std/jpeg: remove decode_idct block-selector arg

name                                             old speed     new speed     delta

wuffs_jpeg_decode_19k_8bpp/clang11               210MB/s ± 0%  213MB/s ± 0%  +1.54%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_progressive/clang11  200MB/s ± 0%  197MB/s ± 0%  -1.31%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_sequential/clang11   362MB/s ± 0%  373MB/s ± 0%  +3.04%  (p=0.008 n=5+5)
wuffs_jpeg_decode_77k_24bpp/clang11              307MB/s ± 0%  318MB/s ± 0%  +3.82%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_420/clang11         318MB/s ± 0%  331MB/s ± 0%  +4.09%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_444/clang11         434MB/s ± 0%  440MB/s ± 0%  +1.34%  (p=0.016 n=4+5)
wuffs_jpeg_decode_4002k_24bpp/clang11            320MB/s ± 0%  340MB/s ± 0%  +6.29%  (p=0.008 n=5+5)

wuffs_jpeg_decode_19k_8bpp/gcc10                 165MB/s ± 0%  172MB/s ± 0%  +4.32%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_progressive/gcc10    224MB/s ± 0%  227MB/s ± 0%  +1.30%  (p=0.008 n=5+5)
wuffs_jpeg_decode_30k_24bpp_sequential/gcc10     454MB/s ± 0%  465MB/s ± 0%  +2.40%  (p=0.008 n=5+5)
wuffs_jpeg_decode_77k_24bpp/gcc10                311MB/s ± 0%  324MB/s ± 0%  +4.38%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_420/gcc10           327MB/s ± 0%  342MB/s ± 0%  +4.52%  (p=0.008 n=5+5)
wuffs_jpeg_decode_552k_24bpp_444/gcc10           384MB/s ± 0%  393MB/s ± 0%  +2.34%  (p=0.008 n=5+5)
wuffs_jpeg_decode_4002k_24bpp/gcc10              324MB/s ± 0%  345MB/s ± 1%  +6.26%  (p=0.008 n=5+5)
This commit is contained in:
Nigel Tao committed 2023-06-10 14:12:08 +10:00
1 parent d248ade0e6
commit c0d7d7d046
9 files changed
+381 -359

No files matched your search

+200 -173
View File
@@ -8827,6 +8827,7 @@ struct wuffs_jpeg__decoder__struct {
uint32_t f_width_in_mcus;
uint32_t f_height_in_mcus;
uint8_t f_call_sequence;
bool f_test_only_interrupt_decode_mcu;
uint8_t f_sof_marker;
uint8_t f_next_restart_marker;
uint8_t f_max_incl_components_h;
@@ -8894,7 +8895,10 @@ struct wuffs_jpeg__decoder__struct {
uint32_t p_prepare_scan[1];
uint32_t p_skip_past_the_next_restart_marker[1];
uint32_t (*choosy_decode_mcu)(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
} private_impl;
struct {
@@ -37469,7 +37473,6 @@ wuffs_jpeg__decoder__decode_idct(
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_dst_buffer,
uint64_t a_dst_stride,
uint32_t a_b,
uint32_t a_q);
static wuffs_base__status
@@ -37597,27 +37600,45 @@ wuffs_jpeg__decoder__swizzle_colorful(
static uint32_t
wuffs_jpeg__decoder__decode_mcu(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
static uint32_t
wuffs_jpeg__decoder__decode_mcu__choosy_default(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_ac_high_bits(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_ac_low_bit(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_dc_high_bits(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_dc_low_bit(
wuffs_jpeg__decoder* self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my);
// ---------------- VTables
@@ -37735,7 +37756,6 @@ wuffs_jpeg__decoder__decode_idct(
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_dst_buffer,
uint64_t a_dst_stride,
uint32_t a_b,
uint32_t a_q) {
uint32_t v_bq0 = 0;
uint32_t v_bq2 = 0;
@@ -37798,14 +37818,14 @@ wuffs_jpeg__decoder__decode_idct(
if (8u > a_dst_stride) {
return wuffs_base__make_empty_struct();
}
if (0u == (self->private_data.f_mcu_blocks[a_b][8u] |
self->private_data.f_mcu_blocks[a_b][16u] |
self->private_data.f_mcu_blocks[a_b][24u] |
self->private_data.f_mcu_blocks[a_b][32u] |
self->private_data.f_mcu_blocks[a_b][40u] |
self->private_data.f_mcu_blocks[a_b][48u] |
self->private_data.f_mcu_blocks[a_b][56u])) {
v_intermediate[0u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][0u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][0u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][8u] |
self->private_data.f_mcu_blocks[0u][16u] |
self->private_data.f_mcu_blocks[0u][24u] |
self->private_data.f_mcu_blocks[0u][32u] |
self->private_data.f_mcu_blocks[0u][40u] |
self->private_data.f_mcu_blocks[0u][48u] |
self->private_data.f_mcu_blocks[0u][56u])) {
v_intermediate[0u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][0u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][0u])))) << 2u));
v_intermediate[8u] = v_intermediate[0u];
v_intermediate[16u] = v_intermediate[0u];
v_intermediate[24u] = v_intermediate[0u];
@@ -37814,23 +37834,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[48u] = v_intermediate[0u];
v_intermediate[56u] = v_intermediate[0u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][16u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][16u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][48u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][48u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][16u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][16u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][48u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][48u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][0u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][0u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][32u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][32u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][0u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][0u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][32u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][32u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][8u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][8u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][24u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][24u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][40u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][40u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][56u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][56u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][8u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][8u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][24u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][24u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][40u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][40u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][56u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][56u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -37859,14 +37879,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[24u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[32u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][9u] |
self->private_data.f_mcu_blocks[a_b][17u] |
self->private_data.f_mcu_blocks[a_b][25u] |
self->private_data.f_mcu_blocks[a_b][33u] |
self->private_data.f_mcu_blocks[a_b][41u] |
self->private_data.f_mcu_blocks[a_b][49u] |
self->private_data.f_mcu_blocks[a_b][57u])) {
v_intermediate[1u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][1u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][1u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][9u] |
self->private_data.f_mcu_blocks[0u][17u] |
self->private_data.f_mcu_blocks[0u][25u] |
self->private_data.f_mcu_blocks[0u][33u] |
self->private_data.f_mcu_blocks[0u][41u] |
self->private_data.f_mcu_blocks[0u][49u] |
self->private_data.f_mcu_blocks[0u][57u])) {
v_intermediate[1u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][1u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][1u])))) << 2u));
v_intermediate[9u] = v_intermediate[1u];
v_intermediate[17u] = v_intermediate[1u];
v_intermediate[25u] = v_intermediate[1u];
@@ -37875,23 +37895,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[49u] = v_intermediate[1u];
v_intermediate[57u] = v_intermediate[1u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][17u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][17u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][49u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][49u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][17u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][17u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][49u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][49u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][1u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][1u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][33u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][33u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][1u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][1u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][33u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][33u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][9u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][9u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][25u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][25u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][41u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][41u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][57u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][57u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][9u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][9u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][25u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][25u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][41u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][41u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][57u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][57u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -37920,14 +37940,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[25u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[33u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][10u] |
