We don't really need the eo_id most of the time, and when we do, it's
very easy to get it. It's better if we just don't save the eo_id on the
stack, and just save if it's an object or a class instead.
It seems that the idea behind that optimisation, is to save object data
fetching when calling functions implemented by the object's class inside
functions implemented by the object's class. This should be rare enough
not to worth the upkeep, memory reads and memory writes, especially
since for all cases apart of mixins (for which this optimisation won't
work for anyway), the upkeep is more costly than fetching the data
again.
As we no longer have an fd handler to listen on the drm fd, we don't
need this function anymore
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: As we already call drmHandleEvent when we pageflip, we don't
need to be using an fd handler to catch them. This should fix T2791
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: If we already have a pending pageflip scheduled for a given
framebuffer, don't reschedule another one. This also includes a minor
fix when mmap'ing the framebuffer (previously was also mapped
PROT_READ).
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: drmHandleEvent will return 0 on success, or -1 on error. We
should trap for the error case so that we can cleanup any allocated
callback structures.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: In efforts to reduce tearing in the gl_drm engine, implement
support for eglSetDamageRegionKHR to mark parts of a surface as being
damaged.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
In efforts to reduce tearing in the gl_drm engine, find and link to
the eglSetDamageRegionKHR function so we can mark damaged regions of a
surface
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: If we are rendering using gl_drm, then we don't need to be
creating extra software dumb buffers.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: If we are running using gl_drm, then we don't need to create
software dumb buffers on the drm device. Since we may not have the
dumb buffers (only used in software rendering), then we should not
always be checking the framebuffer size against the dumb buffer size.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
The engine setup code already checks if we are disabling vsync
(defaults to on), so we should be setting the Outbuf vsync according
to what the engine info has
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary:
Fix Segmentation Fault when TEXT part getting resized and when it has
some params (for example ellipsis).
Fix T2640
@fix
Test Plan: Refer to T2640
Reviewers: raster, Hermet, seoz, herdsman, cedric, reutskiy.v.v, NikaWhite
Subscribers: stefan_schmidt, tasn, cedric
Maniphest Tasks: T2640
Differential Revision: https://phab.enlightenment.org/D2944
We see segfaults on Jenkins with these two test cases. Better check
before setting them as we had similar problems before on this setup
as XDG_RUNTIME_DIR might never be set.
SSLv3 has been compromised a year ago by what is known as POODLE
(https://en.wikipedia.org/wiki/POODLE). Every major browser have now
dropped support for SSLv3 and distribution are starting to do so also.
It is a good timing for us to do so, especially as it breaks build on
some distribution.
Summary: Support for using EVAS_NATIVE_SURFACE_WL was missing/incomplete in the
wayland_egl engine. This commit addresses that issue so that now the
wayland_egl engine can support both EVAS_NATIVE_SURFACE_OPENGL and
EVAS_NATIVE_SURFACE_WL.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: With linux kernels >= 4.2.x, the gl_drm engine was not
functional. This is due to the egl config returning an improper config
which did not match the expected pixel format. This commit fixes that
issue and gl_drm evas engine works again. Should fix ticket T2807
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: With linux kernels >= 4.2.x, we need to use
GBM_FORMAT_XRGB8888 when creating a new canvas else we end up with a
format mismatch when trying to add the framebuffer
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
While cached surfaces is a topic we're discussing recently,
this code is dead right now, and we will have to redesign the
buffer caching better to handle proxies, maps, smart objects, etc...
When the no_render flag was set on a proxy source, the object would
not be visible, but it would also not render inside the proxy surface,
which completely beats the purpose of this flag. This patch makes
the objects render inside a proxy surface.
On Windows, both backslash and slash can be used as file path
separators. Therefore, it is fixed to consider backslash as a file path
separator as well on Windows.
@fix
if a client added $home in the efreet extra desktops dirs, then
efreetd would detect and nuke cache, exit, causing a restart cycle
forever. this makes efreet simply ignore the errant dir so it can keep
working.
@fix
use env vars to determine module init for ecore-imf so you don't have
things like e locking up trying to init scim when running in wl mode.
do the same for ibus and xim modules too. do the inverse for wayland
imf module but here add also the check for ELM_DISPLAY that wasn't
there (like it is now in the other modules) so it is only initted on
wayland.
