dsputil: Make dsputil selectable

Signed-off-by: Martin Storsjö <martin@martin.st>

9 files changed, 64 insertions, 54 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 0b7fb9690f..e6379a515d 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -17,17 +17,13 @@ OBJS = allcodecs.o                                                      \
        bitstream.o                                                      \
        bitstream_filter.o                                               \
        codec_desc.o                                                     \
-       dsputil.o                                                        \
-       faanidct.o                                                       \
        fmtconvert.o                                                     \
        imgconvert.o                                                     \
-       jrevdct.o                                                        \
        log2_tab.o                                                       \
        mathtables.o                                                     \
        options.o                                                        \
        parser.o                                                         \
        raw.o                                                            \
-       simple_idct.o                                                    \
        utils.o                                                          \
 
 # parts needed for many different codecs
@@ -36,6 +32,8 @@ OBJS-$(CONFIG_AC3DSP)                  += ac3dsp.o
 OBJS-$(CONFIG_AUDIO_FRAME_QUEUE)       += audio_frame_queue.o
 OBJS-$(CONFIG_DCT)                     += dct.o dct32_fixed.o dct32_float.o
 OBJS-$(CONFIG_DXVA2)                   += dxva2.o
+OBJS-$(CONFIG_DSPUTIL)                 += dsputil.o faanidct.o          \
+                                          simple_idct.o jrevdct.o
 OBJS-$(CONFIG_ENCODERS)                += faandct.o jfdctfst.o jfdctint.o
 OBJS-$(CONFIG_ERROR_RESILIENCE)        += error_resilience.o
 FFT-OBJS-$(CONFIG_HARDCODED_TABLES)    += cos_tables.o cos_fixed_tables.o
diff --git a/libavcodec/utils.c b/libavcodec/utils.c
index 2b63759b90..46486a2ff7 100644
--- a/libavcodec/utils.c
+++ b/libavcodec/utils.c
@@ -117,7 +117,8 @@ static void avcodec_init(void)
         return;
     initialized = 1;
 
-    ff_dsputil_static_init();
+    if (CONFIG_DSPUTIL)
+        ff_dsputil_static_init();
 }
 
 int av_codec_is_encoder(const AVCodec *codec)
diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index cc60f2f0b3..61672b8041 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -1,4 +1,5 @@
-OBJS                                   += x86/fmtconvert_init.o
+OBJS                                   += x86/constants.o               \
+                                          x86/fmtconvert_init.o         \
 
 OBJS-$(CONFIG_AAC_DECODER)             += x86/sbrdsp_init.o
 OBJS-$(CONFIG_AC3DSP)                  += x86/ac3dsp_init.o
@@ -29,7 +30,7 @@ OBJS-$(CONFIG_VP6_DECODER)             += x86/vp56dsp_init.o
 OBJS-$(CONFIG_VP8_DECODER)             += x86/vp8dsp_init.o
 OBJS-$(CONFIG_XMM_CLOBBER_TEST)        += x86/w64xmmtest.o
 
