Merge "Moving count_mb_ref_frame_usage to macroblock struct"
diff --git a/build/make/ads2gas.pl b/build/make/ads2gas.pl
index ba70242..95be467 100755
--- a/build/make/ads2gas.pl
+++ b/build/make/ads2gas.pl
@@ -61,26 +61,26 @@
s/:SHR:/ >> /g;
# Convert ELSE to .else
- s/ELSE/.else/g;
+ s/\bELSE\b/.else/g;
# Convert ENDIF to .endif
- s/ENDIF/.endif/g;
+ s/\bENDIF\b/.endif/g;
# Convert ELSEIF to .elseif
- s/ELSEIF/.elseif/g;
+ s/\bELSEIF\b/.elseif/g;
# Convert LTORG to .ltorg
- s/LTORG/.ltorg/g;
+ s/\bLTORG\b/.ltorg/g;
# Convert endfunc to nothing.
- s/endfunc//ig;
+ s/\bendfunc\b//ig;
# Convert FUNCTION to nothing.
- s/FUNCTION//g;
- s/function//g;
+ s/\bFUNCTION\b//g;
+ s/\bfunction\b//g;
- s/ENTRY//g;
- s/MSARMASM/0/g;
+ s/\bENTRY\b//g;
+ s/\bMSARMASM\b/0/g;
s/^\s+end\s+$//g;
# Convert IF :DEF:to .if
@@ -149,11 +149,15 @@
s/^([a-zA-Z_0-9\$]+)/$1:/ if !/EQU/;
# ALIGN directive
- s/ALIGN/.balign/g;
+ s/\bALIGN\b/.balign/g;
# ARM code
s/\sARM/.arm/g;
+ # push/pop
+ s/(push\s+)(r\d+)/stmdb sp\!, \{$2\}/g;
+ s/(pop\s+)(r\d+)/ldmia sp\!, \{$2\}/g;
+
# NEON code
s/(vld1.\d+\s+)(q\d+)/$1\{$2\}/g;
s/(vtbl.\d+\s+[^,]+),([^,]+)/$1,\{$2\}/g;
@@ -189,7 +193,7 @@
s/(\S+\s+)EQU(\s+\S+)/.equ $1, $2/;
# Begin macro definition
- if (/MACRO/) {
+ if (/\bMACRO\b/) {
$_ = <STDIN>;
s/^/.macro/;
s/\$//g; # remove formal param reference
@@ -198,7 +202,7 @@
# For macros, use \ to reference formal params
s/\$/\\/g; # End macro definition
- s/MEND/.endm/; # No need to tell it where to stop assembling
+ s/\bMEND\b/.endm/; # No need to tell it where to stop assembling
next if /^\s*END\s*$/;
print;
print "$comment_sub$comment\n" if defined $comment;
diff --git a/build/make/configure.sh b/build/make/configure.sh
index 05bbabe..c99a01c 100755
--- a/build/make/configure.sh
+++ b/build/make/configure.sh
@@ -767,6 +767,7 @@
;;
armv5te)
soft_enable edsp
+ disable fast_unaligned
;;
esac
diff --git a/libs.mk b/libs.mk
index 373c1cd..4115dd8 100644
--- a/libs.mk
+++ b/libs.mk
@@ -20,8 +20,16 @@
CODEC_SRCS-yes += CHANGELOG
CODEC_SRCS-yes += libs.mk
+# If this is a universal (fat) binary, then all the subarchitectures have
+# already been built and our job is to stitch them together. The
+# BUILD_LIBVPX variable indicates whether we should be building
+# (compiling, linking) the library. The LIPO_LIBVPX variable indicates
+# that we're stitching.
