this allows us to enable h/w acceleration on low-end
devices while keeping memory usage down.
Bug: 6557760
Change-Id: I8af2de3038dc2579360b8b73aa452cb7a0e506a9
Hibernating EGL takes a long time (>100 ms) and blocks all other
rendering. During window animations, the outgoing activity begins
hibernation before the animation stops, causing visible stutter.
Hibernation is still available by setting 'BOARD_ALLOW_EGL_HIBERNATION
:= true' in the devices BoardConfig.mk
Change-Id: Iab4e00723a1adcd97481e81b2efdc821b3e9712f
When eglTerminate() is called with a window surface still exists, a
deadlock would occur since egl_display_t::terminate() holds a lock
while destroying the window surface, which calls
onWindowSurfaceDestroyed() which attempts to take the same lock.
This change refactors the hibernation code and data into a separate
object with its own lock, separate from the egl_display_t lock. This
avoids the deadlock and better encapsulates the hibernation logic.
The change also fixes a bug discovered incidentally while debugging:
hibernating after calling eglTerminate() succeeds, but will cause
awakens from subsequent eglInitialize() to fail. We will no longer
hibernate a terminated display.
Change-Id: If55e5bb603d4f8953babc439ffc8d8a60af103d9
If the EGL implementation supports the EGL_IMG_hibernate_process
extension, use it to hibernate (and hopefully release memory or other
resources) when the process isn't actively using EGL or OpenGL ES. The
idleness heuristic used in this change is:
(a) Wake up when entering any EGL API call, and remain awake for the
duration of the call.
(b) Do not hibernate when any window surface exists; this means the
application is very likely in the foreground.
(c) Do not hibernate while any context is made current to a thread.
The app may be using a client API without the EGL layer knowing,
so it is not safe to hibernate.
(d) Only check these conditions and attempt to hibernate after a
window surface is destroyed or a thread's context is detached. By
not attempting to hibernate at the end of every EGL call, we avoid
some transient wakeups/hibernate cycles when the app is mostly idle,
or is starting to become active but hasn't created its window
surface yet.
On a Galaxy Nexus, hibernating frees 1567 VM pages from the process.
Both hibernating and waking can take anywhere from 30ms to over 100ms
-- measurements have been very inconsistent.
Change-Id: Ib555f5d9d069aefccca06e8173a89625b5f32d7e
This is in preparation for a change that will hibernate the underlying
EGL when idle. Instead of a bare egl_display_t*, get_display() now
returns a egl_display_ptr, which acts like a smart pointer. The
"wakecount" counter managed by the smart pointer isn't used for
anything in this change. It will be used to make sure we don't
hibernate when any thread is in an EGL call, without having to hold a
mutex for the duration of the call.
Change-Id: Iee52f3549a51162efc3800e1195d3f76bba2f2ce
This change adds a debug option to EGL to use an EGLSyncKHR each frame to
determine when the GPU finishes rendering the frame.
Change-Id: I09ce071db904b44f07ca814c586c291c8b59385a
This patch adds a function setGlDebugLevel() to libEGL to enable
GL tracing. This will be used by the Java layer to add an option
to "am start" that can enable tracing for a particular application.
Change-Id: Ie1dbdd550f502df8633553595cb33ee9d9ae44e1
this simplify our EGL wrapper implementation a lot.
This wrapping is no longer needed now that we can only
support a single underlaying EGL implementation.
Change-Id: I8213df7ac69daac447f1fe6e37044b78aac4e9a9
this is possible now that we support only a single
EGL implementation. this allows a large code simplification.
Change-Id: I7a6b9db4c5d60f4407c6061e7a68729af63d5242
from now on, the system can only have one EGL
implementation. this means the software and h/w renderer
cannot be used at the same time on a device. Of course, the
h/w renderer is always prefered; in its absence we
default to the software renderer.
Change-Id: Ib579f58055dd0ce4c4a99144131efa11c16ca3d3
this was introduced in a recent change. eglMakeCurrent can
end up calling eglDestroyImageKHR via ANativewWindow::disconnect
when the consumer is in the same process.
we make sure we don't hold the lock while this is happening.
Change-Id: Id17fe4fd76eecf5f962cefb9aa32be41fc1b042d
it would happen when a context was made non-current, in this
case we would call the implementation's eglMakeCurrent() which
would succeed, if we're rescheduled at that point, another
eglMakeCurrent() could make that context current to another thread,
however, when we came back to the original call we would
overwrite egl_context_t internal state.
this is fixed by moving the critical section under
egl_display_t's lock.
