This might not be used as over two consecutive runs all the
same buffers should be used. But it could happen if some
parameters in the filter change (eg. blur radius).
Fixes major (GPU) memory leaks. Reuse mode is still leaking.
An odd-sized image scaled down by 2 was losing 1 pixel during the
downscale, and it was not restored after scaling up. The same
happened with downscaling by 4 except the effect was even more
visible.
This meant that a moving snapshot with a large blur would trigger
some really ugly sampling issues if the content below was precise
(such a text).
If the obscured area in a snapshot object changes a lot, do
not try to keep track of it forever. Instead, redraw the filter
over the entire object region, without obscure.
This fixes a performance issue when an opaque window is moved
above a fixed transparent window (the latter has a snapshot with
blur filter).
This dramatically improves the performance and now seems
to give acceptable results. Eventually we need a quality flag
in order to enable this or not. Alternatively, "gaussian" blur
mode would skip this optimization, while "default" would trigger
it.
When the filtered object is an image, without borders, map,
fill info or anything of this sort, then the filter input
buffer is really just a copy of the original image. We can
skip that to save on memory usage and pixel fetches.
This can help with performance when a large region of the
filtered image (eg. snapshot) is fully hidden by an opaque
object. For instance the window border is hidden by the
opaque window content.
This improves over the previous code for handling
snapshot objects and cutouts. Basically any opaque object
above a snapshot should be obscuring it. That is true
unless a crazy filter is applied, or the snapshot object
is itself the source of a map or proxy.
This also uses eina_tiler instead of a custom (and really
bad) algorithm to compute the obscure regions.
This make save() work on snapshot objects, provided the call
is done from inside render_post.
Also, this saves the filtered output of an image, rather than
its source pixels. Any call to save() on a filtered image must
be done from post-render as well.
Fixes T2102
@feature
If we delete the image that was the target surface for gl
rendering, a crash would occur on the next render cycle.
Unlikely but not impossible to trigger from app side.
@fix
This was a poor attempt at improving the performance but
obviously the root cause isn't fixed (too many texel fetches).
Uniform should (theoretically) work better than an attribute
the for loop. Just a guess here.
This also makes GL blur use a float value as radius, allowing
future extension to non-integer blur radii, as well as using
linear scaling as a fast blur approximation.
This factorizes some of the common code for image render
and resolving is_inside (verifying alpha value of a pixel).
This should also be used by save(), as well as buffer_map().
This patch introduces lots of whitespace changes by using return
instead of long if() {} or else {} blocks.
The situation is clearly visible in the Snapshot test case:
increase the radius and a red glow would appear. This is because
the snapshot object was not marked as needing redraw and so had
no pixels under the opaque rectangle.
This will reuse existing buffers by resetting only the minimum
required in the filter context (also reused). Work in progress,
as the actual reuse is disabled for now.
This avoids creating one more FBO and doing one more draw,
by rendering the image input data directly into the input
buffer. This also makes the code common between SW and GL.
This optimizes the GL blur algorithm by reducing the number of
texel fetches (roughly half the number of before this patch). This
works by exploiting GL's interpolation capabilities.
This skips extra tests with image objects that have the
is_opaque() function. That way, if an object is marked as COPY,
rendering of all objects below it will be skipped.
This can dramatically help with performance when flagging a
snapshot object as COPY. This should not be done if a filter is
applied and is meant to blend with the underlying UI.
When using a snapshot object we have access to exactly all
the pixels below it inside the snapshot surface. So, in order
to produce a nice blur, it is necessary to expand this snapshot
and then clip it otherwise the edges will look a bit ugly.
Unfortunately, there are still places where blurs didn't look
so good, as objects below an opaque region would never get
rendered into the snapshot object. So the edges, inside a
snapshot object, around an opaque region would have blur
artifacts.
This fixes that by shrinking the cutout regions by the radius
of the filter. Eg for blur this is the blur radius.
The test case in elm_test can exhibit this fix very clearly:
a red glow would be visible around the opaque rectangle, but with
these changes we instead see the blurry edges of the objects
below the rectangle.
This will be most useful in a special case, where a filter is
used in a window decoration, applied to a snapshot object.