+$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes)
+
include $(SRC_PATH_BARE)/vpx/vpx_codec.mk
CODEC_SRCS-yes += $(addprefix vpx/,$(call enabled,API_SRCS))
+CODEC_DOC_SRCS += $(addprefix vpx/,$(call enabled,API_DOC_SRCS))
include $(SRC_PATH_BARE)/vpx_mem/vpx_mem.mk
CODEC_SRCS-yes += $(addprefix vpx_mem/,$(call enabled,MEM_SRCS))
@@ -29,17 +37,17 @@
include $(SRC_PATH_BARE)/vpx_scale/vpx_scale.mk
CODEC_SRCS-yes += $(addprefix vpx_scale/,$(call enabled,SCALE_SRCS))
+include $(SRC_PATH_BARE)/vpx_ports/vpx_ports.mk
+CODEC_SRCS-yes += $(addprefix vpx_ports/,$(call enabled,PORTS_SRCS))
+
ifeq ($(CONFIG_VP8_ENCODER),yes)
VP8_PREFIX=vp8/
include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx.mk
CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_CX_SRCS))
CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_CX_EXPORTS))
- CODEC_SRCS-yes += $(VP8_PREFIX)vp8cx.mk vpx/vp8.h vpx/vp8cx.h
- CODEC_SRCS-$(ARCH_ARM) += $(VP8_PREFIX)vp8cx_arm.mk
INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8cx.h
INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
- CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8cx.h
CODEC_DOC_SECTIONS += vp8 vp8_encoder
endif
@@ -48,10 +56,8 @@
include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx.mk
CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_DX_SRCS))
CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_DX_EXPORTS))
- CODEC_SRCS-yes += $(VP8_PREFIX)vp8dx.mk vpx/vp8.h vpx/vp8dx.h
INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8dx.h
INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/%
- CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8dx.h
CODEC_DOC_SECTIONS += vp8 vp8_decoder
endif
@@ -83,29 +89,10 @@
INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Debug/%)
endif
-# If this is a universal (fat) binary, then all the subarchitectures have
-# already been built and our job is to stitch them together. The
-# BUILD_LIBVPX variable indicates whether we should be building
-# (compiling, linking) the library. The LIPO_LIBVPX variable indicates
-# that we're stitching.
-$(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes)
-
CODEC_SRCS-$(BUILD_LIBVPX) += build/make/version.sh
CODEC_SRCS-$(BUILD_LIBVPX) += build/make/rtcd.sh
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx/vpx_integer.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/asm_offsets.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/vpx_timer.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/mem.h
CODEC_SRCS-$(BUILD_LIBVPX) += $(BUILD_PFX)vpx_config.c
INSTALL-SRCS-no += $(BUILD_PFX)vpx_config.c
-ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/emms.asm
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86.h
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_abi_support.asm
-CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_cpuid.c
-endif
-CODEC_SRCS-$(ARCH_ARM) += vpx_ports/arm_cpudetect.c
-CODEC_SRCS-$(ARCH_ARM) += vpx_ports/arm.h
CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com
CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc
CODEC_EXPORTS-$(CONFIG_DECODERS) += vpx/exports_dec
@@ -344,11 +331,6 @@
$(RTCD_OPTIONS) $^ > $@
CLEAN-OBJS += $(BUILD_PFX)vpx_rtcd.h
-CODEC_DOC_SRCS += vpx/vpx_codec.h \
- vpx/vpx_decoder.h \
- vpx/vpx_encoder.h \
- vpx/vpx_image.h
-
##
## libvpx test directives
##
diff --git a/test/datarate_test.cc b/test/datarate_test.cc
index f2a2031..6fbcb64 100644
--- a/test/datarate_test.cc
+++ b/test/datarate_test.cc
@@ -59,9 +59,13 @@
/* Test the buffer model here before subtracting the frame. Do so because
* the way the leaky bucket model works in libvpx is to allow the buffer to
* empty - and then stop showing frames until we've got enough bits to
- * show one. */
- ASSERT_GE(bits_in_buffer_model_, 0) << "Buffer Underrun at frame "
- << pkt->data.frame.pts;
+ * show one. As noted in comment below (issue 495), this does not currently
+ * apply to key frames. For now exclude key frames in condition below. */
+ bool key_frame = (pkt->data.frame.flags & VPX_FRAME_IS_KEY) ? true: false;
+ if (!key_frame) {
+ ASSERT_GE(bits_in_buffer_model_, 0) << "Buffer Underrun at frame "
+ << pkt->data.frame.pts;
+ }
const int frame_size_in_bits = pkt->data.frame.sz * 8;
@@ -125,7 +129,12 @@
::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288,
30, 1, 0, 140);
- for (int i = 70; i < 700; i += 200) {
+ // There is an issue for low bitrates in real-time mode, where the
+ // effective_datarate slightly overshoots the target bitrate.
+ // This is same the issue as noted about (#495).
+ // TODO(jimbankoski/marpan): Update test to run for lower bitrates (< 100),
+ // when the issue is resolved.
