arm: hpeldsp: Move half-pel assembly from dsputil to hpeldsp
authorRonald S. Bultje <rsbultje@gmail.com>
Sun, 10 Mar 2013 23:16:45 +0000 (16:16 -0700)
committerMartin Storsjö <martin@martin.st>
Fri, 19 Apr 2013 20:19:08 +0000 (23:19 +0300)
Signed-off-by: Martin Storsjö <martin@martin.st>
16 files changed:
libavcodec/arm/Makefile
libavcodec/arm/dsputil_arm.S
libavcodec/arm/dsputil_armv6.S
libavcodec/arm/dsputil_init_arm.c
libavcodec/arm/dsputil_init_armv6.c
libavcodec/arm/dsputil_init_neon.c
libavcodec/arm/dsputil_neon.S
libavcodec/arm/hpeldsp_arm.S [new file with mode: 0644]
libavcodec/arm/hpeldsp_arm.h [new file with mode: 0644]
libavcodec/arm/hpeldsp_armv6.S [new file with mode: 0644]
libavcodec/arm/hpeldsp_init_arm.c [new file with mode: 0644]
libavcodec/arm/hpeldsp_init_armv6.c [new file with mode: 0644]
libavcodec/arm/hpeldsp_init_neon.c [new file with mode: 0644]
libavcodec/arm/hpeldsp_neon.S [new file with mode: 0644]
libavcodec/hpeldsp.c
libavcodec/hpeldsp.h

index 1c91d62..5b717b7 100644 (file)
@@ -31,6 +31,11 @@ OBJS-$(CONFIG_H264DSP)                 += arm/h264dsp_init_arm.o
 OBJS-$(CONFIG_H264PRED)                += arm/h264pred_init_arm.o
 OBJS-$(CONFIG_H264QPEL)                += arm/h264qpel_init_arm.o
 
+OBJS-$(CONFIG_HPELDSP)                 += arm/hpeldsp_init_arm.o        \
+                                          arm/hpeldsp_arm.o
+ARMV6-OBJS-$(CONFIG_HPELDSP)           += arm/hpeldsp_init_armv6.o      \
+                                          arm/hpeldsp_armv6.o
+
 OBJS-$(CONFIG_RV30_DECODER)            += arm/rv34dsp_init_arm.o
 OBJS-$(CONFIG_RV40_DECODER)            += arm/rv34dsp_init_arm.o        \
                                           arm/rv40dsp_init_arm.o        \
@@ -84,6 +89,9 @@ NEON-OBJS-$(CONFIG_AAC_DECODER)        += arm/sbrdsp_neon.o             \
 NEON-OBJS-$(CONFIG_DCA_DECODER)        += arm/dcadsp_neon.o             \
                                           arm/synth_filter_neon.o       \
 
+NEON-OBJS-$(CONFIG_HPELDSP)            += arm/hpeldsp_init_neon.o       \
+                                          arm/hpeldsp_neon.o
+
 NEON-OBJS-$(CONFIG_MPEGVIDEO)          += arm/mpegvideo_neon.o
 NEON-OBJS-$(CONFIG_RV30_DECODER)       += arm/rv34dsp_neon.o
 NEON-OBJS-$(CONFIG_RV40_DECODER)       += arm/rv34dsp_neon.o            \
index 8504032..1692a58 100644 (file)
 #define pld @
 #endif
 
-.macro  ALIGN_QWORD_D shift, Rd0, Rd1, Rd2, Rd3, Rn0, Rn1, Rn2, Rn3, Rn4
-        mov             \Rd0, \Rn0, lsr #(\shift * 8)
-        mov             \Rd1, \Rn1, lsr #(\shift * 8)
-        mov             \Rd2, \Rn2, lsr #(\shift * 8)
-        mov             \Rd3, \Rn3, lsr #(\shift * 8)
-        orr             \Rd0, \Rd0, \Rn1, lsl #(32 - \shift * 8)
-        orr             \Rd1, \Rd1, \Rn2, lsl #(32 - \shift * 8)
-        orr             \Rd2, \Rd2, \Rn3, lsl #(32 - \shift * 8)
-        orr             \Rd3, \Rd3, \Rn4, lsl #(32 - \shift * 8)
-.endm
-.macro  ALIGN_DWORD shift, R0, R1, R2
-        mov             \R0, \R0, lsr #(\shift * 8)
-        orr             \R0, \R0, \R1, lsl #(32 - \shift * 8)
-        mov             \R1, \R1, lsr #(\shift * 8)
-        orr             \R1, \R1, \R2, lsl #(32 - \shift * 8)
-.endm
-.macro  ALIGN_DWORD_D shift, Rdst0, Rdst1, Rsrc0, Rsrc1, Rsrc2
-        mov             \Rdst0, \Rsrc0, lsr #(\shift * 8)
-        mov             \Rdst1, \Rsrc1, lsr #(\shift * 8)
-        orr             \Rdst0, \Rdst0, \Rsrc1, lsl #(32 - (\shift * 8))
-        orr             \Rdst1, \Rdst1, \Rsrc2, lsl #(32 - (\shift * 8))
-.endm
-
-.macro  RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
-        @ Rd = (Rn | Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
-        @ Rmask = 0xFEFEFEFE
-        @ Rn = destroy
-        eor             \Rd0, \Rn0, \Rm0
-        eor             \Rd1, \Rn1, \Rm1
-        orr             \Rn0, \Rn0, \Rm0
-        orr             \Rn1, \Rn1, \Rm1
-        and             \Rd0, \Rd0, \Rmask
-        and             \Rd1, \Rd1, \Rmask
-        sub             \Rd0, \Rn0, \Rd0, lsr #1
-        sub             \Rd1, \Rn1, \Rd1, lsr #1
-.endm
-
-.macro  NO_RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
-        @ Rd = (Rn & Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
-        @ Rmask = 0xFEFEFEFE
-        @ Rn = destroy
-        eor             \Rd0, \Rn0, \Rm0
-        eor             \Rd1, \Rn1, \Rm1
-        and             \Rn0, \Rn0, \Rm0
-        and             \Rn1, \Rn1, \Rm1
-        and             \Rd0, \Rd0, \Rmask
-        and             \Rd1, \Rd1, \Rmask
-        add             \Rd0, \Rn0, \Rd0, lsr #1
-        add             \Rd1, \Rn1, \Rd1, lsr #1
-.endm
-
-.macro  JMP_ALIGN tmp, reg
-        ands            \tmp, \reg, #3
-        bic             \reg, \reg, #3
-        beq             1f
-        subs            \tmp, \tmp, #1
-        beq             2f
-        subs            \tmp, \tmp, #1
-        beq             3f
-        b    4f
-.endm
-
-@ ----------------------------------------------------------------
-        .align 5
-function ff_put_pixels16_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r11, lr}
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r7}
-        add             r1,  r1,  r2
-        stm             r0,  {r4-r7}
-        pld             [r1]
-        subs            r3,  r3,  #1
-        add             r0,  r0,  r2
-        bne             1b
-        pop             {r4-r11, pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r8}
-        add             r1,  r1,  r2
-        ALIGN_QWORD_D   1,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r9-r12}
-        add             r0,  r0,  r2
-        bne             2b
-        pop             {r4-r11, pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r8}
-        add             r1,  r1,  r2
-        ALIGN_QWORD_D   2,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r9-r12}
-        add             r0,  r0,  r2
-        bne             3b
-        pop             {r4-r11, pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r8}
-        add             r1,  r1,  r2
-        ALIGN_QWORD_D   3,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r9-r12}
-        add             r0,  r0,  r2
-        bne             4b
-        pop             {r4-r11,pc}
-endfunc
-
-@ ----------------------------------------------------------------
-        .align 5
-function ff_put_pixels8_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r5,lr}
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r5}
-        add             r1,  r1,  r2
-        subs            r3,  r3,  #1
-        pld             [r1]
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             1b
-        pop             {r4-r5,pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r5, r12}
-        add             r1,  r1,  r2
-        ALIGN_DWORD     1,   r4,  r5,  r12
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             2b
-        pop             {r4-r5,pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r5, r12}
-        add             r1,  r1,  r2
-        ALIGN_DWORD     2,   r4,  r5,  r12
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             3b
-        pop             {r4-r5,pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r5, r12}
-        add             r1,  r1,  r2
-        ALIGN_DWORD     3,   r4,  r5,  r12
-        pld             [r1]
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             4b
-        pop             {r4-r5,pc}
-endfunc
-
-@ ----------------------------------------------------------------
-        .align 5
-function ff_put_pixels8_x2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r10,lr}
-        ldr             r12, =0xfefefefe
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
-        pld             [r1]
-        RND_AVG32       r8,  r9,  r4,  r5,  r6,  r7,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             1b
-        pop             {r4-r10,pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
-        ALIGN_DWORD_D   2,   r8,  r9,  r4,  r5,  r10
-        pld             [r1]
-        RND_AVG32       r4,  r5,  r6,  r7,  r8,  r9,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             2b
-        pop             {r4-r10,pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   2,   r6,  r7,  r4,  r5,  r10
-        ALIGN_DWORD_D   3,   r8,  r9,  r4,  r5,  r10
-        pld             [r1]
-        RND_AVG32       r4,  r5,  r6,  r7,  r8,  r9,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             3b
-        pop             {r4-r10,pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   3,   r6,  r7,  r4,  r5,  r10
-        pld             [r1]
-        RND_AVG32       r8,  r9,  r6,  r7,  r5,  r10, r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             4b
-        pop             {r4-r10,pc}
-endfunc
-
-        .align 5
-function ff_put_no_rnd_pixels8_x2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r10,lr}
-        ldr             r12, =0xfefefefe
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
-        pld             [r1]
-        NO_RND_AVG32    r8,  r9,  r4,  r5,  r6,  r7,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             1b
-        pop             {r4-r10,pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
-        ALIGN_DWORD_D   2,   r8,  r9,  r4,  r5,  r10
-        pld             [r1]
-        NO_RND_AVG32    r4,  r5,  r6,  r7,  r8,  r9,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             2b
-        pop             {r4-r10,pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   2,   r6,  r7,  r4,  r5,  r10
-        ALIGN_DWORD_D   3,   r8,  r9,  r4,  r5,  r10
-        pld             [r1]
-        NO_RND_AVG32    r4,  r5,  r6,  r7,  r8,  r9,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bne             3b
-        pop             {r4-r10,pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r5, r10}
-        add             r1,  r1,  r2
-        ALIGN_DWORD_D   3,   r6,  r7,  r4,  r5,  r10
-        pld             [r1]
-        NO_RND_AVG32    r8,  r9,  r6,  r7,  r5,  r10, r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             4b
-        pop             {r4-r10,pc}
-endfunc
-
-
-@ ----------------------------------------------------------------
-        .align 5
-function ff_put_pixels8_y2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r11,lr}
-        mov             r3,  r3,  lsr #1
-        ldr             r12, =0xfefefefe
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r5}
-        add             r1,  r1,  r2
-6:      ldm             r1,  {r6-r7}
-        add             r1,  r1,  r2
-        pld             [r1]
-        RND_AVG32       r8,  r9,  r4,  r5,  r6,  r7,  r12
-        ldm             r1,  {r4-r5}
-        add             r1,  r1,  r2
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        pld             [r1]
-        RND_AVG32       r8,  r9,  r6,  r7,  r4,  r5,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r7,  r8,  r9
-        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r7,  r8,  r9
-        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r7,  r8,  r9
-        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-endfunc
-
-        .align 5
-function ff_put_no_rnd_pixels8_y2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r11,lr}
-        mov             r3,  r3,  lsr #1
-        ldr             r12, =0xfefefefe
-        JMP_ALIGN       r5,  r1
-1:
-        ldm             r1,  {r4-r5}
-        add             r1,  r1,  r2
-6:      ldm             r1,  {r6-r7}
-        add             r1,  r1,  r2
-        pld             [r1]
-        NO_RND_AVG32    r8,  r9,  r4,  r5,  r6,  r7,  r12
-        ldm             r1,  {r4-r5}
-        add             r1,  r1,  r2
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        pld             [r1]
-        NO_RND_AVG32    r8,  r9,  r6,  r7,  r4,  r5,  r12
-        subs            r3,  r3,  #1
-        stm             r0,  {r8-r9}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-2:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r7,  r8,  r9
-        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     1,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-3:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r7,  r8,  r9
-        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     2,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-        .align 5
-4:
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r4,  r5,  r6
-6:      ldm             r1,  {r7-r9}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r7,  r8,  r9
-        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        ldm             r1,  {r4-r6}
-        add             r1,  r1,  r2
-        pld             [r1]
-        ALIGN_DWORD     3,   r4,  r5,  r6
-        subs            r3,  r3,  #1
-        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
-        stm             r0,  {r10-r11}
-        add             r0,  r0,  r2
-        bne             6b
-        pop             {r4-r11,pc}
-endfunc
-
-        .ltorg
-
-@ ----------------------------------------------------------------
-.macro  RND_XY2_IT align, rnd
-        @ l1=  (a & 0x03030303) + (b & 0x03030303) ?(+ 0x02020202)
-        @ h1= ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2)
-.if \align == 0
-        ldm             r1,  {r6-r8}
-.elseif \align == 3
-        ldm             r1,  {r5-r7}
-.else
-        ldm             r1,  {r8-r10}
-.endif
-        add             r1,  r1,  r2
-        pld             [r1]
-.if \align == 0
-        ALIGN_DWORD_D   1,   r4,  r5,  r6,  r7,  r8
-.elseif \align == 1
-        ALIGN_DWORD_D   1,   r4,  r5,  r8,  r9,  r10
-        ALIGN_DWORD_D   2,   r6,  r7,  r8,  r9,  r10
-.elseif \align == 2
-        ALIGN_DWORD_D   2,   r4,  r5,  r8,  r9,  r10
-        ALIGN_DWORD_D   3,   r6,  r7,  r8,  r9,  r10
-.elseif \align == 3
-        ALIGN_DWORD_D   3,   r4,  r5,  r5,  r6,  r7
-.endif
-        ldr             r14, =0x03030303
-        tst             r3,  #1
-        and             r8,  r4,  r14
-        and             r9,  r5,  r14
-        and             r10, r6,  r14
-        and             r11, r7,  r14
-        it              eq
-        andeq           r14, r14, r14, \rnd #1
-        add             r8,  r8,  r10
-        add             r9,  r9,  r11
-        ldr             r12, =0xfcfcfcfc >> 2
-        itt             eq
-        addeq           r8,  r8,  r14
-        addeq           r9,  r9,  r14
-        and             r4,  r12, r4,  lsr #2
-        and             r5,  r12, r5,  lsr #2
-        and             r6,  r12, r6,  lsr #2
-        and             r7,  r12, r7,  lsr #2
-        add             r10, r4,  r6
-        add             r11, r5,  r7
-        subs            r3,  r3,  #1
-.endm
-
-.macro RND_XY2_EXPAND align, rnd
-        RND_XY2_IT      \align, \rnd
-6:      push            {r8-r11}
-        RND_XY2_IT      \align, \rnd
-        pop             {r4-r7}
-        add             r4,  r4,  r8
-        add             r5,  r5,  r9
-        ldr             r14, =0x0f0f0f0f
-        add             r6,  r6,  r10
-        add             r7,  r7,  r11
-        and             r4,  r14, r4,  lsr #2
-        and             r5,  r14, r5,  lsr #2
-        add             r4,  r4,  r6
-        add             r5,  r5,  r7
-        stm             r0,  {r4-r5}
-        add             r0,  r0,  r2
-        bge             6b
-        pop             {r4-r11,pc}
-.endm
-
-        .align 5
-function ff_put_pixels8_xy2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r11,lr} @ R14 is also called LR
-        JMP_ALIGN       r5,  r1
-1:      RND_XY2_EXPAND  0, lsl
-        .align 5
-2:      RND_XY2_EXPAND  1, lsl
-        .align 5
-3:      RND_XY2_EXPAND  2, lsl
-        .align 5
-4:      RND_XY2_EXPAND  3, lsl
-endfunc
-
-        .align 5
-function ff_put_no_rnd_pixels8_xy2_arm, export=1
-        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-        @ block = word aligned, pixles = unaligned
-        pld             [r1]
-        push            {r4-r11,lr}
-        JMP_ALIGN       r5,  r1
-1:      RND_XY2_EXPAND  0, lsr
-        .align 5
-2:      RND_XY2_EXPAND  1, lsr
-        .align 5
-3:      RND_XY2_EXPAND  2, lsr
-        .align 5
-4:      RND_XY2_EXPAND  3, lsr
-endfunc
-
         .align 5
 @ void ff_add_pixels_clamped_arm(int16_t *block, uint8_t *dest, int stride)
 function ff_add_pixels_clamped_arm, export=1
index 61535a1..e667a47 100644 (file)
 
