Merge remote-tracking branch 'qatar/master'

* qatar/master:
  pcm-mpeg: convert to bytestream2 API
  Revert "h264: clear trailing bits in partially parsed NAL units"
  remove iwmmxt optimizations
  mimic: do not continue if swap_buf_size is 0
  mimic: convert to bytestream2 API
  frwu: use MKTAG to check marker instead of AV_RL32
  txd: port to bytestream2 API
  c93: convert to bytestream2 API
  iff: make .long_name more descriptive
  FATE: add test for cdxl demuxer
  rtsp: Fix a typo

Conflicts:
	libavcodec/arm/dsputil_iwmmxt.c
	libavcodec/arm/dsputil_iwmmxt_rnd_template.c
	libavcodec/arm/mpegvideo_iwmmxt.c
	libavcodec/c93.c
	libavcodec/txd.c
	libavutil/arm/cpu.c
	tests/fate/demux.mak

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

15 files changed, 139 insertions, 1584 deletions

diff --git a/libavcodec/arm/Makefile b/libavcodec/arm/Makefile
index 0e45bc0a52..39852c4f4a 100644
--- a/libavcodec/arm/Makefile
+++ b/libavcodec/arm/Makefile
@@ -44,9 +44,6 @@ OBJS-$(HAVE_ARMVFP)                    += arm/dsputil_vfp.o             \
                                           arm/dsputil_init_vfp.o        \
                                           $(VFP-OBJS-yes)
 
-OBJS-$(HAVE_IWMMXT)                    += arm/dsputil_iwmmxt.o          \
-                                          arm/mpegvideo_iwmmxt.o        \
-
 NEON-OBJS-$(CONFIG_FFT)                += arm/fft_neon.o                \
                                           arm/fft_fixed_neon.o          \
 
diff --git a/libavcodec/arm/dsputil_arm.h b/libavcodec/arm/dsputil_arm.h
index b333c70226..b7b5bdc0e6 100644
--- a/libavcodec/arm/dsputil_arm.h
+++ b/libavcodec/arm/dsputil_arm.h
@@ -28,6 +28,5 @@ void ff_dsputil_init_armv5te(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_armv6(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_vfp(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx);
-void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
 
 #endif /* AVCODEC_ARM_DSPUTIL_H */
diff --git a/libavcodec/arm/dsputil_init_arm.c b/libavcodec/arm/dsputil_init_arm.c
index 58577dde9a..82af718b66 100644
--- a/libavcodec/arm/dsputil_init_arm.c
+++ b/libavcodec/arm/dsputil_init_arm.c
@@ -119,7 +119,6 @@ void ff_dsputil_init_arm(DSPContext* c, AVCodecContext *avctx)
 
