| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
| |
The ARM assembly contains a couple of hand-coded PC relative references
to data defined in other object files. Make sure the externs are tagged
as hidden so the linker knows that it can resolve the reference locally
at build time.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@gmail.com>
Bug: 16853291
Change-Id: If7bc8e5bb6a2c63b9df33666d56e6c42f398fd47
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
mInputFrameData and mSliceGroup are allocated memory with
malloc() but released using delete. Changed this to use free()
instead.
Change-Id: Iecf6684d97fabb2229d9abd94279a24faf9ac2d7
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
|
|/
|
|
|
|
|
|
|
|
|
|
|
| |
ULong is used in the encoder code as 32-bit unsigned integer type.
But it has been defined as an unsigned long, which will be 64-bit
on 64-bit platforms. This causes many problems including output
mismatch between 32-bit and 64-bit code and crash in certain cases.
This patch defines ULong as uint32_t. In addition, it uses uintptr_t
at places where an address is cast to an integer to check for alignment.
Change-Id: I2b0086824a5985b9fc4628ce6fbe81db156e37d2
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
|
|
|
|
|
|
|
|
|
| |
The unused ARM assembly files were written for
legacy proprietary ARM compilers. They are not
needed when gcc or clang compiler is used.
Change-Id: I8dfe9ad458efbb34922094e943ad334cc34f59a3
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
|
|
|
|
|
| |
Bug: 13790505
Change-Id: Icfc403bceaefdede1af65dfd70ba3e8b26abab02
|
|
|
|
|
|
|
|
| |
- change internal sized types to use stdint.h
- printf & scanf formats
- size_t or unsigned int for iterators
Change-Id: Id993a70d8bf54c667c5d652b34179a2c727ed446
|
|
|
|
|
|
|
|
|
|
|
| |
%lld -> %" PRId64 " for int64_t
%d -> %zu for size_t
Also fixes some casts from void* to integer types, and some comparisons
between signed and unsigned.
(cherry picked from commit b4a7a2df4c28c3f32b5d877b54831d2cc5d78f81)
Change-Id: I76ba94d0b67776fd7abdc83b43d47c61d6c32f4c
|
|
|
|
| |
Change-Id: I37ef154eea0f77fb7f22faf6891f3025f6ff346e
|
|
|
|
|
|
|
|
| |
(cherry picked from commit 84333e0475bc911adc16417f4ca327c975cf6c36)
Modified by Mark Salyzyn <salyzyn@google.com> to keep merge conflicts
or errors downstream to a minimum.
Change-Id: Ic3b272f9cbf3155001aabd2f79728f1bc31de613
|
|
|
|
| |
Change-Id: I9c41bb34c35595ac19a7ab8faaacde8cb4fa9a15
|
|
|
|
|
|
| |
b/8543366
Change-Id: I746ffed6289486b802a2292bfc492ea6c780ed0b
|
|
|
|
|
|
|
|
|
|
| |
Conceptually it should be the same whether EOS is signalled on the last
buffer holding data, or an empty buffer that follows. Make it so that
this actually behaves the same for mp3, AAC and Vorbis.
b/8747869
Change-Id: Idece8ef45689a3ffaf70fb45d19862d7b93b2f92
|
|
|
|
| |
Change-Id: I88d5fcfc005a2c2acd0246cdd9c08b6c00b6c39b
Signed-off-by: Glenn Kasten <gkasten@android.com>
|
|
|
|
|
|
| |
also 32-bit only command-line apps
Change-Id: I9ac557a8d02bbf6986a9b5c3cdce23d400b306a3
|
|
|
|
|
|
|
|
| |
This patch prevents ARM neon code from being included
for Arm64, where ARM is the 2nd architecture.