 #include "libavutil/arm/asm.S"
 
-.macro  call_2x_pixels  type, subp
-function ff_\type\()_pixels16\subp\()_armv6, export=1
-        push            {r0-r3, lr}
-        bl              ff_\type\()_pixels8\subp\()_armv6
-        pop             {r0-r3, lr}
-        add             r0,  r0,  #8
-        add             r1,  r1,  #8
-        b               ff_\type\()_pixels8\subp\()_armv6
-endfunc
-.endm
-
-call_2x_pixels          avg
-call_2x_pixels          put, _x2
-call_2x_pixels          put, _y2
-call_2x_pixels          put, _x2_no_rnd
-call_2x_pixels          put, _y2_no_rnd
-
-function ff_put_pixels16_armv6, export=1
-        push            {r4-r11}
-1:
-        ldr             r5,  [r1, #4]
-        ldr             r6,  [r1, #8]
-        ldr             r7,  [r1, #12]
-        ldr_post        r4,  r1,  r2
-        strd            r6,  r7,  [r0, #8]
-        ldr             r9,  [r1, #4]
-        strd_post       r4,  r5,  r0,  r2
-        ldr             r10, [r1, #8]
-        ldr             r11, [r1, #12]
-        ldr_post        r8,  r1,  r2
-        strd            r10, r11, [r0, #8]
-        subs            r3,  r3,  #2
-        strd_post       r8,  r9,  r0,  r2
-        bne             1b
-
-        pop             {r4-r11}
-        bx              lr
-endfunc
-
-function ff_put_pixels8_armv6, export=1
-        push            {r4-r7}
-1:
-        ldr             r5,  [r1, #4]
-        ldr_post        r4,  r1,  r2
-        ldr             r7,  [r1, #4]
-        strd_post       r4,  r5,  r0,  r2
-        ldr_post        r6,  r1,  r2
-        subs            r3,  r3,  #2
-        strd_post       r6,  r7,  r0,  r2
-        bne             1b
-
-        pop             {r4-r7}
-        bx              lr
-endfunc
-
-function ff_put_pixels8_x2_armv6, export=1
-        push            {r4-r11, lr}
-        mov             r12, #1
-        orr             r12, r12, r12, lsl #8
-        orr             r12, r12, r12, lsl #16
-1:
-        ldr             r4,  [r1]
-        subs            r3,  r3,  #2
-        ldr             r5,  [r1, #4]
-        ldr             r7,  [r1, #5]
-        lsr             r6,  r4,  #8
-        ldr_pre         r8,  r1,  r2
-        orr             r6,  r6,  r5,  lsl #24
-        ldr             r9,  [r1, #4]
-        ldr             r11, [r1, #5]
-        lsr             r10, r8,  #8
-        add             r1,  r1,  r2
-        orr             r10, r10, r9,  lsl #24
-        eor             r14, r4,  r6
-        uhadd8          r4,  r4,  r6
-        eor             r6,  r5,  r7
-        uhadd8          r5,  r5,  r7
-        and             r14, r14, r12
-        and             r6,  r6,  r12
-        uadd8           r4,  r4,  r14
-        eor             r14, r8,  r10
-        uadd8           r5,  r5,  r6
-        eor             r6,  r9,  r11
-        uhadd8          r8,  r8,  r10
-        and             r14, r14, r12
-        uhadd8          r9,  r9,  r11
-        and             r6,  r6,  r12
-        uadd8           r8,  r8,  r14
-        strd_post       r4,  r5,  r0,  r2
-        uadd8           r9,  r9,  r6
-        strd_post       r8,  r9,  r0,  r2
-        bne             1b
-
-        pop             {r4-r11, pc}
-endfunc
-
-function ff_put_pixels8_y2_armv6, export=1
-        push            {r4-r11}
-        mov             r12, #1
-        orr             r12, r12, r12, lsl #8
-        orr             r12, r12, r12, lsl #16
-        ldr             r4,  [r1]
-        ldr             r5,  [r1, #4]
-        ldr_pre         r6,  r1,  r2
-        ldr             r7,  [r1, #4]
-1:
-        subs            r3,  r3,  #2
-        uhadd8          r8,  r4,  r6
-        eor             r10, r4,  r6
-        uhadd8          r9,  r5,  r7
-        eor             r11, r5,  r7
-        and             r10, r10, r12
-        ldr_pre         r4,  r1,  r2
-        uadd8           r8,  r8,  r10
-        and             r11, r11, r12
-        uadd8           r9,  r9,  r11
-        ldr             r5,  [r1, #4]
-        uhadd8          r10, r4,  r6
-        eor             r6,  r4,  r6
-        uhadd8          r11, r5,  r7
-        and             r6,  r6,  r12
-        eor             r7,  r5,  r7
-        uadd8           r10, r10, r6
-        and             r7,  r7,  r12
-        ldr_pre         r6,  r1,  r2
-        uadd8           r11, r11, r7
-        strd_post       r8,  r9,  r0,  r2
-        ldr             r7,  [r1, #4]
-        strd_post       r10, r11, r0,  r2
-        bne             1b
-
-        pop             {r4-r11}
-        bx              lr
-endfunc
-
-function ff_put_pixels8_x2_no_rnd_armv6, export=1
-        push            {r4-r9, lr}
-1:
-        subs            r3,  r3,  #2
-        ldr             r4,  [r1]
-        ldr             r5,  [r1, #4]
-        ldr             r7,  [r1, #5]
-        ldr_pre         r8,  r1,  r2
-        ldr             r9,  [r1, #4]
-        ldr             r14, [r1, #5]
-        add             r1,  r1,  r2
-        lsr             r6,  r4,  #8
-        orr             r6,  r6,  r5,  lsl #24
-        lsr             r12, r8,  #8
-        orr             r12, r12, r9,  lsl #24
-        uhadd8          r4,  r4,  r6
-        uhadd8          r5,  r5,  r7
-        uhadd8          r8,  r8,  r12
-        uhadd8          r9,  r9,  r14
-        stm             r0,  {r4,r5}
-        add             r0,  r0,  r2
-        stm             r0,  {r8,r9}
-        add             r0,  r0,  r2
-        bne             1b
-
-        pop             {r4-r9, pc}
-endfunc
-
-function ff_put_pixels8_y2_no_rnd_armv6, export=1
-        push            {r4-r9, lr}
-        ldr             r4,  [r1]
-        ldr             r5,  [r1, #4]
-        ldr_pre         r6,  r1,  r2
-        ldr             r7,  [r1, #4]
-1:
-        subs            r3,  r3,  #2
-        uhadd8          r8,  r4,  r6
-        ldr_pre         r4,  r1,  r2
-        uhadd8          r9,  r5,  r7
-        ldr             r5,  [r1, #4]
-        uhadd8          r12, r4,  r6
-        ldr_pre         r6,  r1,  r2
-        uhadd8          r14, r5,  r7
-        ldr             r7,  [r1, #4]
-        stm             r0,  {r8,r9}
-        add             r0,  r0,  r2
-        stm             r0,  {r12,r14}
-        add             r0,  r0,  r2
-        bne             1b
-
-        pop             {r4-r9, pc}
-endfunc
-
-function ff_avg_pixels8_armv6, export=1
-        pld             [r1, r2]
-        push            {r4-r10, lr}
-        mov             lr,  #1
-        orr             lr,  lr,  lr,  lsl #8
-        orr             lr,  lr,  lr,  lsl #16
-        ldrd            r4,  r5,  [r0]
-        ldr             r10, [r1, #4]
-        ldr_post        r9,  r1,  r2
-        subs            r3,  r3,  #2
-1:
-        pld             [r1, r2]
-        eor             r8,  r4,  r9
-        uhadd8          r4,  r4,  r9
-        eor             r12, r5,  r10
-        ldrd_reg        r6,  r7,  r0,  r2
-        uhadd8          r5,  r5,  r10
-        and             r8,  r8,  lr
-        ldr             r10, [r1, #4]
-        and             r12, r12, lr
-        uadd8           r4,  r4,  r8
-        ldr_post        r9,  r1,  r2
-        eor             r8,  r6,  r9
-        uadd8           r5,  r5,  r12
-        pld             [r1, r2,  lsl #1]
-        eor             r12, r7,  r10
-        uhadd8          r6,  r6,  r9
-        strd_post       r4,  r5,  r0,  r2
-        uhadd8          r7,  r7,  r10
-        beq             2f
-        and             r8,  r8,  lr
-        ldrd_reg        r4,  r5,  r0,  r2
-        uadd8           r6,  r6,  r8
-        ldr             r10, [r1, #4]
-        and             r12, r12, lr
-        subs            r3,  r3,  #2
-        uadd8           r7,  r7,  r12
-        ldr_post        r9,  r1,  r2
-        strd_post       r6,  r7,  r0,  r2
-        b               1b
-2:
-        and             r8,  r8,  lr
-        and             r12, r12, lr
-        uadd8           r6,  r6,  r8
-        uadd8           r7,  r7,  r12
-        strd_post       r6,  r7,  r0,  r2
-
-        pop             {r4-r10, pc}
-endfunc
-
 function ff_add_pixels_clamped_armv6, export=1
         push            {r4-r8,lr}
         mov             r3,  #8
index cc24c0a..bb68eb6 100644 (file)
@@ -30,24 +30,6 @@ void ff_simple_idct_arm(int16_t *data);
 static void (*ff_put_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size);
 static void (*ff_add_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size);
 