+ for (int i = 100; i < 800; i += 200) {
cfg_.rc_target_bitrate = i;
ResetModel();
ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
diff --git a/test/decode_test_driver.cc b/test/decode_test_driver.cc
index 3610f025d..84afe7f 100644
--- a/test/decode_test_driver.cc
+++ b/test/decode_test_driver.cc
@@ -9,6 +9,7 @@
*/
#include "test/decode_test_driver.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/register_state_check.h"
#include "test/video_source.h"
namespace libvpx_test {
@@ -21,8 +22,9 @@
ASSERT_EQ(VPX_CODEC_OK, res_init) << DecodeError();
}
- const vpx_codec_err_t res_dec = vpx_codec_decode(&decoder_,
- cxdata, size, NULL, 0);
+ vpx_codec_err_t res_dec;
+ REGISTER_STATE_CHECK(res_dec = vpx_codec_decode(&decoder_,
+ cxdata, size, NULL, 0));
ASSERT_EQ(VPX_CODEC_OK, res_dec) << DecodeError();
}
diff --git a/test/encode_test_driver.cc b/test/encode_test_driver.cc
index ebb3959..56339ca 100644
--- a/test/encode_test_driver.cc
+++ b/test/encode_test_driver.cc
@@ -12,6 +12,7 @@
#if CONFIG_VP8_DECODER
#include "test/decode_test_driver.h"
#endif
+#include "test/register_state_check.h"
#include "test/video_source.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
@@ -58,9 +59,10 @@
}
// Encode the frame
- res = vpx_codec_encode(&encoder_,
- video.img(), video.pts(), video.duration(),
- frame_flags, deadline_);
+ REGISTER_STATE_CHECK(
+ res = vpx_codec_encode(&encoder_,
+ video.img(), video.pts(), video.duration(),
+ frame_flags, deadline_));
ASSERT_EQ(VPX_CODEC_OK, res) << EncoderError();
}
diff --git a/test/idctllm_test.cc b/test/idctllm_test.cc
index dd42e22..1be5fa0 100644
--- a/test/idctllm_test.cc
+++ b/test/idctllm_test.cc
@@ -13,6 +13,7 @@
#include "vpx_config.h"
#include "vpx_rtcd.h"
}
+#include "test/register_state_check.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
typedef void (*idct_fn_t)(short *input, unsigned char *pred_ptr,
@@ -54,7 +55,7 @@
{
int i;
- UUT(input, output, 16, output, 16);
+ REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
for(i=0; i<256; i++)
if((i&0xF) < 4 && i<64)
@@ -68,7 +69,7 @@
int i;
input[0] = 4;
- UUT(input, output, 16, output, 16);
+ REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
for(i=0; i<256; i++)
if((i&0xF) < 4 && i<64)
@@ -85,7 +86,7 @@
predict[i] = i;
input[0] = 4;
- UUT(input, predict, 16, output, 16);
+ REGISTER_STATE_CHECK(UUT(input, predict, 16, output, 16));
for(i=0; i<256; i++)
if((i&0xF) < 4 && i<64)
@@ -101,7 +102,7 @@
for(i=0; i<16; i++)
input[i] = i;
- UUT(input, output, 16, output, 16);
+ REGISTER_STATE_CHECK(UUT(input, output, 16, output, 16));
for(i=0; i<256; i++)
if((i&0xF) > 3 || i>63)
diff --git a/test/intrapred_test.cc b/test/intrapred_test.cc
index d2e0d61..4c16c3f 100644
--- a/test/intrapred_test.cc
+++ b/test/intrapred_test.cc
@@ -11,6 +11,7 @@
#include <string.h>
#include "test/acm_random.h"
+#include "test/register_state_check.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
extern "C" {
#include "vpx_config.h"
@@ -246,8 +247,10 @@
virtual void Predict(MB_PREDICTION_MODE mode) {
mb_.mode_info_context->mbmi.mode = mode;
- pred_fn_(&mb_, data_ptr_[0] - kStride, data_ptr_[0] - 1, kStride,
- data_ptr_[0], kStride);
+ REGISTER_STATE_CHECK(pred_fn_(&mb_,
+ data_ptr_[0] - kStride,
+ data_ptr_[0] - 1, kStride,
+ data_ptr_[0], kStride));
}
intra_pred_y_fn_t pred_fn_;
diff --git a/test/pp_filter_test.cc b/test/pp_filter_test.cc
index af2f3bd..9227449 100644
--- a/test/pp_filter_test.cc
+++ b/test/pp_filter_test.cc
@@ -7,6 +7,7 @@
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
+#include "test/register_state_check.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
extern "C" {
#include "vpx_config.h"
@@ -74,8 +75,8 @@
// Initialize pixels in the output to 99.