Change-Id: I5a283991034f2650a0e7d4227aef967be276fc9c
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the Intra16x16PlanePrediction function, a u32 (unsigned 32-bit
integer) was being used for the "i" variable, and being used with a
value of 7 in the loop "for (i = 0, b = 0; i < 8; i++)" to access the
"above[6-i]" location where "above" is defined as "u8 *". Because "i"
was unsigned there was no sign extension on use with the __LP64__
64-bit pointer, so rather than 0xFFFFFFFF being treated as -1, it was
treated as 4,292,967,295 and that resulted in a SIGSEGV at an invalid
address. By changing the type of "i" to an i32 (signed 32-bit integer)
the expected sign extension occurs and the value is treated as -1.
This change fixes android.media.cts.DecoderTest#testCodeBasicH264 on
64-bit platforms
Change-Id: I85df58b0dc1d39f89ab421d04ab5481356520f0c
Signed-off-by: Marcus Oakland <marcus.oakland@arm.com>
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Section 7.4.1 in the H.264 standard says that the only valid bytes
to follow a sequence that starts with 0x000003 are 0x00, 0x01,
0x02 or 0x03.
This makes EncodeDecodeTest pass properly when decoding using
OMX.google.h264.decoder, which is strict about the forbidden
escape sequences.
Change-Id: Ice113d9b934015003ea9cb10d0b21cee4d18d774
|
|/
|
|
|
|
|
|
|
|
|
|
| |
Contains the necessary changes to make frameworks/av build and work
on a 64-bit machine.
Signed-off-by: Craig Barber <craig.barber@arm.com>
Signed-off-by: Kévin PETIT <kevin.petit@arm.com>
Signed-off-by: Ashok Bhat <ashok.bhat@arm.com>
Signed-off-by: Marcus Oakland <marcus.oakland@arm.com>
Change-Id: I725feaae50ed8eee25ca2c947cf15aee1f395c43
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This makes sure that the right parameters are returned on both
input and output ports if queried after setting.
This also makes sure that the output MediaFormat from the
MediaCodec class contains the right video size.
Change-Id: I0667b3b4c6bb90331ad0ae7d51388e1bca3d1bbd
|
| |
| |
| |
| |
| |
| |
| | |
This makes sure the MediaCodec output MediaFormat contains
the right width and height.
Change-Id: Ic97af3b5850ebaf563533c3d1cae992be3e4d074
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| |
| | |
Allow the bitrate to be updated while the encoder is running.
Bug: 8422347
Change-Id: I8427fe20921b00f92b8f99fe21691709fab354b0
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Set the appropriate OMX flag when the encoder generates a keyframe. This
is necessary for any muxer which needs to indicate which frames are
seekable.
Bug: 8422347
Change-Id: I744a0b3023db24d3de2210bce82f41e50d259505
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The audio object type for parametric stereo and lowdelay was not passed
to the FDK AAC encoder.
Bug 9428126
Change-Id: Ic32822afff8b1da6a2d80c1b65d514f24059fb29
|
| |
| |
| |
| |
| |
| | |
encoder.
Change-Id: I370d5831f7d6037faf361a92521390f19f179cbe
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
even with the software VP8 encoder.
This relies heavily on the fact that the "Nexus" devices use ARGB32 as
the colorspace for the data underlying a surface provided by SurfaceFlinger
(mirroring). Generally there are no such guarantees.
Change-Id: I1de32f591a3bb935ca76151816b3a02665bec40b
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Also change the VP8 encoder role name from
video_encoder.vpx to video_encoder.vp8 for
future VP9 encoder support.
Requires the change in frameworks/native and media_codecs.xml
corresponding to the device.
VP9 decoding test will be added to cts repo later.
|
| |
| |
| |
| |
| |
| |
| | |
enough room was available in the output queue at the time they were
available. No more.
Change-Id: I5957290d40ba31bda7944271ec7f2aa0f1f7043c
|
| |
| |
| |
| |
| |
| |
| |
| | |
Created common base class for all software video decoders to make
adding new features easier.