-void ff_put_pixels8_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-void ff_put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-void ff_put_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-void ff_put_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-
-void ff_put_no_rnd_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-void ff_put_no_rnd_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-void ff_put_no_rnd_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-
-void ff_put_pixels16_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
-
-CALL_2X_PIXELS(ff_put_pixels16_x2_arm,         ff_put_pixels8_x2_arm,        8)
-CALL_2X_PIXELS(ff_put_pixels16_y2_arm,         ff_put_pixels8_y2_arm,        8)
-CALL_2X_PIXELS(ff_put_pixels16_xy2_arm,        ff_put_pixels8_xy2_arm,       8)
-CALL_2X_PIXELS(ff_put_no_rnd_pixels16_x2_arm,  ff_put_no_rnd_pixels8_x2_arm, 8)
-CALL_2X_PIXELS(ff_put_no_rnd_pixels16_y2_arm,  ff_put_no_rnd_pixels8_y2_arm, 8)
-CALL_2X_PIXELS(ff_put_no_rnd_pixels16_xy2_arm, ff_put_no_rnd_pixels8_xy2_arm,8)
-
 void ff_add_pixels_clamped_arm(const int16_t *block, uint8_t *dest,
                                int line_size);
 
@@ -76,7 +58,6 @@ static void simple_idct_arm_add(uint8_t *dest, int line_size, int16_t *block)
 
 av_cold void ff_dsputil_init_arm(DSPContext *c, AVCodecContext *avctx)
 {
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
     int cpu_flags = av_get_cpu_flags();
 
     ff_put_pixels_clamped = c->put_pixels_clamped;
@@ -99,26 +80,6 @@ av_cold void ff_dsputil_init_arm(DSPContext *c, AVCodecContext *avctx)
 
     c->add_pixels_clamped = ff_add_pixels_clamped_arm;
 
-    if (!high_bit_depth) {
-    c->put_pixels_tab[0][0] = ff_put_pixels16_arm;
-    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_arm;
-    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_arm;
-    c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_arm;
-    c->put_pixels_tab[1][0] = ff_put_pixels8_arm;
-    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_arm;
-    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_arm;
-    c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_arm;
-
-    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_arm;
-    c->put_no_rnd_pixels_tab[0][1] = ff_put_no_rnd_pixels16_x2_arm;
-    c->put_no_rnd_pixels_tab[0][2] = ff_put_no_rnd_pixels16_y2_arm;
-    c->put_no_rnd_pixels_tab[0][3] = ff_put_no_rnd_pixels16_xy2_arm;
-    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_arm;
-    c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_arm;
-    c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_arm;
-    c->put_no_rnd_pixels_tab[1][3] = ff_put_no_rnd_pixels8_xy2_arm;
-    }
-
     if (have_armv5te(cpu_flags)) ff_dsputil_init_armv5te(c, avctx);
     if (have_armv6(cpu_flags))   ff_dsputil_init_armv6(c, avctx);
     if (have_neon(cpu_flags))    ff_dsputil_init_neon(c, avctx);
index 5e2c661..4c8ba47 100644 (file)
@@ -27,24 +27,6 @@ void ff_simple_idct_armv6(int16_t *data);
 void ff_simple_idct_put_armv6(uint8_t *dest, int line_size, int16_t *data);
 void ff_simple_idct_add_armv6(uint8_t *dest, int line_size, int16_t *data);
 
-void ff_put_pixels16_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_x2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_y2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_put_pixels16_x2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_y2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_avg_pixels16_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_put_pixels8_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_x2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_y2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_put_pixels8_x2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_y2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_avg_pixels8_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
 void ff_add_pixels_clamped_armv6(const int16_t *block,
                                  uint8_t *restrict pixels,
                                  int line_size);
@@ -82,29 +64,6 @@ av_cold void ff_dsputil_init_armv6(DSPContext *c, AVCodecContext *avctx)
         c->idct_permutation_type = FF_LIBMPEG2_IDCT_PERM;
     }
 
-    if (!high_bit_depth) {
-    c->put_pixels_tab[0][0] = ff_put_pixels16_armv6;
-    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_armv6;
-    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_armv6;
-/*     c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_armv6; */
-    c->put_pixels_tab[1][0] = ff_put_pixels8_armv6;
-    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_armv6;
-    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_armv6;
-/*     c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_armv6; */
-
-    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_armv6;
-    c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_armv6;
-    c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_armv6;
-/*     c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_armv6; */
-    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_armv6;
-    c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_armv6;
-    c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_armv6;
-/*     c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_armv6; */
-
-    c->avg_pixels_tab[0][0] = ff_avg_pixels16_armv6;
-    c->avg_pixels_tab[1][0] = ff_avg_pixels8_armv6;
-    }
-
     if (!high_bit_depth)
         c->get_pixels = ff_get_pixels_armv6;
     c->add_pixels_clamped = ff_add_pixels_clamped_armv6;
index 5866118..793fab1 100644 (file)
@@ -32,33 +32,6 @@ void ff_simple_idct_add_neon(uint8_t *dest, int line_size, int16_t *data);
 void ff_clear_block_neon(int16_t *block);
 void ff_clear_blocks_neon(int16_t *blocks);
 
-void ff_put_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
-void ff_avg_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-void ff_avg_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
-
 void ff_add_pixels_clamped_neon(const int16_t *, uint8_t *, int);
 void ff_put_pixels_clamped_neon(const int16_t *, uint8_t *, int);
 void ff_put_signed_pixels_clamped_neon(const int16_t *, uint8_t *, int);
@@ -92,38 +65,6 @@ av_cold void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
     if (!high_bit_depth) {
         c->clear_block  = ff_clear_block_neon;
         c->clear_blocks = ff_clear_blocks_neon;
-
-        c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
-        c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
-        c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
-        c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
-        c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
-        c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
-        c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
-        c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
-
-        c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
-        c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
-        c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
-        c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
-        c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
-        c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
-        c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
-        c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
-
-        c->avg_pixels_tab[0][0] = ff_avg_pixels16_neon;
-        c->avg_pixels_tab[0][1] = ff_avg_pixels16_x2_neon;
-        c->avg_pixels_tab[0][2] = ff_avg_pixels16_y2_neon;
-        c->avg_pixels_tab[0][3] = ff_avg_pixels16_xy2_neon;
-        c->avg_pixels_tab[1][0] = ff_avg_pixels8_neon;
-        c->avg_pixels_tab[1][1] = ff_avg_pixels8_x2_neon;
-        c->avg_pixels_tab[1][2] = ff_avg_pixels8_y2_neon;
-        c->avg_pixels_tab[1][3] = ff_avg_pixels8_xy2_neon;
-
-        c->avg_no_rnd_pixels_tab[0] = ff_avg_pixels16_neon;
-        c->avg_no_rnd_pixels_tab[1] = ff_avg_pixels16_x2_no_rnd_neon;
-        c->avg_no_rnd_pixels_tab[2] = ff_avg_pixels16_y2_no_rnd_neon;
-        c->avg_no_rnd_pixels_tab[3] = ff_avg_pixels16_xy2_no_rnd_neon;
     }
 
     c->add_pixels_clamped = ff_add_pixels_clamped_neon;
index f33fa33..81662c0 100644 (file)
@@ -37,394 +37,6 @@ function ff_clear_blocks_neon, export=1
         bx              lr
 endfunc
 
