Merge remote-tracking branch 'qatar/master'

* qatar/master:
  dsputil: fix invalid array indexing
  configure: add libavresample to rpath
  build: icc: silence some warnings
  fft-test: add option to set cpuflag mask
  cpu: recognise only cpu flag names pertinent to the architecture
  avutil: add av_parse_cpu_flags() function
  vp8: armv6: fix non-armv6t2 build
  vp8: armv6 optimisations
  vp8: arm: separate ARMv6 functions from NEON
  ARM: add some compatibility macros
  mov: support eac3 audio
  avf: fix faulty check in has_duration

Conflicts:
	configure
	doc/APIchanges
	ffmpeg.c
	libavcodec/arm/Makefile
	libavcodec/arm/asm.S
	libavcodec/arm/vp8dsp_armv6.S
	libavcodec/arm/vp8dsp_init_arm.c
	libavutil/avutil.h

Merged-by: Michael Niedermayer <[email protected]>

1 files changed, 5 insertions, 285 deletions

diff --git a/libavcodec/arm/vp8dsp_init_arm.c b/libavcodec/arm/vp8dsp_init_arm.c
index 2b105dbe1d..5c84ed3231 100644
--- a/libavcodec/arm/vp8dsp_init_arm.c
+++ b/libavcodec/arm/vp8dsp_init_arm.c
@@ -20,294 +20,14 @@
 