Another optimization that might be wanted is passing a list
of update regions (from the proxy or snapshot).
The filters don't support the obscuring region yet, only some
of the high-level logic is implemented.
If anything in the canvas needs redraw and a snapshot object
happens to intersect with the update region then it was redrawn,
even if all objects below it hadn't changed. This has an insane
performance impact when you apply a blur filter on the snapshot
object. Walking the object list will always be cheaper than
rendering the snapshot!
Note: Added a FIXME comment and forced clean_them to be true
because some odd behaviour happens when breaking with GDB and
the array snapshot_objects keeps growing at each frame (I guess
only if we miss a frame or something like that).
By simply splitting X and Y blurs in two passes we can improve
the performance of the blur filter a lot.
There is still much to be done to make it really fast and nice
looking:
- implement true gaussian blur (not sine-based approximation,
right now the actual blurs look different in SW and GL)
- exploit linear interpolation for R tap instead of R*2+1 taps
(a tap being a texel fetch)
- downscale & upscale large images with large blur radii
Wait a second though, this implementation is not only incomplete
(no support for box vs. gaussian blur), it's also insanely bad in
terms of performance. Small radii may work fine, but at least blurs
render properly in GL with this patch (no more glReadPixels!).
The shader needs a lot of love, including in particular:
- support for 1D box blur single pass
- support for 1D gaussian (or sine) blur
- use linear interpolation and N-tap filters
- separation of 2D blur in two passes (high-level logic)
- potentially separation of large 1D blurs in 2 or more passes
knowing that 2sigma == sigma + sigma when it comes to the gaussian
bell curve.
This one was a bit more... "fun". I had to add a new vertex
attribute and obviously using a VertexAttribPointer led to
incomprehensible crashes. But a simple glVertexAttrib2fv makes
it work like a charm!
A rare option is not handled yet.
This reuses the existing mask infrastructure, but adds a color
flag to use the whole RGBA range, rather than just the Alpha
channel.
Filters are still very slow (glReadPixels and non-optimized use of
GL buffers...), but this is progress :)
This corrects two things:
- the blur filter high-level logic, that lead to reusing some
temporary buffers which contained garbage;
- the versatile gl buffer implementation so that it now properly
switches between the RGBA_Image and the FBO content (yes, this
is insanely slow and inefficient... but it works and that was
the only point).
Alright, so this is a massive patch that is the result of
trying to get rid of unused or poorly implemented classes in
ector. Originally ector was meant to support VG but extend to
things like filters as well. At the moment, ector's design
makes it quite hard to plug in the filters.
For now I think it's easier to implement the GL support for
the filters directly in the engine, where I hope to interfere
as little as possible.
This massive patch keeps only the required minimum to support
a versatile gl buffer that can be mapped, drawn or rendered to (FBO).
It's extremely inefficient as it relies on glReadPixels and lots
of texture uploads, as well as conversions between ARGB and Alpha.
Another type of GL buffer is a wrap around an existing GL image,
but that one is read-only (map or draw: no write map, no FBO).
No, all the filters run fine, and the high-level implementation
(evas_filters.c) does not need to know whether the underlying engine
is SW or GL. One problem though appears with the blending or blurring
of some Alpha buffers, the colors are wrong.
This patch removes more lines than it adds so it must be good ;)
This is an attempt at refactoring the filters code so I can
later implement GL support. This patch adds a few extra changes
to remove avoid calling functions of libevas from the software
engine: use the draw functions from static_libs/draw rather
than evas_common APIs.
Summary:
In previous patch, !w should be replaced with EINA_FLT_EQ(w, 0.0), but it was
replaced with !EINA_FLT_EQ(w, 0.0). This breaks rounded rectangle.
T5291
Test Plan: compile and run attached file
Reviewers: cedric, jpeg, Jaehyun_Cho
Reviewed By: Jaehyun_Cho
Differential Revision: https://phab.enlightenment.org/D4787
The only purpose of this commit was to allow efl 1.19 to be
released on macOS wothout crashing on termination. Time to revert
it and see that we can find a real fix for the next release.