-.macro  pixels16        rnd=1, avg=0
-  .if \avg
-        mov             r12, r0
-  .endif
-1:      vld1.8          {q0},     [r1], r2
-        vld1.8          {q1},     [r1], r2
-        vld1.8          {q2},     [r1], r2
-        pld             [r1, r2, lsl #2]
-        vld1.8          {q3},     [r1], r2
-        pld             [r1]
-        pld             [r1, r2]
-        pld             [r1, r2, lsl #1]
-  .if \avg
-        vld1.8          {q8},     [r12,:128], r2
-        vrhadd.u8       q0,  q0,  q8
-        vld1.8          {q9},     [r12,:128], r2
-        vrhadd.u8       q1,  q1,  q9
-        vld1.8          {q10},    [r12,:128], r2
-        vrhadd.u8       q2,  q2,  q10
-        vld1.8          {q11},    [r12,:128], r2
-        vrhadd.u8       q3,  q3,  q11
-  .endif
-        subs            r3,  r3,  #4
-        vst1.64         {q0},     [r0,:128], r2
-        vst1.64         {q1},     [r0,:128], r2
-        vst1.64         {q2},     [r0,:128], r2
-        vst1.64         {q3},     [r0,:128], r2
-        bne             1b
-        bx              lr
-.endm
-
-.macro  pixels16_x2     rnd=1, avg=0
-1:      vld1.8          {d0-d2},  [r1], r2
-        vld1.8          {d4-d6},  [r1], r2
-        pld             [r1]
-        pld             [r1, r2]
-        subs            r3,  r3,  #2
-        vext.8          q1,  q0,  q1,  #1
-        avg             q0,  q0,  q1
-        vext.8          q3,  q2,  q3,  #1
-        avg             q2,  q2,  q3
-  .if \avg
-        vld1.8          {q1},     [r0,:128], r2
-        vld1.8          {q3},     [r0,:128]
-        vrhadd.u8       q0,  q0,  q1
-        vrhadd.u8       q2,  q2,  q3
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {q0},     [r0,:128], r2
-        vst1.8          {q2},     [r0,:128], r2
-        bne             1b
-        bx              lr
-.endm
-
-.macro  pixels16_y2     rnd=1, avg=0
-        sub             r3,  r3,  #2
-        vld1.8          {q0},     [r1], r2
-        vld1.8          {q1},     [r1], r2
-1:      subs            r3,  r3,  #2
-        avg             q2,  q0,  q1
-        vld1.8          {q0},     [r1], r2
-        avg             q3,  q0,  q1
-        vld1.8          {q1},     [r1], r2
-        pld             [r1]
-        pld             [r1, r2]
-  .if \avg
-        vld1.8          {q8},     [r0,:128], r2
-        vld1.8          {q9},     [r0,:128]
-        vrhadd.u8       q2,  q2,  q8
-        vrhadd.u8       q3,  q3,  q9
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {q2},     [r0,:128], r2
-        vst1.8          {q3},     [r0,:128], r2
-        bne             1b
-
-        avg             q2,  q0,  q1
-        vld1.8          {q0},     [r1], r2
-        avg             q3,  q0,  q1
-  .if \avg
-        vld1.8          {q8},     [r0,:128], r2
-        vld1.8          {q9},     [r0,:128]
-        vrhadd.u8       q2,  q2,  q8
-        vrhadd.u8       q3,  q3,  q9
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {q2},     [r0,:128], r2
-        vst1.8          {q3},     [r0,:128], r2
-
-        bx              lr
-.endm
-
-.macro  pixels16_xy2    rnd=1, avg=0
-        sub             r3,  r3,  #2
-        vld1.8          {d0-d2},  [r1], r2
-        vld1.8          {d4-d6},  [r1], r2
-NRND    vmov.i16        q13, #1
-        pld             [r1]
-        pld             [r1, r2]
-        vext.8          q1,  q0,  q1,  #1
-        vext.8          q3,  q2,  q3,  #1
-        vaddl.u8        q8,  d0,  d2
-        vaddl.u8        q10, d1,  d3
-        vaddl.u8        q9,  d4,  d6
-        vaddl.u8        q11, d5,  d7
-1:      subs            r3,  r3,  #2
-        vld1.8          {d0-d2},  [r1], r2
-        vadd.u16        q12, q8,  q9
-        pld             [r1]
-NRND    vadd.u16        q12, q12, q13
-        vext.8          q15, q0,  q1,  #1
-        vadd.u16        q1 , q10, q11
-        shrn            d28, q12, #2
-NRND    vadd.u16        q1,  q1,  q13
-        shrn            d29, q1,  #2
-  .if \avg
-        vld1.8          {q8},     [r0,:128]
-        vrhadd.u8       q14, q14, q8
-  .endif
-        vaddl.u8        q8,  d0,  d30
-        vld1.8          {d2-d4},  [r1], r2
-        vaddl.u8        q10, d1,  d31
-        vst1.8          {q14},    [r0,:128], r2
-        vadd.u16        q12, q8,  q9
-        pld             [r1, r2]
-NRND    vadd.u16        q12, q12, q13
-        vext.8          q2,  q1,  q2,  #1
-        vadd.u16        q0,  q10, q11
-        shrn            d30, q12, #2
-NRND    vadd.u16        q0,  q0,  q13
-        shrn            d31, q0,  #2
-  .if \avg
-        vld1.8          {q9},     [r0,:128]
-        vrhadd.u8       q15, q15, q9
-  .endif
-        vaddl.u8        q9,  d2,  d4
-        vaddl.u8        q11, d3,  d5
-        vst1.8          {q15},    [r0,:128], r2
-        bgt             1b
-
-        vld1.8          {d0-d2},  [r1], r2
-        vadd.u16        q12, q8,  q9
-NRND    vadd.u16        q12, q12, q13
-        vext.8          q15, q0,  q1,  #1
-        vadd.u16        q1 , q10, q11
-        shrn            d28, q12, #2
-NRND    vadd.u16        q1,  q1,  q13
-        shrn            d29, q1,  #2
-  .if \avg
-        vld1.8          {q8},     [r0,:128]
-        vrhadd.u8       q14, q14, q8
-  .endif
-        vaddl.u8        q8,  d0,  d30
-        vaddl.u8        q10, d1,  d31
-        vst1.8          {q14},    [r0,:128], r2
-        vadd.u16        q12, q8,  q9
-NRND    vadd.u16        q12, q12, q13
-        vadd.u16        q0,  q10, q11
-        shrn            d30, q12, #2
-NRND    vadd.u16        q0,  q0,  q13
-        shrn            d31, q0,  #2
-  .if \avg
-        vld1.8          {q9},     [r0,:128]
-        vrhadd.u8       q15, q15, q9
-  .endif
-        vst1.8          {q15},    [r0,:128], r2
-
-        bx              lr
-.endm
-
-.macro  pixels8         rnd=1, avg=0
-1:      vld1.8          {d0},     [r1], r2
-        vld1.8          {d1},     [r1], r2
-        vld1.8          {d2},     [r1], r2
-        pld             [r1, r2, lsl #2]
-        vld1.8          {d3},     [r1], r2
-        pld             [r1]
-        pld             [r1, r2]
-        pld             [r1, r2, lsl #1]
-  .if \avg
-        vld1.8          {d4},     [r0,:64], r2
-        vrhadd.u8       d0,  d0,  d4
-        vld1.8          {d5},     [r0,:64], r2
-        vrhadd.u8       d1,  d1,  d5
-        vld1.8          {d6},     [r0,:64], r2
-        vrhadd.u8       d2,  d2,  d6
-        vld1.8          {d7},     [r0,:64], r2
-        vrhadd.u8       d3,  d3,  d7
-        sub             r0,  r0,  r2,  lsl #2
-  .endif
-        subs            r3,  r3,  #4
-        vst1.8          {d0},     [r0,:64], r2
-        vst1.8          {d1},     [r0,:64], r2
-        vst1.8          {d2},     [r0,:64], r2
-        vst1.8          {d3},     [r0,:64], r2
-        bne             1b
-        bx              lr
-.endm
-
-.macro  pixels8_x2      rnd=1, avg=0
-1:      vld1.8          {q0},     [r1], r2
-        vext.8          d1,  d0,  d1,  #1
-        vld1.8          {q1},     [r1], r2
-        vext.8          d3,  d2,  d3,  #1
-        pld             [r1]
-        pld             [r1, r2]
-        subs            r3,  r3,  #2
-        vswp            d1,  d2
-        avg             q0,  q0,  q1
-  .if \avg
-        vld1.8          {d4},     [r0,:64], r2
-        vld1.8          {d5},     [r0,:64]
-        vrhadd.u8       q0,  q0,  q2
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {d0},     [r0,:64], r2
-        vst1.8          {d1},     [r0,:64], r2
-        bne             1b
-        bx              lr
-.endm
-
-.macro  pixels8_y2      rnd=1, avg=0
-        sub             r3,  r3,  #2
-        vld1.8          {d0},     [r1], r2
-        vld1.8          {d1},     [r1], r2
-1:      subs            r3,  r3,  #2
-        avg             d4,  d0,  d1
-        vld1.8          {d0},     [r1], r2
-        avg             d5,  d0,  d1
-        vld1.8          {d1},     [r1], r2
-        pld             [r1]
-        pld             [r1, r2]
-  .if \avg
-        vld1.8          {d2},     [r0,:64], r2
-        vld1.8          {d3},     [r0,:64]
-        vrhadd.u8       q2,  q2,  q1
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {d4},     [r0,:64], r2
-        vst1.8          {d5},     [r0,:64], r2
-        bne             1b
-
-        avg             d4,  d0,  d1
-        vld1.8          {d0},     [r1], r2
-        avg             d5,  d0,  d1
-  .if \avg
-        vld1.8          {d2},     [r0,:64], r2
-        vld1.8          {d3},     [r0,:64]
-        vrhadd.u8       q2,  q2,  q1
-        sub             r0,  r0,  r2
-  .endif
-        vst1.8          {d4},     [r0,:64], r2
-        vst1.8          {d5},     [r0,:64], r2
-
-        bx              lr
-.endm
-
-.macro  pixels8_xy2     rnd=1, avg=0
-        sub             r3,  r3,  #2
-        vld1.8          {q0},     [r1], r2
-        vld1.8          {q1},     [r1], r2
-NRND    vmov.i16        q11, #1
-        pld             [r1]
-        pld             [r1, r2]
-        vext.8          d4,  d0,  d1,  #1
-        vext.8          d6,  d2,  d3,  #1
-        vaddl.u8        q8,  d0,  d4
-        vaddl.u8        q9,  d2,  d6
-1:      subs            r3,  r3,  #2
-        vld1.8          {q0},     [r1], r2
-        pld             [r1]
-        vadd.u16        q10, q8,  q9
-        vext.8          d4,  d0,  d1,  #1
-NRND    vadd.u16        q10, q10, q11
-        vaddl.u8        q8,  d0,  d4
-        shrn            d5,  q10, #2
-        vld1.8          {q1},     [r1], r2
-        vadd.u16        q10, q8,  q9
-        pld             [r1, r2]
-  .if \avg
-        vld1.8          {d7},     [r0,:64]
-        vrhadd.u8       d5,  d5,  d7
-  .endif
-NRND    vadd.u16        q10, q10, q11
-        vst1.8          {d5},     [r0,:64], r2
-        shrn            d7,  q10, #2
-  .if \avg
-        vld1.8          {d5},     [r0,:64]
-        vrhadd.u8       d7,  d7,  d5
-  .endif
-        vext.8          d6,  d2,  d3,  #1
-        vaddl.u8        q9,  d2,  d6
-        vst1.8          {d7},     [r0,:64], r2
-        bgt             1b
-
-        vld1.8          {q0},     [r1], r2
-        vadd.u16        q10, q8,  q9
-        vext.8          d4,  d0,  d1,  #1
-NRND    vadd.u16        q10, q10, q11
-        vaddl.u8        q8,  d0,  d4
-        shrn            d5,  q10, #2
-        vadd.u16        q10, q8,  q9
-  .if \avg
-        vld1.8          {d7},     [r0,:64]
-        vrhadd.u8       d5,  d5,  d7
-  .endif
-NRND    vadd.u16        q10, q10, q11
-        vst1.8          {d5},     [r0,:64], r2
-        shrn            d7,  q10, #2
-  .if \avg
-        vld1.8          {d5},     [r0,:64]
-        vrhadd.u8       d7,  d7,  d5
-  .endif
-        vst1.8          {d7},     [r0,:64], r2
-
-        bx              lr
-.endm
-
-.macro  pixfunc         pfx, name, suf, rnd=1, avg=0
-  .if \rnd
-    .macro avg  rd, rn, rm
-        vrhadd.u8       \rd, \rn, \rm
-    .endm
-    .macro shrn rd, rn, rm
-        vrshrn.u16      \rd, \rn, \rm
-    .endm
-    .macro NRND insn:vararg
-    .endm
-  .else
-    .macro avg  rd, rn, rm
-        vhadd.u8        \rd, \rn, \rm
-    .endm
-    .macro shrn rd, rn, rm
-        vshrn.u16       \rd, \rn, \rm
-    .endm
-    .macro NRND insn:vararg
-        \insn
-    .endm
-  .endif
-function ff_\pfx\name\suf\()_neon, export=1
-        \name           \rnd, \avg
-endfunc
-        .purgem         avg
-        .purgem         shrn
-        .purgem         NRND
-.endm
-
-.macro  pixfunc2        pfx, name, avg=0
-        pixfunc         \pfx, \name,          rnd=1, avg=\avg
-        pixfunc         \pfx, \name, _no_rnd, rnd=0, avg=\avg