Change-Id: I743c85696e13263d3a9570824940263df0caacdc
when increasing MAX_NUMBER_OF_GL_EXTENSIONS to 256
we also needed to create all the corresponding wrappers.
Change-Id: I90edaaf0885ccdfab48e7a1396bcf88e039cfb25
This extension is always added to the GL_EXTENSIONS
extension string for the current GL context, regardless
of if it's supported by the h/w driver.
The extension itself will be handled by GLES_trace (eventually),
when GLES_trace is not enabled, it'll result to a no-op.
If the h/w implementation has this extension, we'll call that version
instead of our dummy version.
Change-Id: Ie5dd3387c4d45cd5ed5f03b73bda6045620a96bc
This patch adds two improvements:
1. Protobuf messages are buffered and sent in chunks.
2. Multiple EGL contexts are handled properly: Corresponding
to each EGLContext, a GLTraceContext with a unique ID is created.
On eglMakeCurrent, the appropriate GLTraceContext is set and is
used while tracing subsequent GL Calls in that thread.
Change-Id: I34076376d3e5af205c87c7396ea47659844abd6e
This patch provides a framework for tracing GLES 1.0 and 2.0
functions. It is missing a lot of features, but here are the
things it accomplishes:
- Stop building the glesv2dbg library, and build the
glestrace library instead.
- Replace the hooks for glesv2dbg with the ones for glestrace.
- Add the basics for the trace library. Currently, this
traces all GL functions, but not all required data is
sent for all the functions. As a result, it will not
be possible to reconstruct the entire GL state on the
host side.
The files gltrace.pb.* and gltrace_api.* are both generated
using the tools/genapi.py script.
Change-Id: Id60a468f7278657f008bc6ea1df01f9bdfecfdd3
This change explicitly sets swap interval 1 on the window when an
EGLSurface is created to render to it.
Change-Id: I91eb29dbee3ae4a55076b921f084d503fbe94e03
This change introduces two new BoardConfig variables to control the size
limits of the EGL blob cache. MAX_EGL_CACHE_ENTRY_SIZE is the size
limit for values inserted into the cache, and MAX_EGL_CACHE_SIZE is the
size limit for all entries in the entire cache (including both keys and
values). If either of these BoardConfig variables are not defined then
a default size limit is used instead.
Change-Id: I6703d93f966b6389c6499f23d841e42339f9c9d7
- don't advertise extensions that are not supported
by any implementation
- remove EGL_ANDROID_swap_rectangle which is not
implemented by anybody and confuses people
- add some comments about mandatory extensions
Bug: 5428001
Change-Id: Id8dc48116ac1d1eb79ec9ef55d03e29d4257c1f3
the code that validated EGL objects dereferenced the object
to access its EGLDisplay -- needed for validation (!).
This was wrong for two reasons, first we dereferenced the object
before validating it (potentially leading to a crash), secondly
we didn't validate that the object existed in the right EGLDisplay.
We now use the EGLDisplay passed by the user API.
Change-Id: I66f9e851d4f8507892a6b1fee3065f124c4e7138
This change adds the ANDROID suffix to the all the types and functions
defined by the EGL_ANDROID_blob_cache extension.
Change-Id: I087875b96d9a7053efb9c8d5614f9f765eed799d
This does not happen on high end gfx devices. This happens
only if only one EGL context is initialized in the current
process.
Change-Id: Ibd1737efdf84eef8a84108b05795440d1ae9964e
This change causes any insertions into the EGL cache to trigger an
attempt to save the cache contents to disk. The save operation is
deferred to allow multiple cache insertions to be batched up.
Change-Id: I6cfec9c0dbbef94d3f8880860e2a365dccc296c7
This change adds support for saving and loading the contents of the EGL
cache. It also adds some simple tests for the EGL cache.
Change-Id: I18e5e789e0897a0783d29d1c1e64d26de2dd44c4
This change makes the makes the stub EGL_ANDROID_blob_cache callbacks
actually use a BlobCache object.
Bug: 5474671
Change-Id: I5cbaae2dea3aad2fe306c9f57029c3f215a0863a
This change adds a stub cache implementation that gets passed to the
underlying EGL implementation at initialization time.
Change-Id: I14437c5b6f91b7a34a19bb02ad802e6e54f88d2a
This patch modifies the library loaded in libEGL.so to
handle the case of GLES emulation as follows:
- if we detect that we run inside the emulator, check the
GPU emulation status through ro.kernel.qemu.gles, which
will be set to 1 if supported, or 0 otherwise.
