mirror of
https://github.com/Swiftgram/Telegram-iOS.git
synced 2025-07-04 10:30:42 +00:00
Update project
This commit is contained in:
Peter
parent
6c782fb009
commit
c8ede59f9f
@ -116,6 +116,7 @@
|
||||
D010E17A22C237CF009324D4 /* vbrush.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D010E06B22C237CE009324D4 /* vbrush.cpp */; };
|
||||
D010E17F22C23C42009324D4 /* LottieInstance.mm in Sources */ = {isa = PBXBuildFile; fileRef = D010E17E22C23C42009324D4 /* LottieInstance.mm */; };
|
||||
D010E18122C23D57009324D4 /* LottieInstance.h in Headers */ = {isa = PBXBuildFile; fileRef = D010E18022C23D57009324D4 /* LottieInstance.h */; settings = {ATTRIBUTES = (Public, ); }; };
|
||||
D05B16B022CC29F100B85A94 /* pixman-arma64-neon-asm.h in Headers */ = {isa = PBXBuildFile; fileRef = D05B16AE22CC29F100B85A94 /* pixman-arma64-neon-asm.h */; };
|
||||
/* End PBXBuildFile section */
|
||||
|
||||
/* Begin PBXFileReference section */
|
||||
@ -232,6 +233,8 @@
|
||||
D010E06B22C237CE009324D4 /* vbrush.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = vbrush.cpp; sourceTree = "<group>"; };
|
||||
D010E17E22C23C42009324D4 /* LottieInstance.mm */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.objcpp; path = LottieInstance.mm; sourceTree = "<group>"; };
|
||||
D010E18022C23D57009324D4 /* LottieInstance.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = LottieInstance.h; sourceTree = "<group>"; };
|
||||
D05B16AD22CC29F100B85A94 /* pixman-arma64-neon-asm.S */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.asm; path = "pixman-arma64-neon-asm.S"; sourceTree = "<group>"; };
|
||||
D05B16AE22CC29F100B85A94 /* pixman-arma64-neon-asm.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = "pixman-arma64-neon-asm.h"; sourceTree = "<group>"; };
|
||||
/* End PBXFileReference section */
|
||||
|
||||
/* Begin PBXFrameworksBuildPhase section */
|
||||
@ -453,6 +456,8 @@
|
||||
D010E03122C237CE009324D4 /* pixman */ = {
|
||||
isa = PBXGroup;
|
||||
children = (
|
||||
D05B16AE22CC29F100B85A94 /* pixman-arma64-neon-asm.h */,
|
||||
D05B16AD22CC29F100B85A94 /* pixman-arma64-neon-asm.S */,
|
||||
D010E03322C237CE009324D4 /* pixman-arm-neon-asm.h */,
|
||||
D010E03422C237CE009324D4 /* vregion.h */,
|
||||
D010E03522C237CE009324D4 /* pixman-arm-neon-asm.S */,
|
||||
@ -518,6 +523,7 @@
|
||||
D010E12D22C237CE009324D4 /* rapidjson.h in Headers */,
|
||||
D010E11E22C237CE009324D4 /* strtod.h in Headers */,
|
||||
D010E14A22C237CF009324D4 /* config.h in Headers */,
|
||||
D05B16B022CC29F100B85A94 /* pixman-arma64-neon-asm.h in Headers */,
|
||||
D010E17422C237CF009324D4 /* vstackallocator.h in Headers */,
|
||||
D010E10C22C237CE009324D4 /* lottieparser.h in Headers */,
|
||||
D010E15A22C237CF009324D4 /* vpathmesure.h in Headers */,
|
||||
|
||||
@ -0,0 +1,418 @@
|
||||
/*
|
||||
* Copyright © 2009 Nokia Corporation
|
||||
*
|
||||
* Permission is hereby granted, free of charge, to any person obtaining a
|
||||
* copy of this software and associated documentation files (the "Software"),
|
||||
* to deal in the Software without restriction, including without limitation
|
||||
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
|
||||
* and/or sell copies of the Software, and to permit persons to whom the
|
||||
* Software is furnished to do so, subject to the following conditions:
|
||||
*
|
||||
* The above copyright notice and this permission notice (including the next
|
||||
* paragraph) shall be included in all copies or substantial portions of the
|
||||
* Software.