-.endm
-
-function ff_put_h264_qpel16_mc00_neon, export=1
-        mov             r3,  #16
-endfunc
-
-        pixfunc         put_, pixels16,     avg=0
-        pixfunc2        put_, pixels16_x2,  avg=0
-        pixfunc2        put_, pixels16_y2,  avg=0
-        pixfunc2        put_, pixels16_xy2, avg=0
-
-function ff_avg_h264_qpel16_mc00_neon, export=1
-        mov             r3,  #16
-endfunc
-
-        pixfunc         avg_, pixels16,     avg=1
-        pixfunc2        avg_, pixels16_x2,  avg=1
-        pixfunc2        avg_, pixels16_y2,  avg=1
-        pixfunc2        avg_, pixels16_xy2, avg=1
-
-function ff_put_h264_qpel8_mc00_neon, export=1
-        mov             r3,  #8
-endfunc
-
-        pixfunc         put_, pixels8,     avg=0
-        pixfunc2        put_, pixels8_x2,  avg=0
-        pixfunc2        put_, pixels8_y2,  avg=0
-        pixfunc2        put_, pixels8_xy2, avg=0
-
-function ff_avg_h264_qpel8_mc00_neon, export=1
-        mov             r3,  #8
-endfunc
-
-        pixfunc         avg_, pixels8,     avg=1
-        pixfunc         avg_, pixels8_x2,  avg=1
-        pixfunc         avg_, pixels8_y2,  avg=1
-        pixfunc         avg_, pixels8_xy2, avg=1
-
 function ff_put_pixels_clamped_neon, export=1
         vld1.16         {d16-d19}, [r0,:128]!
         vqmovun.s16     d0, q8
diff --git a/libavcodec/arm/hpeldsp_arm.S b/libavcodec/arm/hpeldsp_arm.S
new file mode 100644 (file)
index 0000000..d4f97e3
--- /dev/null
@@ -0,0 +1,611 @@
+@
+@ ARMv4 optimized DSP utils
+@ Copyright (c) 2004 AGAWA Koji <i (AT) atty (DOT) jp>
+@
+@ This file is part of Libav.
+@
+@ Libav is free software; you can redistribute it and/or
+@ modify it under the terms of the GNU Lesser General Public
+@ License as published by the Free Software Foundation; either
+@ version 2.1 of the License, or (at your option) any later version.
+@
+@ Libav is distributed in the hope that it will be useful,
+@ but WITHOUT ANY WARRANTY; without even the implied warranty of
+@ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+@ Lesser General Public License for more details.
+@
+@ You should have received a copy of the GNU Lesser General Public
+@ License along with Libav; if not, write to the Free Software
+@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+@
+
+#include "config.h"
+#include "libavutil/arm/asm.S"
+
+#if !HAVE_ARMV5TE_EXTERNAL
+#define pld @
+#endif
+
+.macro  ALIGN_QWORD_D shift, Rd0, Rd1, Rd2, Rd3, Rn0, Rn1, Rn2, Rn3, Rn4
+        mov             \Rd0, \Rn0, lsr #(\shift * 8)
+        mov             \Rd1, \Rn1, lsr #(\shift * 8)
+        mov             \Rd2, \Rn2, lsr #(\shift * 8)
+        mov             \Rd3, \Rn3, lsr #(\shift * 8)
+        orr             \Rd0, \Rd0, \Rn1, lsl #(32 - \shift * 8)
+        orr             \Rd1, \Rd1, \Rn2, lsl #(32 - \shift * 8)
+        orr             \Rd2, \Rd2, \Rn3, lsl #(32 - \shift * 8)
+        orr             \Rd3, \Rd3, \Rn4, lsl #(32 - \shift * 8)
+.endm
+.macro  ALIGN_DWORD shift, R0, R1, R2
+        mov             \R0, \R0, lsr #(\shift * 8)
+        orr             \R0, \R0, \R1, lsl #(32 - \shift * 8)
+        mov             \R1, \R1, lsr #(\shift * 8)
+        orr             \R1, \R1, \R2, lsl #(32 - \shift * 8)
+.endm
+.macro  ALIGN_DWORD_D shift, Rdst0, Rdst1, Rsrc0, Rsrc1, Rsrc2
+        mov             \Rdst0, \Rsrc0, lsr #(\shift * 8)
+        mov             \Rdst1, \Rsrc1, lsr #(\shift * 8)
+        orr             \Rdst0, \Rdst0, \Rsrc1, lsl #(32 - (\shift * 8))
+        orr             \Rdst1, \Rdst1, \Rsrc2, lsl #(32 - (\shift * 8))
+.endm
+
+.macro  RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
+        @ Rd = (Rn | Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
+        @ Rmask = 0xFEFEFEFE
+        @ Rn = destroy
+        eor             \Rd0, \Rn0, \Rm0
+        eor             \Rd1, \Rn1, \Rm1
+        orr             \Rn0, \Rn0, \Rm0
+        orr             \Rn1, \Rn1, \Rm1
+        and             \Rd0, \Rd0, \Rmask
+        and             \Rd1, \Rd1, \Rmask
+        sub             \Rd0, \Rn0, \Rd0, lsr #1
+        sub             \Rd1, \Rn1, \Rd1, lsr #1
+.endm
+
+.macro  NO_RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
+        @ Rd = (Rn & Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
+        @ Rmask = 0xFEFEFEFE
+        @ Rn = destroy
+        eor             \Rd0, \Rn0, \Rm0
+        eor             \Rd1, \Rn1, \Rm1
+        and             \Rn0, \Rn0, \Rm0
+        and             \Rn1, \Rn1, \Rm1
+        and             \Rd0, \Rd0, \Rmask
+        and             \Rd1, \Rd1, \Rmask
+        add             \Rd0, \Rn0, \Rd0, lsr #1
+        add             \Rd1, \Rn1, \Rd1, lsr #1
+.endm
+
+.macro  JMP_ALIGN tmp, reg
+        ands            \tmp, \reg, #3
+        bic             \reg, \reg, #3
+        beq             1f
+        subs            \tmp, \tmp, #1
+        beq             2f
+        subs            \tmp, \tmp, #1
+        beq             3f
+        b    4f
+.endm
+
+@ ----------------------------------------------------------------
+        .align 5
+function ff_put_pixels16_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r11, lr}
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r7}
+        add             r1,  r1,  r2
+        stm             r0,  {r4-r7}
+        pld             [r1]
+        subs            r3,  r3,  #1
+        add             r0,  r0,  r2
+        bne             1b
+        pop             {r4-r11, pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r8}
+        add             r1,  r1,  r2
+        ALIGN_QWORD_D   1,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r9-r12}
+        add             r0,  r0,  r2
+        bne             2b
+        pop             {r4-r11, pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r8}
+        add             r1,  r1,  r2
+        ALIGN_QWORD_D   2,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r9-r12}
+        add             r0,  r0,  r2
+        bne             3b
+        pop             {r4-r11, pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r8}
+        add             r1,  r1,  r2
+        ALIGN_QWORD_D   3,   r9,  r10, r11, r12, r4,  r5,  r6,  r7,  r8
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r9-r12}
+        add             r0,  r0,  r2
+        bne             4b
+        pop             {r4-r11,pc}
+endfunc
+
+@ ----------------------------------------------------------------
+        .align 5
+function ff_put_pixels8_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r5,lr}
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r5}
+        add             r1,  r1,  r2
+        subs            r3,  r3,  #1
+        pld             [r1]
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             1b
+        pop             {r4-r5,pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r5, r12}
+        add             r1,  r1,  r2
+        ALIGN_DWORD     1,   r4,  r5,  r12
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             2b
+        pop             {r4-r5,pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r5, r12}
+        add             r1,  r1,  r2
+        ALIGN_DWORD     2,   r4,  r5,  r12
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             3b
+        pop             {r4-r5,pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r5, r12}
+        add             r1,  r1,  r2
+        ALIGN_DWORD     3,   r4,  r5,  r12
+        pld             [r1]
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             4b
+        pop             {r4-r5,pc}
+endfunc
+
+@ ----------------------------------------------------------------
+        .align 5
+function ff_put_pixels8_x2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r10,lr}
+        ldr             r12, =0xfefefefe
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
+        pld             [r1]
+        RND_AVG32       r8,  r9,  r4,  r5,  r6,  r7,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             1b
+        pop             {r4-r10,pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
+        ALIGN_DWORD_D   2,   r8,  r9,  r4,  r5,  r10
+        pld             [r1]
+        RND_AVG32       r4,  r5,  r6,  r7,  r8,  r9,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             2b
+        pop             {r4-r10,pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   2,   r6,  r7,  r4,  r5,  r10
+        ALIGN_DWORD_D   3,   r8,  r9,  r4,  r5,  r10
+        pld             [r1]
+        RND_AVG32       r4,  r5,  r6,  r7,  r8,  r9,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             3b
+        pop             {r4-r10,pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   3,   r6,  r7,  r4,  r5,  r10
+        pld             [r1]
+        RND_AVG32       r8,  r9,  r6,  r7,  r5,  r10, r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             4b
+        pop             {r4-r10,pc}
+endfunc
+
+        .align 5
+function ff_put_no_rnd_pixels8_x2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r10,lr}
+        ldr             r12, =0xfefefefe
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
+        pld             [r1]
+        NO_RND_AVG32    r8,  r9,  r4,  r5,  r6,  r7,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             1b
+        pop             {r4-r10,pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   1,   r6,  r7,  r4,  r5,  r10
+        ALIGN_DWORD_D   2,   r8,  r9,  r4,  r5,  r10
+        pld             [r1]
+        NO_RND_AVG32    r4,  r5,  r6,  r7,  r8,  r9,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             2b
+        pop             {r4-r10,pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   2,   r6,  r7,  r4,  r5,  r10
+        ALIGN_DWORD_D   3,   r8,  r9,  r4,  r5,  r10
+        pld             [r1]
+        NO_RND_AVG32    r4,  r5,  r6,  r7,  r8,  r9,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bne             3b
+        pop             {r4-r10,pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r5, r10}
+        add             r1,  r1,  r2
+        ALIGN_DWORD_D   3,   r6,  r7,  r4,  r5,  r10
+        pld             [r1]
+        NO_RND_AVG32    r8,  r9,  r6,  r7,  r5,  r10, r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             4b
+        pop             {r4-r10,pc}
+endfunc
+
+
+@ ----------------------------------------------------------------
+        .align 5
+function ff_put_pixels8_y2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r11,lr}
+        mov             r3,  r3,  lsr #1
+        ldr             r12, =0xfefefefe
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r5}
+        add             r1,  r1,  r2
+6:      ldm             r1,  {r6-r7}
+        add             r1,  r1,  r2
+        pld             [r1]
+        RND_AVG32       r8,  r9,  r4,  r5,  r6,  r7,  r12
+        ldm             r1,  {r4-r5}
+        add             r1,  r1,  r2
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        pld             [r1]
+        RND_AVG32       r8,  r9,  r6,  r7,  r4,  r5,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r7,  r8,  r9
+        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r7,  r8,  r9
+        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r7,  r8,  r9
+        RND_AVG32       r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        RND_AVG32       r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+endfunc
+
+        .align 5
+function ff_put_no_rnd_pixels8_y2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r11,lr}
+        mov             r3,  r3,  lsr #1