@fix
Summary:
According to my understanding of this function, this check of im_old is
redundant, as im_old will never be NULL. For im_old to be NULL, image should be
NULL. But that is checked at line 637. im is assigned image and im is checked for NULL. At line 654 im_old is assigned image and it is not modified till line 673. So this check would always return true and enter if case and would never
enter else case. So removing the redundant code.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Reviewed By: cedric
Differential Revision: https://phab.enlightenment.org/D3233
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Checking for NULL is redundant here, because if cfgs was NULL, then at
line 760 it would fail.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3238
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
This is not an ABI break as we never did provide the broken name symbol, just a typo
in the header that wasn't detected until now.
@fix
Signed-off-by: Vivek Ellur <vivek.ellur@samsung.com>
Reviewers: cedric
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D3248
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: In efforts to debug some non-working drm issues for people,
it would be helpful to know what drivers and versions they are using.
This commit just adds some debug printing for that information
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary; This fixes an issue where maximizing efl/elm apps in Weston
and in Enlightenment would cause extra space to be left around the
window.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: When we are told to resize the canvas, we should also be
resizing the gbm_surface. This should fix the issue of wl_drm module
in E not working when the canvas gets resized (WFM now)
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Summary: If we go to setup a tty and it is already in graphics mode,
then there is no need to exit with a failed setup here. Instead, we
can actually continue to setup the tty.
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
Typos, lack of NULL check, excessive sizeof(type) not matching
the object type, no border set, etc... This all lead to a crash
and then no render (with an error message and then without...).
This also simplifies the implicit loading of ETC1 as ETC2 when
supported by the driver.
@fix
This reverts commit d1f863b272.
My test script was wrong, thus the test for this commit being a pass while it shouldn’t.
Moral: don’t code when you crave some sleep.
i found that intel drm device gives BROKEN timestamps vs system clock
- it is off by about 0.3 to 0.4ms - this means the vsync event is in
the future vs when we actually wake up and do processing - this leads
to bizare timelines and likely odd event and animation handling.
fix this by detecting it and figuring out an average delay and
offsetting events by that in future, but until then, use "now" when
the drm thread wakes up as the timestamp.
@fix
This reverts commit 22b45f220c.
eina_file_cleanup always does an eina_tmpstr_del. This is now capable of doing
double or even triple free in some case.
During a rework this bit got missed and made the build fail if ivi-shell was
enabled. Thanks a lot to aerodynamik for bringing this to our attention and
even pointed out the fix.
Fixes T2798
with eo id indirection on, a nul object is silently ignored as anok
error case (like free(NULL)). but if you turne eoid off in build its
all complaints to here and the black stump. fix this by making the eo
id "off" path match eo id on by making null objects silent.
@fix
so the evas thread renderer didnt START rendering until evas FINISHEd
walking all objects generating a render queue. this means all the cpu
time spend generating commands couldn't allow a parallel thread
actually go and DO the rendering.
this flushes the render thread every render command thus waking up the
render thread to work in parallel to the mainloop generating commands.
this actually means int he traces i see the render thread finished byt
he time evas_render completes thus brinign forward the frame display
by quite a bit.
thanks to evlog for pointing this out.
@fix
removing the klass member meant removing hooks and keeping cache small
but that meant not using it. this meand if the object is not an obj...
i removed the:
call->obj = _eo_class_id_get(call->klass);
line - seemed harmless/pointless. apparently not. so put it back but
use the klass there in local vars and not in call as it's not there
(and not needed).
fix.
in the case of operations which change framespace, rejecting resizes
at this point will cause the canvas to fail at resizing and result in a
partially-rendered canvas; the real canvas geometry must be calculated by
running the entire function in order to determine whether the resize is valid
fixes toggling borderless state of windows
Summary:
Add code to unit test to check if Eolian correctly recognize a struct
name as a struct type when it is used in a method.
Add new method to struct.eo to create this test.
Update struct_ref.c accordingly.
Reviewers: tasn, q66
Differential Revision: https://phab.enlightenment.org/D3213
we pass both the callcache and the op id - both are static and filled
in at runtime, so merge them into the same struct. this should lead to
better alignment/padding with the offset array and the next slot and
op fields, probably saving about 4-8 bytes of rame per method with no
downsides. also pass in only cache ptr, not both cache ptr and opid -
less passing of stuff around and should be better.
Add the edje signal "swallow", "unswallow", "text,set", "text,unset".