     if (HAVE_ARMV5TE) ff_dsputil_init_armv5te(c, avctx);
     if (HAVE_ARMV6)   ff_dsputil_init_armv6(c, avctx);
-    if (HAVE_IWMMXT)  ff_dsputil_init_iwmmxt(c, avctx);
     if (HAVE_ARMVFP)  ff_dsputil_init_vfp(c, avctx);
     if (HAVE_NEON)    ff_dsputil_init_neon(c, avctx);
 }
diff --git a/libavcodec/arm/dsputil_iwmmxt.c b/libavcodec/arm/dsputil_iwmmxt.c
deleted file mode 100644
index 2837af119f..0000000000
--- a/libavcodec/arm/dsputil_iwmmxt.c
+++ /dev/null
@@ -1,210 +0,0 @@
-/*
- * iWMMXt optimized DSP utils
- * Copyright (c) 2004 AGAWA Koji
- *
- * This file is part of FFmpeg.
- *
- * FFmpeg 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.
- *
- * FFmpeg 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 FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "libavutil/cpu.h"
-#include "libavcodec/dsputil.h"
-
-#define DEF(x, y) x ## _no_rnd_ ## y ##_iwmmxt
-#define SET_RND(regd)  __asm__ volatile ("mov r12, #1 \n\t tbcsth " #regd ", r12":::"r12");
-#define WAVG2B "wavg2b"
-#include "dsputil_iwmmxt_rnd_template.c"
-#undef DEF
-#undef SET_RND
-#undef WAVG2B
-
-#define DEF(x, y) x ## _ ## y ##_iwmmxt
-#define SET_RND(regd)  __asm__ volatile ("mov r12, #2 \n\t tbcsth " #regd ", r12":::"r12");
-#define WAVG2B "wavg2br"
-#include "dsputil_iwmmxt_rnd_template.c"
-#undef DEF
-#undef SET_RND
-#undef WAVG2BR
-
-// need scheduling
-#define OP(AVG)                                         \
-    __asm__ volatile (                                      \
-        /* alignment */                                 \
-        "and r12, %[pixels], #7 \n\t"                   \
-        "bic %[pixels], %[pixels], #7 \n\t"             \
-        "tmcr wcgr1, r12 \n\t"                          \
-                                                        \
-        "wldrd wr0, [%[pixels]] \n\t"                   \
-        "wldrd wr1, [%[pixels], #8] \n\t"               \
-        "add %[pixels], %[pixels], %[line_size] \n\t"   \
-        "walignr1 wr4, wr0, wr1 \n\t"                   \
-                                                        \
-        "1: \n\t"                                       \
-                                                        \
-        "wldrd wr2, [%[pixels]] \n\t"                   \
-        "wldrd wr3, [%[pixels], #8] \n\t"               \
-        "add %[pixels], %[pixels], %[line_size] \n\t"   \
-        "pld [%[pixels]] \n\t"                          \
-        "walignr1 wr5, wr2, wr3 \n\t"                   \
-        AVG " wr6, wr4, wr5 \n\t"                       \
-        "wstrd wr6, [%[block]] \n\t"                    \
-        "add %[block], %[block], %[line_size] \n\t"     \
-                                                        \
-        "wldrd wr0, [%[pixels]] \n\t"                   \
-        "wldrd wr1, [%[pixels], #8] \n\t"               \
-        "add %[pixels], %[pixels], %[line_size] \n\t"   \
-        "walignr1 wr4, wr0, wr1 \n\t"                   \
-        "pld [%[pixels]] \n\t"                          \
-        AVG " wr6, wr4, wr5 \n\t"                       \
-        "wstrd wr6, [%[block]] \n\t"                    \
-        "add %[block], %[block], %[line_size] \n\t"     \
-                                                        \
-        "subs %[h], %[h], #2 \n\t"                      \
-        "bne 1b \n\t"                                   \
-        : [block]"+r"(block), [pixels]"+r"(pixels), [h]"+r"(h)  \
-        : [line_size]"r"(line_size) \
-        : "memory", "r12");
-void put_pixels8_y2_iwmmxt(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    OP("wavg2br");
-}
-void put_no_rnd_pixels8_y2_iwmmxt(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    OP("wavg2b");
-}
-#undef OP
-
-void add_pixels_clamped_iwmmxt(const DCTELEM *block, uint8_t *pixels, int line_size)
-{
-    uint8_t *pixels2 = pixels + line_size;
-
-    __asm__ volatile (
-        "mov            r12, #4                 \n\t"
-        "1:                                     \n\t"
-        "pld            [%[pixels], %[line_size2]]              \n\t"
-        "pld            [%[pixels2], %[line_size2]]             \n\t"
-        "wldrd          wr4, [%[pixels]]        \n\t"
-        "wldrd          wr5, [%[pixels2]]       \n\t"
-        "pld            [%[block], #32]         \n\t"
-        "wunpckelub     wr6, wr4                \n\t"
-        "wldrd          wr0, [%[block]]         \n\t"
-        "wunpckehub     wr7, wr4                \n\t"
-        "wldrd          wr1, [%[block], #8]     \n\t"
-        "wunpckelub     wr8, wr5                \n\t"
-        "wldrd          wr2, [%[block], #16]    \n\t"
-        "wunpckehub     wr9, wr5                \n\t"
-        "wldrd          wr3, [%[block], #24]    \n\t"
-        "add            %[block], %[block], #32 \n\t"
-        "waddhss        wr10, wr0, wr6          \n\t"
-        "waddhss        wr11, wr1, wr7          \n\t"
-        "waddhss        wr12, wr2, wr8          \n\t"
-        "waddhss        wr13, wr3, wr9          \n\t"
-        "wpackhus       wr14, wr10, wr11        \n\t"
-        "wpackhus       wr15, wr12, wr13        \n\t"
-        "wstrd          wr14, [%[pixels]]       \n\t"
-        "add            %[pixels], %[pixels], %[line_size2]     \n\t"
-        "subs           r12, r12, #1            \n\t"
-        "wstrd          wr15, [%[pixels2]]      \n\t"
-        "add            %[pixels2], %[pixels2], %[line_size2]   \n\t"
-        "bne            1b                      \n\t"
-        : [block]"+r"(block), [pixels]"+r"(pixels), [pixels2]"+r"(pixels2)
-        : [line_size2]"r"(line_size << 1)
-        : "cc", "memory", "r12");
-}
-
-static void clear_blocks_iwmmxt(DCTELEM *blocks)
-{
-    __asm__ volatile(
-                "wzero wr0                      \n\t"
-                "mov r1, #(128 * 6 / 32)        \n\t"
-                "1:                             \n\t"
-                "wstrd wr0, [%0]                \n\t"
-                "wstrd wr0, [%0, #8]            \n\t"
-                "wstrd wr0, [%0, #16]           \n\t"
-                "wstrd wr0, [%0, #24]           \n\t"
-                "subs r1, r1, #1                \n\t"
-                "add %0, %0, #32                \n\t"
-                "bne 1b                         \n\t"
-                : "+r"(blocks)
-                :
-                : "r1"
-        );
-}
-
-static void nop(uint8_t *block, const uint8_t *pixels, int line_size, int h)
-{
-    return;
-}
-
-/* A run time test is not simple. If this file is compiled in
- * then we should install the functions
- */
-
-void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx)
-{
-    int mm_flags = AV_CPU_FLAG_IWMMXT; /* multimedia extension flags */
-    const int high_bit_depth = avctx->bits_per_raw_sample > 8;
-
-    if (avctx->dsp_mask) {
-        if (avctx->dsp_mask & AV_CPU_FLAG_FORCE)
-            mm_flags |= (avctx->dsp_mask & 0xffff);
-        else
-            mm_flags &= ~(avctx->dsp_mask & 0xffff);
-    }
-
-    if (!(mm_flags & AV_CPU_FLAG_IWMMXT)) return;
-
-    c->add_pixels_clamped = add_pixels_clamped_iwmmxt;
-
-    if (!high_bit_depth) {
-    c->clear_blocks = clear_blocks_iwmmxt;
-
-    c->put_pixels_tab[0][0] = put_pixels16_iwmmxt;
-    c->put_pixels_tab[0][1] = put_pixels16_x2_iwmmxt;
-    c->put_pixels_tab[0][2] = put_pixels16_y2_iwmmxt;
-    c->put_pixels_tab[0][3] = put_pixels16_xy2_iwmmxt;
-    c->put_no_rnd_pixels_tab[0][0] = put_pixels16_iwmmxt;
-    c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_iwmmxt;
-    c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_iwmmxt;
-    c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_iwmmxt;
-
-    c->put_pixels_tab[1][0] = put_pixels8_iwmmxt;
-    c->put_pixels_tab[1][1] = put_pixels8_x2_iwmmxt;
-    c->put_pixels_tab[1][2] = put_pixels8_y2_iwmmxt;
-    c->put_pixels_tab[1][3] = put_pixels8_xy2_iwmmxt;
-    c->put_no_rnd_pixels_tab[1][0] = put_pixels8_iwmmxt;
-    c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_iwmmxt;
-    c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_iwmmxt;
-    c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_iwmmxt;
-
-    c->avg_pixels_tab[0][0] = avg_pixels16_iwmmxt;
-    c->avg_pixels_tab[0][1] = avg_pixels16_x2_iwmmxt;
-    c->avg_pixels_tab[0][2] = avg_pixels16_y2_iwmmxt;
-    c->avg_pixels_tab[0][3] = avg_pixels16_xy2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[0][0] = avg_pixels16_iwmmxt;
-    c->avg_no_rnd_pixels_tab[0][1] = avg_no_rnd_pixels16_x2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[0][2] = avg_no_rnd_pixels16_y2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[0][3] = avg_no_rnd_pixels16_xy2_iwmmxt;
-
-    c->avg_pixels_tab[1][0] = avg_pixels8_iwmmxt;
-    c->avg_pixels_tab[1][1] = avg_pixels8_x2_iwmmxt;
-    c->avg_pixels_tab[1][2] = avg_pixels8_y2_iwmmxt;
-    c->avg_pixels_tab[1][3] = avg_pixels8_xy2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[1][0] = avg_no_rnd_pixels8_iwmmxt;
-    c->avg_no_rnd_pixels_tab[1][1] = avg_no_rnd_pixels8_x2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[1][2] = avg_no_rnd_pixels8_y2_iwmmxt;
-    c->avg_no_rnd_pixels_tab[1][3] = avg_no_rnd_pixels8_xy2_iwmmxt;
-    }
-}
diff --git a/libavcodec/arm/dsputil_iwmmxt_rnd_template.c b/libavcodec/arm/dsputil_iwmmxt_rnd_template.c
deleted file mode 100644
index 35a5a9b8b4..0000000000
--- a/libavcodec/arm/dsputil_iwmmxt_rnd_template.c
+++ /dev/null
@@ -1,1114 +0,0 @@
-/*
- * iWMMXt optimized DSP utils
- * copyright (c) 2004 AGAWA Koji
- *
- * This file is part of FFmpeg.
- *
- * FFmpeg 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.
- *
- * FFmpeg 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 FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-void DEF(put, pixels8)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    __asm__ volatile (
-        "and r12, %[pixels], #7 \n\t"
-        "bic %[pixels], %[pixels], #7 \n\t"
-        "tmcr wcgr1, r12 \n\t"
-        "add r4, %[pixels], %[line_size] \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "1: \n\t"
-        "wldrd wr0, [%[pixels]] \n\t"
-        "subs %[h], %[h], #2 \n\t"
-        "wldrd wr1, [%[pixels], #8] \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr3, [r4] \n\t"
-        "pld [%[pixels]] \n\t"
-        "pld [%[pixels], #32] \n\t"
-        "wldrd wr4, [r4, #8] \n\t"
-        "add r4, r4, %[line_size] \n\t"
-        "walignr1 wr8, wr0, wr1 \n\t"
-        "pld [r4] \n\t"
-        "pld [r4, #32] \n\t"
-        "walignr1 wr10, wr3, wr4 \n\t"
-        "wstrd wr8, [%[block]] \n\t"
-        "add %[block], %[block], %[line_size] \n\t"
-        "wstrd wr10, [r5] \n\t"
-        "add r5, r5, %[line_size] \n\t"
-        "bne 1b \n\t"
-        : [block]"+r"(block), [pixels]"+r"(pixels), [line_size]"+r"(stride), [h]"+r"(h)
-        :
-        : "memory", "r4", "r5", "r12");
-}
-
-void DEF(avg, pixels8)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    __asm__ volatile (
-        "and r12, %[pixels], #7 \n\t"
-        "bic %[pixels], %[pixels], #7 \n\t"
-        "tmcr wcgr1, r12 \n\t"
-        "add r4, %[pixels], %[line_size] \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "1: \n\t"
-        "wldrd wr0, [%[pixels]] \n\t"
-        "subs %[h], %[h], #2 \n\t"
-        "wldrd wr1, [%[pixels], #8] \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr3, [r4] \n\t"
-        "pld [%[pixels]] \n\t"
-        "pld [%[pixels], #32] \n\t"
-        "wldrd wr4, [r4, #8] \n\t"
-        "add r4, r4, %[line_size] \n\t"
-        "walignr1 wr8, wr0, wr1 \n\t"
-        "wldrd wr0, [%[block]] \n\t"
-        "wldrd wr2, [r5] \n\t"
-        "pld [r4] \n\t"
-        "pld [r4, #32] \n\t"
-        "walignr1 wr10, wr3, wr4 \n\t"
-        WAVG2B" wr8, wr8, wr0 \n\t"
-        WAVG2B" wr10, wr10, wr2 \n\t"
-        "wstrd wr8, [%[block]] \n\t"
-        "add %[block], %[block], %[line_size] \n\t"
-        "wstrd wr10, [r5] \n\t"
-        "pld [%[block]] \n\t"
-        "pld [%[block], #32] \n\t"
-        "add r5, r5, %[line_size] \n\t"
-        "pld [r5] \n\t"
-        "pld [r5, #32] \n\t"
-        "bne 1b \n\t"
-        : [block]"+r"(block), [pixels]"+r"(pixels), [line_size]"+r"(stride), [h]"+r"(h)
-        :
-        : "memory", "r4", "r5", "r12");
-}
-
-void DEF(put, pixels16)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    __asm__ volatile (
-        "and r12, %[pixels], #7 \n\t"
-        "bic %[pixels], %[pixels], #7 \n\t"
-        "tmcr wcgr1, r12 \n\t"
-        "add r4, %[pixels], %[line_size] \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "1: \n\t"
-        "wldrd wr0, [%[pixels]] \n\t"
-        "wldrd wr1, [%[pixels], #8] \n\t"
-        "subs %[h], %[h], #2 \n\t"
-        "wldrd wr2, [%[pixels], #16] \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr3, [r4] \n\t"
-        "pld [%[pixels]] \n\t"