+        ldr             r12, =0xfefefefe
+        JMP_ALIGN       r5,  r1
+1:
+        ldm             r1,  {r4-r5}
+        add             r1,  r1,  r2
+6:      ldm             r1,  {r6-r7}
+        add             r1,  r1,  r2
+        pld             [r1]
+        NO_RND_AVG32    r8,  r9,  r4,  r5,  r6,  r7,  r12
+        ldm             r1,  {r4-r5}
+        add             r1,  r1,  r2
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        pld             [r1]
+        NO_RND_AVG32    r8,  r9,  r6,  r7,  r4,  r5,  r12
+        subs            r3,  r3,  #1
+        stm             r0,  {r8-r9}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+2:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r7,  r8,  r9
+        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     1,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+3:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r7,  r8,  r9
+        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     2,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+        .align 5
+4:
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r4,  r5,  r6
+6:      ldm             r1,  {r7-r9}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r7,  r8,  r9
+        NO_RND_AVG32    r10, r11, r4,  r5,  r7,  r8,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        ldm             r1,  {r4-r6}
+        add             r1,  r1,  r2
+        pld             [r1]
+        ALIGN_DWORD     3,   r4,  r5,  r6
+        subs            r3,  r3,  #1
+        NO_RND_AVG32    r10, r11, r7,  r8,  r4,  r5,  r12
+        stm             r0,  {r10-r11}
+        add             r0,  r0,  r2
+        bne             6b
+        pop             {r4-r11,pc}
+endfunc
+
+        .ltorg
+
+@ ----------------------------------------------------------------
+.macro  RND_XY2_IT align, rnd
+        @ l1=  (a & 0x03030303) + (b & 0x03030303) ?(+ 0x02020202)
+        @ h1= ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2)
+.if \align == 0
+        ldm             r1,  {r6-r8}
+.elseif \align == 3
+        ldm             r1,  {r5-r7}
+.else
+        ldm             r1,  {r8-r10}
+.endif
+        add             r1,  r1,  r2
+        pld             [r1]
+.if \align == 0
+        ALIGN_DWORD_D   1,   r4,  r5,  r6,  r7,  r8
+.elseif \align == 1
+        ALIGN_DWORD_D   1,   r4,  r5,  r8,  r9,  r10
+        ALIGN_DWORD_D   2,   r6,  r7,  r8,  r9,  r10
+.elseif \align == 2
+        ALIGN_DWORD_D   2,   r4,  r5,  r8,  r9,  r10
+        ALIGN_DWORD_D   3,   r6,  r7,  r8,  r9,  r10
+.elseif \align == 3
+        ALIGN_DWORD_D   3,   r4,  r5,  r5,  r6,  r7
+.endif
+        ldr             r14, =0x03030303
+        tst             r3,  #1
+        and             r8,  r4,  r14
+        and             r9,  r5,  r14
+        and             r10, r6,  r14
+        and             r11, r7,  r14
+        it              eq
+        andeq           r14, r14, r14, \rnd #1
+        add             r8,  r8,  r10
+        add             r9,  r9,  r11
+        ldr             r12, =0xfcfcfcfc >> 2
+        itt             eq
+        addeq           r8,  r8,  r14
+        addeq           r9,  r9,  r14
+        and             r4,  r12, r4,  lsr #2
+        and             r5,  r12, r5,  lsr #2
+        and             r6,  r12, r6,  lsr #2
+        and             r7,  r12, r7,  lsr #2
+        add             r10, r4,  r6
+        add             r11, r5,  r7
+        subs            r3,  r3,  #1
+.endm
+
+.macro RND_XY2_EXPAND align, rnd
+        RND_XY2_IT      \align, \rnd
+6:      push            {r8-r11}
+        RND_XY2_IT      \align, \rnd
+        pop             {r4-r7}
+        add             r4,  r4,  r8
+        add             r5,  r5,  r9
+        ldr             r14, =0x0f0f0f0f
+        add             r6,  r6,  r10
+        add             r7,  r7,  r11
+        and             r4,  r14, r4,  lsr #2
+        and             r5,  r14, r5,  lsr #2
+        add             r4,  r4,  r6
+        add             r5,  r5,  r7
+        stm             r0,  {r4-r5}
+        add             r0,  r0,  r2
+        bge             6b
+        pop             {r4-r11,pc}
+.endm
+
+        .align 5
+function ff_put_pixels8_xy2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r11,lr} @ R14 is also called LR
+        JMP_ALIGN       r5,  r1
+1:      RND_XY2_EXPAND  0, lsl
+        .align 5
+2:      RND_XY2_EXPAND  1, lsl
+        .align 5
+3:      RND_XY2_EXPAND  2, lsl
+        .align 5
+4:      RND_XY2_EXPAND  3, lsl
+endfunc
+
+        .align 5
+function ff_put_no_rnd_pixels8_xy2_arm, export=1
+        @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
+        @ block = word aligned, pixles = unaligned
+        pld             [r1]
+        push            {r4-r11,lr}
+        JMP_ALIGN       r5,  r1
+1:      RND_XY2_EXPAND  0, lsr
+        .align 5
+2:      RND_XY2_EXPAND  1, lsr
+        .align 5
+3:      RND_XY2_EXPAND  2, lsr
+        .align 5
+4:      RND_XY2_EXPAND  3, lsr
+endfunc
diff --git a/libavcodec/arm/hpeldsp_arm.h b/libavcodec/arm/hpeldsp_arm.h
new file mode 100644 (file)
index 0000000..6b5c1ce
--- /dev/null
@@ -0,0 +1,27 @@
+/*
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_ARM_HPELDSP_H
+#define AVCODEC_ARM_HPELDSP_H
+
+#include "libavcodec/hpeldsp.h"
+
+void ff_hpeldsp_init_armv6(HpelDSPContext *c, int flags);
+void ff_hpeldsp_init_neon(HpelDSPContext *c, int flags);
+
+#endif /* AVCODEC_ARM_HPELDSP_H */
diff --git a/libavcodec/arm/hpeldsp_armv6.S b/libavcodec/arm/hpeldsp_armv6.S
new file mode 100644 (file)
index 0000000..a030d42
--- /dev/null
@@ -0,0 +1,259 @@
+/*
+ * Copyright (c) 2009 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/arm/asm.S"
+
+.macro  call_2x_pixels  type, subp
+function ff_\type\()_pixels16\subp\()_armv6, export=1
+        push            {r0-r3, lr}
+        bl              ff_\type\()_pixels8\subp\()_armv6
+        pop             {r0-r3, lr}
+        add             r0,  r0,  #8
+        add             r1,  r1,  #8
+        b               ff_\type\()_pixels8\subp\()_armv6
+endfunc
+.endm
+
+call_2x_pixels          avg
+call_2x_pixels          put, _x2
+call_2x_pixels          put, _y2
+call_2x_pixels          put, _x2_no_rnd
+call_2x_pixels          put, _y2_no_rnd
+
+function ff_put_pixels16_armv6, export=1
+        push            {r4-r11}
+1:
+        ldr             r5,  [r1, #4]
+        ldr             r6,  [r1, #8]
+        ldr             r7,  [r1, #12]
+        ldr_post        r4,  r1,  r2
+        strd            r6,  r7,  [r0, #8]
+        ldr             r9,  [r1, #4]
+        strd_post       r4,  r5,  r0,  r2
+        ldr             r10, [r1, #8]
+        ldr             r11, [r1, #12]
+        ldr_post        r8,  r1,  r2
+        strd            r10, r11, [r0, #8]
+        subs            r3,  r3,  #2
+        strd_post       r8,  r9,  r0,  r2
+        bne             1b
+
+        pop             {r4-r11}
+        bx              lr
+endfunc
+
+function ff_put_pixels8_armv6, export=1
+        push            {r4-r7}
+1:
+        ldr             r5,  [r1, #4]
+        ldr_post        r4,  r1,  r2
+        ldr             r7,  [r1, #4]
+        strd_post       r4,  r5,  r0,  r2
+        ldr_post        r6,  r1,  r2
+        subs            r3,  r3,  #2
+        strd_post       r6,  r7,  r0,  r2
+        bne             1b
+
+        pop             {r4-r7}
+        bx              lr
+endfunc
+
+function ff_put_pixels8_x2_armv6, export=1
+        push            {r4-r11, lr}
+        mov             r12, #1
+        orr             r12, r12, r12, lsl #8
+        orr             r12, r12, r12, lsl #16
+1:
+        ldr             r4,  [r1]
+        subs            r3,  r3,  #2
+        ldr             r5,  [r1, #4]
+        ldr             r7,  [r1, #5]
+        lsr             r6,  r4,  #8
+        ldr_pre         r8,  r1,  r2
+        orr             r6,  r6,  r5,  lsl #24
+        ldr             r9,  [r1, #4]
+        ldr             r11, [r1, #5]
+        lsr             r10, r8,  #8
+        add             r1,  r1,  r2
+        orr             r10, r10, r9,  lsl #24
+        eor             r14, r4,  r6
+        uhadd8          r4,  r4,  r6
+        eor             r6,  r5,  r7
+        uhadd8          r5,  r5,  r7
+        and             r14, r14, r12
+        and             r6,  r6,  r12
+        uadd8           r4,  r4,  r14
+        eor             r14, r8,  r10
+        uadd8           r5,  r5,  r6
+        eor             r6,  r9,  r11
+        uhadd8          r8,  r8,  r10
+        and             r14, r14, r12
+        uhadd8          r9,  r9,  r11
+        and             r6,  r6,  r12
+        uadd8           r8,  r8,  r14
+        strd_post       r4,  r5,  r0,  r2
+        uadd8           r9,  r9,  r6
+        strd_post       r8,  r9,  r0,  r2
+        bne             1b
+
+        pop             {r4-r11, pc}
+endfunc
+
+function ff_put_pixels8_y2_armv6, export=1
+        push            {r4-r11}
+        mov             r12, #1
+        orr             r12, r12, r12, lsl #8
+        orr             r12, r12, r12, lsl #16
+        ldr             r4,  [r1]
+        ldr             r5,  [r1, #4]
+        ldr_pre         r6,  r1,  r2
+        ldr             r7,  [r1, #4]
+1:
+        subs            r3,  r3,  #2
+        uhadd8          r8,  r4,  r6
+        eor             r10, r4,  r6
+        uhadd8          r9,  r5,  r7
+        eor             r11, r5,  r7
+        and             r10, r10, r12
+        ldr_pre         r4,  r1,  r2
+        uadd8           r8,  r8,  r10
+        and             r11, r11, r12
+        uadd8           r9,  r9,  r11
+        ldr             r5,  [r1, #4]
+        uhadd8          r10, r4,  r6
+        eor             r6,  r4,  r6
+        uhadd8          r11, r5,  r7
+        and             r6,  r6,  r12
+        eor             r7,  r5,  r7
+        uadd8           r10, r10, r6
+        and             r7,  r7,  r12
+        ldr_pre         r6,  r1,  r2
+        uadd8           r11, r11, r7
+        strd_post       r8,  r9,  r0,  r2
+        ldr             r7,  [r1, #4]
+        strd_post       r10, r11, r0,  r2
+        bne             1b
+
+        pop             {r4-r11}
+        bx              lr
+endfunc
+
+function ff_put_pixels8_x2_no_rnd_armv6, export=1
+        push            {r4-r9, lr}
+1:
+        subs            r3,  r3,  #2
+        ldr             r4,  [r1]
+        ldr             r5,  [r1, #4]
+        ldr             r7,  [r1, #5]
+        ldr_pre         r8,  r1,  r2
+        ldr             r9,  [r1, #4]
+        ldr             r14, [r1, #5]
+        add             r1,  r1,  r2
+        lsr             r6,  r4,  #8
+        orr             r6,  r6,  r5,  lsl #24
+        lsr             r12, r8,  #8
+        orr             r12, r12, r9,  lsl #24
+        uhadd8          r4,  r4,  r6
+        uhadd8          r5,  r5,  r7
+        uhadd8          r8,  r8,  r12
+        uhadd8          r9,  r9,  r14
+        stm             r0,  {r4,r5}
+        add             r0,  r0,  r2
+        stm             r0,  {r8,r9}
+        add             r0,  r0,  r2
+        bne             1b
+
+        pop             {r4-r9, pc}
+endfunc
+
+function ff_put_pixels8_y2_no_rnd_armv6, export=1
+        push            {r4-r9, lr}
+        ldr             r4,  [r1]
+        ldr             r5,  [r1, #4]
+        ldr_pre         r6,  r1,  r2
+        ldr             r7,  [r1, #4]
+1:
+        subs            r3,  r3,  #2
+        uhadd8          r8,  r4,  r6
+        ldr_pre         r4,  r1,  r2
+        uhadd8          r9,  r5,  r7