In edc file, the part name take a role of source.
If the layout should be changed when any object is swallowed or any
text is set, use this signal.
@feature
Due to the missing break we would fall into the next case here which might lead
to variables overridden with wrong values.
CID: 1261441, 1261440, 1261438
This seems to come from some intention to fetch dh from openssl somewhow but
it was never implemented. fh always stays 0 since its init and thus we can
remove the code it guards.
CID: 1288930
so. clang is wrong. end of story. it complains that i should add
braces to:
static Eo_Call_Cache ___callcache = { 0 };
WRONG. that is correct c99. 100%. you can add more {}'s and init every
field separately like {{0},{0},{0}} etc. or make it 1 or any value -
it doesn't matter... clang complains. clang is wrong. plain and
simple. this warning should just never exist. it is pointless.
but... peolpe won't shut up about clang warnings until i "fool" clang
into being silent by assuming the default 0 value of static storage.
this silences clang
BEWARE! this breaks eo ABI. _eo_call_resolve and _eo_data_scope_get
are 2 of the biggest cpu users in eo. they easily consume like 10-15%
cpu between them on tests that drive a lot of api - like simply
scrolling a genlist around. this is a lot of overhead for efl. this
fixes that to make them far leaner. In fact this got an overall 10%
cpu usage drop and that includes all of the actual rendering, and code
work, so this would drop the eo overhead of these functions incredibly
low. using this much cpu just on doing call marshalling is a bug and
thus - this is a fix, but ... with an abi break to boot. more abi
breaks may happen before release to try and get them all in this
release so we don't have to do them again later.
note i actually tested 4, 3, 2, and 1 cache slots, and 1 was the
fastest. 2 was very close behind and then it got worse. all were
better than with no cache though.
benchmark test method:
export ELM_ENGINE=gl
export ELM_TEST_AUTOBOUNCE=1
while [ 1 ]; do sync; sync; sync; time elementary_test -to genlist;
sleep 1; done
take the 2nd to the 8th results (7 runs) and total up system and user
time. copmpare this to the same without the cache. with the cache cpu
time used is 90.3% of the cpu time used without - thus a win. at least
in my tests.
@fix
in many cases edje dumbly calls eo_do() or evas_object_xxxx on objects
to set up their porperties when the part never uses that thing at al -
eg filters. the obnject never had filters and will not have them n3ext
- why always set filter to null every time? skip when not needed for
speedups.
so this fixes over-zealous calling whihc adds overhead that is not
needed causing battery drain and cpu usage, heat creation etc. etc.
@fix
1. according to svg path specification, path string may or may not contain ',' as the separator
with current parsing logic we were expecting a ',' after each segment.
2. relative cubic bezier parsing was wrong as we were not adding the current value to all 4 points.
3. refactored the parse_pair, parse_six and parse_quad to use same helper function
path1: "M7.279,2h35.442C45.637,2,48,4.359,48,7.271v35.455C48,45.639,45.637,48,42.723,48H7.279C4.362,47.997,2,45.639,2,42.727V7.271C2,4.359,4.362,2,7.279,2z"
path2: "M-2.073-7h36.147C36.796-7,39-4.793,39-2.073v36.146C39,36.796,36.796,39,34.074,39H-2.073C-4.793,39-7,36.796-7,34.072V-2.073C-7-4.793-4.793-7-2.073-7z"
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Memory leak was caused by using the USE macro. So move the macro before
doing any allocation.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3183
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Assigning to NULL has no effect in the function calling this. So changed it to void.
Some compiler complain about this kind of construct. It is better to use the (void)
construct for silencing unused parameter with different kind of configure option.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3180
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Fix memory leak
Delimiter string is being saved using
eina_stringshare_replace without any del or free
when object is deleted.
@fix
Test Plan: NA
Reviewers: cedric, tasn, herdsman
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D3201
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
when running in a wayland compositor, the ideal mode of operation is to
only prepare/send frames when the compositor has finished with the previous
frame
to achieve this, manual rendering can be toggled upon creating and completing
a frame callback, ensuring that a canvas never has multiple pending buffers at
any given time
fix T2784
The ecore_evas_fb module uses ecore_fb_ts_* functions which are only available to EFL internals and not for the general API.