|
||||
*
|
||||
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
||||
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
||||
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
|
||||
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
||||
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
|
||||
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
|
||||
* DEALINGS IN THE SOFTWARE.
|
||||
*
|
||||
* Author: Siarhei Siamashka (siarhei.siamashka@nokia.com)
|
||||
*/
|
||||
|
||||
/*
|
||||
* This file contains implementations of NEON optimized pixel processing
|
||||
* functions. There is no full and detailed tutorial, but some functions
|
||||
* (those which are exposing some new or interesting features) are
|
||||
* extensively commented and can be used as examples.
|
||||
*
|
||||
* You may want to have a look at the comments for following functions:
|
||||
* - pixman_composite_over_8888_0565_asm_neon
|
||||
* - pixman_composite_over_n_8_0565_asm_neon
|
||||
*/
|
||||
|
||||
/* Prevent the stack from becoming executable for no reason... */
|
||||
#if defined(__linux__) && defined(__ELF__)
|
||||
.section .note.GNU-stack,"",%progbits
|
||||
#endif
|
||||
|
||||
.text
|
||||
.arch armv8-a
|
||||
.altmacro
|
||||
.p2align 2
|
||||
|
||||
/* Supplementary macro for setting function attributes */
|
||||
.macro pixman_asm_function fname
|
||||
.func fname
|
||||
.global fname
|
||||
#ifdef __ELF__
|
||||
.hidden fname
|
||||
.type fname, %function
|
||||
#endif
|
||||
fname:
|
||||
.endm
|
||||
|
||||
#include "pixman-arma64-neon-asm.h"
|
||||
|
||||
/* Global configuration options and preferences */
|
||||
|
||||
/*
|
||||
* The code can optionally make use of unaligned memory accesses to improve
|
||||
* performance of handling leading/trailing pixels for each scanline.
|
||||
* Configuration variable RESPECT_STRICT_ALIGNMENT can be set to 0 for
|
||||
* example in linux if unaligned memory accesses are not configured to
|
||||
* generate.exceptions.
|
||||
*/
|
||||
.set RESPECT_STRICT_ALIGNMENT, 1
|
||||
|
||||
/*
|
||||
* Set default prefetch type. There is a choice between the following options:
|
||||
*
|
||||
* PREFETCH_TYPE_NONE (may be useful for the ARM cores where PLD is set to work
|
||||
* as NOP to workaround some HW bugs or for whatever other reason)
|
||||
*
|
||||
* PREFETCH_TYPE_SIMPLE (may be useful for simple single-issue ARM cores where
|
||||
* advanced prefetch intruduces heavy overhead)
|
||||
*
|
||||
* PREFETCH_TYPE_ADVANCED (useful for superscalar cores such as ARM Cortex-A8
|
||||
* which can run ARM and NEON instructions simultaneously so that extra ARM
|
||||
* instructions do not add (many) extra cycles, but improve prefetch efficiency)
|
||||
*
|
||||
* Note: some types of function can't support advanced prefetch and fallback
|
||||
* to simple one (those which handle 24bpp pixels)
|
||||
*/
|
||||
.set PREFETCH_TYPE_DEFAULT, PREFETCH_TYPE_ADVANCED
|
||||
|
||||
/* Prefetch distance in pixels for simple prefetch */
|
||||
.set PREFETCH_DISTANCE_SIMPLE, 64
|
||||
|
||||
/*
|
||||
* Implementation of pixman_composite_over_8888_0565_asm_neon
|
||||
*
|
||||
* This function takes a8r8g8b8 source buffer, r5g6b5 destination buffer and
|
||||
* performs OVER compositing operation. Function fast_composite_over_8888_0565
|
||||
* from pixman-fast-path.c does the same in C and can be used as a reference.