This reverts commit cd5e755951.
ref T5245
If you define either the macro MY_CLASS_EXTRA_OPS for normal
methods/properties or MY_CLASS_EXTRA_CLASS_OPS for class methods
or properties, which contains a comma-delimited list of ops defs
(i.e. EFL_OBJECT_OP_FUNC(...), ...) right before including the
generated my_class.eo.c file, the definitions from these will
be included in the actual class. This can be used to override
certain things in a class internally without exposing it to
Eolian, or for testing/debugging.
Summary:
When _item_filtered_get is called, block and pan re-calculations
happen, When there is no filter applied, we can skip item filtering to
avoid some unwanted calculations
Reviewers: cedric, raster, SanghyeonLee
Reviewed By: raster
Subscribers: rajeshps, Princekrdubey, cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4759
Summary:
The Ecore_Event_Joystick would be not enough information on user side.
Because the button index such as ECORE_EVENT_JOYSTICK_BUTTON_SELECT/START/META,
etc could be mapped to different button for different named joystick.
Test Plan: Using example
Reviewers: raster, cedric, jpeg
Reviewed By: raster
Differential Revision: https://phab.enlightenment.org/D4669
Summary:
The _ecore_con_post_event_server_upgrade() call adds an event to free
the server_upgrade object, svr, via _ecore_con_server_free(svr) so we
should assume srv is freed after it returns. Thus, perform the
pending_slice processing prior to calling it. Otherwise it triggers an
illegal access (USE_AFTER_FREE) error in Coverity.
@fix CID1373485
Reviewers: barbieri
Reviewed By: barbieri
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4785
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Test Plan:
- There are other grammatical errors, but I'm focusing this patch just
on pluralization-related issues to ease review and make it clearer if
I've introduced any inaccuracies.
Reviewers: cedric
Reviewed By: cedric
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4784
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary:
When popup is timeout, notify call hide func, and send time out event to popup.
After popup receiving time out event, popup hide object again,
so hide animation doesn't show.
And notify hide function doesn't send hide signal to block events part,
(but dismiss function send it)
So add that signal.
Test Plan:
1. elementary_test -to popup
2. click first item, "popup+center+text"
3. compare time out event before this patch and after.
@fix
Reviewers: herb, singh.amitesh, Hermet, cedric, raster, jpeg
Differential Revision: https://phab.enlightenment.org/D4780
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
See the previous commit. efl_provider_find() could not
find a canvas because the event object had no parent.
This restores the slider in E's sound mixer.
Note: Input events may expose the evas canvas object
because of parenting. This will eventually need fixing,
as only efl.ui.win should be exposed.
If 8ff2dffe7c7a21278dis backported to 1.19 then this patch
also needs to be merged as well.
Test scenario:
elementary_test -to "ExtScroller"
Try and modify a slider's value with the mouse wheel. Bad things
were happenning, as the flag ON_HOLD was not properly propagated
from the slider to the scroller. This is because the legacy
event_info structure inside the eo event info was not updated
with the new flag value.
By introducing a new EO only API, which is meant to remain
internal, we can use a single legacy info structure, fixing
this issue.
Note: In the future this API needs to be internal, not protected.
@fix
Prerequisites:
Disable scroll animation (in elementary_config)
Test case:
elementary_test -to Scroller
1. Scroll with the mouse wheel. Scrolling will not loop or anything.
2. Enable loop in Y, scroll to the top and to the bottom, verify
that scrolling loops fine and shows the last and first page in
full.
3. Enable page snapping, and repeat 2.
FIXME: Page snapping doesn't do much if you use thumbscroll or
drag the scrollbars.
FIXME: In the test case, Y +/- 1 is useless as the scroller snaps
to the pages even without snapping enabled.
@fix
The freeze property is a set() only but could internally
be reset to false, after dragging a vertical or horizontal slider.
Test scenario:
elementary_test -to scroller
Click Freeze, test the mouse wheel (can't do anything), drag a
scroller side bar, test the mouse wheel again.
Before this patch, the scroller would scroll. After the patch, the
scroller remains fixed, respecting the value of freeze.
FIXME: It is possible that the proper fix would be to disable
bars drag during freeze, but that is not the case currently.