(void)vpx_memset(dst_image, 99, output_size);
- GetParam()(src_image_ptr, dst_image_ptr, input_stride,
- output_stride, block_width, flimits, 16);
+ REGISTER_STATE_CHECK(GetParam()(src_image_ptr, dst_image_ptr, input_stride,
+ output_stride, block_width, flimits, 16));
static const uint8_t expected_data[block_height] = {
4, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 4
diff --git a/test/register_state_check.h b/test/register_state_check.h
new file mode 100644
index 0000000..fb3f53b
--- /dev/null
+++ b/test/register_state_check.h
@@ -0,0 +1,95 @@
+/*
+ * Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef LIBVPX_TEST_REGISTER_STATE_CHECK_H_
+#define LIBVPX_TEST_REGISTER_STATE_CHECK_H_
+
+#ifdef _WIN64
+
+#define _WIN32_LEAN_AND_MEAN
+#include <windows.h>
+#include <winnt.h>
+
+#include "third_party/googletest/src/include/gtest/gtest.h"
+
+namespace testing {
+namespace internal {
+
+inline bool operator==(const M128A& lhs, const M128A& rhs) {
+ return (lhs.Low == rhs.Low && lhs.High == rhs.High);
+}
+
+} // namespace internal
+} // namespace testing
+
+namespace libvpx_test {
+
+// Compares the state of xmm[6-15] at construction with their state at
+// destruction. These registers should be preserved by the callee on
+// Windows x64.
+// Usage:
+// {
+// RegisterStateCheck reg_check;
+// FunctionToVerify();
+// }
+class RegisterStateCheck {
+ public:
+ RegisterStateCheck() { initialized_ = StoreRegisters(&pre_context_); }
+ ~RegisterStateCheck() { EXPECT_TRUE(Check()); }
+
+ private:
+ static bool StoreRegisters(CONTEXT* const context) {
+ const HANDLE this_thread = GetCurrentThread();
+ EXPECT_TRUE(this_thread != NULL);
+ context->ContextFlags = CONTEXT_FLOATING_POINT;
+ const bool context_saved = GetThreadContext(this_thread, context) == TRUE;
+ EXPECT_TRUE(context_saved) << "GetLastError: " << GetLastError();
+ return context_saved;
+ }
+
+ // Compares the register state. Returns true if the states match.
+ bool Check() const {
+ if (!initialized_) return false;
+ CONTEXT post_context;
+ if (!StoreRegisters(&post_context)) return false;
+
+ const M128A* xmm_pre = &pre_context_.Xmm6;
+ const M128A* xmm_post = &post_context.Xmm6;
+ for (int i = 6; i <= 15; ++i) {
+ EXPECT_EQ(*xmm_pre, *xmm_post) << "xmm" << i << " has been modified!";
+ ++xmm_pre;
+ ++xmm_post;
+ }
+ return !testing::Test::HasNonfatalFailure();
+ }
+
+ bool initialized_;
+ CONTEXT pre_context_;
+};
+
+#define REGISTER_STATE_CHECK(statement) do { \
+ libvpx_test::RegisterStateCheck reg_check; \
+ statement; \
+} while (false)
+
+} // namespace libvpx_test
+
+#else // !_WIN64
+
+namespace libvpx_test {
+
+class RegisterStateCheck {};
+#define REGISTER_STATE_CHECK(statement) statement
+
+} // namespace libvpx_test
+
+#endif // _WIN64
+
+#endif // LIBVPX_TEST_REGISTER_STATE_CHECK_H_
diff --git a/test/sad_test.cc b/test/sad_test.cc
index 2b562e6..5a0653b 100644
--- a/test/sad_test.cc
+++ b/test/sad_test.cc
@@ -21,6 +21,7 @@
}
#include "test/acm_random.h"
+#include "test/register_state_check.h"
#include "test/util.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
@@ -65,9 +66,11 @@
sad_m_by_n_fn_t sad_fn_;
virtual unsigned int SAD(unsigned int max_sad) {
- return sad_fn_(source_data_, source_stride_,
- reference_data_, reference_stride_,
- max_sad);
+ unsigned int ret;
+ REGISTER_STATE_CHECK(ret = sad_fn_(source_data_, source_stride_,
+ reference_data_, reference_stride_,
+ max_sad));
+ return ret;
}
// Sum of Absolute Differences. Given two blocks, calculate the absolute
diff --git a/test/sixtap_predict_test.cc b/test/sixtap_predict_test.cc
index 06f14a1..c9dcceb 100644
--- a/test/sixtap_predict_test.cc
+++ b/test/sixtap_predict_test.cc
@@ -12,6 +12,7 @@
#include <stdlib.h>
#include <string.h>
#include "test/acm_random.h"
+#include "test/register_state_check.h"
#include "test/util.h"
#include "third_party/googletest/src/include/gtest/gtest.h"
extern "C" {
@@ -136,8 +137,8 @@
uint8_t *src = const_cast<uint8_t*>(test_data);
- sixtap_predict_(&src[kSrcStride * 2 + 2 + 1], kSrcStride,
- 2, 2, dst_, kDstStride);
+ REGISTER_STATE_CHECK(sixtap_predict_(&src[kSrcStride * 2 + 2 + 1], kSrcStride,
+ 2, 2, dst_, kDstStride));
for (int i = 0; i < height_; ++i)
for (int j = 0; j < width_; ++j)
@@ -162,8 +163,9 @@
xoffset, yoffset, dst_c_, kDstStride);
// Run test.
