arm: hpeldsp: Move half-pel assembly from dsputil to hpeldsp

Signed-off-by: Martin Storsjö <martin@martin.st>

1 files changed, 259 insertions, 0 deletions

diff --git a/libavcodec/arm/hpeldsp_armv6.S b/libavcodec/arm/hpeldsp_armv6.S
new file mode 100644
index 0000000000..a030d423bf
--- /dev/null
+++ b/libavcodec/arm/hpeldsp_armv6.S
@@ -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