Merge remote-tracking branch 'qatar/master'

* qatar/master:
  v410dec: Implement explode mode support
  zerocodec: fix direct rendering.
  wav: init st to NULL to avoid a false-positive warning.
  wavpack: set bits_per_raw_sample for S32 samples to properly identify 24-bit
  h264: refactor NAL decode loop
  RTMPTE protocol support
  RTMPE protocol support
  rtmp: Add ff_rtmp_calc_digest_pos()
  rtmp: Rename rtmp_calc_digest to ff_rtmp_calc_digest and make it global
  swscale: add missing HAVE_INLINE_ASM check.
  lavfi: place x86 inline assembly under HAVE_INLINE_ASM.
  vc1: Add a test for interlaced field pictures
  swscale: Mark all init functions as av_cold
  swscale: x86: Drop pointless _mmx suffix from filenames
  lavf: use conditional notation for default codec in muxer declarations.
  swscale: place inline assembly bilinear scaler under HAVE_INLINE_ASM.
  dsputil: ppc: cosmetics: pretty-print
  dsputil: x86: add SHUFFLE_MASK_W macro
  configure: respect CC_O setting in check_cc

Conflicts:
	Changelog
	configure
	libavcodec/v410dec.c
	libavcodec/zerocodec.c
	libavformat/asfenc.c
	libavformat/version.h
	libswscale/utils.c
	libswscale/x86/swscale.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

11 files changed, 38 insertions, 26 deletions

diff --git a/libswscale/output.c b/libswscale/output.c
index d2a480c173..8955665f66 100644
--- a/libswscale/output.c
+++ b/libswscale/output.c
@@ -24,6 +24,7 @@
 #include <stdio.h>
 #include <string.h>
 
+#include "libavutil/attributes.h"
 #include "libavutil/avutil.h"
 #include "libavutil/avassert.h"
 #include "libavutil/bswap.h"
@@ -1260,13 +1261,13 @@ YUV2RGBWRAPPERX(yuv2, rgb_full, xrgb32_full, PIX_FMT_ARGB,  0)
 YUV2RGBWRAPPERX(yuv2, rgb_full, bgr24_full,  PIX_FMT_BGR24, 0)
 YUV2RGBWRAPPERX(yuv2, rgb_full, rgb24_full,  PIX_FMT_RGB24, 0)
 
-void ff_sws_init_output_funcs(SwsContext *c,
-                              yuv2planar1_fn *yuv2plane1,
-                              yuv2planarX_fn *yuv2planeX,
-                              yuv2interleavedX_fn *yuv2nv12cX,
-                              yuv2packed1_fn *yuv2packed1,
-                              yuv2packed2_fn *yuv2packed2,
-                              yuv2packedX_fn *yuv2packedX)
+av_cold void ff_sws_init_output_funcs(SwsContext *c,
+                                      yuv2planar1_fn *yuv2plane1,
+                                      yuv2planarX_fn *yuv2planeX,
+                                      yuv2interleavedX_fn *yuv2nv12cX,
+                                      yuv2packed1_fn *yuv2packed1,
+                                      yuv2packed2_fn *yuv2packed2,
+                                      yuv2packedX_fn *yuv2packedX)
 {
     enum PixelFormat dstFormat = c->dstFormat;
 
diff --git a/libswscale/ppc/swscale_altivec.c b/libswscale/ppc/swscale_altivec.c
index 694c8d8b9a..c621866eed 100644
--- a/libswscale/ppc/swscale_altivec.c
+++ b/libswscale/ppc/swscale_altivec.c
@@ -26,6 +26,7 @@
 #include "config.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
+#include "libavutil/attributes.h"
 #include "libavutil/cpu.h"
 #include "yuv2rgb_altivec.h"
 
@@ -310,7 +311,7 @@ static void hScale_altivec_real(SwsContext *c, int16_t *dst, int dstW,
         }
 }
 
-void ff_sws_init_swScale_altivec(SwsContext *c)
+av_cold void ff_sws_init_swScale_altivec(SwsContext *c)
 {
     enum PixelFormat dstFormat = c->dstFormat;
 
diff --git a/libswscale/ppc/yuv2rgb_altivec.c b/libswscale/ppc/yuv2rgb_altivec.c
index fba4162738..ade134d779 100644
--- a/libswscale/ppc/yuv2rgb_altivec.c
+++ b/libswscale/ppc/yuv2rgb_altivec.c
@@ -95,6 +95,7 @@
 #include "libswscale/rgb2rgb.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
+#include "libavutil/attributes.h"
 #include "libavutil/cpu.h"
 #include "libavutil/pixdesc.h"
 #include "yuv2rgb_altivec.h"
@@ -531,7 +532,7 @@ static int altivec_uyvy_rgb32(SwsContext *c, const unsigned char **in,
  *
  * So we just fall back to the C codes for this.
  */
-SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c)
+av_cold SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c)
 {
     if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
         return NULL;
@@ -591,9 +592,11 @@ SwsFunc ff_yuv2rgb_init_altivec(SwsContext *c)
     return NULL;
 }
 