- sixtap_predict_(&src_[kSrcStride * 2 + 2 + 1], kSrcStride,
- xoffset, yoffset, dst_, kDstStride);
+ REGISTER_STATE_CHECK(
+ sixtap_predict_(&src_[kSrcStride * 2 + 2 + 1], kSrcStride,
+ xoffset, yoffset, dst_, kDstStride));
for (int i = 0; i < height_; ++i)
for (int j = 0; j < width_; ++j)
diff --git a/test/subtract_test.cc b/test/subtract_test.cc
index 99363de..60acf81 100644
--- a/test/subtract_test.cc
+++ b/test/subtract_test.cc
@@ -10,6 +10,7 @@
#include "third_party/googletest/src/include/gtest/gtest.h"
#include "test/acm_random.h"
+#include "test/register_state_check.h"
extern "C" {
#include "vpx_config.h"
#include "vpx_rtcd.h"
@@ -77,7 +78,7 @@
predictor += kDiffPredStride;
}
- GetParam()(&be, &bd, kDiffPredStride);
+ REGISTER_STATE_CHECK(GetParam()(&be, &bd, kDiffPredStride));
base_src = *be.base_src;
src_diff = be.src_diff;
diff --git a/test/test.mk b/test/test.mk
index 7a11a27..982be5b 100644
--- a/test/test.mk
+++ b/test/test.mk
@@ -1,4 +1,5 @@
LIBVPX_TEST_SRCS-yes += acm_random.h
+LIBVPX_TEST_SRCS-yes += register_state_check.h
LIBVPX_TEST_SRCS-yes += test.mk
LIBVPX_TEST_SRCS-yes += test_libvpx.cc
LIBVPX_TEST_SRCS-yes += util.h
diff --git a/vp8/common/x86/loopfilter_block_sse2.asm b/vp8/common/x86/loopfilter_block_sse2.asm
index 1c445ef..3d45c61 100644
--- a/vp8/common/x86/loopfilter_block_sse2.asm
+++ b/vp8/common/x86/loopfilter_block_sse2.asm
@@ -150,6 +150,7 @@
push rbp
mov rbp, rsp
+ SAVE_XMM 11
push r12
push r13
mov thresh, arg(4)
@@ -258,6 +259,7 @@
%ifidn __OUTPUT_FORMAT__,x64
pop r13
pop r12
+ RESTORE_XMM
pop rbp
%endif
diff --git a/vp8/common/x86/recon_sse2.asm b/vp8/common/x86/recon_sse2.asm
index fe77450..1434bcd 100644
--- a/vp8/common/x86/recon_sse2.asm
+++ b/vp8/common/x86/recon_sse2.asm
@@ -890,6 +890,7 @@
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 5
+ SAVE_XMM 7
push rsi
push rdi
GET_GOT rbx
@@ -957,6 +958,7 @@
RESTORE_GOT
pop rdi
pop rsi
+ RESTORE_XMM
UNSHADOW_ARGS
pop rbp
ret
diff --git a/vp8/common/x86/subpixel_ssse3.asm b/vp8/common/x86/subpixel_ssse3.asm
index 13bcaf6..c06f245 100644
--- a/vp8/common/x86/subpixel_ssse3.asm
+++ b/vp8/common/x86/subpixel_ssse3.asm
@@ -352,6 +352,7 @@
pop rdi
pop rsi
RESTORE_GOT
+ RESTORE_XMM
UNSHADOW_ARGS
pop rbp
ret
diff --git a/vp8/decoder/threading.c b/vp8/decoder/threading.c
index e52a707..88c06be 100644
--- a/vp8/decoder/threading.c
+++ b/vp8/decoder/threading.c
@@ -29,6 +29,13 @@
#include "error_concealment.h"
#endif
+#define CALLOC_ARRAY(p, n) CHECK_MEM_ERROR((p), vpx_calloc(sizeof(*(p)), (n)))
+#define CALLOC_ARRAY_ALIGNED(p, n, algn) do { \
+ CHECK_MEM_ERROR((p), vpx_memalign((algn), sizeof(*(p)) * (n))); \
+ memset((p), 0, (n) * sizeof(*(p))); \
+} while (0)
+
+
extern void vp8_mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd);
static void setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC *mbrd, int count)
@@ -668,11 +675,10 @@
pbi->b_multithreaded_rd = 1;
pbi->decoding_thread_count = core_count - 1;
- CHECK_MEM_ERROR(pbi->h_decoding_thread, vpx_malloc(sizeof(pthread_t) * pbi->decoding_thread_count));
- CHECK_MEM_ERROR(pbi->h_event_start_decoding, vpx_malloc(sizeof(sem_t) * pbi->decoding_thread_count));