|
||||
*
|
||||
* First we need to have some NEON assembly code which can do the actual
|
||||
* operation on the pixels and provide it to the template macro.
|
||||
*
|
||||
* Template macro quite conveniently takes care of emitting all the necessary
|
||||
* code for memory reading and writing (including quite tricky cases of
|
||||
* handling unaligned leading/trailing pixels), so we only need to deal with
|
||||
* the data in NEON registers.
|
||||
*
|
||||
* NEON registers allocation in general is recommented to be the following:
|
||||
* v0, v1, v2, v3 - contain loaded source pixel data
|
||||
* v4, v5, v6, v7 - contain loaded destination pixels (if they are needed)
|
||||
* v24, v25, v26, v27 - contain loading mask pixel data (if mask is used)
|
||||
* v28, v29, v30, v31 - place for storing the result (destination pixels)
|
||||
*
|
||||
* As can be seen above, four 64-bit NEON registers are used for keeping
|
||||
* intermediate pixel data and up to 8 pixels can be processed in one step
|
||||
* for 32bpp formats (16 pixels for 16bpp, 32 pixels for 8bpp).
|
||||
*
|
||||
* This particular function uses the following registers allocation:
|
||||
* v0, v1, v2, v3 - contain loaded source pixel data
|
||||
* v4, v5 - contain loaded destination pixels (they are needed)
|
||||
* v28, v29 - place for storing the result (destination pixels)
|
||||
*/
|
||||
|
||||
/*
|
||||
* Step one. We need to have some code to do some arithmetics on pixel data.
|
||||
* This is implemented as a pair of macros: '*_head' and '*_tail'. When used
|
||||
* back-to-back, they take pixel data from {v0, v1, v2, v3} and {v4, v5},
|
||||
* perform all the needed calculations and write the result to {v28, v29}.
|
||||
* The rationale for having two macros and not just one will be explained
|
||||
* later. In practice, any single monolitic function which does the work can
|
||||
* be split into two parts in any arbitrary way without affecting correctness.
|
||||
*
|
||||
* There is one special trick here too. Common template macro can optionally
|
||||
* make our life a bit easier by doing R, G, B, A color components
|
||||
* deinterleaving for 32bpp pixel formats (and this feature is used in
|
||||
* 'pixman_composite_over_8888_0565_asm_neon' function). So it means that
|
||||
* instead of having 8 packed pixels in {v0, v1, v2, v3} registers, we
|
||||
* actually use v0 register for blue channel (a vector of eight 8-bit
|
||||
* values), v1 register for green, v2 for red and v3 for alpha. This
|
||||
* simple conversion can be also done with a few NEON instructions:
|
||||
*
|
||||
* Packed to planar conversion: // vuzp8 is a wrapper macro
|
||||
* vuzp8 v0, v1
|
||||
* vuzp8 v2, v3
|
||||
* vuzp8 v1, v3
|
||||
* vuzp8 v0, v2
|
||||
*
|
||||
* Planar to packed conversion: // vzip8 is a wrapper macro
|
||||
* vzip8 v0, v2
|
||||
* vzip8 v1, v3
|
||||
* vzip8 v2, v3
|
||||
* vzip8 v0, v1
|
||||
*
|
||||
* But pixel can be loaded directly in planar format using LD4 / b NEON
|
||||
* instruction. It is 1 cycle slower than LD1 / s, so this is not always
|
||||
* desirable, that's why deinterleaving is optional.