-void ff_yuv2rgb_init_tables_altivec(SwsContext *c, const int inv_table[4],
-                                    int brightness, int contrast,
-                                    int saturation)
+av_cold void ff_yuv2rgb_init_tables_altivec(SwsContext *c,
+                                            const int inv_table[4],
+                                            int brightness,
+                                            int contrast,
+                                            int saturation)
 {
     union {
         DECLARE_ALIGNED(16, signed short, tmp)[8];
diff --git a/libswscale/rgb2rgb.c b/libswscale/rgb2rgb.c
index 70aaee47d8..321e5ff1b8 100644
--- a/libswscale/rgb2rgb.c
+++ b/libswscale/rgb2rgb.c
@@ -25,6 +25,7 @@
 
 #include <inttypes.h>
 
+#include "libavutil/attributes.h"
 #include "libavutil/bswap.h"
 #include "config.h"
 #include "rgb2rgb.h"
@@ -125,7 +126,7 @@ void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  * 32-bit C version, and and&add trick by Michael Niedermayer
  */
 
-void sws_rgb2rgb_init(void)
+av_cold void sws_rgb2rgb_init(void)
 {
     rgb2rgb_init_c();
     if (HAVE_MMX)
diff --git a/libswscale/sparc/yuv2rgb_vis.c b/libswscale/sparc/yuv2rgb_vis.c
index 62f502619c..bb9ab22dbe 100644
--- a/libswscale/sparc/yuv2rgb_vis.c
+++ b/libswscale/sparc/yuv2rgb_vis.c
@@ -22,6 +22,7 @@
 #include <inttypes.h>
 #include <stdlib.h>
 
+#include "libavutil/attributes.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
 
@@ -184,7 +185,7 @@ static int vis_422P_ARGB32(SwsContext *c, uint8_t *src[], int srcStride[],
     return srcSliceH;
 }
 
-SwsFunc ff_yuv2rgb_init_vis(SwsContext *c)
+av_cold SwsFunc ff_yuv2rgb_init_vis(SwsContext *c)
 {
     c->sparc_coeffs[5] = c->yCoeff;
     c->sparc_coeffs[6] = c->vgCoeff;
diff --git a/libswscale/swscale.c b/libswscale/swscale.c
index 1f807852d5..8c4c934dd0 100644
--- a/libswscale/swscale.c
+++ b/libswscale/swscale.c
@@ -544,7 +544,7 @@ static int swScale(SwsContext *c, const uint8_t *src[],
         if (!enough_lines)
             break;  // we can't output a dstY line so let's try with the next slice
 
-#if HAVE_MMX
+#if HAVE_MMX && HAVE_INLINE_ASM
         updateMMXDitherTables(c, dstY, lumBufIndex, chrBufIndex,
                               lastInLumBuf, lastInChrBuf);
 #endif
diff --git a/libswscale/utils.c b/libswscale/utils.c
index a6bb732024..f4d3546845 100644
--- a/libswscale/utils.c
+++ b/libswscale/utils.c
@@ -37,6 +37,7 @@
 #include <windows.h>
 #endif
 
+#include "libavutil/attributes.h"
 #include "libavutil/avassert.h"
 #include "libavutil/avutil.h"
 #include "libavutil/bswap.h"
@@ -598,7 +599,7 @@ fail:
     return ret;
 }
 
-#if HAVE_MMX2
+#if HAVE_MMX2 && HAVE_INLINE_ASM
 static int initMMX2HScaler(int dstW, int xInc, uint8_t *filterCode,
                            int16_t *filter, int32_t *filterPos, int numSplits)
 {
@@ -761,7 +762,7 @@ static int initMMX2HScaler(int dstW, int xInc, uint8_t *filterCode,
 
     return fragmentPos + 1;
 }
-#endif /* HAVE_MMX2 */
+#endif /* HAVE_MMX2 && HAVE_INLINE_ASM */
 
 static void getSubSampleFactors(int *h, int *v, enum PixelFormat format)
 {
@@ -856,7 +857,8 @@ SwsContext *sws_alloc_context(void)
     return c;
 }
 
-int sws_init_context(SwsContext *c, SwsFilter *srcFilter, SwsFilter *dstFilter)
+av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,
+                             SwsFilter *dstFilter)
 {
     int i, j;
     int usesVFilter, usesHFilter;
@@ -1022,7 +1024,7 @@ int sws_init_context(SwsContext *c, SwsFilter *srcFilter, SwsFilter *dstFilter)
         c->srcBpc = 16;
     if (c->dstBpc == 16)
         dst_stride <<= 1;
-    if (HAVE_MMX2 && cpu_flags & AV_CPU_FLAG_MMX2 &&
+    if (HAVE_MMX2 && HAVE_INLINE_ASM && cpu_flags & AV_CPU_FLAG_MMX2 &&
         c->srcBpc == 8 && c->dstBpc <= 14) {
         c->canMMX2BeUsed = (dstW >= srcW && (dstW & 31) == 0 &&
                             (srcW & 15) == 0) ? 1 : 0;
@@ -1061,7 +1063,7 @@ int sws_init_context(SwsContext *c, SwsFilter *srcFilter, SwsFilter *dstFilter)
 