self->private_data.f_mcu_blocks[a_b][18u] |
self->private_data.f_mcu_blocks[a_b][26u] |
self->private_data.f_mcu_blocks[a_b][34u] |
self->private_data.f_mcu_blocks[a_b][42u] |
self->private_data.f_mcu_blocks[a_b][50u] |
self->private_data.f_mcu_blocks[a_b][58u])) {
v_intermediate[2u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][2u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][2u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][10u] |
self->private_data.f_mcu_blocks[0u][18u] |
self->private_data.f_mcu_blocks[0u][26u] |
self->private_data.f_mcu_blocks[0u][34u] |
self->private_data.f_mcu_blocks[0u][42u] |
self->private_data.f_mcu_blocks[0u][50u] |
self->private_data.f_mcu_blocks[0u][58u])) {
v_intermediate[2u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][2u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][2u])))) << 2u));
v_intermediate[10u] = v_intermediate[2u];
v_intermediate[18u] = v_intermediate[2u];
v_intermediate[26u] = v_intermediate[2u];
@@ -37936,23 +37956,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[50u] = v_intermediate[2u];
v_intermediate[58u] = v_intermediate[2u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][18u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][18u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][50u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][50u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][18u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][18u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][50u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][50u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][2u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][2u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][34u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][34u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][2u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][2u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][34u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][34u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][10u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][10u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][26u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][26u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][42u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][42u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][58u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][58u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][10u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][10u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][26u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][26u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][42u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][42u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][58u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][58u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -37981,14 +38001,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[26u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[34u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][11u] |
self->private_data.f_mcu_blocks[a_b][19u] |
self->private_data.f_mcu_blocks[a_b][27u] |
self->private_data.f_mcu_blocks[a_b][35u] |
self->private_data.f_mcu_blocks[a_b][43u] |
self->private_data.f_mcu_blocks[a_b][51u] |
self->private_data.f_mcu_blocks[a_b][59u])) {
v_intermediate[3u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][3u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][3u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][11u] |
self->private_data.f_mcu_blocks[0u][19u] |
self->private_data.f_mcu_blocks[0u][27u] |
self->private_data.f_mcu_blocks[0u][35u] |
self->private_data.f_mcu_blocks[0u][43u] |
self->private_data.f_mcu_blocks[0u][51u] |
self->private_data.f_mcu_blocks[0u][59u])) {
v_intermediate[3u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][3u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][3u])))) << 2u));
v_intermediate[11u] = v_intermediate[3u];
v_intermediate[19u] = v_intermediate[3u];
v_intermediate[27u] = v_intermediate[3u];
@@ -37997,23 +38017,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[51u] = v_intermediate[3u];
v_intermediate[59u] = v_intermediate[3u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][19u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][19u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][51u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][51u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][19u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][19u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][51u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][51u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][3u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][3u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][35u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][35u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][3u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][3u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][35u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][35u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][11u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][11u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][27u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][27u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][43u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][43u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][59u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][59u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][11u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][11u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][27u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][27u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][43u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][43u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][59u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][59u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -38042,14 +38062,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[27u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[35u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][12u] |
self->private_data.f_mcu_blocks[a_b][20u] |
self->private_data.f_mcu_blocks[a_b][28u] |
self->private_data.f_mcu_blocks[a_b][36u] |
self->private_data.f_mcu_blocks[a_b][44u] |
self->private_data.f_mcu_blocks[a_b][52u] |
self->private_data.f_mcu_blocks[a_b][60u])) {
v_intermediate[4u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][4u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][4u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][12u] |
self->private_data.f_mcu_blocks[0u][20u] |
self->private_data.f_mcu_blocks[0u][28u] |
self->private_data.f_mcu_blocks[0u][36u] |
self->private_data.f_mcu_blocks[0u][44u] |
self->private_data.f_mcu_blocks[0u][52u] |
self->private_data.f_mcu_blocks[0u][60u])) {
v_intermediate[4u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][4u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][4u])))) << 2u));
v_intermediate[12u] = v_intermediate[4u];
v_intermediate[20u] = v_intermediate[4u];
v_intermediate[28u] = v_intermediate[4u];
@@ -38058,23 +38078,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[52u] = v_intermediate[4u];
v_intermediate[60u] = v_intermediate[4u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][20u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][20u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][52u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][52u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][20u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][20u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][52u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][52u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][4u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][4u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][36u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][36u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][4u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][4u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][36u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][36u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][12u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][12u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][28u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][28u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][44u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][44u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][60u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][60u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][12u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][12u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][28u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][28u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][44u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][44u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][60u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][60u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -38103,14 +38123,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[28u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[36u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][13u] |