-        "pld [%[pixels], #32] \n\t"
-        "walignr1 wr8, wr0, wr1 \n\t"
-        "wldrd wr4, [r4, #8] \n\t"
-        "walignr1 wr9, wr1, wr2 \n\t"
-        "wldrd wr5, [r4, #16] \n\t"
-        "add r4, r4, %[line_size] \n\t"
-        "pld [r4] \n\t"
-        "pld [r4, #32] \n\t"
-        "walignr1 wr10, wr3, wr4 \n\t"
-        "wstrd wr8, [%[block]] \n\t"
-        "walignr1 wr11, wr4, wr5 \n\t"
-        "wstrd wr9, [%[block], #8] \n\t"
-        "add %[block], %[block], %[line_size] \n\t"
-        "wstrd wr10, [r5] \n\t"
-        "wstrd wr11, [r5, #8] \n\t"
-        "add r5, r5, %[line_size] \n\t"
-        "bne 1b \n\t"
-        : [block]"+r"(block), [pixels]"+r"(pixels), [line_size]"+r"(stride), [h]"+r"(h)
-        :
-        : "memory", "r4", "r5", "r12");
-}
-
-void DEF(avg, pixels16)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    __asm__ volatile (
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r4, %[pixels], %[line_size]\n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "1:                             \n\t"
-        "wldrd wr0, [%[pixels]]         \n\t"
-        "wldrd wr1, [%[pixels], #8]     \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wldrd wr2, [%[pixels], #16]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr3, [r4]                \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr8, wr0, wr1         \n\t"
-        "wldrd wr4, [r4, #8]            \n\t"
-        "walignr1 wr9, wr1, wr2         \n\t"
-        "wldrd wr5, [r4, #16]           \n\t"
-        "add r4, r4, %[line_size]       \n\t"
-        "wldrd wr0, [%[block]]          \n\t"
-        "pld [r4]                       \n\t"
-        "wldrd wr1, [%[block], #8]      \n\t"
-        "pld [r4, #32]                  \n\t"
-        "wldrd wr2, [r5]                \n\t"
-        "walignr1 wr10, wr3, wr4        \n\t"
-        "wldrd wr3, [r5, #8]            \n\t"
-        WAVG2B" wr8, wr8, wr0           \n\t"
-        WAVG2B" wr9, wr9, wr1           \n\t"
-        WAVG2B" wr10, wr10, wr2         \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "walignr1 wr11, wr4, wr5        \n\t"
-        WAVG2B" wr11, wr11, wr3         \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size] \n\t"
-        "wstrd wr10, [r5]               \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "wstrd wr11, [r5, #8]           \n\t"
-        "add r5, r5, %[line_size]       \n\t"
-        "pld [r5]                       \n\t"
-        "pld [r5, #32]                  \n\t"
-        "bne 1b \n\t"
-        : [block]"+r"(block), [pixels]"+r"(pixels), [line_size]"+r"(stride), [h]"+r"(h)
-        :
-        : "memory", "r4", "r5", "r12");
-}
-
-void DEF(put, pixels8_x2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "add r4, %[pixels], %[line_size]\n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr13, [r4]               \n\t"
-        "pld [%[pixels]]                \n\t"
-        "wldrd wr14, [r4, #8]           \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "add r4, r4, %[line_size]       \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "pld [r4]                       \n\t"
-        "pld [r4, #32]                  \n\t"
-        "walignr1 wr2, wr13, wr14       \n\t"
-        "wmoveq wr4, wr11               \n\t"
-        "wmoveq wr6, wr14               \n\t"
-        "walignr2ne wr4, wr10, wr11     \n\t"
-        "walignr2ne wr6, wr13, wr14     \n\t"
-        WAVG2B" wr0, wr0, wr4           \n\t"
-        WAVG2B" wr2, wr2, wr6           \n\t"
-        "wstrd wr0, [%[block]]          \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wstrd wr2, [r5]                \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "add r5, r5, %[line_size]       \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "r4", "r5", "r12", "memory");
-}
-
-void DEF(put, pixels16_x2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "add r4, %[pixels], %[line_size]\n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr13, [r4]               \n\t"
-        "pld [%[pixels]]                \n\t"
-        "wldrd wr14, [r4, #8]           \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wldrd wr15, [r4, #16]          \n\t"
-        "add r4, r4, %[line_size]       \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "pld [r4]                       \n\t"
-        "pld [r4, #32]                  \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-        "walignr1 wr2, wr13, wr14       \n\t"
-        "walignr1 wr3, wr14, wr15       \n\t"
-        "wmoveq wr4, wr11               \n\t"
-        "wmoveq wr5, wr12               \n\t"
-        "wmoveq wr6, wr14               \n\t"
-        "wmoveq wr7, wr15               \n\t"
-        "walignr2ne wr4, wr10, wr11     \n\t"
-        "walignr2ne wr5, wr11, wr12     \n\t"
-        "walignr2ne wr6, wr13, wr14     \n\t"
-        "walignr2ne wr7, wr14, wr15     \n\t"
-        WAVG2B" wr0, wr0, wr4           \n\t"
-        WAVG2B" wr1, wr1, wr5           \n\t"
-        "wstrd wr0, [%[block]]          \n\t"
-        WAVG2B" wr2, wr2, wr6           \n\t"
-        "wstrd wr1, [%[block], #8]      \n\t"
-        WAVG2B" wr3, wr3, wr7           \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "wstrd wr2, [r5]                \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wstrd wr3, [r5, #8]            \n\t"
-        "add r5, r5, %[line_size]       \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "r4", "r5", "r12", "memory");
-}
-
-void DEF(avg, pixels8_x2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "add r4, %[pixels], %[line_size]\n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "pld [r5]                       \n\t"
-        "pld [r5, #32]                  \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr13, [r4]               \n\t"
-        "pld [%[pixels]]                \n\t"
-        "wldrd wr14, [r4, #8]           \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "add r4, r4, %[line_size]       \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "pld [r4]                       \n\t"
-        "pld [r4, #32]                  \n\t"
-        "walignr1 wr2, wr13, wr14       \n\t"
-        "wmoveq wr4, wr11               \n\t"
-        "wmoveq wr6, wr14               \n\t"
-        "walignr2ne wr4, wr10, wr11     \n\t"
-        "wldrd wr10, [%[block]]         \n\t"
-        "walignr2ne wr6, wr13, wr14     \n\t"
-        "wldrd wr12, [r5]               \n\t"
-        WAVG2B" wr0, wr0, wr4           \n\t"
-        WAVG2B" wr2, wr2, wr6           \n\t"
-        WAVG2B" wr0, wr0, wr10          \n\t"
-        WAVG2B" wr2, wr2, wr12          \n\t"
-        "wstrd wr0, [%[block]]          \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wstrd wr2, [r5]                \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "add r5, r5, %[line_size]       \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "pld [r5]                       \n\t"
-        "pld [r5, #32]                  \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "r4", "r5", "r12", "memory");
-}
-
-void DEF(avg, pixels16_x2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "add r4, %[pixels], %[line_size]\n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add r5, %[block], %[line_size] \n\t"
-        "mov %[line_size], %[line_size], lsl #1 \n\t"
-        "pld [r5]                       \n\t"
-        "pld [r5, #32]                  \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "wldrd wr13, [r4]               \n\t"
-        "pld [%[pixels]]                \n\t"
-        "wldrd wr14, [r4, #8]           \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wldrd wr15, [r4, #16]          \n\t"
-        "add r4, r4, %[line_size]       \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "pld [r4]                       \n\t"
-        "pld [r4, #32]                  \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-        "walignr1 wr2, wr13, wr14       \n\t"
-        "walignr1 wr3, wr14, wr15       \n\t"
-        "wmoveq wr4, wr11               \n\t"
-        "wmoveq wr5, wr12               \n\t"
-        "wmoveq wr6, wr14               \n\t"
-        "wmoveq wr7, wr15               \n\t"
-        "walignr2ne wr4, wr10, wr11     \n\t"
-        "walignr2ne wr5, wr11, wr12     \n\t"
-        "walignr2ne wr6, wr13, wr14     \n\t"
-        "walignr2ne wr7, wr14, wr15     \n\t"
-        "wldrd wr10, [%[block]]         \n\t"
-        WAVG2B" wr0, wr0, wr4           \n\t"
-        "wldrd wr11, [%[block], #8]     \n\t"
-        WAVG2B" wr1, wr1, wr5           \n\t"
-        "wldrd wr12, [r5]               \n\t"
-        WAVG2B" wr2, wr2, wr6           \n\t"
-        "wldrd wr13, [r5, #8]           \n\t"
-        WAVG2B" wr3, wr3, wr7           \n\t"
-        WAVG2B" wr0, wr0, wr10          \n\t"
-        WAVG2B" wr1, wr1, wr11          \n\t"
-        WAVG2B" wr2, wr2, wr12          \n\t"
-        WAVG2B" wr3, wr3, wr13          \n\t"
-        "wstrd wr0, [%[block]]          \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wstrd wr1, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "wstrd wr2, [r5]                \n\t"
-        "pld [%[block]]                 \n\t"
-        "wstrd wr3, [r5, #8]            \n\t"
-        "add r5, r5, %[line_size]       \n\t"
-        "pld [%[block], #32]            \n\t"
-        "pld [r5]                       \n\t"
-        "pld [r5, #32]                  \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        :"r4", "r5", "r12", "memory");
-}
-
-void DEF(avg, pixels8_y2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    __asm__ volatile(
-        "pld            [%[pixels]]                             \n\t"
-        "pld            [%[pixels], #32]                        \n\t"
-        "and            r12, %[pixels], #7                      \n\t"
-        "tmcr           wcgr1, r12                              \n\t"
-        "bic            %[pixels], %[pixels], #7                \n\t"
-
-        "wldrd          wr10, [%[pixels]]                       \n\t"
-        "wldrd          wr11, [%[pixels], #8]                   \n\t"
-        "pld            [%[block]]                              \n\t"
-        "add            %[pixels], %[pixels], %[line_size]      \n\t"
-        "walignr1       wr0, wr10, wr11                         \n\t"
-        "pld            [%[pixels]]                             \n\t"
-        "pld            [%[pixels], #32]                        \n\t"
-
-      "1:                                                       \n\t"
-        "wldrd          wr10, [%[pixels]]                       \n\t"
-        "wldrd          wr11, [%[pixels], #8]                   \n\t"
-        "add            %[pixels], %[pixels], %[line_size]      \n\t"
-        "pld            [%[pixels]]                             \n\t"
-        "pld            [%[pixels], #32]                        \n\t"
-        "walignr1       wr4, wr10, wr11                         \n\t"
-        "wldrd          wr10, [%[block]]                        \n\t"
-         WAVG2B"        wr8, wr0, wr4                           \n\t"
-         WAVG2B"        wr8, wr8, wr10                          \n\t"
-        "wstrd          wr8, [%[block]]                         \n\t"
-        "add            %[block], %[block], %[line_size]        \n\t"
-
-        "wldrd          wr10, [%[pixels]]                       \n\t"
-        "wldrd          wr11, [%[pixels], #8]                   \n\t"
-        "pld            [%[block]]                              \n\t"
-        "add            %[pixels], %[pixels], %[line_size]      \n\t"
-        "pld            [%[pixels]]                             \n\t"
-        "pld            [%[pixels], #32]                        \n\t"
-        "walignr1       wr0, wr10, wr11                         \n\t"
-        "wldrd          wr10, [%[block]]                        \n\t"
-         WAVG2B"        wr8, wr0, wr4                           \n\t"
-         WAVG2B"        wr8, wr8, wr10                          \n\t"
-        "wstrd          wr8, [%[block]]                         \n\t"
-        "add            %[block], %[block], %[line_size]        \n\t"
-
-        "subs           %[h], %[h], #2                          \n\t"
-        "pld            [%[block]]                              \n\t"
-        "bne            1b                                      \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "cc", "memory", "r12");
-}
-
-void DEF(put, pixels16_y2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr4, wr10, wr11       \n\t"
-        "walignr1 wr5, wr11, wr12       \n\t"
-        WAVG2B" wr8, wr0, wr4           \n\t"
-        WAVG2B" wr9, wr1, wr5           \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-        WAVG2B" wr8, wr0, wr4           \n\t"