-MMX-OBJS                               += x86/dsputil_mmx.o             \
+MMX-OBJS-$(CONFIG_DSPUTIL)             += x86/dsputil_mmx.o             \
                                           x86/fdct.o                    \
                                           x86/idct_mmx_xvid.o           \
                                           x86/idct_sse2_xvid.o          \
@@ -42,6 +43,10 @@ MMX-OBJS-$(CONFIG_VC1_DECODER)         += x86/vc1dsp_mmx.o
 YASM-OBJS-$(CONFIG_AAC_DECODER)        += x86/sbrdsp.o
 YASM-OBJS-$(CONFIG_AC3DSP)             += x86/ac3dsp.o
 YASM-OBJS-$(CONFIG_DCT)                += x86/dct32.o
+YASM-OBJS-$(CONFIG_DSPUTIL)            += x86/dsputil.o                 \
+                                          x86/hpeldsp.o                 \
+                                          x86/mpeg4qpel.o               \
+                                          x86/qpel.o
 YASM-OBJS-$(CONFIG_ENCODERS)           += x86/dsputilenc.o
 YASM-OBJS-$(CONFIG_FFT)                += x86/fft.o
 YASM-OBJS-$(CONFIG_H263_DECODER)       += x86/h263_loopfilter.o
@@ -57,7 +62,8 @@ YASM-OBJS-$(CONFIG_H264DSP)            += x86/h264_deblock.o            \
 YASM-OBJS-$(CONFIG_H264PRED)           += x86/h264_intrapred.o          \
                                           x86/h264_intrapred_10bit.o
 YASM-OBJS-$(CONFIG_H264QPEL)           += x86/h264_qpel_8bit.o          \
-                                          x86/h264_qpel_10bit.o
+                                          x86/h264_qpel_10bit.o         \
+                                          x86/qpel.o
 YASM-OBJS-$(CONFIG_MPEGAUDIODSP)       += x86/imdct36.o
 YASM-OBJS-$(CONFIG_PNG_DECODER)        += x86/pngdsp.o
 YASM-OBJS-$(CONFIG_PRORES_DECODER)     += x86/proresdsp.o
@@ -71,9 +77,5 @@ YASM-OBJS-$(CONFIG_VP3DSP)             += x86/vp3dsp.o
 YASM-OBJS-$(CONFIG_VP6_DECODER)        += x86/vp56dsp.o
 YASM-OBJS-$(CONFIG_VP8_DECODER)        += x86/vp8dsp.o
 
-YASM-OBJS                              += x86/dsputil.o                 \
-                                          x86/deinterlace.o             \
+YASM-OBJS                              += x86/deinterlace.o             \
                                           x86/fmtconvert.o              \
-                                          x86/hpeldsp.o                 \
-                                          x86/mpeg4qpel.o               \
-                                          x86/qpel.o                    \
diff --git a/libavcodec/x86/constants.c b/libavcodec/x86/constants.c
new file mode 100644
index 0000000000..f4d04729ae
--- /dev/null
+++ b/libavcodec/x86/constants.c
@@ -0,0 +1,42 @@
+/*
+ * MMX/SSE constants used across x86 dsp optimizations.
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/mem.h"
+#include "libavutil/x86/asm.h" // for xmm_reg
+
+DECLARE_ALIGNED(8,  const uint64_t, ff_bone) = 0x0101010101010101ULL;
+DECLARE_ALIGNED(8,  const uint64_t, ff_wtwo) = 0x0002000200020002ULL;
+
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_1)    = { 0x0001000100010001ULL, 0x0001000100010001ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_2)    = { 0x0002000200020002ULL, 0x0002000200020002ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_3)    = { 0x0003000300030003ULL, 0x0003000300030003ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_4)    = { 0x0004000400040004ULL, 0x0004000400040004ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_5)    = { 0x0005000500050005ULL, 0x0005000500050005ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_8)    = { 0x0008000800080008ULL, 0x0008000800080008ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_9)    = { 0x0009000900090009ULL, 0x0009000900090009ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_16)   = { 0x0010001000100010ULL, 0x0010001000100010ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_18)   = { 0x0012001200120012ULL, 0x0012001200120012ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_32)   = { 0x0020002000200020ULL, 0x0020002000200020ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_64)   = { 0x0040004000400040ULL, 0x0040004000400040ULL };
+
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_0)    = { 0x0000000000000000ULL, 0x0000000000000000ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_1)    = { 0x0101010101010101ULL, 0x0101010101010101ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_3)    = { 0x0303030303030303ULL, 0x0303030303030303ULL };
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_80)   = { 0x8080808080808080ULL, 0x8080808080808080ULL };
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index aa7b3984aa..7af11b0e68 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -36,47 +36,22 @@
 //#include <assert.h>
 