self->private_data.f_mcu_blocks[a_b][21u] |
self->private_data.f_mcu_blocks[a_b][29u] |
self->private_data.f_mcu_blocks[a_b][37u] |
self->private_data.f_mcu_blocks[a_b][45u] |
self->private_data.f_mcu_blocks[a_b][53u] |
self->private_data.f_mcu_blocks[a_b][61u])) {
v_intermediate[5u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][5u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][5u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][13u] |
self->private_data.f_mcu_blocks[0u][21u] |
self->private_data.f_mcu_blocks[0u][29u] |
self->private_data.f_mcu_blocks[0u][37u] |
self->private_data.f_mcu_blocks[0u][45u] |
self->private_data.f_mcu_blocks[0u][53u] |
self->private_data.f_mcu_blocks[0u][61u])) {
v_intermediate[5u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][5u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][5u])))) << 2u));
v_intermediate[13u] = v_intermediate[5u];
v_intermediate[21u] = v_intermediate[5u];
v_intermediate[29u] = v_intermediate[5u];
@@ -38119,23 +38139,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[53u] = v_intermediate[5u];
v_intermediate[61u] = v_intermediate[5u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][21u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][21u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][53u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][53u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][21u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][21u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][53u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][53u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][5u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][5u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][37u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][37u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][5u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][5u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][37u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][37u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][13u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][13u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][29u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][29u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][45u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][45u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][61u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][61u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][13u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][13u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][29u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][29u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][45u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][45u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][61u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][61u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -38164,14 +38184,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[29u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[37u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][14u] |
self->private_data.f_mcu_blocks[a_b][22u] |
self->private_data.f_mcu_blocks[a_b][30u] |
self->private_data.f_mcu_blocks[a_b][38u] |
self->private_data.f_mcu_blocks[a_b][46u] |
self->private_data.f_mcu_blocks[a_b][54u] |
self->private_data.f_mcu_blocks[a_b][62u])) {
v_intermediate[6u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][6u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][6u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][14u] |
self->private_data.f_mcu_blocks[0u][22u] |
self->private_data.f_mcu_blocks[0u][30u] |
self->private_data.f_mcu_blocks[0u][38u] |
self->private_data.f_mcu_blocks[0u][46u] |
self->private_data.f_mcu_blocks[0u][54u] |
self->private_data.f_mcu_blocks[0u][62u])) {
v_intermediate[6u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][6u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][6u])))) << 2u));
v_intermediate[14u] = v_intermediate[6u];
v_intermediate[22u] = v_intermediate[6u];
v_intermediate[30u] = v_intermediate[6u];
@@ -38180,23 +38200,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[54u] = v_intermediate[6u];
v_intermediate[62u] = v_intermediate[6u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][22u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][22u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][54u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][54u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][22u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][22u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][54u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][54u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][6u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][6u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][38u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][38u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][6u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][6u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][38u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][38u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][14u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][14u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][30u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][30u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][46u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][46u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][62u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][62u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][14u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][14u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][30u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][30u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][46u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][46u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][62u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][62u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -38225,14 +38245,14 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[30u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 + v_ck7)) + 1024u)), 11u);
v_intermediate[38u] = wuffs_base__utility__sign_extend_rshift_u32(((uint32_t)(((uint32_t)(v_cd3 - v_ck7)) + 1024u)), 11u);
}
if (0u == (self->private_data.f_mcu_blocks[a_b][15u] |
self->private_data.f_mcu_blocks[a_b][23u] |
self->private_data.f_mcu_blocks[a_b][31u] |
self->private_data.f_mcu_blocks[a_b][39u] |
self->private_data.f_mcu_blocks[a_b][47u] |
self->private_data.f_mcu_blocks[a_b][55u] |
self->private_data.f_mcu_blocks[a_b][63u])) {
v_intermediate[7u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][7u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][7u])))) << 2u));
if (0u == (self->private_data.f_mcu_blocks[0u][15u] |
self->private_data.f_mcu_blocks[0u][23u] |
self->private_data.f_mcu_blocks[0u][31u] |
self->private_data.f_mcu_blocks[0u][39u] |
self->private_data.f_mcu_blocks[0u][47u] |
self->private_data.f_mcu_blocks[0u][55u] |
self->private_data.f_mcu_blocks[0u][63u])) {
v_intermediate[7u] = ((uint32_t)(((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][7u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][7u])))) << 2u));
v_intermediate[15u] = v_intermediate[7u];
v_intermediate[23u] = v_intermediate[7u];
v_intermediate[31u] = v_intermediate[7u];
@@ -38241,23 +38261,23 @@ wuffs_jpeg__decoder__decode_idct(
v_intermediate[55u] = v_intermediate[7u];
v_intermediate[63u] = v_intermediate[7u];
} else {
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][23u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][23u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][55u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][55u]))));
v_bq2 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][23u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][23u]))));
v_bq6 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][55u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][55u]))));
v_ca = ((uint32_t)(((uint32_t)(v_bq2 + v_bq6)) * 4433u));
v_cb2 = ((uint32_t)(v_ca + ((uint32_t)(v_bq2 * 6270u))));
v_cb6 = ((uint32_t)(v_ca - ((uint32_t)(v_bq6 * 15137u))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][7u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][7u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][39u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][39u]))));