-        WAVG2B" wr9, wr1, wr5           \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        "subs %[h], %[h], #2            \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "r4", "r5", "r12", "memory");
-}
-
-void DEF(avg, pixels16_y2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    int stride = line_size;
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "and r12, %[pixels], #7         \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "pld [%[block]]                 \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-
-        "1:                             \n\t"
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr4, wr10, wr11       \n\t"
-        "walignr1 wr5, wr11, wr12       \n\t"
-        "wldrd wr10, [%[block]]         \n\t"
-        "wldrd wr11, [%[block], #8]     \n\t"
-        WAVG2B" wr8, wr0, wr4           \n\t"
-        WAVG2B" wr9, wr1, wr5           \n\t"
-        WAVG2B" wr8, wr8, wr10          \n\t"
-        WAVG2B" wr9, wr9, wr11          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        "wldrd wr10, [%[pixels]]        \n\t"
-        "wldrd wr11, [%[pixels], #8]    \n\t"
-        "pld [%[block]]                 \n\t"
-        "wldrd wr12, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr0, wr10, wr11       \n\t"
-        "walignr1 wr1, wr11, wr12       \n\t"
-        "wldrd wr10, [%[block]]         \n\t"
-        "wldrd wr11, [%[block], #8]     \n\t"
-        WAVG2B" wr8, wr0, wr4           \n\t"
-        WAVG2B" wr9, wr1, wr5           \n\t"
-        WAVG2B" wr8, wr8, wr10          \n\t"
-        WAVG2B" wr9, wr9, wr11          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        "subs %[h], %[h], #2            \n\t"
-        "pld [%[block]]                 \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block), [line_size]"+r"(stride)
-        :
-        : "r4", "r5", "r12", "memory");
-}
-
-void DEF(put, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "mov r12, #2                    \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "tmcr wcgr0, r12                \n\t" /* for shift value */
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "add r12, r12, #1               \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "cmp r12, #8                    \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-
-        "1:                             \n\t"
-        // [wr0 wr1 wr2 wr3]
-        // [wr4 wr5 wr6 wr7] <= *
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr6, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr4, wr6            \n\t"
-        "wunpckehub wr5, wr6            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr4, wr4, wr8          \n\t"
-        "waddhus wr5, wr5, wr9          \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block)
-        : [line_size]"r"(line_size)
-        : "r12", "memory");
-}
-
-void DEF(put, pixels16_xy2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[pixels]]                \n\t"
-        "mov r12, #2                    \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "tmcr wcgr0, r12                \n\t" /* for shift value */
-        /* alignment */
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "tmcr wcgr2, r12                \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr3, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr2, wr3            \n\t"
-        "wunpckehub wr3, wr3            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr2, wr2, wr10         \n\t"
-        "waddhus wr3, wr3, wr11         \n\t"
-
-        "1:                             \n\t"
-        // [wr0 wr1 wr2 wr3]
-        // [wr4 wr5 wr6 wr7] <= *
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr6, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr7, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr4, wr6            \n\t"
-        "wunpckehub wr5, wr6            \n\t"
-        "wunpckelub wr6, wr7            \n\t"
-        "wunpckehub wr7, wr7            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr4, wr4, wr8          \n\t"
-        "waddhus wr5, wr5, wr9          \n\t"
-        "waddhus wr6, wr6, wr10         \n\t"
-        "waddhus wr7, wr7, wr11         \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr10, wr2, wr6         \n\t"
-        "waddhus wr11, wr3, wr7         \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "waddhus wr10, wr10, wr15       \n\t"
-        "waddhus wr11, wr11, wr15       \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wsrlhg wr10, wr10, wcgr0       \n\t"
-        "wsrlhg wr11, wr11, wcgr0       \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "wpackhus wr9, wr10, wr11       \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr3, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr2, wr3            \n\t"
-        "wunpckehub wr3, wr3            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr2, wr2, wr10         \n\t"
-        "waddhus wr3, wr3, wr11         \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr10, wr2, wr6         \n\t"
-        "waddhus wr11, wr3, wr7         \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "waddhus wr10, wr10, wr15       \n\t"
-        "waddhus wr11, wr11, wr15       \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wsrlhg wr10, wr10, wcgr0       \n\t"
-        "wsrlhg wr11, wr11, wcgr0       \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "wpackhus wr9, wr10, wr11       \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        "subs %[h], %[h], #2            \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block)
-        : [line_size]"r"(line_size)
-        : "r12", "memory");
-}
-
-void DEF(avg, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "pld [%[pixels]]                \n\t"
-        "mov r12, #2                    \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "tmcr wcgr0, r12                \n\t" /* for shift value */
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7   \n\t"
-        "tmcr wcgr1, r12                \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "add r12, r12, #1               \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "tmcr wcgr2, r12                \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "cmp r12, #8                    \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-
-        "1:                             \n\t"
-        // [wr0 wr1 wr2 wr3]
-        // [wr4 wr5 wr6 wr7] <= *
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr6, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr4, wr6            \n\t"
-        "wunpckehub wr5, wr6            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr4, wr4, wr8          \n\t"
-        "waddhus wr5, wr5, wr9          \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "wldrd wr12, [%[block]]         \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        WAVG2B" wr8, wr8, wr12          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "wldrd wr12, [%[block]]         \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        WAVG2B" wr8, wr8, wr12          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block)
-        : [line_size]"r"(line_size)
-        : "r12", "memory");
-}
-
-void DEF(avg, pixels16_xy2)(uint8_t *block, const uint8_t *pixels, const int line_size, int h)
-{
-    // [wr0 wr1 wr2 wr3] for previous line
-    // [wr4 wr5 wr6 wr7] for current line
-    SET_RND(wr15); // =2 for rnd  and  =1 for no_rnd version
-    __asm__ volatile(
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "pld [%[pixels]]                \n\t"
-        "mov r12, #2                    \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "tmcr wcgr0, r12                \n\t" /* for shift value */
-        /* alignment */
-        "and r12, %[pixels], #7         \n\t"
-        "bic %[pixels], %[pixels], #7           \n\t"
-        "tmcr wcgr1, r12                \n\t"
-        "add r12, r12, #1               \n\t"
-        "tmcr wcgr2, r12                \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "pld [%[pixels]]                \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr3, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr2, wr3            \n\t"
-        "wunpckehub wr3, wr3            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr2, wr2, wr10         \n\t"
-        "waddhus wr3, wr3, wr11         \n\t"
-
-        "1:                             \n\t"
-        // [wr0 wr1 wr2 wr3]
-        // [wr4 wr5 wr6 wr7] <= *
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "cmp r12, #8                    \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr6, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr7, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr4, wr6            \n\t"
-        "wunpckehub wr5, wr6            \n\t"
-        "wunpckelub wr6, wr7            \n\t"
-        "wunpckehub wr7, wr7            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr4, wr4, wr8          \n\t"
-        "waddhus wr5, wr5, wr9          \n\t"
-        "waddhus wr6, wr6, wr10         \n\t"
-        "waddhus wr7, wr7, wr11         \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr10, wr2, wr6         \n\t"
-        "waddhus wr11, wr3, wr7         \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "waddhus wr10, wr10, wr15       \n\t"
-        "waddhus wr11, wr11, wr15       \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wldrd wr12, [%[block]]         \n\t"
-        "wldrd wr13, [%[block], #8]     \n\t"
-        "wsrlhg wr10, wr10, wcgr0       \n\t"
-        "wsrlhg wr11, wr11, wcgr0       \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "wpackhus wr9, wr10, wr11       \n\t"
-        WAVG2B" wr8, wr8, wr12          \n\t"
-        WAVG2B" wr9, wr9, wr13          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-
-        // [wr0 wr1 wr2 wr3] <= *
-        // [wr4 wr5 wr6 wr7]
-        "wldrd wr12, [%[pixels]]        \n\t"
-        "pld [%[block]]                 \n\t"
-        "wldrd wr13, [%[pixels], #8]    \n\t"
-        "pld [%[block], #32]            \n\t"
-        "wldrd wr14, [%[pixels], #16]   \n\t"
-        "add %[pixels], %[pixels], %[line_size] \n\t"
-        "walignr1 wr2, wr12, wr13       \n\t"
-        "pld [%[pixels]]                \n\t"
-        "pld [%[pixels], #32]           \n\t"
-        "walignr1 wr3, wr13, wr14       \n\t"
-        "wmoveq wr10, wr13              \n\t"
-        "wmoveq wr11, wr14              \n\t"
-        "walignr2ne wr10, wr12, wr13    \n\t"
-        "walignr2ne wr11, wr13, wr14    \n\t"
-        "wunpckelub wr0, wr2            \n\t"
-        "wunpckehub wr1, wr2            \n\t"
-        "wunpckelub wr2, wr3            \n\t"
-        "wunpckehub wr3, wr3            \n\t"
-        "wunpckelub wr8, wr10           \n\t"
-        "wunpckehub wr9, wr10           \n\t"
-        "wunpckelub wr10, wr11          \n\t"
-        "wunpckehub wr11, wr11          \n\t"
-        "waddhus wr0, wr0, wr8          \n\t"
-        "waddhus wr1, wr1, wr9          \n\t"
-        "waddhus wr2, wr2, wr10         \n\t"
-        "waddhus wr3, wr3, wr11         \n\t"
-        "waddhus wr8, wr0, wr4          \n\t"
-        "waddhus wr9, wr1, wr5          \n\t"
-        "waddhus wr10, wr2, wr6         \n\t"
-        "waddhus wr11, wr3, wr7         \n\t"
-        "waddhus wr8, wr8, wr15         \n\t"
-        "waddhus wr9, wr9, wr15         \n\t"
-        "waddhus wr10, wr10, wr15       \n\t"
-        "waddhus wr11, wr11, wr15       \n\t"
-        "wsrlhg wr8, wr8, wcgr0         \n\t"
-        "wsrlhg wr9, wr9, wcgr0         \n\t"
-        "wldrd wr12, [%[block]]         \n\t"
-        "wldrd wr13, [%[block], #8]     \n\t"
-        "wsrlhg wr10, wr10, wcgr0       \n\t"
-        "wsrlhg wr11, wr11, wcgr0       \n\t"
-        "wpackhus wr8, wr8, wr9         \n\t"
-        "wpackhus wr9, wr10, wr11       \n\t"
-        WAVG2B" wr8, wr8, wr12          \n\t"
-        WAVG2B" wr9, wr9, wr13          \n\t"
-        "wstrd wr8, [%[block]]          \n\t"
-        "wstrd wr9, [%[block], #8]      \n\t"
-        "add %[block], %[block], %[line_size]   \n\t"
-        "subs %[h], %[h], #2            \n\t"
-        "pld [%[block]]                 \n\t"
-        "pld [%[block], #32]            \n\t"
-        "bne 1b                         \n\t"
-        : [h]"+r"(h), [pixels]"+r"(pixels), [block]"+r"(block)
-        : [line_size]"r"(line_size)
-        : "r12", "memory");
-}
diff --git a/libavcodec/arm/mpegvideo_arm.c b/libavcodec/arm/mpegvideo_arm.c
index b2b254e2d3..73d03c1a5e 100644
--- a/libavcodec/arm/mpegvideo_arm.c
+++ b/libavcodec/arm/mpegvideo_arm.c
@@ -40,16 +40,9 @@ void ff_dct_unquantize_h263_intra_neon(MpegEncContext *s, DCTELEM *block,
 