|
||||
*
|
||||
* But anyway, here is the code:
|
||||
*/
|
||||
|
||||
.macro pixman_composite_out_reverse_8888_8888_process_pixblock_head
|
||||
mvn v24.8b, v3.8b /* get inverted alpha */
|
||||
/* do alpha blending */
|
||||
umull v8.8h, v24.8b, v4.8b
|
||||
umull v9.8h, v24.8b, v5.8b
|
||||
umull v10.8h, v24.8b, v6.8b
|
||||
umull v11.8h, v24.8b, v7.8b
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_out_reverse_8888_8888_process_pixblock_tail
|
||||
urshr v14.8h, v8.8h, #8
|
||||
urshr v15.8h, v9.8h, #8
|
||||
urshr v16.8h, v10.8h, #8
|
||||
urshr v17.8h, v11.8h, #8
|
||||
raddhn v28.8b, v14.8h, v8.8h
|
||||
raddhn v29.8b, v15.8h, v9.8h
|
||||
raddhn v30.8b, v16.8h, v10.8h
|
||||
raddhn v31.8b, v17.8h, v11.8h
|
||||
.endm
|
||||
|
||||
/******************************************************************************/
|
||||
|
||||
.macro pixman_composite_over_8888_8888_process_pixblock_head
|
||||
pixman_composite_out_reverse_8888_8888_process_pixblock_head
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_over_8888_8888_process_pixblock_tail
|
||||
pixman_composite_out_reverse_8888_8888_process_pixblock_tail
|
||||
uqadd v28.8b, v0.8b, v28.8b
|
||||
uqadd v29.8b, v1.8b, v29.8b
|
||||
uqadd v30.8b, v2.8b, v30.8b
|
||||
uqadd v31.8b, v3.8b, v31.8b
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_over_8888_8888_process_pixblock_tail_head
|
||||
ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32
|
||||
urshr v14.8h, v8.8h, #8
|
||||
PF add PF_X, PF_X, #8
|
||||
PF tst PF_CTL, #0xF
|
||||
urshr v15.8h, v9.8h, #8
|
||||
urshr v16.8h, v10.8h, #8
|
||||
urshr v17.8h, v11.8h, #8
|
||||
PF beq 10f
|
||||
PF add PF_X, PF_X, #8
|
||||
PF sub PF_CTL, PF_CTL, #1
|
||||
10:
|
||||
raddhn v28.8b, v14.8h, v8.8h
|
||||
raddhn v29.8b, v15.8h, v9.8h
|
||||
PF cmp PF_X, ORIG_W
|
||||
raddhn v30.8b, v16.8h, v10.8h
|
||||
raddhn v31.8b, v17.8h, v11.8h
|
||||
uqadd v28.8b, v0.8b, v28.8b
|
||||
uqadd v29.8b, v1.8b, v29.8b
|
||||
uqadd v30.8b, v2.8b, v30.8b
|
||||
uqadd v31.8b, v3.8b, v31.8b
|
||||
fetch_src_pixblock
|
||||
PF lsl DUMMY, PF_X, #src_bpp_shift
|
||||
PF prfm PREFETCH_MODE, [PF_SRC, DUMMY]
|
||||
mvn v22.8b, v3.8b
|
||||
PF lsl DUMMY, PF_X, #dst_bpp_shift
|
||||
PF prfm PREFETCH_MODE, [PF_DST, DUMMY]
|
||||
st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32
|
||||
PF ble 10f
|
||||
PF sub PF_X, PF_X, ORIG_W
|
||||
10:
|
||||
umull v8.8h, v22.8b, v4.8b
|
||||
PF ble 10f
|
||||
PF subs PF_CTL, PF_CTL, #0x10
|
||||
10:
|
||||
umull v9.8h, v22.8b, v5.8b
|
||||
PF ble 10f
|
||||
PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift
|
||||
PF ldrsb DUMMY, [PF_SRC, DUMMY]
|
||||
PF add PF_SRC, PF_SRC, #1
|
||||
10:
|
||||
umull v10.8h, v22.8b, v6.8b
|
||||
PF ble 10f
|
||||
PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift
|
||||
PF ldrsb DUMMY, [PF_DST, DUMMY]
|
||||
PF add PF_DST, PF_DST, #1
|
||||
10:
|
||||
umull v11.8h, v22.8b, v7.8b
|
||||
.endm
|
||||
|
||||
generate_composite_function \
|
||||
pixman_composite_over_8888_8888_asm_neon, 32, 0, 32, \
|
||||
FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \
|
||||
8, /* number of pixels, processed in a single block */ \