     /* precalculate horizontal scaler filter coefficients */
     {
-#if HAVE_MMX2
+#if HAVE_MMX2 && HAVE_INLINE_ASM
 // can't downscale !!!
         if (c->canMMX2BeUsed && (flags & SWS_FAST_BILINEAR)) {
             c->lumMmx2FilterCodeSize = initMMX2HScaler(dstW, c->lumXInc, NULL,
@@ -1105,7 +1107,7 @@ int sws_init_context(SwsContext *c, SwsFilter *srcFilter, SwsFilter *dstFilter)
             mprotect(c->chrMmx2FilterCode, c->chrMmx2FilterCodeSize, PROT_EXEC | PROT_READ);
 #endif
         } else
-#endif /* HAVE_MMX2 */
+#endif /* HAVE_MMX2 && HAVE_INLINE_ASM */
         {
             const int filterAlign =
                 (HAVE_MMX && cpu_flags & AV_CPU_FLAG_MMX) ? 4 :
diff --git a/libswscale/x86/Makefile b/libswscale/x86/Makefile
index 68d5381701..7d219b458b 100644
--- a/libswscale/x86/Makefile
+++ b/libswscale/x86/Makefile
@@ -3,8 +3,8 @@ $(SUBDIR)x86/swscale_mmx.o: CFLAGS += $(NOREDZONE_FLAGS)
 OBJS-$(CONFIG_XMM_CLOBBER_TEST) += x86/w64xmmtest.o
 
 MMX-OBJS                        += x86/rgb2rgb.o                        \
-                                   x86/swscale_mmx.o                    \
-                                   x86/yuv2rgb_mmx.o                    \
+                                   x86/swscale.o                        \
+                                   x86/yuv2rgb.o                        \
 
 YASM-OBJS                       += x86/input.o                          \
                                    x86/output.o                         \
diff --git a/libswscale/x86/rgb2rgb.c b/libswscale/x86/rgb2rgb.c
index 92deda2f5f..cd46df719d 100644
--- a/libswscale/x86/rgb2rgb.c
+++ b/libswscale/x86/rgb2rgb.c
@@ -26,6 +26,7 @@
 #include <stdint.h>
 
 #include "config.h"
+#include "libavutil/attributes.h"
 #include "libavutil/x86_cpu.h"
 #include "libavutil/cpu.h"
 #include "libavutil/bswap.h"
@@ -130,7 +131,7 @@ DECLARE_ASM_CONST(8, uint64_t, mul16_mid)    = 0x2080208020802080ULL;
 
 #endif /* HAVE_INLINE_ASM */
 
-void rgb2rgb_init_x86(void)
+av_cold void rgb2rgb_init_x86(void)
 {
 #if HAVE_INLINE_ASM
     int cpu_flags = av_get_cpu_flags();
diff --git a/libswscale/x86/swscale_mmx.c b/libswscale/x86/swscale.c
index 48d1a8e9c7..eb33788168 100644
--- a/libswscale/x86/swscale_mmx.c
+++ b/libswscale/x86/swscale.c
@@ -22,6 +22,7 @@
 #include "config.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
+#include "libavutil/attributes.h"
 #include "libavutil/avassert.h"
 #include "libavutil/intreadwrite.h"
 #include "libavutil/x86_cpu.h"
@@ -367,7 +368,7 @@ INPUT_FUNCS(sse2);
 INPUT_FUNCS(ssse3);
 INPUT_FUNCS(avx);
 
-void ff_sws_init_swScale_mmx(SwsContext *c)
+av_cold void ff_sws_init_swScale_mmx(SwsContext *c)
 {
     int cpu_flags = av_get_cpu_flags();
 
diff --git a/libswscale/x86/yuv2rgb_mmx.c b/libswscale/x86/yuv2rgb.c
index 2107dd1ca4..2a76716335 100644
--- a/libswscale/x86/yuv2rgb_mmx.c
+++ b/libswscale/x86/yuv2rgb.c
@@ -33,6 +33,7 @@
 #include "libswscale/rgb2rgb.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
+#include "libavutil/attributes.h"
 #include "libavutil/x86_cpu.h"
 #include "libavutil/cpu.h"
 
@@ -68,7 +69,7 @@ DECLARE_ASM_CONST(8, uint64_t, pb_07) = 0x0707070707070707ULL;
 
 #endif /* HAVE_INLINE_ASM */
 
-SwsFunc ff_yuv2rgb_init_mmx(SwsContext *c)
+av_cold SwsFunc ff_yuv2rgb_init_mmx(SwsContext *c)
 {
 #if HAVE_INLINE_ASM
     int cpu_flags = av_get_cpu_flags();