 #include "libavutil/arm/cpu.h"
 #include "libavcodec/vp8dsp.h"
-
-void ff_vp8_luma_dc_wht_dc_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
-
-#define idct_funcs(opt) \
-void ff_vp8_luma_dc_wht_ ## opt(DCTELEM block[4][4][16], DCTELEM dc[16]); \
-void ff_vp8_idct_add_ ## opt(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride); \
-void ff_vp8_idct_dc_add_ ## opt(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride); \
-void ff_vp8_idct_dc_add4y_ ## opt(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride); \
-void ff_vp8_idct_dc_add4uv_ ## opt(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)
-
-idct_funcs(neon);
-idct_funcs(armv6);
-
-void ff_vp8_v_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
-                                 int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
-                                 int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_v_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                  int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                  int flim_E, int flim_I, int hev_thresh);
-
-void ff_vp8_v_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
-                                       int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_h_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
-                                       int flim_E, int flim_I, int hev_thresh);
-void ff_vp8_v_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
-                                        ptrdiff_t stride, int flim_E, int flim_I,
-                                        int hev_thresh);
-void ff_vp8_h_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
-                                        ptrdiff_t stride, int flim_E, int flim_I,
-                                        int hev_thresh);
-
-void ff_vp8_v_loop_filter_inner_armv6(uint8_t *dst, ptrdiff_t stride,
-                                      int flim_E, int flim_I,
-                                      int hev_thresh, int count);
-void ff_vp8_h_loop_filter_inner_armv6(uint8_t *dst, ptrdiff_t stride,
-                                      int flim_E, int flim_I,
-                                      int hev_thresh, int count);
-void ff_vp8_v_loop_filter_armv6(uint8_t *dst, ptrdiff_t stride,
-                                int flim_E, int flim_I,
-                                int hev_thresh, int count);
-void ff_vp8_h_loop_filter_armv6(uint8_t *dst, ptrdiff_t stride,
-                                int flim_E, int flim_I,
-                                int hev_thresh, int count);
-
-static void ff_vp8_v_loop_filter16_armv6(uint8_t *dst, ptrdiff_t stride,
-                                         int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_v_loop_filter_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
-}
-
-static void ff_vp8_h_loop_filter16_armv6(uint8_t *dst, ptrdiff_t stride,
-                                         int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_h_loop_filter_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
-}
-
-static void ff_vp8_v_loop_filter8uv_armv6(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                          int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_v_loop_filter_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
-    ff_vp8_v_loop_filter_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
-}
-
-static void ff_vp8_h_loop_filter8uv_armv6(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
-                                          int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_h_loop_filter_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
-    ff_vp8_h_loop_filter_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
-}
-
-static void ff_vp8_v_loop_filter16_inner_armv6(uint8_t *dst, ptrdiff_t stride,
-                                               int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_v_loop_filter_inner_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
-}
-
-static void ff_vp8_h_loop_filter16_inner_armv6(uint8_t *dst, ptrdiff_t stride,
-                                               int flim_E, int flim_I, int hev_thresh)
-{
-    ff_vp8_h_loop_filter_inner_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
-}
-
-static void ff_vp8_v_loop_filter8uv_inner_armv6(uint8_t *dstU, uint8_t *dstV,
-                                                ptrdiff_t stride, int flim_E, int flim_I,
-                                                int hev_thresh)
-{
-    ff_vp8_v_loop_filter_inner_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
-    ff_vp8_v_loop_filter_inner_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
-}
-
-static void ff_vp8_h_loop_filter8uv_inner_armv6(uint8_t *dstU, uint8_t *dstV,
-                                                ptrdiff_t stride, int flim_E, int flim_I,
-                                                int hev_thresh)
-{
-    ff_vp8_h_loop_filter_inner_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
-    ff_vp8_h_loop_filter_inner_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
-}
-
-#define simple_lf_funcs(opt) \
-void ff_vp8_v_loop_filter16_simple_ ## opt(uint8_t *dst, ptrdiff_t stride, int flim); \
-void ff_vp8_h_loop_filter16_simple_ ## opt(uint8_t *dst, ptrdiff_t stride, int flim)
-
-simple_lf_funcs(neon);
-simple_lf_funcs(armv6);
-
-#define VP8_MC_OPT(n, opt)                                               \
-    void ff_put_vp8_##n##_##opt(uint8_t *dst, ptrdiff_t dststride,       \
-                                uint8_t *src, ptrdiff_t srcstride,       \
-                                int h, int x, int y)
-
-#define VP8_MC(n) \
-    VP8_MC_OPT(n, neon)
-
-#define VP8_EPEL(w)                             \
-    VP8_MC(epel ## w ## _h4);                   \
-    VP8_MC(epel ## w ## _h6);                   \
-    VP8_MC(epel ## w ## _h4v4);                 \
-    VP8_MC(epel ## w ## _h6v4);                 \
-    VP8_MC(epel ## w ## _v4);                   \
-    VP8_MC(epel ## w ## _v6);                   \
-    VP8_MC(epel ## w ## _h4v6);                 \
-    VP8_MC(epel ## w ## _h6v6)
-
-VP8_EPEL(16);
-VP8_MC(pixels16);
-VP8_MC_OPT(pixels16, armv6);
-VP8_EPEL(8);
-VP8_MC(pixels8);
-VP8_MC_OPT(pixels8,  armv6);
-VP8_EPEL(4);
-VP8_MC_OPT(pixels4,  armv6);
-
-VP8_MC(bilin16_h);
-VP8_MC(bilin16_v);
-VP8_MC(bilin16_hv);
-VP8_MC(bilin8_h);
-VP8_MC(bilin8_v);
-VP8_MC(bilin8_hv);
-VP8_MC(bilin4_h);
-VP8_MC(bilin4_v);
-VP8_MC(bilin4_hv);
-
-#define VP8_V6_MC(n) \
-void ff_put_vp8_##n##_armv6(uint8_t *dst, int dststride, uint8_t *src, \
-                            int srcstride, int w, int h, int mxy)
-
-VP8_V6_MC(epel_v6);
-VP8_V6_MC(epel_h6);
-VP8_V6_MC(epel_v4);
-VP8_V6_MC(epel_h4);
-VP8_V6_MC(bilin_v);
-VP8_V6_MC(bilin_h);
-
-#define VP8_EPEL_HV(SIZE, TAPNUMX, TAPNUMY, NAME, HNAME, VNAME, MAXHEIGHT) \
-static void ff_put_vp8_##NAME##SIZE##_##HNAME##VNAME##_armv6( \
-                                        uint8_t *dst, int dststride, uint8_t *src, \
-                                        int srcstride, int h, int mx, int my) \
-{ \
-    DECLARE_ALIGNED(4, uint8_t, tmp)[SIZE * (MAXHEIGHT + TAPNUMY - 1)]; \
-    uint8_t *tmpptr = tmp + SIZE * (TAPNUMY / 2 - 1); \
-    src -= srcstride * (TAPNUMY / 2 - 1); \
-    ff_put_vp8_ ## NAME ## _ ## HNAME ## _armv6(tmp, SIZE,      src,    srcstride, \
-                                                SIZE, h + TAPNUMY - 1,  mx); \
-    ff_put_vp8_ ## NAME ## _ ## VNAME ## _armv6(dst, dststride, tmpptr, SIZE, \
-                                                SIZE, h,                my); \
-}
-
-VP8_EPEL_HV(16, 6, 6, epel,  h6, v6, 16);
-VP8_EPEL_HV(16, 2, 2, bilin, h,  v,  16);
-VP8_EPEL_HV(8,  6, 6, epel,  h6, v6, 16);
-VP8_EPEL_HV(8,  4, 6, epel,  h4, v6, 16);
-VP8_EPEL_HV(8,  6, 4, epel,  h6, v4, 16);
-VP8_EPEL_HV(8,  4, 4, epel,  h4, v4, 16);
-VP8_EPEL_HV(8,  2, 2, bilin, h,  v,  16);
-VP8_EPEL_HV(4,  6, 6, epel,  h6, v6, 8);
-VP8_EPEL_HV(4,  4, 6, epel,  h4, v6, 8);
-VP8_EPEL_HV(4,  6, 4, epel,  h6, v4, 8);
-VP8_EPEL_HV(4,  4, 4, epel,  h4, v4, 8);
-VP8_EPEL_HV(4,  2, 2, bilin, h,  v,  8);
-
-extern void put_vp8_epel4_v6_c(uint8_t *dst, int d, uint8_t *src, int s, int h, int mx, int my);
-#undef printf
-#define VP8_EPEL_H_OR_V(SIZE, NAME, HV) \
-static void ff_put_vp8_##NAME##SIZE##_##HV##_armv6( \
-                                        uint8_t *dst, int dststride, uint8_t *src, \
-                                        int srcstride, int h, int mx, int my) \
-{ \
-    ff_put_vp8_## NAME ## _ ## HV ## _armv6(dst, dststride, src, srcstride, \
-                                            SIZE, h, mx | my); \
-}
-
-VP8_EPEL_H_OR_V(4,  epel,  h6);
-VP8_EPEL_H_OR_V(4,  epel,  h4);
-VP8_EPEL_H_OR_V(4,  epel,  v6);
-VP8_EPEL_H_OR_V(4,  epel,  v4);
-VP8_EPEL_H_OR_V(4,  bilin, v);
-VP8_EPEL_H_OR_V(4,  bilin, h);
-VP8_EPEL_H_OR_V(8,  epel,  h6);
-VP8_EPEL_H_OR_V(8,  epel,  h4);
-VP8_EPEL_H_OR_V(8,  epel,  v6);
-VP8_EPEL_H_OR_V(8,  epel,  v4);
-VP8_EPEL_H_OR_V(8,  bilin, v);
-VP8_EPEL_H_OR_V(8,  bilin, h);
-VP8_EPEL_H_OR_V(16, epel,  h6);
-VP8_EPEL_H_OR_V(16, epel,  v6);
-VP8_EPEL_H_OR_V(16, bilin, v);
-VP8_EPEL_H_OR_V(16, bilin, h);
+#include "vp8dsp.h"
 