NOTE: freeze, hold, movement_block, lock_x/y have very similar
meanings. The doc really needs clarification here, and some
property might be removed. Also, freeze and hold have no
getter, only a setter. drag_lock_x/y is part of elm_widget,
and not specific to scrollers.
If the scroll animation is disabled, we ended up with an
immediate call from inside a post-event callback to modify
the canvas geometry which led to feeding events. Since
99d21f6d9c and 54e5841b2f it is basically forbidden
to modify the canvas or feed events from the post-event cb.
This is because feeding events from inside the post-event
callback can break the logical order of operations between
post-event cb and event cb.
Note: This also implements no-animation scrolling for page
scroll, in case scroll animation is disabled (unifying the
code did that).
Fixes T5289 (abort inside E)
Test scenario:
elementary_test -to "Scroller 2"
Use the mouse wheel to scroll inside the horizontal scroller
(the one with many "...Horizontal scrolling..." buttons). This
scroller should scroll horizontally. When reaching the end of
this scroller, the main vertical scroller should take over
and scroll vertically, but only after a 0.5s timeout has passed.
Before this patch, you could wait forever and scrolling inside
the horizontal scroller would never trigger a scroll in the main
vertical scroller, despite reaching the end point.
In 1.18 both the main and the horizontal scrollers scroll
simultaneously. The imbricated vertical scrollers seem to work
as designed, but not H inside V.
@fix
Summary:
Before, rsc->current_ctx is always same with ctx.
So checking context change was meaningless.
From now, it has meaning.
Test Plan: App call evas_gl_make_current more than twice in pixels callback. Those surfaces are indirect rendering surface.
Reviewers: jpeg, dkdk, wonsik
Reviewed By: jpeg
Subscribers: cedric
Differential Revision: https://phab.enlightenment.org/D4773
Summary:
There's problem in Tizen3.0.
1. Clip set mask_obj to obj for masking.
2. Unset mask_obj from obj, and del mask_obj.
3. obj has clip.mask still. So obj is trying to do mask_subrender() for freeed mask_obj.
So reset clip.mask to NULL, If there isn't clipper.
Now, there's no routine for reseting clip.mask when clipper object is freed. isn't it?
Actually I'm not sure that clip.mask=NULL should be there as this patch.
Test Plan: Tizen3.0 wearable
Reviewers: cedric, raster, wonsik, jpeg
Subscribers: scholb.kim, dkdk
Differential Revision: https://phab.enlightenment.org/D4721
Signed-off-by: Jean-Philippe Andre <jp.andre@samsung.com>
Theoretical bug that may become real in the coming days - ecore_imf's
wayland module has the interface for the text input protocol. The
imf module gets unloaded after destroying the protocol object but
before receiving all events from the server destined for it.
Currently that's fine, but I've posted some wayland patches for
review that require the interface to remain present while the text
input object is a zombie. These patches cause some EFL apps to
crash on shutdown when they unmap the wayland imf module.
This moves the wayland interface bits into ecore_wl2 which doesn't
get unmapped.
Since 9b7ac51943 evas map tries to avoid recalculating
stuff when the map parameters have not changed. Unfortunately the
code in elementary_test -to "Flip Page" was badly written. It was
modifying a constant's internal value (after ugly cast). So the
memcmp() and all other checks would return successfully, as the
exact same pointer was being compared to itself.
So, I've fixed the comparison by adding some forgotten parameters
(perspective) but most importantly I fixed the map API usage in the
test case.
There are reports of crashes when y < 0. This case seems
abnormal in case of filters, as I don't know how to reproduce it,
but it's happened.
Thanks Youngbok Shin for the report.
@fix
An unfortunately very common misuse of clip is as follows:
- Layout A is created (edje object / elm_layout)
- Object B is swallowed inside A
- Clipper C is set to clip B
This is a invalid usage, as layout A takes control over the clip
property of B (just like it does for geometry, visibility, color...).
Since 75ec3a7338 edje_recalc resets the clip at every calc
loop, as it can change between states.
In the past, edje_recalc did not reset the clip so anyone could
(wrongly) swallow an object and then change its clip from C to modify
its color, mask it, blend it, etc... Even though this was not proper
use of the API, this is not very clearly documented, and since it
worked, it has been (ab)used a lot already.