 /* pixel operations */
-DECLARE_ALIGNED(8,  const uint64_t, ff_bone) = 0x0101010101010101ULL;
-DECLARE_ALIGNED(8,  const uint64_t, ff_wtwo) = 0x0002000200020002ULL;
-
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_1)    = { 0x0001000100010001ULL, 0x0001000100010001ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_2)    = { 0x0002000200020002ULL, 0x0002000200020002ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_3)    = { 0x0003000300030003ULL, 0x0003000300030003ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_4)    = { 0x0004000400040004ULL, 0x0004000400040004ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_5)    = { 0x0005000500050005ULL, 0x0005000500050005ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_8)    = { 0x0008000800080008ULL, 0x0008000800080008ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_9)    = { 0x0009000900090009ULL, 0x0009000900090009ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_15)   =   0x000F000F000F000FULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_16)   = { 0x0010001000100010ULL, 0x0010001000100010ULL };
 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_17)   = { 0x0011001100110011ULL, 0x0011001100110011ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_18)   = { 0x0012001200120012ULL, 0x0012001200120012ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_20)   =   0x0014001400140014ULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_27)   = { 0x001B001B001B001BULL, 0x001B001B001B001BULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_28)   = { 0x001C001C001C001CULL, 0x001C001C001C001CULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_32)   = { 0x0020002000200020ULL, 0x0020002000200020ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_42)   =   0x002A002A002A002AULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_53)   =   0x0035003500350035ULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_63)   = { 0x003F003F003F003FULL, 0x003F003F003F003FULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_64)   = { 0x0040004000400040ULL, 0x0040004000400040ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_96)   =   0x0060006000600060ULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_128)  =   0x0080008000800080ULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_255)  =   0x00ff00ff00ff00ffULL;
 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_512)  = { 0x0200020002000200ULL, 0x0200020002000200ULL };
 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_1019) = { 0x03FB03FB03FB03FBULL, 0x03FB03FB03FB03FBULL };
 
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_0)    = { 0x0000000000000000ULL, 0x0000000000000000ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_1)    = { 0x0101010101010101ULL, 0x0101010101010101ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_3)    = { 0x0303030303030303ULL, 0x0303030303030303ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_4)    = { 0x0404040404040404ULL, 0x0404040404040404ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_7)    =   0x0707070707070707ULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_1F)   =   0x1F1F1F1F1F1F1F1FULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_3F)   =   0x3F3F3F3F3F3F3F3FULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_80)   = { 0x8080808080808080ULL, 0x8080808080808080ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_81)   =   0x8181818181818181ULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_A1)   = { 0xA1A1A1A1A1A1A1A1ULL, 0xA1A1A1A1A1A1A1A1ULL };
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_F8)   = { 0xF8F8F8F8F8F8F8F8ULL, 0xF8F8F8F8F8F8F8F8ULL };
 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_FC)   =   0xFCFCFCFCFCFCFCFCULL;
-DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_FE)   = { 0xFEFEFEFEFEFEFEFEULL, 0xFEFEFEFEFEFEFEFEULL };
 
 DECLARE_ALIGNED(16, const double, ff_pd_1)[2] = { 1.0, 1.0 };
 DECLARE_ALIGNED(16, const double, ff_pd_2)[2] = { 2.0, 2.0 };
diff --git a/libavcodec/x86/dsputil_mmx.h b/libavcodec/x86/dsputil_mmx.h
index e4880c0976..d4a7754fc9 100644
--- a/libavcodec/x86/dsputil_mmx.h
+++ b/libavcodec/x86/dsputil_mmx.h
@@ -28,8 +28,6 @@
 #include "libavcodec/dsputil.h"
 #include "libavutil/x86/asm.h"
 
-typedef struct xmm_reg { uint64_t a, b; } xmm_reg;
-
 extern const uint64_t ff_bone;
 extern const uint64_t ff_wtwo;
 