Change-Id: Id89964e572d5cc5add02662273e6ae96c6b7eb12
Signed-off-by: Lajos Molnar <lajos@google.com>
|
| |
| |
| |
| |
| |
| | |
Change-Id: I7772e3afec020f889dea80fd6372afbc36cd68d6
Signed-off-by: Lajos Molnar <lajos@google.com>
Bug: 9285553
(cherry picked from commit e113aa1f078cb3d5f8182058e144fd14ce945fca)
|
| |
| |
| |
| |
| |
| |
| |
| | |
Otherwise, state may be undefined for subsequent frames.
Change-Id: Icdc0126247e1422eba21f2008a04cf7867d93f5d
Signed-off-by: Lajos Molnar <lajos@google.com>
Bug: 9284771
(cherry picked from commit 0f15875b8e80fb49a3b18d88964b063326f307b9)
|
| |
| |
| |
| |
| |
| |
| | |
properly on a transition from idle->loaded.
Change-Id: I56ccfeef24c391e50e42b522194206e35c7ab700
related-to-bug: 9105408
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
SoftVP8 decoder ignored frame content if EOS flag was set on
input frame. Now, decode the frame first, unless it is empty.
Change-Id: Id105a9eb86103a61390af3de60cae2507028e2d1
Signed-off-by: Lajos Molnar <lajos@google.com>
Bug: 9091495
|
| |\
| | |
| | |
| | |
| | |
| | |
| | | |
signals an error if client" into jb-mr2-dev
* commit 'bfc5b96c326a6321f167a59de4a201e2b61f5ca4':
The software FLAC encoder now properly signals an error if client
|
| | |\
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
if client" into jb-mr2-dev
* commit '09f842b03d8f9c0ab5e4541937f8061a7d1bf53d':
The software FLAC encoder now properly signals an error if client
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
attempts to configure it with too large an input buffer size.
Previously this would lead to memory corruption during encoding due to a
typo.
Change-Id: I229b07b7dbe87fb8424419706671b66a8d58ec6b
related-to-bug: 8778893
|
| |\ \ \
| | |/ /
| | | |
| | | |
| | | |
| | | |
| | | | |
camera recording"
* commit 'b0ec4e15704b77847c1d24e611029b6311d67dfe':
Fixed the wrong timestamps issue about camera recording
|
| | |\ \
| | | |/
| | |/|
| | | |
| | | | |
* commit '3030f344e77d14cf688c34f79c83a045c888825d':
Fixed the wrong timestamps issue about camera recording
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
On emulator, the camera video recording crashes randomly.
It's caused by wrong timestamp when encoding the video frame.
The element of the vector is erased before accessing it.
Therefore, when reading the timestamp from the element,
the memory of the element has been changed.
Change-Id: I9750417fbb9dc4697ccae1e78e286b23925d3764
Signed-off-by: Panfeng Chang <panfengx.chang@intel.com>
Signed-off-by: Jun Tian <jun.j.tian@intel.com>
|
| |\ \ \
| | |/ /
| | | |
| | | |
| | | |
| | | |
| | | | |
into jb-mr2-dev
* commit '4707c9b04891e5cf7772de74e9c229bce028c87d':
Make sure the drain buffer is big enough
|
| | |\ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
b/8614909
Change-Id: I9d973dcd74100b793791359c262b821207ff9ddd
|
| |\ \ \ \
| | |/ / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
was configured with the wrong" into jb-mr2-dev
* commit 'c9e18491b4dd6d82d52ec7baa2101732ab617db6':
Apparently the mp3 decoder glitches if it was configured with the wrong
|
| | |/ /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
sample rate. Make sure we tell it about the one the extractor provides.
Change-Id: Ice5b9cf55485ca1a8f099a14be0202be9540e781
related-to-bug: 8621639
|
| |\ \ \
| | |/ /
| | | |
| | | |
| | | | |
* commit '844c8b039feb7d2dd9b951a244966edd5e6fab67':
Better default AAC DRC setting
|
| | |\ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
By default, use boost and attenuation together.
bug 7140036
Change-Id: Ie0565f9e5cd24f7a59b5a218b4e6186661832601
|