The result now is that a clipper set from C will become visible
as an opaque white rectangle covering the entire UI. Booh.
This patch is a workaround that should have no impact on well
written applications. As a bonus this avoids an extra call to
clip_set() from edje.
@fix
Summary:
If the last item before ellipsis item has bigger width than its advance,
evas_common_font_query_last_up_to_pos() function can find wrong ellipsis position.
When Evas finds a position for non last item, Evas must care about additionally
available space for glyph's width of the given x position.
ex) the last item's glyph before ellipsis item has a tail to draw above the ellipsis item.
@fix
Test Plan:
Test case will added as comment.
(Becasue of font license problem.)
Reviewers: herdsman, raster, jpeg, woohyun
Subscribers: cedric, Blackmole
Differential Revision: https://phab.enlightenment.org/D4727
if you set config via environment variables then sa config
update/reload may end up overriding these again whenever it is updated
and thus may result in scaling or other things suddely changing
@fix
just match the ref with the drop anddon't also manually free the image
as this leads to accessing of freed memory which is not a good thing.
this should fix T5374
@fix
for dialers we should also monitor "resolved" from inner TCP socket
and emit that ourselves, letting people know that we have an IP
address.
this is important for ecore_con_legacy.c, since the svr->ip is only
decoded and stored once when this signal is emitted.
Since efl_net_ssl_context is immutable for a dialer and we create the
dialer with the context, the SSL cases uses a trick to postpone dialer
creation using a job, then it allows one main loop iteration for the
user to call various ecore_con_server_ssl_*() methods.
However this breaks immediate ecore_con_server_send() after
ecore_con_server_connect() as used to be allowed and used by
azy/erssd.
Most people wouldn't notice that, since the most common case is to
either use ecore_con_url (which uses cURL and a complete different
code path) OR they would wait for ECORE_CON_EVENT_SERVER_ADD prior to
sending data.
Nonetheless it was a compatibility issue and must be fixed.
Fixes T5339
We should be setting this to the context version we understand, not
the highest version supported by the library.
From Daniel Stone's recent intel-gpu-tools commit fixing the same bug:
With libdrm 2.4.78, setting a higher context version than 2 will attempt
to call the page_flip_handler2 vfunc if it was non-NULL, which being a
random chunk of stack memory, it might well have been.
The efl_ui_win overrides _elm_interface_atspi_component_extents_get to give
correct value if screen_coord is EINA_FALSE which means relative position.
The efl_ui_win has given its object geometry value + ecore evas geometry value.
The object geometry value was equal to the ecore evas geometry value, so the
relative position was not correct(twice of ecore evas geometry value).
old_clipper = evas_object_clip_get(objA);
evas_object_clip_unset(obj_A);
evas_object_clip_set(obj_A, new_clipper);
evas_object_del(old_clipper);
when old_clipper deleted, _clipper_del_cb unset the current new_clipper of obj_A.
Summary:
By 403b0ecfa6 the coreaudio
support was dropped.
So 'HAVE_COREAUDIO' define is invalid.
Reviewers: jpeg
Subscribers: cedric, woohyun
Differential Revision: https://phab.enlightenment.org/D4732
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Summary: Pointer ed->callbacks may have NULL value in callback add functions.
This reduce the chance of continue to kind of work for longer in case of memory
constrain. Maybe using Eina macro would be better.
Reviewers: jpeg, jypark
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4761
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Reviewers: jpeg, jypark
This could lead to a crash as we were checking against NULL on
the wrong pointer.
@fix
Subscribers: cedric, jpeg
Differential Revision: https://phab.enlightenment.org/D4762
Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
Jean-Philippe Andre
Jeeyong Um
Cedric BAIL
Stefan Schmidt
Daniel Kolesa
Bryce Harrington
Godly T.Alias
Shinwoo Kim
JinYong Park
Minkyoung Kim
Carsten Haitzler
Derek Foreman
Youngbok Shin
Gustavo Sverzut Barbieri
Andy Williams
Mike Blumenkrantz
Jiyoun Park
Jiwon Kim
JEONGHYUN YUN