v_bq0 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][7u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][7u]))));
v_bq4 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][39u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][39u]))));
v_ccp = ((uint32_t)(((uint32_t)(v_bq0 + v_bq4)) << 13u));
v_ccm = ((uint32_t)(((uint32_t)(v_bq0 - v_bq4)) << 13u));
v_cd0 = ((uint32_t)(v_ccp + v_cb2));
v_cd1 = ((uint32_t)(v_ccm + v_cb6));
v_cd2 = ((uint32_t)(v_ccm - v_cb6));
v_cd3 = ((uint32_t)(v_ccp - v_cb2));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][15u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][15u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][31u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][31u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][47u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][47u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[a_b][63u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][63u]))));
v_bq1 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][15u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][15u]))));
v_bq3 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][31u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][31u]))));
v_bq5 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][47u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][47u]))));
v_bq7 = ((uint32_t)(wuffs_base__utility__sign_extend_convert_u16_u32(self->private_data.f_mcu_blocks[0u][63u]) * ((uint32_t)(self->private_impl.f_quant_tables[a_q][63u]))));
v_ci51 = ((uint32_t)(v_bq5 + v_bq1));
v_ci53 = ((uint32_t)(v_bq5 + v_bq3));
v_ci71 = ((uint32_t)(v_bq7 + v_bq1));
@@ -40472,10 +40492,6 @@ wuffs_jpeg__decoder__decode_sos(
uint32_t v_mx = 0;
uint32_t v_decode_mcu_result = 0;
uint32_t v_bitstream_length = 0;
uint32_t v_b = 0;
uint8_t v_csel = 0;
uint64_t v_stride = 0;
uint64_t v_offset = 0;
uint32_t coro_susp_point = self->private_impl.p_decode_sos[0];
if (coro_susp_point) {
@@ -40514,11 +40530,9 @@ wuffs_jpeg__decoder__decode_sos(
self->private_impl.f_mcu_zig_index = ((uint32_t)(self->private_impl.f_scan_ss));
if (self->private_impl.f_sof_marker >= 194u) {
wuffs_jpeg__decoder__load_mcu_blocks(self, v_mx, v_my, a_workbuf);
} else {
wuffs_base__bulk_memset(&self->private_data.f_mcu_blocks[0], self->private_impl.f_mcu_num_blocks * (size_t)128u, 0u);
}
while (true) {
v_decode_mcu_result = wuffs_jpeg__decoder__decode_mcu(self);
v_decode_mcu_result = wuffs_jpeg__decoder__decode_mcu(self, a_workbuf, v_mx, v_my);
if (v_decode_mcu_result == 0u) {
goto label__decode_mcu__break;
} else if (v_decode_mcu_result != 1u) {
@@ -40539,21 +40553,6 @@ wuffs_jpeg__decoder__decode_sos(
label__decode_mcu__break:;
if (self->private_impl.f_sof_marker >= 194u) {
wuffs_jpeg__decoder__save_mcu_blocks(self, v_mx, v_my, a_workbuf);
} else {
v_b = 0u;
while (v_b < self->private_impl.f_mcu_num_blocks) {
v_csel = self->private_impl.f_scan_comps_cselector[self->private_impl.f_mcu_blocks_sselector[v_b]];
v_stride = ((uint64_t)(self->private_impl.f_components_workbuf_widths[v_csel]));
v_offset = (self->private_impl.f_mcu_blocks_offset[v_b] + (((uint64_t)(self->private_impl.f_mcu_blocks_mx_mul[v_b])) * ((uint64_t)(v_mx))) + (((uint64_t)(self->private_impl.f_mcu_blocks_my_mul[v_b])) * ((uint64_t)(v_my))));
if (v_offset <= ((uint64_t)(a_workbuf.len))) {
wuffs_jpeg__decoder__decode_idct(self,
wuffs_base__slice_u8__subslice_i(a_workbuf, v_offset),
v_stride,
v_b,
((uint32_t)(self->private_impl.f_components_tq[v_csel])));
}
v_b += 1u;
}
}
if (self->private_impl.f_restarts_remaining > 0u) {
#if defined(__GNUC__)
@@ -41190,11 +41189,7 @@ wuffs_jpeg__decoder__apply_progressive_idct(
v_stride = ((uint64_t)(self->private_impl.f_components_workbuf_widths[v_csel]));
v_offset = (self->private_impl.f_components_workbuf_offsets[v_csel] + (((uint64_t)(v_mcu_blocks_mx_mul_0)) * ((uint64_t)(v_mx))) + (((uint64_t)(v_mcu_blocks_my_mul_0)) * ((uint64_t)(v_my))));
if (v_offset <= ((uint64_t)(a_workbuf.len))) {
wuffs_jpeg__decoder__decode_idct(self,
wuffs_base__slice_u8__subslice_i(a_workbuf, v_offset),
v_stride,
0u,
((uint32_t)(self->private_impl.f_components_tq[v_csel])));
wuffs_jpeg__decoder__decode_idct(self, wuffs_base__slice_u8__subslice_i(a_workbuf, v_offset), v_stride, ((uint32_t)(self->private_impl.f_components_tq[v_csel])));
}
v_mx += 1u;
}
@@ -41505,13 +41500,19 @@ wuffs_jpeg__decoder__workbuf_len(
static uint32_t
wuffs_jpeg__decoder__decode_mcu(
wuffs_jpeg__decoder* self) {
return (*self->private_impl.choosy_decode_mcu)(self);
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
return (*self->private_impl.choosy_decode_mcu)(self, a_workbuf, a_mx, a_my);
}
static uint32_t
wuffs_jpeg__decoder__decode_mcu__choosy_default(
wuffs_jpeg__decoder* self) {
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
uint32_t v_ret = 0;
uint64_t v_bits = 0;
uint32_t v_n_bits = 0;
@@ -41542,6 +41543,9 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
uint32_t v_ac_rrrr = 0;
uint32_t v_ac_ssss = 0;
uint32_t v_z = 0;
uint32_t v_mcb = 0;
uint64_t v_stride = 0;
uint64_t v_offset = 0;
v_bits = self->private_impl.f_bitstream_bits;
v_n_bits = self->private_impl.f_bitstream_n_bits;
@@ -41567,6 +41571,7 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
while (true) {
while (self->private_impl.f_mcu_current_block < self->private_impl.f_mcu_num_blocks) {
while (self->private_impl.f_mcu_zig_index <= 0u) {
wuffs_base__bulk_memset(&self->private_data.f_mcu_blocks[0], 1u * (size_t)128u, 0u);
if (((uint64_t)(io2_v_r - iop_v_r)) < 8u) {
v_ret = 1u;
goto label__goto_done__break;
@@ -41611,7 +41616,7 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
v_n_bits -= v_dc_symbol;
v_csel = self->private_impl.f_scan_comps_cselector[self->private_impl.f_mcu_blocks_sselector[self->private_impl.f_mcu_current_block]];
self->private_impl.f_mcu_previous_dc_values[v_csel] += ((uint16_t)(v_dc_value));
self->private_data.f_mcu_blocks[self->private_impl.f_mcu_current_block][0u] = self->private_impl.f_mcu_previous_dc_values[v_csel];
self->private_data.f_mcu_blocks[0u][0u] = self->private_impl.f_mcu_previous_dc_values[v_csel];
self->private_impl.f_mcu_zig_index = 1u;
goto label__dc_component__break;
}
@@ -41663,7 +41668,7 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
v_bits <<= v_ac_ssss;
v_n_bits -= v_ac_ssss;
v_z = (self->private_impl.f_mcu_zig_index + v_ac_rrrr);
self->private_data.f_mcu_blocks[self->private_impl.f_mcu_current_block][WUFFS_JPEG__UNZIG[v_z]] = ((uint16_t)(v_ac_value));
self->private_data.f_mcu_blocks[0u][WUFFS_JPEG__UNZIG[v_z]] = ((uint16_t)(v_ac_value));
if ((v_z + 1u) > 63u) {
goto label__ac_components__break;
}
@@ -41678,7 +41683,17 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
}
label__ac_components__break:;
self->private_impl.f_mcu_zig_index = 0u;
v_mcb = self->private_impl.f_mcu_current_block;
self->private_impl.f_mcu_current_block += 1u;
if (self->private_impl.f_test_only_interrupt_decode_mcu) {
goto label__goto_done__break;
}
v_csel = self->private_impl.f_scan_comps_cselector[self->private_impl.f_mcu_blocks_sselector[v_mcb]];
v_stride = ((uint64_t)(self->private_impl.f_components_workbuf_widths[v_csel]));
v_offset = (self->private_impl.f_mcu_blocks_offset[v_mcb] + (((uint64_t)(self->private_impl.f_mcu_blocks_mx_mul[v_mcb])) * ((uint64_t)(a_mx))) + (((uint64_t)(self->private_impl.f_mcu_blocks_my_mul[v_mcb])) * ((uint64_t)(a_my))));
if (v_offset <= ((uint64_t)(a_workbuf.len))) {
wuffs_jpeg__decoder__decode_idct(self, wuffs_base__slice_u8__subslice_i(a_workbuf, v_offset), v_stride, ((uint32_t)(self->private_impl.f_components_tq[v_csel])));
}
}
self->private_impl.f_mcu_current_block = 0u;
goto label__goto_done__break;
@@ -41705,7 +41720,10 @@ wuffs_jpeg__decoder__decode_mcu__choosy_default(
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_ac_high_bits(
wuffs_jpeg__decoder* self) {
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
uint32_t v_ret = 0;
uint64_t v_bits = 0;
uint32_t v_n_bits = 0;
@@ -41857,7 +41875,10 @@ wuffs_jpeg__decoder__decode_mcu_progressive_ac_high_bits(
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_ac_low_bit(
wuffs_jpeg__decoder* self) {
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
uint32_t v_ret = 0;
uint64_t v_bits = 0;
uint32_t v_n_bits = 0;
@@ -42107,7 +42128,10 @@ wuffs_jpeg__decoder__decode_mcu_progressive_ac_low_bit(
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_dc_high_bits(
wuffs_jpeg__decoder* self) {
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
uint32_t v_ret = 0;
uint64_t v_bits = 0;
uint32_t v_n_bits = 0;
@@ -42227,7 +42251,10 @@ wuffs_jpeg__decoder__decode_mcu_progressive_dc_high_bits(
static uint32_t
wuffs_jpeg__decoder__decode_mcu_progressive_dc_low_bit(
wuffs_jpeg__decoder* self) {
wuffs_jpeg__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mx,
uint32_t a_my) {
uint32_t v_ret = 0;
uint64_t v_bits = 0;
uint32_t v_n_bits = 0;
+129 -129
View File
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b: base.u32[..= 9], q: base.u32[..= 3]) {
pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, q: base.u32[..= 3]) {
// This method implements the same algorithm as libjpeg-turbo's jidctint.c.
var bq0 : base.u32
@@ -120,17 +120,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 0.