|
||||
5, /* prefetch distance */ \
|
||||
default_init, \
|
||||
default_cleanup, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_head, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_tail, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_tail_head
|
||||
|
||||
generate_composite_function_single_scanline \
|
||||
pixman_composite_scanline_over_asm_neon, 32, 0, 32, \
|
||||
FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \
|
||||
8, /* number of pixels, processed in a single block */ \
|
||||
default_init, \
|
||||
default_cleanup, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_head, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_tail, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_tail_head
|
||||
|
||||
/******************************************************************************/
|
||||
|
||||
.macro pixman_composite_over_n_8888_process_pixblock_head
|
||||
/* deinterleaved source pixels in {v0, v1, v2, v3} */
|
||||
/* inverted alpha in {v24} */
|
||||
/* destination pixels in {v4, v5, v6, v7} */
|
||||
umull v8.8h, v24.8b, v4.8b
|
||||
umull v9.8h, v24.8b, v5.8b
|
||||
umull v10.8h, v24.8b, v6.8b
|
||||
umull v11.8h, v24.8b, v7.8b
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_over_n_8888_process_pixblock_tail
|
||||
urshr v14.8h, v8.8h, #8
|
||||
urshr v15.8h, v9.8h, #8
|
||||
urshr v16.8h, v10.8h, #8
|
||||
urshr v17.8h, v11.8h, #8
|
||||
raddhn v28.8b, v14.8h, v8.8h
|
||||
raddhn v29.8b, v15.8h, v9.8h
|
||||
raddhn v30.8b, v16.8h, v10.8h
|
||||
raddhn v31.8b, v17.8h, v11.8h
|
||||
uqadd v28.8b, v0.8b, v28.8b
|
||||
uqadd v29.8b, v1.8b, v29.8b
|
||||
uqadd v30.8b, v2.8b, v30.8b
|
||||
uqadd v31.8b, v3.8b, v31.8b
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_over_n_8888_process_pixblock_tail_head
|
||||
urshr v14.8h, v8.8h, #8
|
||||
urshr v15.8h, v9.8h, #8
|
||||
urshr v16.8h, v10.8h, #8
|
||||
urshr v17.8h, v11.8h, #8
|
||||
raddhn v28.8b, v14.8h, v8.8h
|
||||
raddhn v29.8b, v15.8h, v9.8h
|
||||
raddhn v30.8b, v16.8h, v10.8h
|
||||
raddhn v31.8b, v17.8h, v11.8h
|
||||
ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32
|
||||
uqadd v28.8b, v0.8b, v28.8b
|
||||
PF add PF_X, PF_X, #8
|
||||
PF tst PF_CTL, #0x0F
|
||||
PF beq 10f
|
||||
PF add PF_X, PF_X, #8
|
||||
PF sub PF_CTL, PF_CTL, #1
|
||||
10:
|
||||
uqadd v29.8b, v1.8b, v29.8b
|
||||
uqadd v30.8b, v2.8b, v30.8b
|
||||
uqadd v31.8b, v3.8b, v31.8b
|
||||
PF cmp PF_X, ORIG_W
|
||||
umull v8.8h, v24.8b, v4.8b
|
||||
PF lsl DUMMY, PF_X, #dst_bpp_shift
|
||||
PF prfm PREFETCH_MODE, [PF_DST, DUMMY]
|
||||
umull v9.8h, v24.8b, v5.8b
|
||||
PF ble 10f
|
||||
PF sub PF_X, PF_X, ORIG_W
|
||||
10:
|
||||
umull v10.8h, v24.8b, v6.8b
|
||||
PF subs PF_CTL, PF_CTL, #0x10
|
||||
umull v11.8h, v24.8b, v7.8b
|
||||
PF ble 10f
|
||||
PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift
|
||||
PF ldrsb DUMMY, [PF_DST, DUMMY]
|
||||
PF add PF_DST, PF_DST, #1
|
||||
10:
|
||||
st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_over_n_8888_init
|
||||
mov v3.s[0], w4
|
||||
dup v0.8b, v3.b[0]
|
||||
dup v1.8b, v3.b[1]
|
||||
dup v2.8b, v3.b[2]
|