 void ff_MPV_common_init_arm(MpegEncContext *s)
 {
-    /* IWMMXT support is a superset of armv5te, so
-     * allow optimized functions for armv5te unless
-     * a better iwmmxt function exists
-     */
 #if HAVE_ARMV5TE
     ff_MPV_common_init_armv5te(s);
 #endif
-#if HAVE_IWMMXT
-    ff_MPV_common_init_iwmmxt(s);
-#endif
 
     if (HAVE_NEON) {
         s->dct_unquantize_h263_intra = ff_dct_unquantize_h263_intra_neon;
diff --git a/libavcodec/arm/mpegvideo_arm.h b/libavcodec/arm/mpegvideo_arm.h
index e9de979300..4ff93b76b9 100644
--- a/libavcodec/arm/mpegvideo_arm.h
+++ b/libavcodec/arm/mpegvideo_arm.h
@@ -21,7 +21,6 @@
 
 #include "libavcodec/mpegvideo.h"
 
-void ff_MPV_common_init_iwmmxt(MpegEncContext *s);
 void ff_MPV_common_init_armv5te(MpegEncContext *s);
 
 #endif /* AVCODEC_ARM_MPEGVIDEO_H */
diff --git a/libavcodec/arm/mpegvideo_iwmmxt.c b/libavcodec/arm/mpegvideo_iwmmxt.c
deleted file mode 100644
index c85e6dc8b1..0000000000
--- a/libavcodec/arm/mpegvideo_iwmmxt.c
+++ /dev/null
@@ -1,101 +0,0 @@
-/*
- * copyright (c) 2004 AGAWA Koji
- *
- * This file is part of FFmpeg.
- *
- * FFmpeg 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.
- *
- * FFmpeg 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 FFmpeg; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "libavutil/cpu.h"
-#include "libavcodec/avcodec.h"
-#include "libavcodec/dsputil.h"
-#include "libavcodec/mpegvideo.h"
-#include "mpegvideo_arm.h"
-
-static void dct_unquantize_h263_intra_iwmmxt(MpegEncContext *s,
-                                             DCTELEM *block, int n, int qscale)
-{
-    int level, qmul, qadd;
-    int nCoeffs;
-    DCTELEM *block_orig = block;
-
-    assert(s->block_last_index[n]>=0);
-
-    qmul = qscale << 1;
-
-    if (!s->h263_aic) {
-        if (n < 4)
-            level = block[0] * s->y_dc_scale;
-        else
-            level = block[0] * s->c_dc_scale;
-        qadd = (qscale - 1) | 1;
-    }else{
-        qadd = 0;
-        level = block[0];
-    }
-    if(s->ac_pred)
-        nCoeffs=63;
-    else
-        nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ];
-
-    __asm__ volatile (
-/*      "movd %1, %%mm6                 \n\t" //qmul */
-/*      "packssdw %%mm6, %%mm6          \n\t" */
-/*      "packssdw %%mm6, %%mm6          \n\t" */
-        "tbcsth wr6, %[qmul]            \n\t"
-/*      "movd %2, %%mm5                 \n\t" //qadd */
-/*      "packssdw %%mm5, %%mm5          \n\t" */
-/*      "packssdw %%mm5, %%mm5          \n\t" */
-        "tbcsth wr5, %[qadd]            \n\t"
-        "wzero wr7                      \n\t" /* "pxor %%mm7, %%mm7             \n\t" */
-        "wzero wr4                      \n\t" /* "pxor %%mm4, %%mm4             \n\t" */
-        "wsubh wr7, wr5, wr7            \n\t" /* "psubw %%mm5, %%mm7            \n\t" */
-        "1:                             \n\t"
-        "wldrd wr2, [%[block]]          \n\t" /* "movq (%0, %3), %%mm0          \n\t" */
-        "wldrd wr3, [%[block], #8]      \n\t" /* "movq 8(%0, %3), %%mm1         \n\t" */
-        "wmulsl wr0, wr6, wr2           \n\t" /* "pmullw %%mm6, %%mm0           \n\t" */
-        "wmulsl wr1, wr6, wr3           \n\t" /* "pmullw %%mm6, %%mm1           \n\t" */
-/*      "movq (%0, %3), %%mm2           \n\t" */
-/*      "movq 8(%0, %3), %%mm3          \n\t" */
-        "wcmpgtsh wr2, wr4, wr2         \n\t" /* "pcmpgtw %%mm4, %%mm2          \n\t" // block[i] < 0 ? -1 : 0 */
-        "wcmpgtsh wr3, wr4, wr2         \n\t" /* "pcmpgtw %%mm4, %%mm3          \n\t" // block[i] < 0 ? -1 : 0 */
-        "wxor wr0, wr2, wr0             \n\t" /* "pxor %%mm2, %%mm0             \n\t" */
-        "wxor wr1, wr3, wr1             \n\t" /* "pxor %%mm3, %%mm1             \n\t" */
-        "waddh wr0, wr7, wr0            \n\t" /* "paddw %%mm7, %%mm0            \n\t" */
-        "waddh wr1, wr7, wr1            \n\t" /* "paddw %%mm7, %%mm1            \n\t" */
-        "wxor wr2, wr0, wr2             \n\t" /* "pxor %%mm0, %%mm2             \n\t" */
-        "wxor wr3, wr1, wr3             \n\t" /* "pxor %%mm1, %%mm3             \n\t" */
-        "wcmpeqh wr0, wr7, wr0          \n\t" /* "pcmpeqw %%mm7, %%mm0          \n\t" // block[i] == 0 ? -1 : 0 */
-        "wcmpeqh wr1, wr7, wr1          \n\t" /* "pcmpeqw %%mm7, %%mm1          \n\t" // block[i] == 0 ? -1 : 0 */
-        "wandn wr0, wr2, wr0            \n\t" /* "pandn %%mm2, %%mm0            \n\t" */
-        "wandn wr1, wr3, wr1            \n\t" /* "pandn %%mm3, %%mm1            \n\t" */
-        "wstrd wr0, [%[block]]          \n\t" /* "movq %%mm0, (%0, %3)          \n\t" */
-        "wstrd wr1, [%[block], #8]      \n\t" /* "movq %%mm1, 8(%0, %3)         \n\t" */
-        "add %[block], %[block], #16    \n\t" /* "addl $16, %3                  \n\t" */
-        "subs %[i], %[i], #1            \n\t"
-        "bne 1b                         \n\t" /* "jng 1b                                \n\t" */
-        :[block]"+r"(block)
-        :[i]"r"((nCoeffs + 8) / 8), [qmul]"r"(qmul), [qadd]"r"(qadd)
-        :"memory");
-
-    block_orig[0] = level;
-}
-
-void ff_MPV_common_init_iwmmxt(MpegEncContext *s)
-{
-    if (!(mm_flags & AV_CPU_FLAG_IWMMXT)) return;
-
-    s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_iwmmxt;
-}
diff --git a/libavcodec/c93.c b/libavcodec/c93.c
index ad2dc0a7bf..733e80cfbb 100644
--- a/libavcodec/c93.c
+++ b/libavcodec/c93.c
@@ -125,8 +125,9 @@ static int decode_frame(AVCodecContext *avctx, void *data,
     AVFrame * const newpic = &c93->pictures[c93->currentpic];
     AVFrame * const oldpic = &c93->pictures[c93->currentpic^1];
     AVFrame *picture = data;
+    GetByteContext gb;
     uint8_t *out;
-    int stride, i, x, y, bt = 0;
+    int stride, i, x, y, b, bt = 0;
 
     c93->currentpic ^= 1;
 
