The basic hashtable is intended to be used to support a variety
of different datastructures such as map, set, multimap,
multiset, linkedmap, generationcache, etc.
Consequently its interface is fairly primitive.
The basic hashtable supports copy-on-write style functionality
using SharedBuffer.
The change introduces a simple generic function in TypeHelpers for
specifying hash functions. The idea is to add template
specializations of hash_type<T> next to the relevant data structures
such as String8, String16, sp<T>, etc.
Change-Id: I2c479229e9d4527b4fbfe3b8b04776a2fd32c973
* commit 'c25972950c2ea62fb085524dbe737c2bf0f08f4a':
SurfaceTexture: fix a couple tests
EGL: default to swap interval 1
SurfaceTexture: clean up some tests
This change fixes some robustness issues with the
EglDestroySurfaceUnrefsBuffers and
EglDestroySurfaceAfterAbandonUnrefsBuffers tests. The tests previously
depended upon GL implementation details that should not have been relied
upon.
Change-Id: I37fd43b56568efe1dbe69d85e892be8a1cf44d20
This change explicitly sets swap interval 1 on the window when an
EGLSurface is created to render to it.
Change-Id: I91eb29dbee3ae4a55076b921f084d503fbe94e03
This change cleans a few things up in the SurfaceTexture tests:
- Wraps a few long lines.
- Refactors the multithreading portions of SurfaceTextureGLToGLTest into
a new test fixture called SurfaceTextureGLThreadToGLTest.
- Changes some of the tests that were creating their own EGLSurface to
use the SurfaceTextureGLToGLTest fixture.
- Reorders the test functions so that they are immediately below to the
test fixture that they use.
Change-Id: I0491ce3528a7ff2b4f1e83602ba290269c087297
this happens often with CPU Surfaces, which disconnect long
after their surfacetexture has been abandoned.
Change-Id: If49da03b72f99130e01b2b9bcbd444bb38f7ed4e
this flag should be enabled for all targets, but currently
some have issues with it, so we're turning it on only for
tested targets.
this will hopefully resolve some performance issues.
Bug: 5553562, 5631630
Change-Id: I54c7a9e2068586898ab13e405d95534669260537
the working theory here is that a Surface object has become non-promotable
because it lost its last reference; later Surface::readFromParcel is called
the previous surface is found in the cache, but can't be promoted. this causes
a new Surface object to be created which will promptly try to connect to the
CPU_API -- this in turn will fail because the previous (now dead) surface is
still connected.
To fix this, we make sure to disconnect from the SurfaceTexture when
Surface[TextureClient] is destroyed.
Change-Id: I422234868a05d7b7d283e9d5a85f7ab79e65d8a9
This change fixes an issue involving buffer reallocation and the
ALLOW_DEQUEUE_CURRENT_BUFFER mode in SurfaceTexture. The bug happened
when the buffer slot currently attached to the GL texture was selected
for dequeuing, but the dequeue operation caused the buffer to be
reallocated. Because the buffer is new, the image producer could fill
the buffer and queue it before an updateTexImage call, which would
result in the "slot %d is current" error in queueBuffer.
Bug: 5631630
Change-Id: Icdd8bc5cad3c7db43953446d9be2603aaea11a8d
Stop using system properties to publish information about
the key character map path. Instead, we can retrieve it
on demand by asking the window manager.
It was possible to exhaust the supply of system properties
when repeatedly adding and removing input devices.
Bug: 5532806
Change-Id: Idd361a24ad7db2edc185c8546db7fb05f9c28669
Surface Texture dequeue logic is modified to return the oldest of the
free buffers to Client on dequeue call.
Currently dequeue method is returning the first buffer index which is free.
The parsing is done in ascending order of the buffer slot indices.
This leads to returning the buffer which has been just queued to composer,
and hence display, and this defeats the purpose of having minimum dequeue count
as 2 in asynchrnouse mode.
This is fixed by checking all the free slots and returning the oldest buffer.
Change-Id: Ibbac10593c3994c278c601af0480b171635ecdd4
Signed-off-by: Sunita Nadampalli <sunitan@ti.com>
this flag should be enabled for all targets, but currently
some have issues with it, so we're turning it on only for
tested targets.
this will hopefully resolve some performance issues.
Bug: 5553562
Change-Id: I939992b4cd0debea980dec0127c72be2dff33af8
Was mistakenly assuming that Parcel::writeFileDescriptor took
ownership of the fd that was passed in. It does not!
Added some comments and a default parameter to allow the caller
to specify whether it wishes the Parcel to take ownership.
Bug: 5563374
Change-Id: I5a12f51d582bf246ce90133cce7690bb9bca93f6
This change adds serialization and deserialization functionality to
BlobCache, conforming to the Flattenable interface.
Change-Id: Ibc99cb1c3d015f363d57d0713eabccec07ff975e
This change removes the mutex from the BlobCache class. The caller must
be responsible for thread synchronization in order to properly implement
the Flattenable interface, which is coming soon. Otherwise would be the
potential for the cache contents to change between the call to the
getFlattenedSize and flatten methods. Because the caller must do this
synchronization anyway there's no reason to also some synchronization
inside BlobCache.
Change-Id: Ie1f4f6f82b78744f46a41ce863cac0cad276a20e
On user-debug and eng builds, you can set the
"db.log.slow_query_threshold" system property to queries that
take longer than the specified number of milliseconds.
Set it to 0 to log all queries.
This property has been around for a while but it was implemented
poorly. In particular, it *changed* the behavior of the query
by calling getCount() while holding the Db connection.
In normal operation, the query will not actually run until later.
By putting the timing logic into fillWindow() instead, we ensure
that we only measure queries that actually ran. We also capture
cases where the cursor window gets filled multiple times.
Bug: 5520301
Change-Id: I174f5e1ea15831a1d22a36e9a804d7755f230b38
Bug: 5520301
When an application requests a row from a SQLiteCursor that
is not in the window, instead of filling from the requested
row position onwards, fill from a little bit ahead of the
requested row position.
This fixes a problem with applications that seek backwards
in large cursor windows. Previously the application could
end up refilling the window every time it moved back
one position.
We try to fill about 1/3 before the requested position and
2/3 after which substantially improves scrolling responsiveness
when the list is bound to a data set that does not fit
entirely within one cursor window.
Change-Id: I168ff1d3aed1a41ac96267be34a026c108590e52