- CHECK_MEM_ERROR(pbi->mb_row_di, vpx_memalign(32, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count));
- vpx_memset(pbi->mb_row_di, 0, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count);
- CHECK_MEM_ERROR(pbi->de_thread_data, vpx_malloc(sizeof(DECODETHREAD_DATA) * pbi->decoding_thread_count));
+ CALLOC_ARRAY(pbi->h_decoding_thread, pbi->decoding_thread_count);
+ CALLOC_ARRAY(pbi->h_event_start_decoding, pbi->decoding_thread_count);
+ CALLOC_ARRAY_ALIGNED(pbi->mb_row_di, pbi->decoding_thread_count, 32);
+ CALLOC_ARRAY(pbi->de_thread_data, pbi->decoding_thread_count);
for (ithread = 0; ithread < pbi->decoding_thread_count; ithread++)
{
@@ -796,32 +802,32 @@
uv_width = width >>1;
/* Allocate an int for each mb row. */
- CHECK_MEM_ERROR(pbi->mt_current_mb_col, vpx_malloc(sizeof(int) * pc->mb_rows));
+ CALLOC_ARRAY(pbi->mt_current_mb_col, pc->mb_rows);
/* Allocate memory for above_row buffers. */
- CHECK_MEM_ERROR(pbi->mt_yabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_yabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_yabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (width + (VP8BORDERINPIXELS<<1))));
- CHECK_MEM_ERROR(pbi->mt_uabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_uabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_uabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS)));
- CHECK_MEM_ERROR(pbi->mt_vabove_row, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_vabove_row, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_vabove_row[i], vpx_memalign(16,sizeof(unsigned char) * (uv_width + VP8BORDERINPIXELS)));
/* Allocate memory for left_col buffers. */
- CHECK_MEM_ERROR(pbi->mt_yleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_yleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_yleft_col[i], vpx_calloc(sizeof(unsigned char) * 16, 1));
- CHECK_MEM_ERROR(pbi->mt_uleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_uleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_uleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
- CHECK_MEM_ERROR(pbi->mt_vleft_col, vpx_malloc(sizeof(unsigned char *) * pc->mb_rows));
- for (i=0; i< pc->mb_rows; i++)
+ CALLOC_ARRAY(pbi->mt_vleft_col, pc->mb_rows);
+ for (i = 0; i < pc->mb_rows; i++)
CHECK_MEM_ERROR(pbi->mt_vleft_col[i], vpx_calloc(sizeof(unsigned char) * 8, 1));
}
}
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 6af1ef8..1a234ca 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -33,7 +33,7 @@
#endif
#include "encodeframe.h"
-extern void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) ;
+extern void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) ;
extern void vp8_calc_ref_frame_costs(int *ref_frame_cost,
int prob_intra,
int prob_last,
@@ -1389,7 +1389,7 @@
}
else
{
- vp8_stuff_mb(cpi, xd, t);
+ vp8_stuff_mb(cpi, x, t);
}
}
diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c
index 77c1c5a..d2b7872 100644
--- a/vp8/encoder/ratectrl.c
+++ b/vp8/encoder/ratectrl.c
@@ -1109,7 +1109,9 @@
}
else
{
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ if (cpi->oxcf.number_of_layers == 1 &&
+ (cpi->common.refresh_alt_ref_frame ||
+ cpi->common.refresh_golden_frame))
rate_correction_factor = cpi->gf_rate_correction_factor;
else
rate_correction_factor = cpi->rate_correction_factor;