@@ -140,7 +141,9 @@ static int decode_frame(AVCodecContext *avctx, void *data,
 
     stride = newpic->linesize[0];
 
-    if (buf[0] & C93_FIRST_FRAME) {
+    bytestream2_init(&gb, buf, buf_size);
+    b = bytestream2_get_byte(&gb);
+    if (b & C93_FIRST_FRAME) {
         newpic->pict_type = AV_PICTURE_TYPE_I;
         newpic->key_frame = 1;
     } else {
@@ -148,17 +151,6 @@ static int decode_frame(AVCodecContext *avctx, void *data,
         newpic->key_frame = 0;
     }
 
-    if (*buf++ & C93_HAS_PALETTE) {
-        uint32_t *palette = (uint32_t *) newpic->data[1];
-        const uint8_t *palbuf = buf + buf_size - 768 - 1;
-        for (i = 0; i < 256; i++) {
-            palette[i] = 0xFF << 24 | bytestream_get_be24(&palbuf);
-        }
-    } else {
-        if (oldpic->data[1])
-            memcpy(newpic->data[1], oldpic->data[1], 256 * 4);
-    }
-
     for (y = 0; y < HEIGHT; y += 8) {
         out = newpic->data[0] + y * stride;
         for (x = 0; x < WIDTH; x += 8) {
@@ -168,12 +160,12 @@ static int decode_frame(AVCodecContext *avctx, void *data,
             C93BlockType block_type;
 
             if (!bt)
-                bt = *buf++;
+                bt = bytestream2_get_byte(&gb);
 
             block_type= bt & 0x0F;
             switch (block_type) {
             case C93_8X8_FROM_PREV:
-                offset = bytestream_get_le16(&buf);
+                offset = bytestream2_get_le16(&gb);
                 if (copy_block(avctx, out, copy_from, offset, 8, stride))
                     return -1;
                 break;
@@ -183,7 +175,7 @@ static int decode_frame(AVCodecContext *avctx, void *data,
             case C93_4X4_FROM_PREV:
                 for (j = 0; j < 8; j += 4) {
                     for (i = 0; i < 8; i += 4) {
-                        offset = bytestream_get_le16(&buf);
+                        offset = bytestream2_get_le16(&gb);
                         if (copy_block(avctx, &out[j*stride+i],
                                            copy_from, offset, 4, stride))
                             return -1;
@@ -192,10 +184,10 @@ static int decode_frame(AVCodecContext *avctx, void *data,
                 break;
 
             case C93_8X8_2COLOR:
-                bytestream_get_buffer(&buf, cols, 2);
+                bytestream2_get_buffer(&gb, cols, 2);
                 for (i = 0; i < 8; i++) {
                     draw_n_color(out + i*stride, stride, 8, 1, 1, cols,
-                                     NULL, *buf++);
+                                     NULL, bytestream2_get_byte(&gb));
                 }
 
                 break;
@@ -206,17 +198,17 @@ static int decode_frame(AVCodecContext *avctx, void *data,
                 for (j = 0; j < 8; j += 4) {
                     for (i = 0; i < 8; i += 4) {
                         if (block_type == C93_4X4_2COLOR) {
-                            bytestream_get_buffer(&buf, cols, 2);
+                            bytestream2_get_buffer(&gb, cols, 2);
                             draw_n_color(out + i + j*stride, stride, 4, 4,
-                                    1, cols, NULL, bytestream_get_le16(&buf));
+                                    1, cols, NULL, bytestream2_get_le16(&gb));
                         } else if (block_type == C93_4X4_4COLOR) {
-                            bytestream_get_buffer(&buf, cols, 4);
+                            bytestream2_get_buffer(&gb, cols, 4);
                             draw_n_color(out + i + j*stride, stride, 4, 4,
-                                    2, cols, NULL, bytestream_get_le32(&buf));
+                                    2, cols, NULL, bytestream2_get_le32(&gb));
                         } else {
-                            bytestream_get_buffer(&buf, grps, 4);
+                            bytestream2_get_buffer(&gb, grps, 4);
                             draw_n_color(out + i + j*stride, stride, 4, 4,
-                                    1, cols, grps, bytestream_get_le16(&buf));
+                                    1, cols, grps, bytestream2_get_le16(&gb));
                         }
                     }
                 }
@@ -227,7 +219,7 @@ static int decode_frame(AVCodecContext *avctx, void *data,
 
             case C93_8X8_INTRA:
                 for (j = 0; j < 8; j++)
-                    bytestream_get_buffer(&buf, out + j*stride, 8);
+                    bytestream2_get_buffer(&gb, out + j*stride, 8);
                 break;
 
             default:
@@ -240,6 +232,16 @@ static int decode_frame(AVCodecContext *avctx, void *data,
         }
     }
 
+    if (b & C93_HAS_PALETTE) {
+        uint32_t *palette = (uint32_t *) newpic->data[1];
+        for (i = 0; i < 256; i++) {
+            palette[i] = 0xFFU << 24 | bytestream2_get_be24(&gb);
+        }
+    } else {
+        if (oldpic->data[1])
+            memcpy(newpic->data[1], oldpic->data[1], 256 * 4);
+    }
+
     *picture = *newpic;
     *data_size = sizeof(AVFrame);
 
diff --git a/libavcodec/frwu.c b/libavcodec/frwu.c
index b47cb717f2..1363b53d92 100644
--- a/libavcodec/frwu.c
+++ b/libavcodec/frwu.c
@@ -22,7 +22,6 @@
 
 #include "avcodec.h"
 #include "bytestream.h"
-#include "libavutil/intreadwrite.h"
 
 static av_cold int decode_init(AVCodecContext *avctx)
 {
@@ -54,7 +53,7 @@ static int decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         av_log(avctx, AV_LOG_ERROR, "Packet is too small.\n");
         return AVERROR_INVALIDDATA;
     }
-    if (bytestream_get_le32(&buf) != AV_RL32("FRW1")) {
+    if (bytestream_get_le32(&buf) != MKTAG('F', 'R', 'W', '1')) {
         av_log(avctx, AV_LOG_ERROR, "incorrect marker\n");
         return AVERROR_INVALIDDATA;
     }
diff --git a/libavcodec/mimic.c b/libavcodec/mimic.c
index 51ae5be8bc..4f085b4e5d 100644
--- a/libavcodec/mimic.c
+++ b/libavcodec/mimic.c
@@ -306,24 +306,26 @@ static int mimic_decode_frame(AVCodecContext *avctx, void *data,
     const uint8_t *buf = avpkt->data;
     int buf_size = avpkt->size;
     MimicContext *ctx = avctx->priv_data;
+    GetByteContext gb;
     int is_pframe;
     int width, height;
     int quality, num_coeffs;
     int swap_buf_size = buf_size - MIMIC_HEADER_SIZE;
 
-    if(buf_size < MIMIC_HEADER_SIZE) {
+    if (buf_size <= MIMIC_HEADER_SIZE) {
         av_log(avctx, AV_LOG_ERROR, "insufficient data\n");
         return -1;
     }
 
-    buf       += 2; /* some constant (always 256) */
-    quality    = bytestream_get_le16(&buf);
-    width      = bytestream_get_le16(&buf);
-    height     = bytestream_get_le16(&buf);
-    buf       += 4; /* some constant */
-    is_pframe  = bytestream_get_le32(&buf);
-    num_coeffs = bytestream_get_byte(&buf);
-    buf       += 3; /* some constant */
+    bytestream2_init(&gb, buf, MIMIC_HEADER_SIZE);
+    bytestream2_skip(&gb, 2); /* some constant (always 256) */
+    quality    = bytestream2_get_le16u(&gb);
+    width      = bytestream2_get_le16u(&gb);
+    height     = bytestream2_get_le16u(&gb);
+    bytestream2_skip(&gb, 4); /* some constant */
+    is_pframe  = bytestream2_get_le32u(&gb);
+    num_coeffs = bytestream2_get_byteu(&gb);
+    bytestream2_skip(&gb, 3); /* some constant */
 
     if(!ctx->avctx) {
         int i;
@@ -372,7 +374,7 @@ static int mimic_decode_frame(AVCodecContext *avctx, void *data,
         return AVERROR(ENOMEM);
 
     ctx->dsp.bswap_buf(ctx->swap_buf,
-                        (const uint32_t*) buf,
+                        (const uint32_t*) (buf + MIMIC_HEADER_SIZE),
                         swap_buf_size>>2);
     init_get_bits(&ctx->gb, ctx->swap_buf, swap_buf_size << 3);
 
diff --git a/libavcodec/pcm-mpeg.c b/libavcodec/pcm-mpeg.c
index aea3ff79c6..9c49a0d9ec 100644
--- a/libavcodec/pcm-mpeg.c
+++ b/libavcodec/pcm-mpeg.c
@@ -141,6 +141,7 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
     const uint8_t *src = avpkt->data;
     int buf_size = avpkt->size;
     PCMBRDecode *s = avctx->priv_data;
+    GetByteContext gb;
     int num_source_channels, channel, retval;
     int sample_size, samples;
     int16_t *dst16;
@@ -156,6 +157,8 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
     src += 4;
     buf_size -= 4;
 