@@ -41,12 +39,9 @@ extern const uint64_t ff_pw_15;
 extern const xmm_reg  ff_pw_16;
 extern const xmm_reg  ff_pw_18;
 extern const uint64_t ff_pw_20;
-extern const xmm_reg  ff_pw_27;
-extern const xmm_reg  ff_pw_28;
 extern const xmm_reg  ff_pw_32;
 extern const uint64_t ff_pw_42;
 extern const uint64_t ff_pw_53;
-extern const xmm_reg  ff_pw_63;
 extern const xmm_reg  ff_pw_64;
 extern const uint64_t ff_pw_96;
 extern const uint64_t ff_pw_128;
@@ -58,10 +53,8 @@ extern const uint64_t ff_pb_7;
 extern const uint64_t ff_pb_1F;
 extern const uint64_t ff_pb_3F;
 extern const uint64_t ff_pb_81;
-extern const xmm_reg  ff_pb_A1;
 extern const xmm_reg  ff_pb_F8;
 extern const uint64_t ff_pb_FC;
-extern const xmm_reg  ff_pb_FE;
 
 extern const double ff_pd_1[2];
 extern const double ff_pd_2[2];
diff --git a/libavcodec/x86/h264_chromamc.asm b/libavcodec/x86/h264_chromamc.asm
index 440a473c6f..b7b18e03f8 100644
--- a/libavcodec/x86/h264_chromamc.asm
+++ b/libavcodec/x86/h264_chromamc.asm
@@ -60,7 +60,7 @@ rnd_rv40_1d_tbl: times 4 dw  0
 cextern pw_3
 cextern pw_4
 cextern pw_8
-cextern pw_28
+pw_28: times 8 dw 28
 cextern pw_32
 cextern pw_64
 
diff --git a/libavcodec/x86/h264_deblock.asm b/libavcodec/x86/h264_deblock.asm
index 67f9c02464..fc6c983052 100644
--- a/libavcodec/x86/h264_deblock.asm
+++ b/libavcodec/x86/h264_deblock.asm
@@ -28,6 +28,7 @@
 
 SECTION_RODATA
 
+pb_A1: times 16 db 0xA1
 pb_3_1: times 4 db 3, 1
 
 SECTION .text
@@ -35,7 +36,6 @@ SECTION .text
 cextern pb_0
 cextern pb_1
 cextern pb_3
-cextern pb_A1
 
 ; expands to [base],...,[base+7*stride]
 %define PASS8ROWS(base, base3, stride, stride3) \
diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
index caf2cd679a..1d7aadc809 100644
--- a/libavcodec/x86/vp8dsp.asm
+++ b/libavcodec/x86/vp8dsp.asm
@@ -143,11 +143,15 @@ filter_h6_shuf1: db 0, 5, 1, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11,  7, 12
 filter_h6_shuf2: db 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,  7, 7,  8,  8,  9
 filter_h6_shuf3: db 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8,  9, 9, 10, 10, 11
 
-pw_256:  times 8 dw 256
-
+pw_27:    times 8 dw 27
+pw_63:    times 8 dw 63
+pw_256:   times 8 dw 256
 pw_20091: times 4 dw 20091
 pw_17734: times 4 dw 17734
 
+pb_4:     times 16 db 4
+pb_F8:    times 16 db 0xF8
+pb_FE:    times 16 db 0xFE
 pb_27_63: times 8 db 27, 63
 pb_18_63: times 8 db 18, 63
 pb_9_63:  times 8 db  9, 63
@@ -156,15 +160,10 @@ cextern pb_1
 cextern pw_3
 cextern pb_3
 cextern pw_4
-cextern pb_4
 cextern pw_9
 cextern pw_18
-cextern pw_27
-cextern pw_63
 cextern pw_64
 cextern pb_80
-cextern pb_F8
-cextern pb_FE
 
 SECTION .text