long after they are gone to get more information on what they were.
So if for example you have an object id that you want to query after the
object has been deleted, you can.
I also plan on having a way to disable/enable certain debug mode
features via env vars, and maybe make the test suite link against this
one instead of the normal one, and possibly add more internal hooks for
the test suite to better inspect internal state?
P.S: The amount of errors spewed out when running it on elementary_test
makes me wish I wrote this earlier. :(
@feature
In commit 7b90e11474 the event name changed but
this module was not updated. Luckily we are building it on Jenkins or it would
have gone unnoticed for a long time.
The events for (async) image download will be redefined by
@cedric later. So, remove them from eo now and only keep their
legacy implementation.
Also, improve elm_test example and add docs.
Summary:
- elm,state,units,visible signal is emitted only when
unit is set, but not when unit_format_func is set.
- Since default unit has been set, this signal is emitted
but signal will not be emitted after unit is set to NULL.
Test Plan:
1. Create a progressbar.
2. elm_progressbar_unit_format_set(obj, NULL);
3. set unit_format_func by elm_progressbar_unit_format_function_set()
and observe elm.text.status part visible.
Reviewers: Hermet, jpeg, cedric
Reviewed By: cedric
Subscribers: D-TAU, eunue, conr2d, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4210
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This patch provides proper parent-child relationship for elm_list and elm_toolbar
while atsapi_mode is set for icon and end element.
This patch is moved from:
bf188e59431ad9c4ca877b2632884d3d430de6b1
Change-Id: Iae855aacf29bef3808a0b5ec159f46cbf0f4539d
Reviewers: stanluk, cedric
Reviewed By: cedric
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4259
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This patch changes the value of 1 to 1ULL in STATE_TYPE macros
to signal the compiler that the value must be considered
as a unsigned long long, it has to be done cause state_set variable
can be longer than 32 bits.
This patch is moved. Orginal commit hash:
a559e473c21c8da7c4e5a87b9c8583ce519cc35e
Change-Id: Ida89f3be185736f61543d37010d0f5cb8d80a751
Reviewers: cedric, stanluk
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4260
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
If ef is null, have to return before _set_material_to_eet_file_from_mesh api.
New momory will be allocated in the api.
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4305
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
alpha_texture need to be set EINA_TRUE to support ETC1+Alpha
@fix
Test Plan:
Create an EDC file with png image with Alpha.
compress image with ETC1 and ETC2.
Observe Alpha is properly applied in both case.
Reviewers: jpeg, Hermet, cedric
Reviewed By: cedric
Subscribers: conr2d
Differential Revision: https://phab.enlightenment.org/D4307
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: There is wrong comparing while using strcmp function. Should be inverted.
Reviewers: cedric, raster, NikaWhite
Reviewed By: NikaWhite
Subscribers: cedric, NikaWhite, jpeg
Differential Revision: https://phab.enlightenment.org/D4291
Ref T4623
v40 bytecode interpreter is official as of freetype 2.7.
The results don't look so good at the moment. The text looks and glyph
positioning seem worse than they were with the previous v35 interpreter.
So, in the meantime we'll keep using v35, just so everything looks
normal again.
Although the v40 is relevant since around 2.6.3, I rather not do any
FREETYPE_MINOR checks in this patch, because distributions might ship
previous versions with the other (v38) interpreter enabled.
this moves a lot of logic that is rare away from the linear/flat asm
path of code so we et fewer l1 cache misses when executing chuncks of
our code. this also reduces the code size and takes some funcs like in
eina_inline_lock_posix.x and makes them real functions to reduce code
size thus better l1 cache usage - only for new/free of locks.
spinlocks, semaphores etc. as these will have no advantage being
inlined but simply bloat out code size instead.
overall this actually reduces efl lib binary sizes 0.4%, so that's a
good sign.
this passes make check and i think i got it right... let me know if i
didn't. i'm also not sure i should just keep the static inlines and
not make the formerly static inline funcs full EAPI ones now... good q.
it's important because some tweens of loaded edje group might use image sets
instead of images, so now making edje_edit API to work with them correctly
@fix
We've been pinning the render thread for every EFL process to core 0.
This is a bit silly in the first place, but some big.LITTLE arm systems,
such as exynos 5422, have the LITTLE cores first.
On those systems we put all the render threads on a slow core.
This attempts to fix that by using a random core from the pool of fast
cores.
If we can't determine which cores are fast (ie: we're not on a
linux kernel with cpufreq enabled) then we'll continue doing what we've
always done - pin to core 0.
libproxy allows various means to configure a proxy, will load from
gnome and kde configuration settings, envvars, macos and even windows
registry.
curl still doesn't use it, but we can make that later.
This test has been failing on Jenkins again and again. After adding the debug
a while ago it now shows that the value is between 0.01 and 0.02 in all cases
I have seen. Relaxing the timeout here a bit to make it pass in situation where
our CI is under load.