diff options
Diffstat (limited to 'src/libffmpeg/libavcodec/alpha')
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/asm.h | 70 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/dsputil_alpha.c | 16 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S | 50 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/motion_est_alpha.c | 12 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/motion_est_mvi_asm.S | 4 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/mpegvideo_alpha.c | 14 | ||||
| -rw-r--r-- | src/libffmpeg/libavcodec/alpha/simple_idct_alpha.c | 12 |
7 files changed, 88 insertions, 90 deletions
diff --git a/src/libffmpeg/libavcodec/alpha/asm.h b/src/libffmpeg/libavcodec/alpha/asm.h index 6dc997b37..056e043f3 100644 --- a/src/libffmpeg/libavcodec/alpha/asm.h +++ b/src/libffmpeg/libavcodec/alpha/asm.h @@ -14,7 +14,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #ifndef LIBAVCODEC_ALPHA_ASM_H @@ -84,24 +84,24 @@ static inline uint64_t WORD_VEC(uint64_t x) } *) (p))->__l) = l; \ } while (0) 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) +#define ldq_u(p) (*(const uint64_t *) (((uint64_t) (p)) & ~7ul)) +#define uldq(a) (((const struct unaligned_long *) (a))->l) #if GNUC_PREREQ(3,3) #define prefetch(p) __builtin_prefetch((p), 0, 1) #define prefetch_en(p) __builtin_prefetch((p), 0, 0) #define prefetch_m(p) __builtin_prefetch((p), 1, 1) #define prefetch_men(p) __builtin_prefetch((p), 1, 0) -#define cmpbge __builtin_alpha_cmpbge +#define cmpbge __builtin_alpha_cmpbge /* Avoid warnings. */ -#define extql(a, b) __builtin_alpha_extql(a, (uint64_t) (b)) -#define extwl(a, b) __builtin_alpha_extwl(a, (uint64_t) (b)) -#define extqh(a, b) __builtin_alpha_extqh(a, (uint64_t) (b)) -#define zap __builtin_alpha_zap -#define zapnot __builtin_alpha_zapnot -#define amask __builtin_alpha_amask -#define implver __builtin_alpha_implver -#define rpcc __builtin_alpha_rpcc +#define extql(a, b) __builtin_alpha_extql(a, (uint64_t) (b)) +#define extwl(a, b) __builtin_alpha_extwl(a, (uint64_t) (b)) +#define extqh(a, b) __builtin_alpha_extqh(a, (uint64_t) (b)) +#define zap __builtin_alpha_zap +#define zapnot __builtin_alpha_zapnot +#define amask __builtin_alpha_amask +#define implver __builtin_alpha_implver +#define rpcc __builtin_alpha_rpcc #else #define prefetch(p) asm volatile("ldl $31,%0" : : "m"(*(const char *) (p)) : "memory") #define prefetch_en(p) asm volatile("ldq $31,%0" : : "m"(*(const char *) (p)) : "memory") @@ -113,26 +113,26 @@ struct unaligned_long { uint64_t l; } __attribute__((packed)); #define extqh(a, b) ({ uint64_t __r; asm ("extqh %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; }) #define zap(a, b) ({ uint64_t __r; asm ("zap %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; }) #define zapnot(a, b) ({ uint64_t __r; asm ("zapnot %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; }) -#define amask(a) ({ uint64_t __r; asm ("amask %1,%0" : "=r" (__r) : "rI" (a)); __r; }) -#define implver() ({ uint64_t __r; asm ("implver %0" : "=r" (__r)); __r; }) -#define rpcc() ({ uint64_t __r; asm volatile ("rpcc %0" : "=r" (__r)); __r; }) +#define amask(a) ({ uint64_t __r; asm ("amask %1,%0" : "=r" (__r) : "rI" (a)); __r; }) +#define implver() ({ uint64_t __r; asm ("implver %0" : "=r" (__r)); __r; }) +#define rpcc() ({ uint64_t __r; asm volatile ("rpcc %0" : "=r" (__r)); __r; }) #endif #define wh64(p) asm volatile("wh64 (%0)" : : "r"(p) : "memory") #if GNUC_PREREQ(3,3) && defined(__alpha_max__) -#define minub8 __builtin_alpha_minub8 -#define minsb8 __builtin_alpha_minsb8 -#define minuw4 __builtin_alpha_minuw4 -#define minsw4 __builtin_alpha_minsw4 -#define maxub8 __builtin_alpha_maxub8 -#define maxsb8 __builtin_alpha_maxsb8 -#define maxuw4 __builtin_alpha_maxuw4 -#define maxsw4 __builtin_alpha_maxsw4 -#define perr __builtin_alpha_perr -#define pklb __builtin_alpha_pklb -#define pkwb __builtin_alpha_pkwb -#define unpkbl __builtin_alpha_unpkbl -#define unpkbw __builtin_alpha_unpkbw +#define minub8 __builtin_alpha_minub8 +#define minsb8 __builtin_alpha_minsb8 +#define minuw4 __builtin_alpha_minuw4 +#define minsw4 __builtin_alpha_minsw4 +#define maxub8 __builtin_alpha_maxub8 +#define maxsb8 __builtin_alpha_maxsb8 +#define maxuw4 __builtin_alpha_maxuw4 +#define maxsw4 __builtin_alpha_maxsw4 +#define perr __builtin_alpha_perr +#define pklb __builtin_alpha_pklb +#define pkwb __builtin_alpha_pkwb +#define unpkbl __builtin_alpha_unpkbl +#define unpkbw __builtin_alpha_unpkbw #else #define minub8(a, b) ({ uint64_t __r; asm (".arch ev6; minub8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; }) #define minsb8(a, b) ({ uint64_t __r; asm (".arch ev6; minsb8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; }) @@ -143,13 +143,13 @@ struct unaligned_long { uint64_t l; } __attribute__((packed)); #define maxuw4(a, b) ({ uint64_t __r; asm (".arch ev6; maxuw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; }) #define maxsw4(a, b) ({ uint64_t __r; asm (".arch ev6; maxsw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; }) #define perr(a, b) ({ uint64_t __r; asm (".arch ev6; perr %r1,%r2,%0" : "=r" (__r) : "%rJ" (a), "rJ" (b)); __r; }) -#define pklb(a) ({ uint64_t __r; asm (".arch ev6; pklb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) -#define pkwb(a) ({ uint64_t __r; asm (".arch ev6; pkwb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) -#define unpkbl(a) ({ uint64_t __r; asm (".arch ev6; unpkbl %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) -#define unpkbw(a) ({ uint64_t __r; asm (".arch ev6; unpkbw %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) +#define pklb(a) ({ uint64_t __r; asm (".arch ev6; pklb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) +#define pkwb(a) ({ uint64_t __r; asm (".arch ev6; pkwb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) +#define unpkbl(a) ({ uint64_t __r; asm (".arch ev6; unpkbl %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) +#define unpkbw(a) ({ uint64_t __r; asm (".arch ev6; unpkbw %r1,%0" : "=r" (__r) : "rJ" (a)); __r; }) #endif -#elif defined(__DECC) /* Digital/Compaq/hp "ccc" compiler */ +#elif defined(__DECC) /* Digital/Compaq/hp "ccc" compiler */ #include <c_asm.h> #define ldq(p) (*(const uint64_t *) (p)) @@ -157,7 +157,7 @@ struct unaligned_long { uint64_t l; } __attribute__((packed)); #define stq(l, p) do { *(uint64_t *) (p) = (l); } while (0) #define stl(l, p) do { *(int32_t *) (p) = (l); } while (0) #define ldq_u(a) asm ("ldq_u %v0,0(%a0)", a) -#define uldq(a) (*(const __unaligned uint64_t *) (a)) +#define uldq(a) (*(const __unaligned uint64_t *) (a)) #define cmpbge(a, b) asm ("cmpbge %a0,%a1,%v0", a, b) #define extql(a, b) asm ("extql %a0,%a1,%v0", a, b) #define extwl(a, b) asm ("extwl %a0,%a1,%v0", a, b) @@ -166,7 +166,7 @@ struct unaligned_long { uint64_t l; } __attribute__((packed)); #define zapnot(a, b) asm ("zapnot %a0,%a1,%v0", a, b) #define amask(a) asm ("amask %a0,%v0", a) #define implver() asm ("implver %v0") -#define rpcc() asm ("rpcc %v0") +#define rpcc() asm ("rpcc %v0") #define minub8(a, b) asm ("minub8 %a0,%a1,%v0", a, b) #define minsb8(a, b) asm ("minsb8 %a0,%a1,%v0", a, b) #define minuw4(a, b) asm ("minuw4 %a0,%a1,%v0", a, b) diff --git a/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c b/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c index 496f46120..299a25dc4 100644 --- a/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c +++ b/src/libffmpeg/libavcodec/alpha/dsputil_alpha.c @@ -14,7 +14,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #include "asm.h" @@ -28,11 +28,11 @@ void put_pixels_axp_asm(uint8_t *block, const uint8_t *pixels, int line_size, int h); void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, int line_size); -void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, +void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, int line_size); void (*put_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, int line_size); -void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, +void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, int line_size); void get_pixels_mvi(DCTELEM *restrict block, @@ -48,7 +48,7 @@ int pix_abs16x16_xy2_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, i #if 0 /* These functions were the base for the optimized assembler routines, and remain here for documentation purposes. */ -static void put_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, +static void put_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, int line_size) { int i = 8; @@ -72,7 +72,7 @@ static void put_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, } while (--i); } -void add_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, +void add_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, int line_size) { int h = 8; @@ -97,7 +97,7 @@ void add_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels, shorts0 ^= signs0; /* Clamp. */ shorts0 = maxsw4(shorts0, 0); - shorts0 = minsw4(shorts0, clampmask); + shorts0 = minsw4(shorts0, clampmask); /* Next 4. */ pix1 = unpkbw(ldl(pixels + 4)); @@ -142,7 +142,7 @@ static inline uint64_t avg2_no_rnd(uint64_t a, uint64_t b) static inline uint64_t avg2(uint64_t a, uint64_t b) { - return (a | b) - (((a ^ b) & BYTE_VEC(0xfe)) >> 1); + return (a | b) - (((a ^ b) & BYTE_VEC(0xfe)) >> 1); } #if 0 @@ -353,7 +353,7 @@ void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx) put_pixels_clamped_axp_p = c->put_pixels_clamped; add_pixels_clamped_axp_p = c->add_pixels_clamped; - + c->idct_put = simple_idct_put_axp; c->idct_add = simple_idct_add_axp; c->idct = simple_idct_axp; diff --git a/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S b/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S index 6519a9590..d555b874c 100644 --- a/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S +++ b/src/libffmpeg/libavcodec/alpha/dsputil_alpha_asm.S @@ -34,7 +34,7 @@ #define tf a4 #define tg a3 #define th v0 - + .set noat .set noreorder .arch pca56 @@ -71,7 +71,7 @@ $unaligned: addq a1, a2, a1 nop - ldq_u t4, 0(a1) + ldq_u t4, 0(a1) ldq_u t5, 8(a1) addq a1, a2, a1 nop @@ -120,25 +120,25 @@ $aligned: addq a1, a2, a1 ldq t3, 0(a1) - addq a0, a2, t4 - addq a1, a2, a1 - addq t4, a2, t5 - subq a3, 4, a3 + addq a0, a2, t4 + addq a1, a2, a1 + addq t4, a2, t5 + subq a3, 4, a3 + + stq t0, 0(a0) + addq t5, a2, t6 + stq t1, 0(t4) + addq t6, a2, a0 - stq t0, 0(a0) - addq t5, a2, t6 - stq t1, 0(t4) - addq t6, a2, a0 + stq t2, 0(t5) + stq t3, 0(t6) - stq t2, 0(t5) - stq t3, 0(t6) - - bne a3, $aligned + bne a3, $aligned ret .end put_pixels_axp_asm /************************************************************************ - * void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, + * void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, * int line_size) */ .align 6 @@ -172,17 +172,17 @@ put_pixels_clamped_mvi_asm: addq a1, a2, ta maxsw4 t3, zero, t3 minsw4 t0, t8, t0 - + minsw4 t1, t8, t1 minsw4 t2, t8, t2 minsw4 t3, t8, t3 pkwb t0, t0 - + pkwb t1, t1 pkwb t2, t2 pkwb t3, t3 stl t0, 0(a1) - + stl t1, 4(a1) addq ta, a2, a1 stl t2, 0(ta) @@ -193,7 +193,7 @@ put_pixels_clamped_mvi_asm: .end put_pixels_clamped_mvi_asm /************************************************************************ - * void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, + * void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels, * int line_size) */ .align 6 @@ -236,18 +236,18 @@ add_pixels_clamped_mvi_asm: bic t0, tg, t0 # 0 2 unpkbw t7, t7 # 2 0 and t3, tg, t5 # 1 1 - addq t0, t1, t0 # 0 3 + addq t0, t1, t0 # 0 3 xor t0, t2, t0 # 0 4 unpkbw ta, ta # 3 0 and t6, tg, t8 # 2 1 maxsw4 t0, zero, t0 # 0 5 - + bic t3, tg, t3 # 1 2 bic t6, tg, t6 # 2 2 minsw4 t0, tf, t0 # 0 6 addq t3, t4, t3 # 1 3 - + pkwb t0, t0 # 0 7 xor t3, t5, t3 # 1 4 maxsw4 t3, zero, t3 # 1 5 @@ -260,14 +260,14 @@ add_pixels_clamped_mvi_asm: maxsw4 t6, zero, t6 # 2 5 addq t9, ta, t9 # 3 3 - stl t0, 0(a1) # 0 8 + stl t0, 0(a1) # 0 8 minsw4 t6, tf, t6 # 2 6 xor t9, tb, t9 # 3 4 maxsw4 t9, zero, t9 # 3 5 lda a0, 32(a0) # block += 16; pkwb t3, t3 # 1 7 - + minsw4 t9, tf, t9 # 3 6 subq th, 2, th pkwb t6, t6 # 2 7 @@ -279,5 +279,5 @@ add_pixels_clamped_mvi_asm: stl t9, 4(te) # 3 8 bne th, 1b - ret + ret .end add_pixels_clamped_mvi_asm diff --git a/src/libffmpeg/libavcodec/alpha/motion_est_alpha.c b/src/libffmpeg/libavcodec/alpha/motion_est_alpha.c index 8b8a0a25c..ea8580be7 100644 --- a/src/libffmpeg/libavcodec/alpha/motion_est_alpha.c +++ b/src/libffmpeg/libavcodec/alpha/motion_est_alpha.c @@ -14,7 +14,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #include "asm.h" @@ -30,7 +30,7 @@ void get_pixels_mvi(DCTELEM *restrict block, p = ldq(pixels); stq(unpkbw(p), block); - stq(unpkbw(p >> 32), block + 4); + stq(unpkbw(p >> 32), block + 4); pixels += line_size; block += 8; @@ -116,7 +116,7 @@ int pix_abs8x8_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) return result; } -#if 0 /* now done in assembly */ +#if 0 /* now done in assembly */ int pix_abs16x16_mvi(uint8_t *pix1, uint8_t *pix2, int line_size) { int result = 0; @@ -187,7 +187,7 @@ int pix_abs16x16_x2_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, in /* |.......l|lllllllr|rrrrrrr*| This case is special because disalign1 would be 8, which gets treated as 0 by extqh. At least it is a bit faster - that way :) */ + that way :) */ do { uint64_t p1_l, p1_r, p2_l, p2_r; uint64_t l, m, r; @@ -201,7 +201,7 @@ int pix_abs16x16_x2_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, in p2_r = avg2(extql(m, disalign) | extqh(r, disalign), r); pix1 += line_size; pix2 += line_size; - + result += perr(p1_l, p2_l) + perr(p1_r, p2_r); } while (--h); @@ -288,7 +288,7 @@ int pix_abs16x16_y2_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, in int pix_abs16x16_xy2_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) { int result = 0; - + uint64_t p1_l, p1_r; uint64_t p2_l, p2_r, p2_x; diff --git a/src/libffmpeg/libavcodec/alpha/motion_est_mvi_asm.S b/src/libffmpeg/libavcodec/alpha/motion_est_mvi_asm.S index 9e6b75f53..276d310ef 100644 --- a/src/libffmpeg/libavcodec/alpha/motion_est_mvi_asm.S +++ b/src/libffmpeg/libavcodec/alpha/motion_est_mvi_asm.S @@ -29,7 +29,7 @@ #define tf a4 #define tg a3 #define th v0 - + .set noat .set noreorder .arch pca56 @@ -91,7 +91,7 @@ $unaligned: ldq t4, 8(a0) # ref right addq a0, a2, a0 # pix1 addq a1, a2, a1 # pix2 - /* load line 1 */ + /* load line 1 */ ldq_u t5, 0(a1) # left_u ldq_u t6, 8(a1) # mid ldq_u t7, 16(a1) # right_u diff --git a/src/libffmpeg/libavcodec/alpha/mpegvideo_alpha.c b/src/libffmpeg/libavcodec/alpha/mpegvideo_alpha.c index f64fb7472..4c512451e 100644 --- a/src/libffmpeg/libavcodec/alpha/mpegvideo_alpha.c +++ b/src/libffmpeg/libavcodec/alpha/mpegvideo_alpha.c @@ -14,7 +14,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #include "asm.h" @@ -28,22 +28,22 @@ static void dct_unquantize_h263_intra_axp(MpegEncContext *s, DCTELEM *block, uint64_t qmul, qadd; uint64_t correction; DCTELEM *orig_block = block; - DCTELEM block0; + DCTELEM block0; /* might not be used uninitialized */ qadd = WORD_VEC((qscale - 1) | 1); qmul = qscale << 1; - /* This mask kills spill from negative subwords to the next subword. */ + /* This mask kills spill from negative subwords to the next subword. */ correction = WORD_VEC((qmul - 1) + 1); /* multiplication / addition */ if (!s->h263_aic) { - if (n < 4) + if (n < 4) block0 = block[0] * s->y_dc_scale; else block0 = block[0] * s->c_dc_scale; } else { qadd = 0; } - n_coeffs = 63; // does not always use zigzag table + n_coeffs = 63; // does not always use zigzag table for(i = 0; i <= n_coeffs; block += 4, i += 4) { uint64_t levels, negmask, zeros, add; @@ -92,12 +92,10 @@ static void dct_unquantize_h263_inter_axp(MpegEncContext *s, DCTELEM *block, int i, n_coeffs; uint64_t qmul, qadd; uint64_t correction; - DCTELEM *orig_block = block; - DCTELEM block0; qadd = WORD_VEC((qscale - 1) | 1); qmul = qscale << 1; - /* This mask kills spill from negative subwords to the next subword. */ + /* This mask kills spill from negative subwords to the next subword. */ correction = WORD_VEC((qmul - 1) + 1); /* multiplication / addition */ n_coeffs = s->intra_scantable.raster_end[s->block_last_index[n]]; diff --git a/src/libffmpeg/libavcodec/alpha/simple_idct_alpha.c b/src/libffmpeg/libavcodec/alpha/simple_idct_alpha.c index 293a2f970..3a5db009b 100644 --- a/src/libffmpeg/libavcodec/alpha/simple_idct_alpha.c +++ b/src/libffmpeg/libavcodec/alpha/simple_idct_alpha.c @@ -15,7 +15,7 @@ * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA * * based upon some outcommented c code from mpeg2dec (idct_mmx.c * written by Aaron Holtzman <aholtzma@ess.engr.uvic.ca>) @@ -29,7 +29,7 @@ extern void (*put_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, int line_size); -extern void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, +extern void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels, int line_size); // cos(i * M_PI / 16) * sqrt(2) * (1 << 14) @@ -55,7 +55,7 @@ static inline int idct_row(DCTELEM *row) if (l == 0 && r == 0) return 0; - + a0 = W4 * sextw(l) + (1 << (ROW_SHIFT - 1)); if (((l & ~0xffffUL) | r) == 0) { @@ -63,7 +63,7 @@ static inline int idct_row(DCTELEM *row) t2 = (uint16_t) a0; t2 |= t2 << 16; t2 |= t2 << 32; - + stq(t2, row); stq(t2, row + 4); return 1; @@ -123,7 +123,7 @@ static inline int idct_row(DCTELEM *row) b3 -= W5 * t; } - + t = extwl(r, 2); /* row[5] */ if (t) { t = sextw(t); @@ -285,7 +285,7 @@ void simple_idct_axp(DCTELEM *block) stq(v, block + 1 * 4); stq(w, block + 2 * 4); stq(w, block + 3 * 4); - block += 4 * 4; + block += 4 * 4; } } else { for (i = 0; i < 8; i++) |