+        ldr             r5,  [r1, #4]
+        uhadd8          r12, r4,  r6
+        ldr_pre         r6,  r1,  r2
+        uhadd8          r14, r5,  r7
+        ldr             r7,  [r1, #4]
+        stm             r0,  {r8,r9}
+        add             r0,  r0,  r2
+        stm             r0,  {r12,r14}
+        add             r0,  r0,  r2
+        bne             1b
+
+        pop             {r4-r9, pc}
+endfunc
+
+function ff_avg_pixels8_armv6, export=1
+        pld             [r1, r2]
+        push            {r4-r10, lr}
+        mov             lr,  #1
+        orr             lr,  lr,  lr,  lsl #8
+        orr             lr,  lr,  lr,  lsl #16
+        ldrd            r4,  r5,  [r0]
+        ldr             r10, [r1, #4]
+        ldr_post        r9,  r1,  r2
+        subs            r3,  r3,  #2
+1:
+        pld             [r1, r2]
+        eor             r8,  r4,  r9
+        uhadd8          r4,  r4,  r9
+        eor             r12, r5,  r10
+        ldrd_reg        r6,  r7,  r0,  r2
+        uhadd8          r5,  r5,  r10
+        and             r8,  r8,  lr
+        ldr             r10, [r1, #4]
+        and             r12, r12, lr
+        uadd8           r4,  r4,  r8
+        ldr_post        r9,  r1,  r2
+        eor             r8,  r6,  r9
+        uadd8           r5,  r5,  r12
+        pld             [r1, r2,  lsl #1]
+        eor             r12, r7,  r10
+        uhadd8          r6,  r6,  r9
+        strd_post       r4,  r5,  r0,  r2
+        uhadd8          r7,  r7,  r10
+        beq             2f
+        and             r8,  r8,  lr
+        ldrd_reg        r4,  r5,  r0,  r2
+        uadd8           r6,  r6,  r8
+        ldr             r10, [r1, #4]
+        and             r12, r12, lr
+        subs            r3,  r3,  #2
+        uadd8           r7,  r7,  r12
+        ldr_post        r9,  r1,  r2
+        strd_post       r6,  r7,  r0,  r2
+        b               1b
+2:
+        and             r8,  r8,  lr
+        and             r12, r12, lr
+        uadd8           r6,  r6,  r8
+        uadd8           r7,  r7,  r12
+        strd_post       r6,  r7,  r0,  r2
+
+        pop             {r4-r10, pc}
+endfunc
diff --git a/libavcodec/arm/hpeldsp_init_arm.c b/libavcodec/arm/hpeldsp_init_arm.c
new file mode 100644 (file)
index 0000000..8176afe
--- /dev/null
@@ -0,0 +1,71 @@
+/*
+ * ARM optimized DSP utils
+ * Copyright (c) 2001 Lionel Ulmer
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/arm/cpu.h"
+#include "libavutil/attributes.h"
+#include "libavcodec/rnd_avg.h"
+#include "hpeldsp_arm.h"
+
+void ff_put_pixels8_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+void ff_put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+void ff_put_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+void ff_put_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+
+void ff_put_no_rnd_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+void ff_put_no_rnd_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+void ff_put_no_rnd_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+
+void ff_put_pixels16_arm(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h);
+
+CALL_2X_PIXELS(ff_put_pixels16_x2_arm,         ff_put_pixels8_x2_arm,        8)
+CALL_2X_PIXELS(ff_put_pixels16_y2_arm,         ff_put_pixels8_y2_arm,        8)
+CALL_2X_PIXELS(ff_put_pixels16_xy2_arm,        ff_put_pixels8_xy2_arm,       8)
+CALL_2X_PIXELS(ff_put_no_rnd_pixels16_x2_arm,  ff_put_no_rnd_pixels8_x2_arm, 8)
+CALL_2X_PIXELS(ff_put_no_rnd_pixels16_y2_arm,  ff_put_no_rnd_pixels8_y2_arm, 8)
+CALL_2X_PIXELS(ff_put_no_rnd_pixels16_xy2_arm, ff_put_no_rnd_pixels8_xy2_arm,8)
+
+av_cold void ff_hpeldsp_init_arm(HpelDSPContext *c, int flags)
+{
+    int cpu_flags = av_get_cpu_flags();
+
+    c->put_pixels_tab[0][0] = ff_put_pixels16_arm;
+    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_arm;
+    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_arm;
+    c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_arm;
+    c->put_pixels_tab[1][0] = ff_put_pixels8_arm;
+    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_arm;
+    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_arm;
+    c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_arm;
+
+    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_arm;
+    c->put_no_rnd_pixels_tab[0][1] = ff_put_no_rnd_pixels16_x2_arm;
+    c->put_no_rnd_pixels_tab[0][2] = ff_put_no_rnd_pixels16_y2_arm;
+    c->put_no_rnd_pixels_tab[0][3] = ff_put_no_rnd_pixels16_xy2_arm;
+    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_arm;
+    c->put_no_rnd_pixels_tab[1][1] = ff_put_no_rnd_pixels8_x2_arm;
+    c->put_no_rnd_pixels_tab[1][2] = ff_put_no_rnd_pixels8_y2_arm;
+    c->put_no_rnd_pixels_tab[1][3] = ff_put_no_rnd_pixels8_xy2_arm;
+
+    if (have_armv6(cpu_flags))
+        ff_hpeldsp_init_armv6(c, flags);
+    if (have_neon(cpu_flags))
+        ff_hpeldsp_init_neon(c, flags);
+}
diff --git a/libavcodec/arm/hpeldsp_init_armv6.c b/libavcodec/arm/hpeldsp_init_armv6.c
new file mode 100644 (file)
index 0000000..67a500d
--- /dev/null
@@ -0,0 +1,67 @@
+/*
+ * Copyright (c) 2009 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stddef.h>
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "hpeldsp_arm.h"
+
+void ff_put_pixels16_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_x2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_y2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_put_pixels16_x2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_y2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_avg_pixels16_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_put_pixels8_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_x2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_y2_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_put_pixels8_x2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_y2_no_rnd_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_avg_pixels8_armv6(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+av_cold void ff_hpeldsp_init_armv6(HpelDSPContext *c, int flags)
+{
+    c->put_pixels_tab[0][0] = ff_put_pixels16_armv6;
+    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_armv6;
+    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_armv6;
+/*     c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_armv6; */
+    c->put_pixels_tab[1][0] = ff_put_pixels8_armv6;
+    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_armv6;
+    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_armv6;
+/*     c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_armv6; */
+
+    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_armv6;
+    c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_armv6;
+    c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_armv6;
+/*     c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_armv6; */
+    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_armv6;
+    c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_armv6;
+    c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_armv6;
+/*     c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_armv6; */
+
+    c->avg_pixels_tab[0][0] = ff_avg_pixels16_armv6;
+    c->avg_pixels_tab[1][0] = ff_avg_pixels8_armv6;
+}
diff --git a/libavcodec/arm/hpeldsp_init_neon.c b/libavcodec/arm/hpeldsp_init_neon.c
new file mode 100644 (file)
index 0000000..76d4eaf
--- /dev/null
@@ -0,0 +1,88 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stddef.h>
+#include <stdint.h>
+
+#include "libavutil/attributes.h"
+#include "hpeldsp_arm.h"
+
+void ff_put_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+void ff_avg_pixels16_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels8_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels8_x2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels8_y2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels8_xy2_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+void ff_avg_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, ptrdiff_t, int);
+
+av_cold void ff_hpeldsp_init_neon(HpelDSPContext *c, int flags)
+{
+    c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
+    c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
+    c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
+    c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
+    c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
+    c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
+    c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
+    c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
+
+    c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
+    c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
+    c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
+    c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
+
+    c->avg_pixels_tab[0][0] = ff_avg_pixels16_neon;
+    c->avg_pixels_tab[0][1] = ff_avg_pixels16_x2_neon;
+    c->avg_pixels_tab[0][2] = ff_avg_pixels16_y2_neon;
+    c->avg_pixels_tab[0][3] = ff_avg_pixels16_xy2_neon;
+    c->avg_pixels_tab[1][0] = ff_avg_pixels8_neon;
+    c->avg_pixels_tab[1][1] = ff_avg_pixels8_x2_neon;
+    c->avg_pixels_tab[1][2] = ff_avg_pixels8_y2_neon;
+    c->avg_pixels_tab[1][3] = ff_avg_pixels8_xy2_neon;
+
+    c->avg_no_rnd_pixels_tab[0] = ff_avg_pixels16_neon;
+    c->avg_no_rnd_pixels_tab[1] = ff_avg_pixels16_x2_no_rnd_neon;
+    c->avg_no_rnd_pixels_tab[2] = ff_avg_pixels16_y2_no_rnd_neon;
+    c->avg_no_rnd_pixels_tab[3] = ff_avg_pixels16_xy2_no_rnd_neon;
+}
diff --git a/libavcodec/arm/hpeldsp_neon.S b/libavcodec/arm/hpeldsp_neon.S
new file mode 100644 (file)
index 0000000..90bc3cb
--- /dev/null
@@ -0,0 +1,410 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/arm/asm.S"
+
+.macro  pixels16        rnd=1, avg=0
+  .if \avg
+        mov             r12, r0
+  .endif
+1:      vld1.8          {q0},     [r1], r2
+        vld1.8          {q1},     [r1], r2
+        vld1.8          {q2},     [r1], r2
+        pld             [r1, r2, lsl #2]
+        vld1.8          {q3},     [r1], r2
+        pld             [r1]
+        pld             [r1, r2]
+        pld             [r1, r2, lsl #1]
+  .if \avg
+        vld1.8          {q8},     [r12,:128], r2
+        vrhadd.u8       q0,  q0,  q8
+        vld1.8          {q9},     [r12,:128], r2
+        vrhadd.u8       q1,  q1,  q9
+        vld1.8          {q10},    [r12,:128], r2
+        vrhadd.u8       q2,  q2,  q10
+        vld1.8          {q11},    [r12,:128], r2
+        vrhadd.u8       q3,  q3,  q11
+  .endif
+        subs            r3,  r3,  #4
+        vst1.64         {q0},     [r0,:128], r2
+        vst1.64         {q1},     [r0,:128], r2
+        vst1.64         {q2},     [r0,:128], r2
+        vst1.64         {q3},     [r0,:128], r2
+        bne             1b
+        bx              lr
+.endm
+
+.macro  pixels16_x2     rnd=1, avg=0
+1:      vld1.8          {d0-d2},  [r1], r2
+        vld1.8          {d4-d6},  [r1], r2
+        pld             [r1]
+        pld             [r1, r2]
+        subs            r3,  r3,  #2
+        vext.8          q1,  q0,  q1,  #1
+        avg             q0,  q0,  q1
+        vext.8          q3,  q2,  q3,  #1
+        avg             q2,  q2,  q3
+  .if \avg
+        vld1.8          {q1},     [r0,:128], r2
+        vld1.8          {q3},     [r0,:128]
+        vrhadd.u8       q0,  q0,  q1
+        vrhadd.u8       q2,  q2,  q3
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {q0},     [r0,:128], r2
+        vst1.8          {q2},     [r0,:128], r2
+        bne             1b
+        bx              lr
