- sync ffmpeg to cvs (sorry Mike it still doesn't decode your teststream --

something must be broken at ffmpeg, also happens with mplayer)
- added priority sorted lists, now autoprobing should work again.
- fixed infinite loop in plugin loader.
obs: latest ffmpeg contains ppc optimizations, someone will have to enable these
though.

CVS patchset: 2676
CVS date: 2002/09/16 21:49:34

3 files changed, 77 insertions, 62 deletions

diff --git a/src/libffmpeg/libavcodec/alpha/asm.h b/src/libffmpeg/libavcodec/alpha/asm.h
index ceaf0be4a..2fdbdf13d 100644
--- a/src/libffmpeg/libavcodec/alpha/asm.h
+++ b/src/libffmpeg/libavcodec/alpha/asm.h
@@ -52,7 +52,7 @@ struct unaligned_long { uint64_t l; } __attribute__((packed));
 #define ldq_u(p)     (*(const uint64_t *) (((uint64_t) (p)) & ~7ul))
 #define uldq(a)	     (((const struct unaligned_long *) (a))->l)
 
-#if __GNUC__ >= 3 && __GNUC_MINOR__ >= 2
+#if __GNUC__ >= 3 && __GNUC_MINOR__ >= 3
 #define cmpbge	__builtin_alpha_cmpbge
 /* Avoid warnings.  */
 #define extql(a, b)	__builtin_alpha_extql(a, (uint64_t) (b))
diff --git a/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c b/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c
index 7ed015521..fef86fe64 100644
--- a/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c
+++ b/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c
@@ -224,16 +224,24 @@ static inline uint64_t avg4(uint64_t l1, uint64_t l2, uint64_t l3, uint64_t l4)
         } while (--h);                                                      \
     } while (0)
 
-#define MAKE_OP(OPNAME, SUFF, OPKIND, STORE)                            \
-static void OPNAME ## _pixels ## SUFF ## _axp                           \
-        (uint8_t *restrict block, const uint8_t *restrict pixels,       \
-         int line_size, int h)                                          \
-{                                                                       \
-    if ((size_t) pixels & 0x7) {                                        \
-        OPKIND(uldq, STORE);                                            \
-    } else {                                                            \
-        OPKIND(ldq, STORE);                                             \
-    }                                                                   \
+#define MAKE_OP(OPNAME, SUFF, OPKIND, STORE)                                \
+static void OPNAME ## _pixels ## SUFF ## _axp                               \
+        (uint8_t *restrict block, const uint8_t *restrict pixels,           \
+         int line_size, int h)                                              \
+{                                                                           \
+    if ((size_t) pixels & 0x7) {                                            \
+        OPKIND(uldq, STORE);                                                \
+    } else {                                                                \
+        OPKIND(ldq, STORE);                                                 \
+    }                                                                       \
+}                                                                           \
+                                                                            \
+static void OPNAME ## _pixels16 ## SUFF ## _axp                             \
+        (uint8_t *restrict block, const uint8_t *restrict pixels,           \
+         int line_size, int h)                                              \
+{                                                                           \
+    OPNAME ## _pixels ## SUFF ## _axp(block,     pixels,     line_size, h); \
+    OPNAME ## _pixels ## SUFF ## _axp(block + 8, pixels + 8, line_size, h); \
 }
 
 #define PIXOP(OPNAME, STORE)                    \
@@ -268,27 +276,54 @@ PIXOP(put_no_rnd, STORE);
 #define STORE(l, b) stq(AVG2(l, ldq(b)), b);
 PIXOP(avg_no_rnd, STORE);
 