if (0 == (
this.mcu_blocks[args.b][0x08] |
this.mcu_blocks[args.b][0x10] |
this.mcu_blocks[args.b][0x18] |
this.mcu_blocks[args.b][0x20] |
this.mcu_blocks[args.b][0x28] |
this.mcu_blocks[args.b][0x30] |
this.mcu_blocks[args.b][0x38])) {
this.mcu_blocks[0][0x08] |
this.mcu_blocks[0][0x10] |
this.mcu_blocks[0][0x18] |
this.mcu_blocks[0][0x20] |
this.mcu_blocks[0][0x28] |
this.mcu_blocks[0][0x30] |
this.mcu_blocks[0][0x38])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x00] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x00]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x00]) ~mod*
(this.quant_tables[args.q][0x00] as base.u32)) ~mod<< 2
intermediate[0x08] = intermediate[0x00]
intermediate[0x10] = intermediate[0x00]
@@ -143,16 +143,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x10]) ~mod* (this.quant_tables[args.q][0x10] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x30]) ~mod* (this.quant_tables[args.q][0x30] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x10]) ~mod* (this.quant_tables[args.q][0x10] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x30]) ~mod* (this.quant_tables[args.q][0x30] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x00]) ~mod* (this.quant_tables[args.q][0x00] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x20]) ~mod* (this.quant_tables[args.q][0x20] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x00]) ~mod* (this.quant_tables[args.q][0x00] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x20]) ~mod* (this.quant_tables[args.q][0x20] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -164,10 +164,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x08]) ~mod* (this.quant_tables[args.q][0x08] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x18]) ~mod* (this.quant_tables[args.q][0x18] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x28]) ~mod* (this.quant_tables[args.q][0x28] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x38]) ~mod* (this.quant_tables[args.q][0x38] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x08]) ~mod* (this.quant_tables[args.q][0x08] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x18]) ~mod* (this.quant_tables[args.q][0x18] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x28]) ~mod* (this.quant_tables[args.q][0x28] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x38]) ~mod* (this.quant_tables[args.q][0x38] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -209,17 +209,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 1.
if (0 == (
this.mcu_blocks[args.b][0x09] |
this.mcu_blocks[args.b][0x11] |
this.mcu_blocks[args.b][0x19] |
this.mcu_blocks[args.b][0x21] |
this.mcu_blocks[args.b][0x29] |
this.mcu_blocks[args.b][0x31] |
this.mcu_blocks[args.b][0x39])) {
this.mcu_blocks[0][0x09] |
this.mcu_blocks[0][0x11] |
this.mcu_blocks[0][0x19] |
this.mcu_blocks[0][0x21] |
this.mcu_blocks[0][0x29] |
this.mcu_blocks[0][0x31] |
this.mcu_blocks[0][0x39])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x01] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x01]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x01]) ~mod*
(this.quant_tables[args.q][0x01] as base.u32)) ~mod<< 2
intermediate[0x09] = intermediate[0x01]
intermediate[0x11] = intermediate[0x01]
@@ -232,16 +232,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x11]) ~mod* (this.quant_tables[args.q][0x11] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x31]) ~mod* (this.quant_tables[args.q][0x31] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x11]) ~mod* (this.quant_tables[args.q][0x11] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x31]) ~mod* (this.quant_tables[args.q][0x31] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x01]) ~mod* (this.quant_tables[args.q][0x01] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x21]) ~mod* (this.quant_tables[args.q][0x21] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x01]) ~mod* (this.quant_tables[args.q][0x01] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x21]) ~mod* (this.quant_tables[args.q][0x21] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -253,10 +253,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x09]) ~mod* (this.quant_tables[args.q][0x09] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x19]) ~mod* (this.quant_tables[args.q][0x19] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x29]) ~mod* (this.quant_tables[args.q][0x29] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x39]) ~mod* (this.quant_tables[args.q][0x39] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x09]) ~mod* (this.quant_tables[args.q][0x09] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x19]) ~mod* (this.quant_tables[args.q][0x19] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x29]) ~mod* (this.quant_tables[args.q][0x29] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x39]) ~mod* (this.quant_tables[args.q][0x39] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -298,17 +298,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 2.
if (0 == (
this.mcu_blocks[args.b][0x0A] |
this.mcu_blocks[args.b][0x12] |
this.mcu_blocks[args.b][0x1A] |
this.mcu_blocks[args.b][0x22] |
this.mcu_blocks[args.b][0x2A] |
this.mcu_blocks[args.b][0x32] |
this.mcu_blocks[args.b][0x3A])) {
this.mcu_blocks[0][0x0A] |
this.mcu_blocks[0][0x12] |
this.mcu_blocks[0][0x1A] |
this.mcu_blocks[0][0x22] |
this.mcu_blocks[0][0x2A] |
this.mcu_blocks[0][0x32] |
this.mcu_blocks[0][0x3A])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x02] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x02]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x02]) ~mod*
(this.quant_tables[args.q][0x02] as base.u32)) ~mod<< 2
intermediate[0x0A] = intermediate[0x02]
intermediate[0x12] = intermediate[0x02]
@@ -321,16 +321,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x12]) ~mod* (this.quant_tables[args.q][0x12] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x32]) ~mod* (this.quant_tables[args.q][0x32] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x12]) ~mod* (this.quant_tables[args.q][0x12] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x32]) ~mod* (this.quant_tables[args.q][0x32] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x02]) ~mod* (this.quant_tables[args.q][0x02] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x22]) ~mod* (this.quant_tables[args.q][0x22] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x02]) ~mod* (this.quant_tables[args.q][0x02] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x22]) ~mod* (this.quant_tables[args.q][0x22] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -342,10 +342,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0A]) ~mod* (this.quant_tables[args.q][0x0A] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1A]) ~mod* (this.quant_tables[args.q][0x1A] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2A]) ~mod* (this.quant_tables[args.q][0x2A] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3A]) ~mod* (this.quant_tables[args.q][0x3A] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0A]) ~mod* (this.quant_tables[args.q][0x0A] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1A]) ~mod* (this.quant_tables[args.q][0x1A] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2A]) ~mod* (this.quant_tables[args.q][0x2A] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3A]) ~mod* (this.quant_tables[args.q][0x3A] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -387,17 +387,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 3.
if (0 == (
this.mcu_blocks[args.b][0x0B] |
this.mcu_blocks[args.b][0x13] |
this.mcu_blocks[args.b][0x1B] |
this.mcu_blocks[args.b][0x23] |
this.mcu_blocks[args.b][0x2B] |
this.mcu_blocks[args.b][0x33] |
this.mcu_blocks[args.b][0x3B])) {
this.mcu_blocks[0][0x0B] |
this.mcu_blocks[0][0x13] |
this.mcu_blocks[0][0x1B] |
this.mcu_blocks[0][0x23] |
this.mcu_blocks[0][0x2B] |
this.mcu_blocks[0][0x33] |
this.mcu_blocks[0][0x3B])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x03] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x03]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x03]) ~mod*
(this.quant_tables[args.q][0x03] as base.u32)) ~mod<< 2
intermediate[0x0B] = intermediate[0x03]
intermediate[0x13] = intermediate[0x03]
@@ -410,16 +410,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x13]) ~mod* (this.quant_tables[args.q][0x13] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x33]) ~mod* (this.quant_tables[args.q][0x33] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x13]) ~mod* (this.quant_tables[args.q][0x13] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x33]) ~mod* (this.quant_tables[args.q][0x33] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x03]) ~mod* (this.quant_tables[args.q][0x03] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x23]) ~mod* (this.quant_tables[args.q][0x23] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x03]) ~mod* (this.quant_tables[args.q][0x03] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x23]) ~mod* (this.quant_tables[args.q][0x23] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -431,10 +431,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0B]) ~mod* (this.quant_tables[args.q][0x0B] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1B]) ~mod* (this.quant_tables[args.q][0x1B] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2B]) ~mod* (this.quant_tables[args.q][0x2B] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3B]) ~mod* (this.quant_tables[args.q][0x3B] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0B]) ~mod* (this.quant_tables[args.q][0x0B] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1B]) ~mod* (this.quant_tables[args.q][0x1B] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2B]) ~mod* (this.quant_tables[args.q][0x2B] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3B]) ~mod* (this.quant_tables[args.q][0x3B] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -476,17 +476,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 4.