@@ -1186,7 +1188,9 @@
cpi->key_frame_rate_correction_factor = rate_correction_factor;
else
{
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ if (cpi->oxcf.number_of_layers == 1 &&
+ (cpi->common.refresh_alt_ref_frame ||
+ cpi->common.refresh_golden_frame))
cpi->gf_rate_correction_factor = rate_correction_factor;
else
cpi->rate_correction_factor = rate_correction_factor;
@@ -1209,11 +1213,13 @@
{
Q = cpi->oxcf.key_q;
}
- else if (cpi->common.refresh_alt_ref_frame)
+ else if (cpi->oxcf.number_of_layers == 1 &&
+ cpi->common.refresh_alt_ref_frame)
{
Q = cpi->oxcf.alt_q;
}
- else if (cpi->common.refresh_golden_frame)
+ else if (cpi->oxcf.number_of_layers == 1 &&
+ cpi->common.refresh_golden_frame)
{
Q = cpi->oxcf.gold_q;
}
@@ -1232,7 +1238,9 @@
correction_factor = cpi->key_frame_rate_correction_factor;
else
{
- if (cpi->common.refresh_alt_ref_frame || cpi->common.refresh_golden_frame)
+ if (cpi->oxcf.number_of_layers == 1 &&
+ (cpi->common.refresh_alt_ref_frame ||
+ cpi->common.refresh_golden_frame))
correction_factor = cpi->gf_rate_correction_factor;
else
correction_factor = cpi->rate_correction_factor;
@@ -1281,7 +1289,10 @@
if (cpi->common.frame_type == KEY_FRAME)
zbin_oqmax = 0;
- else if (cpi->common.refresh_alt_ref_frame || (cpi->common.refresh_golden_frame && !cpi->source_alt_ref_active))
+ else if (cpi->oxcf.number_of_layers == 1 &&
+ (cpi->common.refresh_alt_ref_frame ||
+ (cpi->common.refresh_golden_frame &&
+ !cpi->source_alt_ref_active)))
zbin_oqmax = 16;
else
zbin_oqmax = ZBIN_OQ_MAX;
diff --git a/vp8/vp8cx.mk b/vp8/vp8cx.mk
index 5976297..0ae2f10 100644
--- a/vp8/vp8cx.mk
+++ b/vp8/vp8cx.mk
@@ -22,16 +22,9 @@
include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx_arm.mk
endif
-VP8_CX_SRCS-yes += vp8_cx_iface.c
+VP8_CX_SRCS-yes += vp8cx.mk
-# encoder
-#INCLUDES += algo/vpx_common/vpx_mem/include
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += common
-#INCLUDES += algo/vpx_ref/cpu_id/include
-#INCLUDES += common
-#INCLUDES += encoder
+VP8_CX_SRCS-yes += vp8_cx_iface.c
VP8_CX_SRCS-yes += encoder/asm_enc_offsets.c
VP8_CX_SRCS-yes += encoder/defaultcoefcounts.h
diff --git a/vp8/vp8cx_arm.mk b/vp8/vp8cx_arm.mk
index b16615d..b030ee5 100644
--- a/vp8/vp8cx_arm.mk
+++ b/vp8/vp8cx_arm.mk
@@ -9,7 +9,7 @@
##
-#VP8_CX_SRCS list is modified according to different platforms.
+VP8_CX_SRCS-$(ARCH_ARM) += vp8cx_arm.mk
#File list for arm
# encoder
diff --git a/vp8/vp8dx.mk b/vp8/vp8dx.mk
index 2cfd280..dd39190 100644
--- a/vp8/vp8dx.mk
+++ b/vp8/vp8dx.mk
@@ -18,6 +18,8 @@
VP8_DX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes)
VP8_DX_SRCS_REMOVE-no += $(VP8_COMMON_SRCS_REMOVE-no)
+VP8_DX_SRCS-yes += vp8dx.mk
+
VP8_DX_SRCS-yes += vp8_dx_iface.c
# common
diff --git a/vpx/vpx_codec.h b/vpx/vpx_codec.h
index d92e165..243b7a5 100644
--- a/vpx/vpx_codec.h
+++ b/vpx/vpx_codec.h
@@ -49,15 +49,22 @@
#ifndef DEPRECATED
#if defined(__GNUC__) && __GNUC__
#define DEPRECATED __attribute__ ((deprecated))
-#define DECLSPEC_DEPRECATED /**< \copydoc #DEPRECATED */
#elif defined(_MSC_VER)
#define DEPRECATED
-#define DECLSPEC_DEPRECATED __declspec(deprecated) /**< \copydoc #DEPRECATED */
#else
#define DEPRECATED
+#endif