+    bytestream2_init(&gb, src, buf_size);
+
     /* There's always an even number of channels in the source */
     num_source_channels = FFALIGN(avctx->channels, 2);
     sample_size = (num_source_channels * (avctx->sample_fmt == AV_SAMPLE_FMT_S16 ? 16 : 24)) >> 3;
@@ -179,15 +182,15 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
             samples *= num_source_channels;
             if (AV_SAMPLE_FMT_S16 == avctx->sample_fmt) {
 #if HAVE_BIGENDIAN
-                memcpy(dst16, src, buf_size);
+                bytestream2_get_buffer(&gb, dst16, buf_size);
 #else
                 do {
-                    *dst16++ = bytestream_get_be16(&src);
+                    *dst16++ = bytestream2_get_be16u(&gb);
                 } while (--samples);
 #endif
             } else {
                 do {
-                    *dst32++ = bytestream_get_be24(&src) << 8;
+                    *dst32++ = bytestream2_get_be24u(&gb) << 8;
                 } while (--samples);
             }
             break;
@@ -199,24 +202,23 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
             if (AV_SAMPLE_FMT_S16 == avctx->sample_fmt) {
                 do {
 #if HAVE_BIGENDIAN
-                    memcpy(dst16, src, avctx->channels * 2);
+                    bytestream2_get_buffer(&gb, dst16, avctx->channels * 2);
                     dst16 += avctx->channels;
-                    src += sample_size;
 #else
                     channel = avctx->channels;
                     do {
-                        *dst16++ = bytestream_get_be16(&src);
+                        *dst16++ = bytestream2_get_be16u(&gb);
                     } while (--channel);
-                    src += 2;
 #endif
+                    bytestream2_skip(&gb, 2);
                 } while (--samples);
             } else {
                 do {
                     channel = avctx->channels;
                     do {
-                        *dst32++ = bytestream_get_be24(&src) << 8;
+                        *dst32++ = bytestream2_get_be24u(&gb) << 8;
                     } while (--channel);
-                    src += 3;
+                    bytestream2_skip(&gb, 3);
                 } while (--samples);
             }
             break;
@@ -224,22 +226,22 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
         case AV_CH_LAYOUT_5POINT1:
             if (AV_SAMPLE_FMT_S16 == avctx->sample_fmt) {
                 do {
-                    dst16[0] = bytestream_get_be16(&src);
-                    dst16[1] = bytestream_get_be16(&src);
-                    dst16[2] = bytestream_get_be16(&src);
-                    dst16[4] = bytestream_get_be16(&src);
-                    dst16[5] = bytestream_get_be16(&src);
-                    dst16[3] = bytestream_get_be16(&src);
+                    dst16[0] = bytestream2_get_be16u(&gb);
+                    dst16[1] = bytestream2_get_be16u(&gb);
+                    dst16[2] = bytestream2_get_be16u(&gb);
+                    dst16[4] = bytestream2_get_be16u(&gb);
+                    dst16[5] = bytestream2_get_be16u(&gb);
+                    dst16[3] = bytestream2_get_be16u(&gb);
                     dst16 += 6;
                 } while (--samples);
             } else {
                 do {
-                    dst32[0] = bytestream_get_be24(&src) << 8;
-                    dst32[1] = bytestream_get_be24(&src) << 8;
-                    dst32[2] = bytestream_get_be24(&src) << 8;
-                    dst32[4] = bytestream_get_be24(&src) << 8;
-                    dst32[5] = bytestream_get_be24(&src) << 8;
-                    dst32[3] = bytestream_get_be24(&src) << 8;
+                    dst32[0] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[1] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[2] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[4] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[5] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[3] = bytestream2_get_be24u(&gb) << 8;
                     dst32 += 6;
                 } while (--samples);
             }
@@ -248,27 +250,27 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
         case AV_CH_LAYOUT_7POINT0:
             if (AV_SAMPLE_FMT_S16 == avctx->sample_fmt) {
                 do {
-                    dst16[0] = bytestream_get_be16(&src);
-                    dst16[1] = bytestream_get_be16(&src);
-                    dst16[2] = bytestream_get_be16(&src);
-                    dst16[5] = bytestream_get_be16(&src);
-                    dst16[3] = bytestream_get_be16(&src);
-                    dst16[4] = bytestream_get_be16(&src);
-                    dst16[6] = bytestream_get_be16(&src);
+                    dst16[0] = bytestream2_get_be16u(&gb);
+                    dst16[1] = bytestream2_get_be16u(&gb);
+                    dst16[2] = bytestream2_get_be16u(&gb);
+                    dst16[5] = bytestream2_get_be16u(&gb);
+                    dst16[3] = bytestream2_get_be16u(&gb);
+                    dst16[4] = bytestream2_get_be16u(&gb);
+                    dst16[6] = bytestream2_get_be16u(&gb);
                     dst16 += 7;
-                    src += 2;
+                    bytestream2_skip(&gb, 2);
                 } while (--samples);
             } else {
                 do {
-                    dst32[0] = bytestream_get_be24(&src) << 8;
-                    dst32[1] = bytestream_get_be24(&src) << 8;
-                    dst32[2] = bytestream_get_be24(&src) << 8;
-                    dst32[5] = bytestream_get_be24(&src) << 8;
-                    dst32[3] = bytestream_get_be24(&src) << 8;
-                    dst32[4] = bytestream_get_be24(&src) << 8;
-                    dst32[6] = bytestream_get_be24(&src) << 8;
+                    dst32[0] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[1] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[2] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[5] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[3] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[4] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[6] = bytestream2_get_be24u(&gb) << 8;
                     dst32 += 7;
-                    src += 3;
+                    bytestream2_skip(&gb, 3);
                 } while (--samples);
             }
             break;
@@ -276,26 +278,26 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
         case AV_CH_LAYOUT_7POINT1:
             if (AV_SAMPLE_FMT_S16 == avctx->sample_fmt) {
                 do {
-                    dst16[0] = bytestream_get_be16(&src);
-                    dst16[1] = bytestream_get_be16(&src);
-                    dst16[2] = bytestream_get_be16(&src);
-                    dst16[6] = bytestream_get_be16(&src);
-                    dst16[4] = bytestream_get_be16(&src);
-                    dst16[5] = bytestream_get_be16(&src);
-                    dst16[7] = bytestream_get_be16(&src);
-                    dst16[3] = bytestream_get_be16(&src);
+                    dst16[0] = bytestream2_get_be16u(&gb);
+                    dst16[1] = bytestream2_get_be16u(&gb);
+                    dst16[2] = bytestream2_get_be16u(&gb);
+                    dst16[6] = bytestream2_get_be16u(&gb);
+                    dst16[4] = bytestream2_get_be16u(&gb);
+                    dst16[5] = bytestream2_get_be16u(&gb);
+                    dst16[7] = bytestream2_get_be16u(&gb);
+                    dst16[3] = bytestream2_get_be16u(&gb);
                     dst16 += 8;
                 } while (--samples);
             } else {
                 do {
-                    dst32[0] = bytestream_get_be24(&src) << 8;
-                    dst32[1] = bytestream_get_be24(&src) << 8;
-                    dst32[2] = bytestream_get_be24(&src) << 8;
-                    dst32[6] = bytestream_get_be24(&src) << 8;
-                    dst32[4] = bytestream_get_be24(&src) << 8;
-                    dst32[5] = bytestream_get_be24(&src) << 8;
-                    dst32[7] = bytestream_get_be24(&src) << 8;
-                    dst32[3] = bytestream_get_be24(&src) << 8;
+                    dst32[0] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[1] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[2] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[6] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[4] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[5] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[7] = bytestream2_get_be24u(&gb) << 8;
+                    dst32[3] = bytestream2_get_be24u(&gb) << 8;
                     dst32 += 8;
                 } while (--samples);
             }
@@ -306,7 +308,7 @@ static int pcm_bluray_decode_frame(AVCodecContext *avctx, void *data,
     *got_frame_ptr   = 1;
     *(AVFrame *)data = s->frame;
 
-    retval = src - avpkt->data;
+    retval = bytestream2_tell(&gb);
     if (avctx->debug & FF_DEBUG_BITSTREAM)
         av_dlog(avctx, "pcm_bluray_decode_frame: decoded %d -> %d bytes\n",
                 retval, buf_size);
diff --git a/libavcodec/s3tc.c b/libavcodec/s3tc.c
index 8e979a84ac..4e791c86c0 100644
--- a/libavcodec/s3tc.c
+++ b/libavcodec/s3tc.c
@@ -21,19 +21,19 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "libavutil/intreadwrite.h"
+#include "libavcodec/bytestream.h"
 #include "avcodec.h"
 #include "s3tc.h"
 
-static inline void dxt1_decode_pixels(const uint8_t *s, uint32_t *d,
+static inline void dxt1_decode_pixels(GetByteContext *gb, uint32_t *d,
                                       unsigned int qstride, unsigned int flag,
                                       uint64_t alpha) {
     unsigned int x, y, c0, c1, a = (!flag * 255u) << 24;
     unsigned int rb0, rb1, rb2, rb3, g0, g1, g2, g3;
     uint32_t colors[4], pixels;
 