if (0 == (
this.mcu_blocks[args.b][0x0C] |
this.mcu_blocks[args.b][0x14] |
this.mcu_blocks[args.b][0x1C] |
this.mcu_blocks[args.b][0x24] |
this.mcu_blocks[args.b][0x2C] |
this.mcu_blocks[args.b][0x34] |
this.mcu_blocks[args.b][0x3C])) {
this.mcu_blocks[0][0x0C] |
this.mcu_blocks[0][0x14] |
this.mcu_blocks[0][0x1C] |
this.mcu_blocks[0][0x24] |
this.mcu_blocks[0][0x2C] |
this.mcu_blocks[0][0x34] |
this.mcu_blocks[0][0x3C])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x04] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x04]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x04]) ~mod*
(this.quant_tables[args.q][0x04] as base.u32)) ~mod<< 2
intermediate[0x0C] = intermediate[0x04]
intermediate[0x14] = intermediate[0x04]
@@ -499,16 +499,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x14]) ~mod* (this.quant_tables[args.q][0x14] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x34]) ~mod* (this.quant_tables[args.q][0x34] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x14]) ~mod* (this.quant_tables[args.q][0x14] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x34]) ~mod* (this.quant_tables[args.q][0x34] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x04]) ~mod* (this.quant_tables[args.q][0x04] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x24]) ~mod* (this.quant_tables[args.q][0x24] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x04]) ~mod* (this.quant_tables[args.q][0x04] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x24]) ~mod* (this.quant_tables[args.q][0x24] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -520,10 +520,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0C]) ~mod* (this.quant_tables[args.q][0x0C] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1C]) ~mod* (this.quant_tables[args.q][0x1C] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2C]) ~mod* (this.quant_tables[args.q][0x2C] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3C]) ~mod* (this.quant_tables[args.q][0x3C] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0C]) ~mod* (this.quant_tables[args.q][0x0C] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1C]) ~mod* (this.quant_tables[args.q][0x1C] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2C]) ~mod* (this.quant_tables[args.q][0x2C] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3C]) ~mod* (this.quant_tables[args.q][0x3C] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -565,17 +565,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 5.
if (0 == (
this.mcu_blocks[args.b][0x0D] |
this.mcu_blocks[args.b][0x15] |
this.mcu_blocks[args.b][0x1D] |
this.mcu_blocks[args.b][0x25] |
this.mcu_blocks[args.b][0x2D] |
this.mcu_blocks[args.b][0x35] |
this.mcu_blocks[args.b][0x3D])) {
this.mcu_blocks[0][0x0D] |
this.mcu_blocks[0][0x15] |
this.mcu_blocks[0][0x1D] |
this.mcu_blocks[0][0x25] |
this.mcu_blocks[0][0x2D] |
this.mcu_blocks[0][0x35] |
this.mcu_blocks[0][0x3D])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x05] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x05]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x05]) ~mod*
(this.quant_tables[args.q][0x05] as base.u32)) ~mod<< 2
intermediate[0x0D] = intermediate[0x05]
intermediate[0x15] = intermediate[0x05]
@@ -588,16 +588,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x15]) ~mod* (this.quant_tables[args.q][0x15] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x35]) ~mod* (this.quant_tables[args.q][0x35] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x15]) ~mod* (this.quant_tables[args.q][0x15] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x35]) ~mod* (this.quant_tables[args.q][0x35] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x05]) ~mod* (this.quant_tables[args.q][0x05] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x25]) ~mod* (this.quant_tables[args.q][0x25] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x05]) ~mod* (this.quant_tables[args.q][0x05] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x25]) ~mod* (this.quant_tables[args.q][0x25] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -609,10 +609,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0D]) ~mod* (this.quant_tables[args.q][0x0D] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1D]) ~mod* (this.quant_tables[args.q][0x1D] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2D]) ~mod* (this.quant_tables[args.q][0x2D] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3D]) ~mod* (this.quant_tables[args.q][0x3D] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0D]) ~mod* (this.quant_tables[args.q][0x0D] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1D]) ~mod* (this.quant_tables[args.q][0x1D] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2D]) ~mod* (this.quant_tables[args.q][0x2D] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3D]) ~mod* (this.quant_tables[args.q][0x3D] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -654,17 +654,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 6.
if (0 == (
this.mcu_blocks[args.b][0x0E] |
this.mcu_blocks[args.b][0x16] |
this.mcu_blocks[args.b][0x1E] |
this.mcu_blocks[args.b][0x26] |
this.mcu_blocks[args.b][0x2E] |
this.mcu_blocks[args.b][0x36] |
this.mcu_blocks[args.b][0x3E])) {
this.mcu_blocks[0][0x0E] |
this.mcu_blocks[0][0x16] |
this.mcu_blocks[0][0x1E] |
this.mcu_blocks[0][0x26] |
this.mcu_blocks[0][0x2E] |
this.mcu_blocks[0][0x36] |
this.mcu_blocks[0][0x3E])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x06] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x06]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x06]) ~mod*
(this.quant_tables[args.q][0x06] as base.u32)) ~mod<< 2
intermediate[0x0E] = intermediate[0x06]
intermediate[0x16] = intermediate[0x06]
@@ -677,16 +677,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x16]) ~mod* (this.quant_tables[args.q][0x16] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x36]) ~mod* (this.quant_tables[args.q][0x36] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x16]) ~mod* (this.quant_tables[args.q][0x16] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x36]) ~mod* (this.quant_tables[args.q][0x36] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x06]) ~mod* (this.quant_tables[args.q][0x06] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x26]) ~mod* (this.quant_tables[args.q][0x26] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x06]) ~mod* (this.quant_tables[args.q][0x06] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x26]) ~mod* (this.quant_tables[args.q][0x26] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -698,10 +698,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0E]) ~mod* (this.quant_tables[args.q][0x0E] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1E]) ~mod* (this.quant_tables[args.q][0x1E] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2E]) ~mod* (this.quant_tables[args.q][0x2E] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3E]) ~mod* (this.quant_tables[args.q][0x3E] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0E]) ~mod* (this.quant_tables[args.q][0x0E] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1E]) ~mod* (this.quant_tables[args.q][0x1E] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2E]) ~mod* (this.quant_tables[args.q][0x2E] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3E]) ~mod* (this.quant_tables[args.q][0x3E] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
@@ -743,17 +743,17 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// ==== First pass, column 7.
