Merge commit '896a5bff64264f4d01ed98eacc97a67260c1e17e'

* commit '896a5bff64264f4d01ed98eacc97a67260c1e17e':
  arm: check if AS supports .dn

Conflicts:
	configure
	libavcodec/arm/vc1dsp_init_neon.c

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

3 files changed, 28 insertions, 15 deletions

diff --git a/configure b/configure
index 7fe6994257..495eb00fa2 100755
--- a/configure
+++ b/configure
@@ -1718,6 +1718,7 @@ SYSTEM_FUNCS="
 "
 
 TOOLCHAIN_FEATURES="
+    as_dn_directive
     asm_mod_q
     attribute_may_alias
     attribute_packed
@@ -4267,6 +4268,20 @@ unsigned int endian = 'B' << 24 | 'I' << 16 | 'G' << 8 | 'E';
 EOF
 od -t x1 $TMPO | grep -q '42 *49 *47 *45' && enable bigendian
 
+
+if enabled asm; then
+    enabled     arm         && nogas=die
+    enabled_all ppc altivec && nogas=warn
+    as=${gas:=$as}
+    check_as <<EOF && enable gnu_as || \
+        $nogas "GNU assembler not found, install gas-preprocessor"
+.macro m n
+\n: .int 0
+.endm
+m x
+EOF
+fi
+
 check_inline_asm inline_asm_labels '"1:\n"'
 
 check_inline_asm inline_asm_nonlocal_labels '"Label:\n"'
@@ -4290,7 +4305,6 @@ float func(float a, float b){ return a+b; }
 EOF
 
     enabled thumb && check_cflags -mthumb || check_cflags -marm
-    nogas=die
 
     if     check_cpp_condition stddef.h "defined __ARM_PCS_VFP"; then
         enable vfp_args
@@ -4318,6 +4332,11 @@ EOF
 
     check_inline_asm asm_mod_q '"add r0, %Q0, %R0" :: "r"((long long)0)'
 
+    check_as <<EOF && enable as_dn_directive
+ra .dn d0.i16
+.unreq ra
+EOF
+
     [ $target_os != win32 ] && enabled_all armv6t2 shared !pic && enable_weak_pic
 
 elif enabled mips; then
@@ -4351,9 +4370,6 @@ elif enabled ppc; then
 
     # AltiVec flags: The FSF version of GCC differs from the Apple version
     if enabled altivec; then
-        if ! enabled_any pic ppc64; then
-            nogas=warn
-        fi
         check_cflags -maltivec -mabi=altivec &&
         { check_header altivec.h && inc_altivec_h="#include <altivec.h>" ; } ||
         check_cflags -faltivec
@@ -4439,17 +4455,6 @@ EOF
 
 fi
 
-if enabled asm; then
-    as=${gas:=$as}
-    check_as <<EOF && enable gnu_as || \
-        $nogas "GNU assembler not found, install gas-preprocessor"
-.macro m n
-\n: .int 0
-.endm
-m x
-EOF
-fi
-
 check_ldflags -Wl,--as-needed
 
 if check_func dlopen; then
diff --git a/libavcodec/arm/vc1dsp_init_neon.c b/libavcodec/arm/vc1dsp_init_neon.c
index 3b280770c5..bb873e687e 100644
--- a/libavcodec/arm/vc1dsp_init_neon.c
+++ b/libavcodec/arm/vc1dsp_init_neon.c
@@ -22,6 +22,8 @@
 #include "libavcodec/vc1dsp.h"
 #include "vc1dsp.h"
 
+#include "config.h"
+
 void ff_vc1_inv_trans_8x8_neon(int16_t *block);
 void ff_vc1_inv_trans_4x8_neon(uint8_t *dest, int linesize, int16_t *block);
 void ff_vc1_inv_trans_8x4_neon(uint8_t *dest, int linesize, int16_t *block);
@@ -93,6 +95,7 @@ av_cold void ff_vc1dsp_init_neon(VC1DSPContext *dsp)
     dsp->vc1_inv_trans_4x4_dc = ff_vc1_inv_trans_4x4_dc_neon;
 
     dsp->put_vc1_mspel_pixels_tab[1][ 0] = ff_put_pixels8x8_neon;
+    if (HAVE_AS_DN_DIRECTIVE) {
     FN_ASSIGN(1, 0);
     FN_ASSIGN(2, 0);
     FN_ASSIGN(3, 0);
@@ -111,6 +114,7 @@ av_cold void ff_vc1dsp_init_neon(VC1DSPContext *dsp)
     FN_ASSIGN(1, 3);
     FN_ASSIGN(2, 3);
     FN_ASSIGN(3, 3);
+    }
 
     dsp->put_no_rnd_vc1_chroma_pixels_tab[0] = ff_put_vc1_chroma_mc8_neon;
     dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] = ff_avg_vc1_chroma_mc8_neon;
diff --git a/libavcodec/arm/vc1dsp_neon.S b/libavcodec/arm/vc1dsp_neon.S
index 3fba41fa66..c4f4db9c8e 100644
--- a/libavcodec/arm/vc1dsp_neon.S
+++ b/libavcodec/arm/vc1dsp_neon.S
@@ -24,6 +24,8 @@
 #include "libavutil/arm/asm.S"
 #include "neon.S"
 
+#include "config.h"
+
 @ Transpose rows into columns of a matrix of 16-bit elements. For 4x4, pass
 @ double-word registers, for 8x4, pass quad-word registers.
 .macro transpose16 r0, r1, r2, r3
@@ -661,6 +663,7 @@ function ff_vc1_inv_trans_4x4_neon, export=1
         bx              lr
 endfunc
 
+#if HAVE_AS_DN_DIRECTIVE
 @ The absolute value of multiplication constants from vc1_mspel_filter and vc1_mspel_{ver,hor}_filter_16bits.
 @ The sign is embedded in the code below that carries out the multiplication (mspel_filter{,.16}).
 #define MSPEL_MODE_1_MUL_CONSTANTS  4 53 18 3
@@ -984,6 +987,7 @@ PUT_VC1_MSPEL_MC_V_ONLY(2)
 PUT_VC1_MSPEL_MC_V_ONLY(3)
 
 #undef PUT_VC1_MSPEL_MC_V_ONLY
+#endif
 
 function ff_put_pixels8x8_neon, export=1
         vld1.64         {d0}, [r1], r2