However, EAPI not being defined in ecore_fb_private.h made the symbols not being exported at all, which in return made execution-time linking not finding them and thus failing.
in the case where a surface is created before the compositor binds its shell(s),
a shell surface would never be created
fixes case where internal windows would not create frames in enlightenment
@fix
Summary:
Fix memory leak
Delimiter string is being saved using
eina_stringshare_replace without any del or free
when object is deleted.
@fix
Test Plan: N/A
Reviewers: tasn, herdsman
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D3175
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
The assigned value has no effect outside the function. So removing it.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3181
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
If realloc fails, lst would be NULL. So handling it.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3182
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
Null assignment has no effect in the caller function. So removed it.
Signed-off-by: Srivardhan Hebbar <sri.hebbar@samsung.com>
Reviewers: cedric
Differential Revision: https://phab.enlightenment.org/D3184
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
In a different scaling environment, wrong comparison of min values causes resize issue as original size is compared instead of scaled size.
Signed-off-by: Shilpa Singh <shilpa.singh@samsung.com>
Signed-off-by: Subodh Kumar <s7158.kumar@samsung.com>
@fix
Test Plan:
Create a layout with some min size and swallow a resizable layout inside the layout.
the parent layout will not expand even when the height has crossed its min size.
Reviewers: cedric, tasn, raster
Subscribers: subodh6129
Differential Revision: https://phab.enlightenment.org/D3185
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
In evas table that homogeneous mode is turned off,
the size of items in cells, whose rowspan or colspan is larger than 1
and horizontal or vertical padding exists, are miscalculatd.
T2655
@fix
Test Plan: elementary_test "Table Padding"
Reviewers: Hermet, cedric
Subscribers: cedric, DaveMDS, Hermet
Differential Revision: https://phab.enlightenment.org/D3192
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
we use a bit more memory than we need by having unused fields for
shaders. the old binary data ptrs and size fields we just have not
used for years now as we dont compile in shader binaries anymore as no
living drivers need this anymore. so this removes that with no actual
side-effects.
no shortcuts. we used the same img sharder for pixles as for native
surf. so need new shaders to do the swiszzling. add them all,
generation scripts metadata and enums etc.
@fix (along with previous 3 commits)
this fixes rendering on ppc (bigendian) where we have thnigs swizzled
oddly. not bgra -> argb but rgba -> grab ...
so generate a bigendian shader file and use if on bigendian.
this should fix T2721
it fixes it in the visual screenshots i can get remotely.
so we do a bit of error handling like does a stack fail to allocate,
does setting the tls var fail, have the stack frames been nulled or
not allocated, etc. - these acutally cost every call because they mean
some extra compare and branches, but ore because they cause a lot fo
extra code to be generated, thus polluting instruction cache with code
and cacheline fetches of code that we rarely take - if ever.
every if () and DBG, ERR etc. does cost something. in really hotpath
code like this, i think it's best we realize that these checks will
basically never be triggered, because if a stack fails to grow... we
likely alreayd blew our REAL stack for the C/C++ side and that can't
allocate anymore and has already just crashed (no magic message there -
just segv). so in this case i think this checking is pointless and
just costs us rather than gets us anything.
when a render occurs, frame callbacks must be managed in order to ensure
successful rendering for future frames. the best place to do this is in the
engine here, since this is the lowest-level place which has access to both
the wl_surface as well as the evas rendering state
ref T2784
This reverts commit bd83d4c03a.
adding an animator (and then not managing its ticks) causes the animator
to fire constantly. in this case, it was causing 100% cpu usage and forcing a
compositor re-render for every frame regardless of damages
This causes a significant speed up (around 10% here) and is definitely
worth it. The way it's done lets the compiler cache the value across
different eo_do calls, and across the parts of eo_do. Start and end.
This breaks ABI.
This may look like an insignificant change, but it doubles the speed of
this function, and since this function is called so often, it actually
improves my benchmarks by around 8%.
This breaks ABI in a harmless way, and it will give us the ability to
drastically improve Eo in the future without breaking ABI again, thus
allowing us to declare Eo stable for this release if we choose to.
My previous patch to this piece of code
(37f84b7e96), caused a significant
performance regression. This is such a hot path, that even accessing the
strings when we don't have to slows things down drastically. It makes
more sense to just store it in the structure.
This commit breaks ABI (though most people probably won't even need to
recompile anything else because of the memory layout).
It was discussed on IRC and was decided this is a big enough issue to
warrant a fix during the freeze.
@fix