if (0 == (
this.mcu_blocks[args.b][0x0F] |
this.mcu_blocks[args.b][0x17] |
this.mcu_blocks[args.b][0x1F] |
this.mcu_blocks[args.b][0x27] |
this.mcu_blocks[args.b][0x2F] |
this.mcu_blocks[args.b][0x37] |
this.mcu_blocks[args.b][0x3F])) {
this.mcu_blocks[0][0x0F] |
this.mcu_blocks[0][0x17] |
this.mcu_blocks[0][0x1F] |
this.mcu_blocks[0][0x27] |
this.mcu_blocks[0][0x2F] |
this.mcu_blocks[0][0x37] |
this.mcu_blocks[0][0x3F])) {
// Fast path when the 1-dimensional AC terms are all zero.
intermediate[0x07] =
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x07]) ~mod*
(this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x07]) ~mod*
(this.quant_tables[args.q][0x07] as base.u32)) ~mod<< 2
intermediate[0x0F] = intermediate[0x07]
intermediate[0x17] = intermediate[0x07]
@@ -766,16 +766,16 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
} else {
// Even rows.
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x17]) ~mod* (this.quant_tables[args.q][0x17] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x37]) ~mod* (this.quant_tables[args.q][0x37] as base.u32)
bq2 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x17]) ~mod* (this.quant_tables[args.q][0x17] as base.u32)
bq6 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x37]) ~mod* (this.quant_tables[args.q][0x37] as base.u32)
ca = (bq2 ~mod+ bq6) ~mod* 0x1151
cb2 = ca ~mod+ (bq2 ~mod* 0x187E)
cb6 = ca ~mod- (bq6 ~mod* 0x3B21)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x07]) ~mod* (this.quant_tables[args.q][0x07] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x27]) ~mod* (this.quant_tables[args.q][0x27] as base.u32)
bq0 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x07]) ~mod* (this.quant_tables[args.q][0x07] as base.u32)
bq4 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x27]) ~mod* (this.quant_tables[args.q][0x27] as base.u32)
ccp = (bq0 ~mod+ bq4) ~mod<< 13
ccm = (bq0 ~mod- bq4) ~mod<< 13
@@ -787,10 +787,10 @@ pri func decoder.decode_idct!(dst_buffer: slice base.u8, dst_stride: base.u64, b
// Odd rows.
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x0F]) ~mod* (this.quant_tables[args.q][0x0F] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x1F]) ~mod* (this.quant_tables[args.q][0x1F] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x2F]) ~mod* (this.quant_tables[args.q][0x2F] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[args.b][0x3F]) ~mod* (this.quant_tables[args.q][0x3F] as base.u32)
bq1 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x0F]) ~mod* (this.quant_tables[args.q][0x0F] as base.u32)
bq3 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x1F]) ~mod* (this.quant_tables[args.q][0x1F] as base.u32)
bq5 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x2F]) ~mod* (this.quant_tables[args.q][0x2F] as base.u32)
bq7 = this.util.sign_extend_convert_u16_u32(a: this.mcu_blocks[0][0x3F]) ~mod* (this.quant_tables[args.q][0x3F] as base.u32)
ci51 = bq5 ~mod+ bq1
ci53 = bq5 ~mod+ bq3
+6 -33
View File
@@ -49,6 +49,8 @@ pub struct decoder? implements base.image_decoder(
// (/doc/std/image-decoders-call-sequence.md).
call_sequence : base.u8,
test_only_interrupt_decode_mcu : base.bool,
sof_marker : base.u8,
next_restart_marker : base.u8[..= 7],
@@ -289,6 +291,9 @@ pub struct decoder? implements base.image_decoder(
) + (
bitstream_buffer : array[0x800] base.u8, // 0x800 = 2048.
// Decoding sequential JPEGs only uses the first element of this array,
// since sequential decode_mcu calls decode_idct straight away.
// Decoding progressive JPEGs can use all of it.
mcu_blocks : array[10] array[64] base.u16,
swizzle_ycck_scratch_buffer_2k : array[2048] base.u8,
@@ -1135,11 +1140,6 @@ pri func decoder.decode_sos?(dst: ptr base.pixel_buffer, src: base.io_reader, wo
var decode_mcu_result : base.u32
var bitstream_length : base.u32
var b : base.u32
var csel : base.u8[..= 3]
var stride : base.u64[..= 0x1_0008]
var offset : base.u64
if this.scan_count >= 64 {
return "#unsupported scan count"
}
@@ -1173,16 +1173,13 @@ pri func decoder.decode_sos?(dst: ptr base.pixel_buffer, src: base.io_reader, wo
if this.sof_marker >= 0xC2 {
this.load_mcu_blocks!(mx: mx, my: my, workbuf: args.workbuf)
} else {
// Clear this.mcu_blocks.
this.mcu_blocks[.. this.mcu_num_blocks].bulk_memset!(byte_value: 0)
}
while.decode_mcu true,
inv my < 0x2000,
inv mx < 0x2000,
{
decode_mcu_result = this.decode_mcu!()
decode_mcu_result = this.decode_mcu!(workbuf: args.workbuf, mx: mx, my: my)
if decode_mcu_result == 0 {
break.decode_mcu
} else if decode_mcu_result <> 1 {
@@ -1204,29 +1201,6 @@ pri func decoder.decode_sos?(dst: ptr base.pixel_buffer, src: base.io_reader, wo
if this.sof_marker >= 0xC2 {
this.save_mcu_blocks!(mx: mx, my: my, workbuf: args.workbuf)
} else {
// Apply IDCT.
b = 0
while b < this.mcu_num_blocks,
inv my < 0x2000,
inv mx < 0x2000,
{
assert b < 10 via "a < b: a < c; c <= b"(c: this.mcu_num_blocks)
csel = this.scan_comps_cselector[this.mcu_blocks_sselector[b]]
stride = this.components_workbuf_widths[csel] as base.u64
offset = this.mcu_blocks_offset[b] +
((this.mcu_blocks_mx_mul[b] as base.u64) * (mx as base.u64)) +
((this.mcu_blocks_my_mul[b] as base.u64) * (my as base.u64))
if offset <= args.workbuf.length() {
this.decode_idct!(
dst_buffer: args.workbuf[offset ..],
dst_stride: stride,
b: b,
q: this.components_tq[csel] as base.u32)
}
b += 1
} endwhile
}
// Check for a restart.