+#endif /* DEPRECATED */
+
+#ifndef DECLSPEC_DEPRECATED
+#if defined(__GNUC__) && __GNUC__
+#define DECLSPEC_DEPRECATED /**< \copydoc #DEPRECATED */
+#elif defined(_MSC_VER)
+#define DECLSPEC_DEPRECATED __declspec(deprecated) /**< \copydoc #DEPRECATED */
+#else
#define DECLSPEC_DEPRECATED /**< \copydoc #DEPRECATED */
#endif
-#endif
+#endif /* DECLSPEC_DEPRECATED */
/*!\brief Decorator indicating a function is potentially unused */
#ifdef UNUSED
diff --git a/vpx/vpx_codec.mk b/vpx/vpx_codec.mk
index 427fd0f..ffa123f 100644
--- a/vpx/vpx_codec.mk
+++ b/vpx/vpx_codec.mk
@@ -11,6 +11,21 @@
API_EXPORTS += exports
+API_SRCS-$(CONFIG_VP8_ENCODER) += vp8.h
+API_SRCS-$(CONFIG_VP8_ENCODER) += vp8cx.h
+API_DOC_SRCS-$(CONFIG_VP8_ENCODER) += vp8.h
+API_DOC_SRCS-$(CONFIG_VP8_ENCODER) += vp8cx.h
+
+API_SRCS-$(CONFIG_VP8_DECODER) += vp8.h
+API_SRCS-$(CONFIG_VP8_DECODER) += vp8dx.h
+API_DOC_SRCS-$(CONFIG_VP8_DECODER) += vp8.h
+API_DOC_SRCS-$(CONFIG_VP8_DECODER) += vp8dx.h
+
+API_DOC_SRCS-yes += vpx_codec.h
+API_DOC_SRCS-yes += vpx_decoder.h
+API_DOC_SRCS-yes += vpx_encoder.h
+API_DOC_SRCS-yes += vpx_image.h
+
API_SRCS-yes += src/vpx_decoder.c
API_SRCS-yes += vpx_decoder.h
API_SRCS-yes += src/vpx_encoder.c
@@ -23,3 +38,4 @@
API_SRCS-yes += vpx_codec_impl_bottom.h
API_SRCS-yes += vpx_codec_impl_top.h
API_SRCS-yes += vpx_image.h
+API_SRCS-$(BUILD_LIBVPX) += vpx_integer.h
diff --git a/vpx_ports/arm_cpudetect.c b/vpx_ports/arm_cpudetect.c
index 8ff95a1..74d8a1f 100644
--- a/vpx_ports/arm_cpudetect.c
+++ b/vpx_ports/arm_cpudetect.c
@@ -125,7 +125,11 @@
}
#elif defined(__ANDROID__) /* end _MSC_VER */
+#if defined(CHROMIUM_BUILD)
+#include <machine/cpu-features.h>
+#else
#include <cpu-features.h>
+#endif
int arm_cpu_caps(void)
{
diff --git a/vpx_ports/vpx_ports.mk b/vpx_ports/vpx_ports.mk
new file mode 100644
index 0000000..e6cb52f
--- /dev/null
+++ b/vpx_ports/vpx_ports.mk
@@ -0,0 +1,26 @@
+##
+## Copyright (c) 2012 The WebM project authors. All Rights Reserved.
+##
+## Use of this source code is governed by a BSD-style license
+## that can be found in the LICENSE file in the root of the source
+## tree. An additional intellectual property rights grant can be found
+## in the file PATENTS. All contributing project authors may
+## be found in the AUTHORS file in the root of the source tree.
+##
+
+
+PORTS_SRCS-yes += vpx_ports.mk
+
+PORTS_SRCS-$(BUILD_LIBVPX) += asm_offsets.h
+PORTS_SRCS-$(BUILD_LIBVPX) += mem.h
+PORTS_SRCS-$(BUILD_LIBVPX) += vpx_timer.h
+
+ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
+PORTS_SRCS-$(BUILD_LIBVPX) += emms.asm
+PORTS_SRCS-$(BUILD_LIBVPX) += x86.h
+PORTS_SRCS-$(BUILD_LIBVPX) += x86_abi_support.asm
+PORTS_SRCS-$(BUILD_LIBVPX) += x86_cpuid.c
+endif
+
+PORTS_SRCS-$(ARCH_ARM) += arm_cpudetect.c
+PORTS_SRCS-$(ARCH_ARM) += arm.h
diff --git a/vpxenc.c b/vpxenc.c
index 7449e6c..c9547ea 100644
--- a/vpxenc.c
+++ b/vpxenc.c
@@ -2284,8 +2284,9 @@
break;
case VPX_CODEC_STATS_PKT:
stream->frames_out++;
- fprintf(stderr, " %6luS",
- (unsigned long)pkt->data.twopass_stats.sz);
+ if (!global->quiet)
+ fprintf(stderr, " %6luS",
+ (unsigned long)pkt->data.twopass_stats.sz);
stats_write(&stream->stats,
pkt->data.twopass_stats.buf,
pkt->data.twopass_stats.sz);