-    c0 = AV_RL16(s);
-    c1 = AV_RL16(s+2);
+    c0 = bytestream2_get_le16(gb);
+    c1 = bytestream2_get_le16(gb);
 
     rb0  = (c0<<3 | c0<<8) & 0xf800f8;
     rb1  = (c1<<3 | c1<<8) & 0xf800f8;
@@ -61,7 +61,7 @@ static inline void dxt1_decode_pixels(const uint8_t *s, uint32_t *d,
 
     colors[2] = rb2 + g2 + a;
 
-    pixels = AV_RL32(s+4);
+    pixels = bytestream2_get_le32(gb);
     for (y=0; y<4; y++) {
         for (x=0; x<4; x++) {
             a        = (alpha & 0x0f) << 28;
@@ -74,24 +74,24 @@ static inline void dxt1_decode_pixels(const uint8_t *s, uint32_t *d,
     }
 }
 
-void ff_decode_dxt1(const uint8_t *s, uint8_t *dst,
+void ff_decode_dxt1(GetByteContext *gb, uint8_t *dst,
                     const unsigned int w, const unsigned int h,
                     const unsigned int stride) {
     unsigned int bx, by, qstride = stride/4;
     uint32_t *d = (uint32_t *) dst;
 
     for (by=0; by < h/4; by++, d += stride-w)
-        for (bx=0; bx < w/4; bx++, s+=8, d+=4)
-            dxt1_decode_pixels(s, d, qstride, 0, 0LL);
+        for (bx = 0; bx < w / 4; bx++, d += 4)
+            dxt1_decode_pixels(gb, d, qstride, 0, 0LL);
 }
 
-void ff_decode_dxt3(const uint8_t *s, uint8_t *dst,
+void ff_decode_dxt3(GetByteContext *gb, uint8_t *dst,
                     const unsigned int w, const unsigned int h,
                     const unsigned int stride) {
     unsigned int bx, by, qstride = stride/4;
     uint32_t *d = (uint32_t *) dst;
 
     for (by=0; by < h/4; by++, d += stride-w)
-        for (bx=0; bx < w/4; bx++, s+=16, d+=4)
-            dxt1_decode_pixels(s+8, d, qstride, 1, AV_RL64(s));
+        for (bx = 0; bx < w / 4; bx++, d += 4)
+            dxt1_decode_pixels(gb, d, qstride, 1, bytestream2_get_le64(gb));
 }
diff --git a/libavcodec/s3tc.h b/libavcodec/s3tc.h
index 5116dc80f6..4378bd9483 100644
--- a/libavcodec/s3tc.h
+++ b/libavcodec/s3tc.h
@@ -29,24 +29,24 @@
 
 /**
  * Decode DXT1 encoded data to RGB32
- * @param src source buffer, has to be aligned on a 4-byte boundary
+ * @param gb GetByteContext
  * @param dst destination buffer
  * @param w width of output image
  * @param h height of output image
  * @param stride line size of output image
  */
-void ff_decode_dxt1(const uint8_t *src, uint8_t *dst,
+void ff_decode_dxt1(GetByteContext *gb, uint8_t *dst,
                     const unsigned int w, const unsigned int h,
                     const unsigned int stride);
 /**
  * Decode DXT3 encoded data to RGB32
- * @param src source buffer, has to be aligned on a 4-byte boundary
+ * @param gb GetByteContext
  * @param dst destination buffer
  * @param w width of output image
  * @param h height of output image
  * @param stride line size of output image
  */
-void ff_decode_dxt3(const uint8_t *src, uint8_t *dst,
+void ff_decode_dxt3(GetByteContext *gb, uint8_t *dst,
                     const unsigned int w, const unsigned int h,
                     const unsigned int stride);
 
diff --git a/libavcodec/txd.c b/libavcodec/txd.c
index ca07b6ce17..cf88a9b3bb 100644
--- a/libavcodec/txd.c
+++ b/libavcodec/txd.c
@@ -25,6 +25,7 @@
 #include "libavutil/imgutils.h"
 #include "bytestream.h"
 #include "avcodec.h"
+#include "bytestream.h"
 #include "s3tc.h"
 
 typedef struct TXDContext {
@@ -42,28 +43,25 @@ static av_cold int txd_init(AVCodecContext *avctx) {
 
 static int txd_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
                             AVPacket *avpkt) {
-    const uint8_t *buf = avpkt->data;
-    const uint8_t *buf_end = avpkt->data + avpkt->size;
     TXDContext * const s = avctx->priv_data;
+    GetByteContext gb;
     AVFrame *picture = data;
     AVFrame * const p = &s->picture;
     unsigned int version, w, h, d3d_format, depth, stride, mipmap_count, flags;
     unsigned int y, v;
     uint8_t *ptr;
-    const uint8_t *cur = buf;
-    const uint32_t *palette = (const uint32_t *)(cur + 88);
     uint32_t *pal;
 
-    if (buf_end - cur < 92)
-        return AVERROR_INVALIDDATA;
-    version         = AV_RL32(cur);
-    d3d_format      = AV_RL32(cur+76);
-    w               = AV_RL16(cur+80);
-    h               = AV_RL16(cur+82);
-    depth           = AV_RL8 (cur+84);
-    mipmap_count    = AV_RL8 (cur+85);
-    flags           = AV_RL8 (cur+87);
-    cur            += 92;
+    bytestream2_init(&gb, avpkt->data, avpkt->size);
+    version         = bytestream2_get_le32(&gb);
+    bytestream2_skip(&gb, 72);
+    d3d_format      = bytestream2_get_le32(&gb);
+    w               = bytestream2_get_le16(&gb);
+    h               = bytestream2_get_le16(&gb);
+    depth           = bytestream2_get_byte(&gb);
+    mipmap_count    = bytestream2_get_byte(&gb);
+    bytestream2_skip(&gb, 1);
+    flags           = bytestream2_get_byte(&gb);
 
     if (version < 8 || version > 9) {
         av_log(avctx, AV_LOG_ERROR, "texture data version %i is unsupported\n",
@@ -73,12 +71,9 @@ static int txd_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
 
     if (depth == 8) {
         avctx->pix_fmt = PIX_FMT_PAL8;
-        if (buf_end - cur < 1024)
-            return AVERROR_INVALIDDATA;
-        cur += 1024;
-    } else if (depth == 16 || depth == 32)
+    } else if (depth == 16 || depth == 32) {
         avctx->pix_fmt = PIX_FMT_RGB32;
-    else {
+    } else {
         av_log(avctx, AV_LOG_ERROR, "depth of %i is unsupported\n", depth);
         return -1;
     }
@@ -102,31 +97,32 @@ static int txd_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
 
     if (depth == 8) {
         pal = (uint32_t *) p->data[1];
-        for (y=0; y<256; y++) {
-            v = AV_RB32(palette+y);
-            pal[y] = (v>>8) + (v<<24);
+        for (y = 0; y < 256; y++) {
+            v = bytestream2_get_be32(&gb);
+            pal[y] = (v >> 8) + (v << 24);
         }
-        if (buf_end - cur < w * h)
+        if (bytestream2_get_bytes_left(&gb) < w * h)
             return AVERROR_INVALIDDATA;
+        bytestream2_skip(&gb, 4);
         for (y=0; y<h; y++) {
-            memcpy(ptr, cur, w);
+            bytestream2_get_buffer(&gb, ptr, w);
             ptr += stride;
-            cur += w;
         }
     } else if (depth == 16) {
+        bytestream2_skip(&gb, 4);
         switch (d3d_format) {
         case 0:
             if (!(flags & 1))
                 goto unsupported;
         case FF_S3TC_DXT1:
-            if (buf_end - cur < (w/4) * (h/4) * 8)
+            if (bytestream2_get_bytes_left(&gb) < (w/4) * (h/4) * 8)
                 return AVERROR_INVALIDDATA;
-            ff_decode_dxt1(cur, ptr, w, h, stride);
+            ff_decode_dxt1(&gb, ptr, w, h, stride);
             break;
         case FF_S3TC_DXT3:
-            if (buf_end - cur < (w/4) * (h/4) * 16)
+            if (bytestream2_get_bytes_left(&gb) < (w/4) * (h/4) * 16)
                 return AVERROR_INVALIDDATA;
-            ff_decode_dxt3(cur, ptr, w, h, stride);
+            ff_decode_dxt3(&gb, ptr, w, h, stride);
             break;
         default:
             goto unsupported;
@@ -135,12 +131,11 @@ static int txd_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         switch (d3d_format) {
         case 0x15:
         case 0x16:
-            if (buf_end - cur < h * w * 4)
+            if (bytestream2_get_bytes_left(&gb) < h * w * 4)
                 return AVERROR_INVALIDDATA;
             for (y=0; y<h; y++) {
-                memcpy(ptr, cur, w*4);
+                bytestream2_get_buffer(&gb, ptr, w * 4);
                 ptr += stride;
-                cur += w*4;
             }
             break;
         default:
@@ -148,17 +143,10 @@ static int txd_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         }
     }
 
-    for (; mipmap_count > 1 && buf_end - cur >= 4; mipmap_count--) {
-        uint32_t length = bytestream_get_le32(&cur);
-        if (buf_end - cur < length)
-            break;
-        cur += length;
-    }
-
     *picture   = s->picture;
     *data_size = sizeof(AVPicture);
 
-    return cur - buf;
+    return avpkt->size;
 
 unsupported:
     av_log(avctx, AV_LOG_ERROR, "unsupported d3d format (%08x)\n", d3d_format);