||||
dup v3.8b, v3.b[3]
|
||||
mvn v24.8b, v3.8b /* get inverted alpha */
|
||||
.endm
|
||||
|
||||
generate_composite_function \
|
||||
pixman_composite_over_n_8888_asm_neon, 0, 0, 32, \
|
||||
FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \
|
||||
8, /* number of pixels, processed in a single block */ \
|
||||
5, /* prefetch distance */ \
|
||||
pixman_composite_over_n_8888_init, \
|
||||
default_cleanup, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_head, \
|
||||
pixman_composite_over_8888_8888_process_pixblock_tail, \
|
||||
pixman_composite_over_n_8888_process_pixblock_tail_head
|
||||
|
||||
/******************************************************************************/
|
||||
|
||||
.macro pixman_composite_src_n_8888_process_pixblock_head
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_n_8888_process_pixblock_tail
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_n_8888_process_pixblock_tail_head
|
||||
st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_n_8888_init
|
||||
mov v0.s[0], w4
|
||||
dup v3.2s, v0.s[0]
|
||||
dup v2.2s, v0.s[0]
|
||||
dup v1.2s, v0.s[0]
|
||||
dup v0.2s, v0.s[0]
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_n_8888_cleanup
|
||||
.endm
|
||||
|
||||
generate_composite_function \
|
||||
pixman_composite_src_n_8888_asm_neon, 0, 0, 32, \
|
||||
FLAG_DST_WRITEONLY, \
|
||||
8, /* number of pixels, processed in a single block */ \
|
||||
0, /* prefetch distance */ \
|
||||
pixman_composite_src_n_8888_init, \
|
||||
pixman_composite_src_n_8888_cleanup, \
|
||||
pixman_composite_src_n_8888_process_pixblock_head, \
|
||||
pixman_composite_src_n_8888_process_pixblock_tail, \
|
||||
pixman_composite_src_n_8888_process_pixblock_tail_head, \
|
||||
0, /* dst_w_basereg */ \
|
||||
0, /* dst_r_basereg */ \
|
||||
0, /* src_basereg */ \
|
||||
0 /* mask_basereg */
|
||||
|
||||
/******************************************************************************/
|
||||
|
||||
.macro pixman_composite_src_8888_8888_process_pixblock_head
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_8888_8888_process_pixblock_tail
|
||||
.endm
|
||||
|
||||
.macro pixman_composite_src_8888_8888_process_pixblock_tail_head
|
||||
st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32
|
||||
fetch_src_pixblock
|
||||
cache_preload 8, 8
|
||||
.endm
|
||||
|
||||
generate_composite_function \
|
||||
pixman_composite_src_8888_8888_asm_neon, 32, 0, 32, \
|
||||
FLAG_DST_WRITEONLY, \
|
||||
8, /* number of pixels, processed in a single block */ \
|
||||
10, /* prefetch distance */ \
|
||||
default_init, \
|
||||
default_cleanup, \
|
||||
pixman_composite_src_8888_8888_process_pixblock_head, \
|
||||
pixman_composite_src_8888_8888_process_pixblock_tail, \
|
||||
pixman_composite_src_8888_8888_process_pixblock_tail_head, \
|
||||
0, /* dst_w_basereg */ \
|
||||
0, /* dst_r_basereg */ \
|
||||
0, /* src_basereg */ \
|
||||
0 /* mask_basereg */
|
||||
|
||||
/******************************************************************************/
|
||||
File diff suppressed because it is too large
Load Diff
Loading…
x
Reference in New Issue
Block a user