@@ -1715,7 +1689,6 @@ pri func decoder.apply_progressive_idct!(workbuf: slice base.u8) {
this.decode_idct!(
dst_buffer: args.workbuf[offset ..],
dst_stride: stride,
b: 0,
q: this.components_tq[csel] as base.u32)
}
+27 -3
View File
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_mcu!() base.u32,
pri func decoder.decode_mcu!(workbuf: slice base.u8, mx: base.u32[..= 0x1FFF], my: base.u32[..= 0x1FFF]) base.u32,
choosy,
{
var ret : base.u32
@@ -51,6 +51,10 @@ pri func decoder.decode_mcu!() base.u32,
var z : base.u32[..= 78] // 78 = 63 + 15.
var mcb : base.u32[..= 9]
var stride : base.u64[..= 0x1_0008]
var offset : base.u64
bits = this.bitstream_bits
n_bits = this.bitstream_n_bits
@@ -65,6 +69,8 @@ pri func decoder.decode_mcu!() base.u32,
while.dc_component this.mcu_zig_index <= 0,
inv this.mcu_current_block < 10,
{
this.mcu_blocks[.. 1].bulk_memset!(byte_value: 0)
// Load at least 31 bits, by keeping 56 .. 64 loaded.
if r.length() < 8 {
ret = 1 // Request another fill_bitstream call.
@@ -118,7 +124,7 @@ pri func decoder.decode_mcu!() base.u32,
n_bits ~mod-= dc_symbol
csel = this.scan_comps_cselector[this.mcu_blocks_sselector[this.mcu_current_block]]
this.mcu_previous_dc_values[csel] ~mod+= (dc_value & 0xFFFF) as base.u16
this.mcu_blocks[this.mcu_current_block][0] =
this.mcu_blocks[0][0] =
this.mcu_previous_dc_values[csel]
this.mcu_zig_index = 1
@@ -183,7 +189,7 @@ pri func decoder.decode_mcu!() base.u32,
bits ~mod<<= ac_ssss
n_bits ~mod-= ac_ssss
z = this.mcu_zig_index + ac_rrrr
this.mcu_blocks[this.mcu_current_block][UNZIG[z]] =
this.mcu_blocks[0][UNZIG[z]] =
(ac_value & 0xFFFF) as base.u16
if (z + 1) > 63 {
break.ac_components
@@ -200,7 +206,25 @@ pri func decoder.decode_mcu!() base.u32,
this.mcu_zig_index = 0
assert this.mcu_current_block < 10
mcb = this.mcu_current_block
this.mcu_current_block += 1
if this.test_only_interrupt_decode_mcu {
break.goto_done
}
// Apply IDCT.
csel = this.scan_comps_cselector[this.mcu_blocks_sselector[mcb]]
stride = this.components_workbuf_widths[csel] as base.u64
offset = this.mcu_blocks_offset[mcb] +
((this.mcu_blocks_mx_mul[mcb] as base.u64) * (args.mx as base.u64)) +
((this.mcu_blocks_my_mul[mcb] as base.u64) * (args.my as base.u64))
if offset <= args.workbuf.length() {
this.decode_idct!(
dst_buffer: args.workbuf[offset ..],
dst_stride: stride,
q: this.components_tq[csel] as base.u32)
}
} endwhile.block
this.mcu_current_block = 0
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_mcu_progressive_ac_high_bits!() base.u32 {
pri func decoder.decode_mcu_progressive_ac_high_bits!(workbuf: slice base.u8, mx: base.u32[..= 0x1FFF], my: base.u32[..= 0x1FFF]) base.u32 {
var ret : base.u32
var bits : base.u64
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_mcu_progressive_ac_low_bit!() base.u32 {
pri func decoder.decode_mcu_progressive_ac_low_bit!(workbuf: slice base.u8, mx: base.u32[..= 0x1FFF], my: base.u32[..= 0x1FFF]) base.u32 {
var ret : base.u32
var bits : base.u64
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_mcu_progressive_dc_high_bits!() base.u32 {
pri func decoder.decode_mcu_progressive_dc_high_bits!(workbuf: slice base.u8, mx: base.u32[..= 0x1FFF], my: base.u32[..= 0x1FFF]) base.u32 {
var ret : base.u32
var bits : base.u64
@@ -12,7 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pri func decoder.decode_mcu_progressive_dc_low_bit!() base.u32 {
pri func decoder.decode_mcu_progressive_dc_low_bit!(workbuf: slice base.u8, mx: base.u32[..= 0x1FFF], my: base.u32[..= 0x1FFF]) base.u32 {
var ret : base.u32
var bits : base.u64
+15 -17
View File
@@ -579,15 +579,14 @@ test_wuffs_jpeg_decode_idct() {
&dec, sizeof dec, WUFFS_VERSION,
WUFFS_INITIALIZE__DEFAULT_OPTIONS));
const uint32_t b = 0;
memcpy(&dec.private_data.f_mcu_blocks[b], mcu_block, sizeof(mcu_block));
memcpy(&dec.private_data.f_mcu_blocks[0], mcu_block, sizeof(mcu_block));
const uint32_t q = 0;
memcpy(&dec.private_impl.f_quant_tables[q], quant_table, sizeof(quant_table));
uint8_t dst_array[64] = {0};
wuffs_jpeg__decoder__decode_idct(
&dec, wuffs_base__make_slice_u8(&dst_array[0], 64), 8, b, q);
&dec, wuffs_base__make_slice_u8(&dst_array[0], 64), 8, q);
wuffs_base__io_buffer have =
wuffs_base__ptr_u8__reader(&dst_array[0], 64, true);
@@ -651,14 +650,8 @@ test_wuffs_jpeg_decode_mcu() {
}
}
// Decode and compare-to-golden the first MCU (Minimum Coded Unit).
if (wuffs_jpeg__decoder__decode_mcu(&dec)) {
RETURN_FAIL("decode_mcu failed");
}
// 4:2:0 chroma subsampling means that the MCU has 4 Y, 1 Cb and 1 Cr blocks.
uint16_t wants[10][64] = {
uint16_t wants[6][64] = {
{
0xFFC9, 0xFFD8, 0x0014, 0xFFF7, 0x0002, 0x0000, 0x0000, 0x0000, //
0x006A, 0xFFE3, 0x001C, 0xFFF9, 0x0002, 0x0000, 0x0000, 0x0000, //
@@ -719,16 +712,21 @@ test_wuffs_jpeg_decode_mcu() {
0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, //
0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, //
},
{0}, //
{0}, //
{0}, //
{0}, //
};
for (int b = 0; b < 10; b++) {
// Decode and compare-to-golden the first MCU (Minimum Coded Unit).
dec.private_impl.f_test_only_interrupt_decode_mcu = true;
for (int b = 0; b < 6; b++) {
const uint32_t mx = 0;
const uint32_t my = 0;
if (wuffs_jpeg__decoder__decode_mcu(&dec, wuffs_base__empty_slice_u8(), mx,
my)) {
RETURN_FAIL("decode_mcu failed");
}
wuffs_base__io_buffer have = wuffs_base__ptr_u8__reader( //
(void*)(&dec.private_data.f_mcu_blocks[b]), //
sizeof(dec.private_data.f_mcu_blocks[b]), true);
(void*)(&dec.private_data.f_mcu_blocks[0]), //
sizeof(dec.private_data.f_mcu_blocks[0]), true);
wuffs_base__io_buffer want = wuffs_base__ptr_u8__reader( //
(void*)(&wants[b]), //
sizeof(wants[b]), true);