 av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
 {
-#define set_func_ptrs(opt) \
-        dsp->vp8_luma_dc_wht    = ff_vp8_luma_dc_wht_##opt; \
-        dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6; \
- \
-        dsp->vp8_idct_add       = ff_vp8_idct_add_##opt; \
-        dsp->vp8_idct_dc_add    = ff_vp8_idct_dc_add_##opt; \
-        dsp->vp8_idct_dc_add4y  = ff_vp8_idct_dc_add4y_##opt; \
-        dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_##opt; \
- \
-        dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_##opt; \
-        dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_##opt; \
-        dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_##opt; \
-        dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_##opt; \
- \
-        dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_##opt; \
-        dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_##opt; \
-        dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_##opt; \
-        dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_##opt; \
- \
-        dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_##opt; \
-        dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_##opt; \
- \
-        dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_##opt; \
-        dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[0][2][2] = ff_put_vp8_epel16_h6v6_##opt; \
- \
-        dsp->put_vp8_epel_pixels_tab[1][0][0] = ff_put_vp8_pixels8_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][0][2] = ff_put_vp8_epel8_h6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][1][1] = ff_put_vp8_epel8_h4v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][1][2] = ff_put_vp8_epel8_h6v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][2][1] = ff_put_vp8_epel8_h4v6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[1][2][2] = ff_put_vp8_epel8_h6v6_##opt; \
- \
-        dsp->put_vp8_epel_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6; \
-        dsp->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][0][2] = ff_put_vp8_epel4_h6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][1][1] = ff_put_vp8_epel4_h4v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][1][2] = ff_put_vp8_epel4_h6v4_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][2][1] = ff_put_vp8_epel4_h4v6_##opt; \
-        dsp->put_vp8_epel_pixels_tab[2][2][2] = ff_put_vp8_epel4_h6v6_##opt; \
- \
-        dsp->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[0][0][2] = ff_put_vp8_bilin16_h_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[0][2][0] = ff_put_vp8_bilin16_v_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[0][2][2] = ff_put_vp8_bilin16_hv_##opt; \
- \
-        dsp->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][0][1] = ff_put_vp8_bilin8_h_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][0][2] = ff_put_vp8_bilin8_h_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][1][0] = ff_put_vp8_bilin8_v_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][1][1] = ff_put_vp8_bilin8_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][1][2] = ff_put_vp8_bilin8_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][2][0] = ff_put_vp8_bilin8_v_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][2][1] = ff_put_vp8_bilin8_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[1][2][2] = ff_put_vp8_bilin8_hv_##opt; \
- \
-        dsp->put_vp8_bilinear_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6; \
-        dsp->put_vp8_bilinear_pixels_tab[2][0][1] = ff_put_vp8_bilin4_h_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][0][2] = ff_put_vp8_bilin4_h_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][1][0] = ff_put_vp8_bilin4_v_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][1][1] = ff_put_vp8_bilin4_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][1][2] = ff_put_vp8_bilin4_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][2][0] = ff_put_vp8_bilin4_v_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_##opt; \
-        dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_##opt
     int cpu_flags = av_get_cpu_flags();
 
-    if (have_neon(cpu_flags)) {
-        set_func_ptrs(neon);
-    } else if (HAVE_ARMV6) {
-        set_func_ptrs(armv6);
-    }
+    if (have_armv6(cpu_flags))
+        ff_vp8dsp_init_armv6(dsp);
+    if (have_neon(cpu_flags))
+        ff_vp8dsp_init_neon(dsp);
 }