+.endm
+
+.macro  pixels16_y2     rnd=1, avg=0
+        sub             r3,  r3,  #2
+        vld1.8          {q0},     [r1], r2
+        vld1.8          {q1},     [r1], r2
+1:      subs            r3,  r3,  #2
+        avg             q2,  q0,  q1
+        vld1.8          {q0},     [r1], r2
+        avg             q3,  q0,  q1
+        vld1.8          {q1},     [r1], r2
+        pld             [r1]
+        pld             [r1, r2]
+  .if \avg
+        vld1.8          {q8},     [r0,:128], r2
+        vld1.8          {q9},     [r0,:128]
+        vrhadd.u8       q2,  q2,  q8
+        vrhadd.u8       q3,  q3,  q9
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {q2},     [r0,:128], r2
+        vst1.8          {q3},     [r0,:128], r2
+        bne             1b
+
+        avg             q2,  q0,  q1
+        vld1.8          {q0},     [r1], r2
+        avg             q3,  q0,  q1
+  .if \avg
+        vld1.8          {q8},     [r0,:128], r2
+        vld1.8          {q9},     [r0,:128]
+        vrhadd.u8       q2,  q2,  q8
+        vrhadd.u8       q3,  q3,  q9
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {q2},     [r0,:128], r2
+        vst1.8          {q3},     [r0,:128], r2
+
+        bx              lr
+.endm
+
+.macro  pixels16_xy2    rnd=1, avg=0
+        sub             r3,  r3,  #2
+        vld1.8          {d0-d2},  [r1], r2
+        vld1.8          {d4-d6},  [r1], r2
+NRND    vmov.i16        q13, #1
+        pld             [r1]
+        pld             [r1, r2]
+        vext.8          q1,  q0,  q1,  #1
+        vext.8          q3,  q2,  q3,  #1
+        vaddl.u8        q8,  d0,  d2
+        vaddl.u8        q10, d1,  d3
+        vaddl.u8        q9,  d4,  d6
+        vaddl.u8        q11, d5,  d7
+1:      subs            r3,  r3,  #2
+        vld1.8          {d0-d2},  [r1], r2
+        vadd.u16        q12, q8,  q9
+        pld             [r1]
+NRND    vadd.u16        q12, q12, q13
+        vext.8          q15, q0,  q1,  #1
+        vadd.u16        q1 , q10, q11
+        shrn            d28, q12, #2
+NRND    vadd.u16        q1,  q1,  q13
+        shrn            d29, q1,  #2
+  .if \avg
+        vld1.8          {q8},     [r0,:128]
+        vrhadd.u8       q14, q14, q8
+  .endif
+        vaddl.u8        q8,  d0,  d30
+        vld1.8          {d2-d4},  [r1], r2
+        vaddl.u8        q10, d1,  d31
+        vst1.8          {q14},    [r0,:128], r2
+        vadd.u16        q12, q8,  q9
+        pld             [r1, r2]
+NRND    vadd.u16        q12, q12, q13
+        vext.8          q2,  q1,  q2,  #1
+        vadd.u16        q0,  q10, q11
+        shrn            d30, q12, #2
+NRND    vadd.u16        q0,  q0,  q13
+        shrn            d31, q0,  #2
+  .if \avg
+        vld1.8          {q9},     [r0,:128]
+        vrhadd.u8       q15, q15, q9
+  .endif
+        vaddl.u8        q9,  d2,  d4
+        vaddl.u8        q11, d3,  d5
+        vst1.8          {q15},    [r0,:128], r2
+        bgt             1b
+
+        vld1.8          {d0-d2},  [r1], r2
+        vadd.u16        q12, q8,  q9
+NRND    vadd.u16        q12, q12, q13
+        vext.8          q15, q0,  q1,  #1
+        vadd.u16        q1 , q10, q11
+        shrn            d28, q12, #2
+NRND    vadd.u16        q1,  q1,  q13
+        shrn            d29, q1,  #2
+  .if \avg
+        vld1.8          {q8},     [r0,:128]
+        vrhadd.u8       q14, q14, q8
+  .endif
+        vaddl.u8        q8,  d0,  d30
+        vaddl.u8        q10, d1,  d31
+        vst1.8          {q14},    [r0,:128], r2
+        vadd.u16        q12, q8,  q9
+NRND    vadd.u16        q12, q12, q13
+        vadd.u16        q0,  q10, q11
+        shrn            d30, q12, #2
+NRND    vadd.u16        q0,  q0,  q13
+        shrn            d31, q0,  #2
+  .if \avg
+        vld1.8          {q9},     [r0,:128]
+        vrhadd.u8       q15, q15, q9
+  .endif
+        vst1.8          {q15},    [r0,:128], r2
+
+        bx              lr
+.endm
+
+.macro  pixels8         rnd=1, avg=0
+1:      vld1.8          {d0},     [r1], r2
+        vld1.8          {d1},     [r1], r2
+        vld1.8          {d2},     [r1], r2
+        pld             [r1, r2, lsl #2]
+        vld1.8          {d3},     [r1], r2
+        pld             [r1]
+        pld             [r1, r2]
+        pld             [r1, r2, lsl #1]
+  .if \avg
+        vld1.8          {d4},     [r0,:64], r2
+        vrhadd.u8       d0,  d0,  d4
+        vld1.8          {d5},     [r0,:64], r2
+        vrhadd.u8       d1,  d1,  d5
+        vld1.8          {d6},     [r0,:64], r2
+        vrhadd.u8       d2,  d2,  d6
+        vld1.8          {d7},     [r0,:64], r2
+        vrhadd.u8       d3,  d3,  d7
+        sub             r0,  r0,  r2,  lsl #2
+  .endif
+        subs            r3,  r3,  #4
+        vst1.8          {d0},     [r0,:64], r2
+        vst1.8          {d1},     [r0,:64], r2
+        vst1.8          {d2},     [r0,:64], r2
+        vst1.8          {d3},     [r0,:64], r2
+        bne             1b
+        bx              lr
+.endm
+
+.macro  pixels8_x2      rnd=1, avg=0
+1:      vld1.8          {q0},     [r1], r2
+        vext.8          d1,  d0,  d1,  #1
+        vld1.8          {q1},     [r1], r2
+        vext.8          d3,  d2,  d3,  #1
+        pld             [r1]
+        pld             [r1, r2]
+        subs            r3,  r3,  #2
+        vswp            d1,  d2
+        avg             q0,  q0,  q1
+  .if \avg
+        vld1.8          {d4},     [r0,:64], r2
+        vld1.8          {d5},     [r0,:64]
+        vrhadd.u8       q0,  q0,  q2
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {d0},     [r0,:64], r2
+        vst1.8          {d1},     [r0,:64], r2
+        bne             1b
+        bx              lr
+.endm
+
+.macro  pixels8_y2      rnd=1, avg=0
+        sub             r3,  r3,  #2
+        vld1.8          {d0},     [r1], r2
+        vld1.8          {d1},     [r1], r2
+1:      subs            r3,  r3,  #2
+        avg             d4,  d0,  d1
+        vld1.8          {d0},     [r1], r2
+        avg             d5,  d0,  d1
+        vld1.8          {d1},     [r1], r2
+        pld             [r1]
+        pld             [r1, r2]
+  .if \avg
+        vld1.8          {d2},     [r0,:64], r2
+        vld1.8          {d3},     [r0,:64]
+        vrhadd.u8       q2,  q2,  q1
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {d4},     [r0,:64], r2
+        vst1.8          {d5},     [r0,:64], r2
+        bne             1b
+
+        avg             d4,  d0,  d1
+        vld1.8          {d0},     [r1], r2
+        avg             d5,  d0,  d1
+  .if \avg
+        vld1.8          {d2},     [r0,:64], r2
+        vld1.8          {d3},     [r0,:64]
+        vrhadd.u8       q2,  q2,  q1
+        sub             r0,  r0,  r2
+  .endif
+        vst1.8          {d4},     [r0,:64], r2
+        vst1.8          {d5},     [r0,:64], r2
+
+        bx              lr
+.endm
+
+.macro  pixels8_xy2     rnd=1, avg=0
+        sub             r3,  r3,  #2
+        vld1.8          {q0},     [r1], r2
+        vld1.8          {q1},     [r1], r2
+NRND    vmov.i16        q11, #1
+        pld             [r1]
+        pld             [r1, r2]
+        vext.8          d4,  d0,  d1,  #1
+        vext.8          d6,  d2,  d3,  #1
+        vaddl.u8        q8,  d0,  d4
+        vaddl.u8        q9,  d2,  d6
+1:      subs            r3,  r3,  #2
+        vld1.8          {q0},     [r1], r2
+        pld             [r1]
+        vadd.u16        q10, q8,  q9
+        vext.8          d4,  d0,  d1,  #1
+NRND    vadd.u16        q10, q10, q11
+        vaddl.u8        q8,  d0,  d4
+        shrn            d5,  q10, #2
+        vld1.8          {q1},     [r1], r2
+        vadd.u16        q10, q8,  q9
+        pld             [r1, r2]
+  .if \avg
+        vld1.8          {d7},     [r0,:64]
+        vrhadd.u8       d5,  d5,  d7
+  .endif
+NRND    vadd.u16        q10, q10, q11
+        vst1.8          {d5},     [r0,:64], r2
+        shrn            d7,  q10, #2
+  .if \avg
+        vld1.8          {d5},     [r0,:64]
+        vrhadd.u8       d7,  d7,  d5
+  .endif
+        vext.8          d6,  d2,  d3,  #1
+        vaddl.u8        q9,  d2,  d6
+        vst1.8          {d7},     [r0,:64], r2
+        bgt             1b
+
+        vld1.8          {q0},     [r1], r2
+        vadd.u16        q10, q8,  q9
+        vext.8          d4,  d0,  d1,  #1
+NRND    vadd.u16        q10, q10, q11
+        vaddl.u8        q8,  d0,  d4
+        shrn            d5,  q10, #2
+        vadd.u16        q10, q8,  q9
+  .if \avg
+        vld1.8          {d7},     [r0,:64]
+        vrhadd.u8       d5,  d5,  d7
+  .endif
+NRND    vadd.u16        q10, q10, q11
+        vst1.8          {d5},     [r0,:64], r2
+        shrn            d7,  q10, #2
+  .if \avg
+        vld1.8          {d5},     [r0,:64]
+        vrhadd.u8       d7,  d7,  d5
+  .endif
+        vst1.8          {d7},     [r0,:64], r2
+
+        bx              lr
+.endm
+
+.macro  pixfunc         pfx, name, suf, rnd=1, avg=0
+  .if \rnd
+    .macro avg  rd, rn, rm
+        vrhadd.u8       \rd, \rn, \rm
+    .endm
+    .macro shrn rd, rn, rm
+        vrshrn.u16      \rd, \rn, \rm
+    .endm
+    .macro NRND insn:vararg
+    .endm
+  .else
+    .macro avg  rd, rn, rm
+        vhadd.u8        \rd, \rn, \rm
+    .endm
+    .macro shrn rd, rn, rm
+        vshrn.u16       \rd, \rn, \rm
+    .endm
+    .macro NRND insn:vararg
+        \insn
+    .endm
+  .endif
+function ff_\pfx\name\suf\()_neon, export=1
+        \name           \rnd, \avg
+endfunc
+        .purgem         avg
+        .purgem         shrn
+        .purgem         NRND
+.endm
+
+.macro  pixfunc2        pfx, name, avg=0
+        pixfunc         \pfx, \name,          rnd=1, avg=\avg
+        pixfunc         \pfx, \name, _no_rnd, rnd=0, avg=\avg
+.endm
+
+function ff_put_h264_qpel16_mc00_neon, export=1
+        mov             r3,  #16
+endfunc
+
+        pixfunc         put_, pixels16,     avg=0
+        pixfunc2        put_, pixels16_x2,  avg=0
+        pixfunc2        put_, pixels16_y2,  avg=0
+        pixfunc2        put_, pixels16_xy2, avg=0
+
+function ff_avg_h264_qpel16_mc00_neon, export=1
+        mov             r3,  #16
+endfunc
+
+        pixfunc         avg_, pixels16,     avg=1
+        pixfunc2        avg_, pixels16_x2,  avg=1
+        pixfunc2        avg_, pixels16_y2,  avg=1
+        pixfunc2        avg_, pixels16_xy2, avg=1
+
+function ff_put_h264_qpel8_mc00_neon, export=1
+        mov             r3,  #8
+endfunc
+
+        pixfunc         put_, pixels8,     avg=0
+        pixfunc2        put_, pixels8_x2,  avg=0
+        pixfunc2        put_, pixels8_y2,  avg=0
+        pixfunc2        put_, pixels8_xy2, avg=0
+
+function ff_avg_h264_qpel8_mc00_neon, export=1
+        mov             r3,  #8
+endfunc
+
+        pixfunc         avg_, pixels8,     avg=1
+        pixfunc         avg_, pixels8_x2,  avg=1
+        pixfunc         avg_, pixels8_y2,  avg=1
+        pixfunc         avg_, pixels8_xy2, avg=1
index 09e4576..9756645 100644 (file)
@@ -54,6 +54,8 @@ av_cold void ff_hpeldsp_init(HpelDSPContext *c, int flags)
     hpel_funcs(avg, [3],  2);
     hpel_funcs(avg_no_rnd,, 16);
 
+    if (ARCH_ARM)
+        ff_hpeldsp_init_arm(c, flags);
     if (ARCH_PPC)
         ff_hpeldsp_init_ppc(c, flags);
     if (ARCH_X86)
index eeb7a1c..837a11b 100644 (file)
@@ -94,6 +94,7 @@ typedef struct HpelDSPContext {
 
 void ff_hpeldsp_init(HpelDSPContext *c, int flags);
 
+void ff_hpeldsp_init_arm(HpelDSPContext *c, int flags);
 void ff_hpeldsp_init_ppc(HpelDSPContext *c, int flags);
 void ff_hpeldsp_init_x86(HpelDSPContext *c, int flags);