When trying to run on an older version of the emulator,
the kernel parameter will not be defined at all.
- if GPU emulation is supported, use egl.cfg as usual.
It will contain a line like "0 0 emulation" that will
load libEGL_android.so appropriately.
- nothing is changed if we don't run inside the emulator.
NOTE: Ideally, we would modify libEGL_emulation.so to
redirect all calls to libEGL_android.so in this case.
However, this turns out to be extremely tedious to implement
(too many functions with different signatures).
As such, it is much simpler to make the check before
loading the library.
Change-Id: I9930bc168d9013cc8700feedc57b979384467c37
This change moves the call to native_window_api_disconnect from
eglDestroySurface to the egl_surface_t destructor. The egl_surface_t
can outlive the external EGLSurface if eglDestroySurface is called while
the surface is made current on a thread.
Change-Id: I0df6117a5633c2a19935fe356579abdd76fc471f
the original connect/disconnect hooks are deprecated
and replace by api_connect/api_disconnect. the original
hooks are no no-ops.
api_connect/api_disconnect is now only called from the
android framework.
Bug: 5057915
Change-Id: I8ca64cd1acd6cabf915bf54689ec2e5f6dfa495a
This change fixes how the Android EGL layer overrides the size and
format of an ANativeWindow in eglCreateWindowSurface. The new behavior
is to leave the size untouched when overriding the format. The previous
behavior was to reset the ANativeWindow to use the default size set by
the ANativeWindow implementation.
It also adds two new 'perform' methods to the ANativeWindow interface:
set_buffers_dimensions and set_buffers_format, and redefines the
behavior of set_buffers_geometry to be the combination of these two new
methods.
Additionally, this change adds an error check for the return value of
the new native_window_set_buffers_format call, which required adding a
(stub) handler for to FramebufferNativeWindow.
Change-Id: I805c7ccd8d4730dfb132d10d8bc3fb058a0b9df1
as specified by the EGL specification, terminated objects's
handles become invalid, the objects themselves are destroyed
when they're not current to some thread.
Change-Id: Id3a4a5736a5bbc3926a9ae8385d43772edb88eeb
This change removes the check that disallowed the creation of an
EGLSurface that would send frames to a SurfaceTexture.
Change-Id: I44c6d5df503cc676a88144d72d39b414692ce4c9
setprop debug.egl.debug_forceUseFile 1: always use file
setprop debug.egl.debug_maxFileSize <int MB>: exit when reached
setprop debug.egl.debug_filePath <path>: may need to make the file
writable first.
setprop debug.egl.debug_port <unsigned short>
setprop debug.egl.debug_proc <proc cmdline>: to match debuggee process
Change-Id: I34ca1f3092f3abf17e2ded9d1689d0cdef6e07e8
Signed-off-by: David Li <davidxli@google.com>
Move EGL objects (contexts, surfaces, ...) collection from global
static into EGL display object. Since EGL implementation currently
supports only one display, this doesn't change current logic. This
aims at preparing for next coming changes in eglTerminate() to
support per-display cleanup.
Bug: 4152714
Change-Id: I5e424781b89c3c275f9922f943463d5f6b1f6bc6
Check for initialized display, and set EGL error to EGL_NOT_INITIALIZED
whenever it isn't, for all EGL functions taking an EGLDisplay as argument
and requiring it to have been initialized.
Bug: 3366011
Change-Id: Ib1d8a3a207257995518f4430fe97f8c406c7fc13
We were leaking all EGLSyncKHR objects, over time the list
would grow and become very slow to operate on.
Bug: 4107131
Change-Id: I6c82daf49fe3189d04550781d64d57c30c95f552
This change makes the ANativeWindow_lock NDK function error out if it is
passed an ANativeWindow with a concrete type that is not Surface. It
also makes eglCreateWindowSurface fail if it is passed a
SurfaceTextureClient as its 'window' argument.
Bug: 4087277
Change-Id: Ie68c50c52d88f72d8a387f6c094908044c83a88c
In message loop, use select to check for available commands from client,
rather than always expecting commands in eglSwapBuffers.
Change-Id: Ifc34dd77c2528c8b9c71f594e3eda4f93400cd2b
Signed-off-by: David Li <davidxli@google.com>
Send VBO related commands to client, which tracks the state.
Maintain index buffer content and vertex attrib pointer/buffer state on server.
During glDrawArrays/Elements, send user memory data to client.
Change-Id: Ia920e90479329b301ae4b5735e833eeb20293c94
Signed-off-by: David Li <davidxli@google.com>
Also added timing mode option using utils/Timers.h.
