Summary:
Patch from a series of patches to rename EAPI symbols to specific
library DSOs.
= The Rationale =
EAPI was designed to be able to pass
`__attribute__ ((visibility ("default")))` for symbols with
GCC, which would mean that even if -fvisibility=hidden was used
when compiling the library, the needed symbols would get exported.
MSVC __almost__ works like GCC (or mingw) in which you can
declare everything as export and it will just work (slower, but
it will work). But there's a caveat: global variables will not
work the same way for MSVC, but works for mingw and GCC.
For global variables (as opposed to functions), MSVC requires
correct DSO visibility for MSVC: instead of declaring a symbol as
export for everything, you need to declare it as import when
importing from another DSO and export when defining it locally.
With current EAPI definitions, we get the following example
working in mingw and MSVC (observe it doesn't define any global
variables as exported symbols).
Example 1:
dll1:
```
EAPI void foo(void);
EAPI void bar()
{
foo();
}
```
dll2:
```
EAPI void foo()
{
printf ("foo\n");
}
```
This works fine with API defined as __declspec(dllexport) in both
cases and for gcc defining as
`__atttribute__((visibility("default")))`.
However, the following:
Example 2:
dll1:
```
EAPI extern int foo;
EAPI void foobar(void);
EAPI void bar()
{
foo = 5;
foobar();
}
```
dll2:
```
EAPI int foo = 0;
EAPI void foobar()
{
printf ("foo %d\n", foo);
}
```
This will work on mingw but will not work for MSVC. And that's why
EAPI is the only solution that worked for MSVC.
Co-authored-by: João Paulo Taylor Ienczak Zanette <jpaulotiz@gmail.com>
Co-authored-by: Ricardo Campos <ricardo.campos@expertise.dev>
Co-authored-by: Lucas Cavalcante de Sousa <lucks.sousa@gmail.com>
Reviewers: vtorri, jptiz, woohyun, lucas, SPAM-smith78899
Reviewed By: vtorri, SPAM-smith78899
Subscribers: SPAM-smith78899, raster, SPAM-cabanacatalogs, cedric, #reviewers, #committers
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D12228
also.. note the badness of the code design mixing a global singleton
with a "per struct" set of data like fd handlers for the same devices
initted only once but... anyway. it's messy.
Small patch to check if the video driver supports drmWaitVBlank. If
not, we can fall back to timer based animators and avoid freezing (due
to usage of vsync'd animators).
@fix
Summary:
This patch removes the ability to do hardware rotation (it's broken
anyway), and refactors the _drm_rotation_do function to just use
software (ecore_evas & evas) rotation.
ref T7690
Depends on D8115
Reviewers: raster, cedric, zmike
Subscribers: cedric
Tags: #efl, #do_not_merge
Maniphest Tasks: T7690
Differential Revision: https://phab.enlightenment.org/D8116
This patch does a small refactor of _ecore_evas_new_internal function
to separate canvas setup for gl & software. This was done so that when
we create a canvas on a per-output bases (multi-output support), the
code inside _ecore_evas_new_internal will be cleaner and easier to
follow.
NB: No real functional changes here until muti-output lands
Differential Revision: https://phab.enlightenment.org/D7679
In order to support per-output ticking, the drmModePageFlip and
drmModeAtomicCommit functions need to pass the actual Output as data
to the pageflip callback so that the pageflip callback function can
pass the proper rectangle to ecore_evas_animator_tick
Differential Revision: https://phab.enlightenment.org/D7678
Summary:
The animator timestamps were only getting the offset applied when they
came from page flips. The "early tick" logic failed to apply the offset.
This likely only changes behaviour on vmware's graphics stack, and
only the env var ECORE_EVAS_DRM_GPU_CLOCK_WRONG is set.
Reviewers: devilhorns
Reviewed By: devilhorns
Subscribers: bu5hm4n, cedric, #reviewers, #committers, zmike
Tags: #efl
Differential Revision: https://phab.enlightenment.org/D6792
We should only have to actually force a tick if we can't set up
the tick job, as the evas changed callback or a following render
from the tick job's tick will cause a flip.
Should remove a potential for a missed tick near the start of
ticking.
Similarly to commit 8b962d24ac prevent
extra flips in the drm engine.
These are potentially more harmful to DRM as they can cause frame
drops and missed render opportunities.
Small patch to allow specifying a different seat via XDG_SEAT env
variable. If no seat name is passed as ecore_evas options, then we
would always default to 'seat0'. As this may not be desirable in all
situations, allow for an override via XDG_SEAT env var.
ref T6455
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This patch adds support for doing output rotations via hardware. This
is implemented inside the Ecore_Evas engine so that it is transparent
to the caller of ecore_evas_rotation_set.
ref T5999
Signed-off-by: Chris Michael <cp.michael@samsung.com>
If we're not ticking already when a new animator is started then we have
to wait for a vblank to get a tick. That's not great.
If we can, use the time of the last vblank to generate an immediate tick
to avoid this latency.
Since the engines can call the flip functions, we need to protect the
ticker from missing those flips. Thus, we let ecore_drm2, which obviously
sees all flips, track them.
Intended to simplify the upcoming commit that merges device find and
device open into a single function that returns a device.
The fd is something callers shouldn't really need to get their hands on,
right now there are still a few places where it's needed, but those will
be gone soon too.
So vmware's graphics driver reports the MONOTONIC drm cap, yet uses
CLOCK_REALTIME instead. This leaves us with a gigantic offset between the
gpu timestamp and the times ecore_time_get() gets from CLOCK_MONOTONIC.
Since ticking screws directly with loop time this results in some long
distance clock jumping.
This commit fixes drm/gl_drm operation under vmware.
This lets us do a blocking wait for a vsync. Something we should try to
do as infrequently as possible, but in some cases we need it one time at
startup to catch graphics driver bugs.
Engines that provide their own tickers may need to be able to provide the
time of the last tick even if they weren't sending ticks to EFL at the
time.
This is a feature added during freeze as it's necessary to resolve a bug.
ref T5462
BAILing here can break animators permanently, so best to just log the
issue and move on (and potentially allow the ticker to wake us from DPMS).
ref T5462
Partially reverts a795629e8c
With atomic mode setting this has the hilarious side effect of waking
up the display from dpms.
Attempting to tick when manual render is set is likely a bug, so
log an ERR.
fix T5462
As we will now use static_libs/libdrm for building ecore_evas_drm, we
no longer need to include this bit of borrowed four_cc code from libdrm
Signed-off-by: Chris Michael <cp.michael@samsung.com>
This is a quick and harmless hack to make sure we don't come back to a
dead compositor on a vc switch.
A proper fix will follow eventually, I promise.
Really. Would I lie?
this is still semi-broken if a seat has many pointer-ish type devices since
pointer devices in ecore-evas were never correctly implemented to be 1:1 with
seat:cursor relationships
@feature
If we are in async render and a call to rotation_set has been made,
then we should delay setting that rotation until rendering is
completed. This patch fixes that issue by setting ee->delayed.rotation
and handling it during _drm_render_updates
@fix
Signed-off-by: Chris Michael <cp.michael@samsung.com>