+void put_pixels16_axp_asm(uint8_t *block, const uint8_t *pixels,
+                          int line_size, int h)
+{
+    put_pixels_axp_asm(block,     pixels,     line_size, h);
+    put_pixels_axp_asm(block + 8, pixels + 8, line_size, h);
+}
+
 void dsputil_init_alpha(void)
 {
-    put_pixels_tab[0] = put_pixels_axp_asm;
-    put_pixels_tab[1] = put_pixels_x2_axp;
-    put_pixels_tab[2] = put_pixels_y2_axp;
-    put_pixels_tab[3] = put_pixels_xy2_axp;
-
-    put_no_rnd_pixels_tab[0] = put_pixels_axp_asm;
-    put_no_rnd_pixels_tab[1] = put_no_rnd_pixels_x2_axp;
-    put_no_rnd_pixels_tab[2] = put_no_rnd_pixels_y2_axp;
-    put_no_rnd_pixels_tab[3] = put_no_rnd_pixels_xy2_axp;
-
-    avg_pixels_tab[0] = avg_pixels_axp;
-    avg_pixels_tab[1] = avg_pixels_x2_axp;
-    avg_pixels_tab[2] = avg_pixels_y2_axp;
-    avg_pixels_tab[3] = avg_pixels_xy2_axp;
-
-    avg_no_rnd_pixels_tab[0] = avg_no_rnd_pixels_axp;
-    avg_no_rnd_pixels_tab[1] = avg_no_rnd_pixels_x2_axp;
-    avg_no_rnd_pixels_tab[2] = avg_no_rnd_pixels_y2_axp;
-    avg_no_rnd_pixels_tab[3] = avg_no_rnd_pixels_xy2_axp;
+    put_pixels_tab[0][0] = put_pixels16_axp_asm;
+    put_pixels_tab[0][1] = put_pixels16_x2_axp;
+    put_pixels_tab[0][2] = put_pixels16_y2_axp;
+    put_pixels_tab[0][3] = put_pixels16_xy2_axp;
+
+    put_no_rnd_pixels_tab[0][0] = put_pixels16_axp_asm;
+    put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_axp;
+    put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_axp;
+    put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_axp;
+
+    avg_pixels_tab[0][0] = avg_pixels16_axp;
+    avg_pixels_tab[0][1] = avg_pixels16_x2_axp;
+    avg_pixels_tab[0][2] = avg_pixels16_y2_axp;
+    avg_pixels_tab[0][3] = avg_pixels16_xy2_axp;
+
+    avg_no_rnd_pixels_tab[0][0] = avg_no_rnd_pixels16_axp;
+    avg_no_rnd_pixels_tab[0][1] = avg_no_rnd_pixels16_x2_axp;
+    avg_no_rnd_pixels_tab[0][2] = avg_no_rnd_pixels16_y2_axp;
+    avg_no_rnd_pixels_tab[0][3] = avg_no_rnd_pixels16_xy2_axp;
+
+    put_pixels_tab[1][0] = put_pixels_axp_asm;
+    put_pixels_tab[1][1] = put_pixels_x2_axp;
+    put_pixels_tab[1][2] = put_pixels_y2_axp;
+    put_pixels_tab[1][3] = put_pixels_xy2_axp;
+
+    put_no_rnd_pixels_tab[1][0] = put_pixels_axp_asm;
+    put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels_x2_axp;
+    put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels_y2_axp;
+    put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels_xy2_axp;
+
+    avg_pixels_tab[1][0] = avg_pixels_axp;
+    avg_pixels_tab[1][1] = avg_pixels_x2_axp;
+    avg_pixels_tab[1][2] = avg_pixels_y2_axp;
+    avg_pixels_tab[1][3] = avg_pixels_xy2_axp;
+
+    avg_no_rnd_pixels_tab[1][0] = avg_no_rnd_pixels_axp;
+    avg_no_rnd_pixels_tab[1][1] = avg_no_rnd_pixels_x2_axp;
+    avg_no_rnd_pixels_tab[1][2] = avg_no_rnd_pixels_y2_axp;
+    avg_no_rnd_pixels_tab[1][3] = avg_no_rnd_pixels_xy2_axp;
 
     clear_blocks = clear_blocks_axp;
 
diff --git a/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S b/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S
index 5349e443c..7ec6757d7 100644
--- a/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S
+++ b/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S
@@ -122,41 +122,21 @@ $aligned:
         ldq     t2, 0(a1)
         addq    a1, a2, a1
         ldq     t3, 0(a1)
-        addq    a1, a2, a1
-
-        ldq     t4, 0(a1)
-        addq    a1, a2, a1
-        ldq     t5, 0(a1)
-        addq    a1, a2, a1
-
-        ldq     t6, 0(a1)
-        addq    a1, a2, a1
-        ldq     t7, 0(a1)
-        addq    a1, a2, a1
-
-        addq    a0, a2, t8
-        stq     t0, 0(a0)
-        addq    t8, a2, t9
-        stq     t1, 0(t8)
-
-        addq    t9, a2, ta
-        stq     t2, 0(t9)
-        addq    ta, a2, tb
-        stq     t3, 0(ta)
-
-        addq    tb, a2, tc
-        stq     t4, 0(tb)
-        addq    tc, a2, td
-        stq     t5, 0(tc)
 
-        addq    td, a2, te
-        stq     t6, 0(td)
-        addq    te, a2, a0
-        stq     t7, 0(te)
+	addq	a0, a2, t4
+	addq    a1, a2, a1
+	addq	t4, a2, t5
+	subq    a3, 4, a3
 
-        subq    a3, 8, a3
-        bne     a3, $aligned
+	stq	t0, 0(a0)
+	addq	t5, a2, t6
+	stq	t1, 0(t4)
+	addq	t6, a2, a0
 
+	stq	t2, 0(t5)
+	stq	t3, 0(t6)
+	
+	bne     a3, $aligned
         ret
         .end put_pixels_axp_asm