Factored out common code to reduce size.
Improved Protobuf message.
Uploads data from glBufferData and glBufferSubData.
Change-Id: Iaae5e706235d942df81c7eada7223fb0b0583911
Signed-off-by: David Li <davidxli@google.com>
Use debug.egl.debug_proc property to match process cmdline.
Binds to TCP:5039 and waits for client connection.
Sends function call parameters, textures and shaders using Protobuf.
Java Eclipse client plug-in is next.
Change-Id: I183b755263663f87e86dde1ad12f527d0445fd57
Signed-off-by: David Li <davidxli@google.com>
HAVE_ANDROID_OS was defined as "1" for targets, but never defined as "0"
for non-targets. Changing them to #ifdef should be safe and matches
all the other uses of HAVE_ANDROID_OS throughout the system.
Change-Id: I82257325a8ae5e4e4371ddfc4dbf51cea8ea0abb
This changes the clearError function in the EGL wrapper layer to simply
call eglGetError(). That should clear any pending errors from all the
underlying EGL implementations, which is needed to correctly report the
error for the most recently called EGL function.
Change-Id: Iad19c69f0c5305e873f3c2f96d353280d31f7b61
Clear the current thread error state on entry to API functions as
mandated by the EGL 1.4 specification, section 3.1. glGetError
returns additional information about the "most recent EGL function".
Change-Id: Ic24c64b39294ffc1a4e43fa72663e076b9d7decf
Return wrappers from GetProcAddress for glEGLImageTargetTexture2DOES
and glEGLImageTargetRenderbufferStorageOES which unwrap the EGLimage
handle before calling through to the implementation.
Change-Id: I2f5b180ab3ccdb28a4f510a2bd8c2eee941a84df
Add EGL_KHR_gl_texture_cubemap_image and EGL_KHR_gl_renderbuffer_image
to the static EGL extension string since these require no new APIs
beyond what is already required.
Change-Id: I2d1fde86b19bb9eee46b3b977f51784a17cfcc3c
eglMakeCurrent() would only deref the previous surfaces if the old and
new contexts were the same. eglTerminate() should not touch TLS.
eglReleaseThread() needs to unbind the current context.
Change-Id: I213b8be77b1a23b5a8a6afaac60643662c8aa010
(there are multiple bugs this should fix)
we now use the EGL_NATIVE_VISUAL_ID of a config to set
the ANativeWindow's format from eglCreateWindowSurface(),
this guarantees that the surface's format will match
whatever EGLConfig the user chose.
this should fix all current and future config/surface format
mismatch and allow users to easily select 32-bits surfaces.
Change-Id: I3835d0eb70c75eeecded3c3509a0a8207554c98b
We just make sure eglGetProcAddress() will return NULL for
glEGLImageTargetTexture2DOES
glEGLImageTargetRenderbufferStorageOES
which is better than returning the address of the wrong implementation.
the correct fix is more involved.
Change-Id: I585a1f40e564f862e5dd382224609ccd069cd3b5
Merge commit '4eb1ad5e98c7b36f7ac4ec8c3270f9763afd107e'
* commit '4eb1ad5e98c7b36f7ac4ec8c3270f9763afd107e':
better fix for [3028370] GL get error should return a valid error if no context is bound.
it turns out that we cannot return INVALID_OPERATION from glGetError() because the
GL spec says that it must be called in a loop until it returns GL_NO_ERROR.
now, we always return 0 from GL functions called from a thread with no
context bound. This means that glGetError() will return NO_ERROR in this case,
which is better than returning a random value (which could trap the app in a loop).
if this happens in the main thread of a process, we LOG an error message once.
Change-Id: Id59620e675a890286ef62a257c02b06e0fdcaf69
Merge commit '9c4d7d677097981a349c718902e29050dad3d59f'
* commit '9c4d7d677097981a349c718902e29050dad3d59f':
fix [3028370] GL get error should return a valid error if no context is bound.
glGetError() will now always return GL_INVALID_OPERATION if called from a thread
with no GL context bound.
Change-Id: I28ba458871db051bb4f5a26668a1fa123526869c
make sure to clear our EGL implementation's error when returning
an error from an underlying implementation
Change-Id: Ibce4726cef1f900e4c7f16002345d7a07f8cdf41
Merge commit '5c7465220d52c64c87fe4566fe109c649ec5c4f7'
* commit '5c7465220d52c64c87fe4566fe109c649ec5c4f7':
support loading EGL libraries from /vendor/lib/egl as well as /system/lib/egl
