/* * Copyright (c) 2016-2016 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * * This file contains Original Code and/or Modifications of Original Code * as defined in and that are subject to the Apple Public Source License * Version 2.0 (the 'License'). You may not use this file except in * compliance with the License. The rights granted to you under the License * may not be used to create, or enable the creation or redistribution of, * unlawful or unlicensed copies of an Apple operating system, or to * circumvent, violate, or enable the circumvention or violation of, any * terms of an Apple operating system software license agreement. * * Please obtain a copy of the License at * http://www.opensource.apple.com/apsl/ and read it before using this file. * * The Original Code and all software distributed under the License are * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. * Please see the License for the specific language governing rights and * limitations under the License. * * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ */ #include #include #include #if LZ4_ENABLE_ASSEMBLY_ENCODE_ARM64 /* void lz4_encode_2gb(uint8_t ** dst_ptr, size_t dst_size, const uint8_t ** src_ptr, const uint8_t * src_begin, size_t src_size, lz4_hash_entry_t hash_table[LZ4_COMPRESS_HASH_ENTRIES], int skip_final_literals) */ .globl _lz4_encode_2gb #define dst_ptr x0 #define dst_size x1 #define src_ptr x2 #define src_begin x3 #define src_size x4 #define hash_table x5 #define skip_final_literals x6 .text .p2align 4 _lz4_encode_2gb: // esteblish frame ARM64_STACK_PROLOG stp fp, lr, [sp, #-16]! mov fp, sp stp x19, x20, [sp, #-16]! stp x21, x22, [sp, #-16]! stp x23, x24, [sp, #-16]! stp x25, x26, [sp, #-16]! stp x27, x28, [sp, #-16]! // constant registers adr x7, L_constant ldr w28, [x7, #4] // x28 = 0x80808081 (magic number to cmopute 1/255) ldr w7, [x7] // x7 = LZ4_COMPRESS_HASH_MULTIPLY mov x27, #-1 // x27 = 0xffffffffffffffff dup.4s v1, w27 // q1 = {0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff} // x9 - is current dst // x10 - dst_end - safety_margin ldr x9, [x0] // dst add x10, x9, x1 // dst_end sub x10, x10, #LZ4_GOFAST_SAFETY_MARGIN // dst_end - safety_margin cmp x10, x9 // if dst_size < safety_margin abort b.lt L_done // x11 - is current src // x12 - is src_end - safety margin ldr x11, [x2] // src add x12, x11, x4 // src_end sub x12, x12, #LZ4_GOFAST_SAFETY_MARGIN // src_end - safety_margin cmp x12, x11 // if src_size < safety_margin skip to trailing_literals b.lt L_trailing_literals // this block search for the next available match // set match_begin to current src (which is also where last match ended) L_search_next_available_match: mov x13, x11 // match_begin = src sub x14, x13, x3 // match_postion = match_begin - src_begin // compute hash value for the next 5 "quads" // hash distance need to be 0 < D < 0x10000 L_hash_match: ldr x15, [x13] // match_first_4_bytes umull x20, w7, w15 // match_bytes * LZ4_COMPRESS_HASH_MULTIPLY lsr w20, w20, #LZ4_COMPRESS_HASH_SHIFT // use LZ4_COMPRESS_HASH_BITS MSbits as index add x20, x5, x20, lsl #3 // hash_table_entry ptr (hash + 8*index) ldp w19, w22, [x20] // read entry values (w19 - pos, w22 - 4 bytes at pos) stp w14, w15, [x20] // write entry values (w14 - current pos, w15 - current 4 bytes) add x26, x14, #1 // next_match pos lsr x25, x15, #8 // next_match_first_4_bytes umull x21, w7, w25 // match_bytes * LZ4_COMPRESS_HASH_MULTIPLY lsr w21, w21, #LZ4_COMPRESS_HASH_SHIFT // use LZ4_COMPRESS_HASH_BITS MSbits as index add x21, x5, x21, lsl #3 // hash_table_entry ptr (hash + 8*index) ldp w23, w24, [x21] // read entry values (w23 - pos, w24 - 4 bytes at pos) stp w26, w25, [x21] // write entry values (w26 - next pos, w25 - next 4 bytes) cmp w15, w22 b.ne L_try_next_match_0 // compare the 4 bytes to see if there is a match sub w19, w14, w19 // x19 - match_dist (current_pos - match_pos) cmp w19, #0x10000 ccmp w19, #0, #0xf, lo b.eq L_try_next_match_0 // verify the 0 < dist < 0x10000 b L_found_valid_match L_try_next_match_0: add x13, x13, #1 add x14, x14, #1 add x26, x14, #1 // next_match pos lsr x15, x15, #16 // next_match_first_4_bytes umull x20, w7, w15 // match_bytes * LZ4_COMPRESS_HASH_MULTIPLY lsr w20, w20, #LZ4_COMPRESS_HASH_SHIFT // use LZ4_COMPRESS_HASH_BITS MSbits as index add x20, x5, x20, lsl #3 // hash_table_entry ptr (hash + 8*index) ldp w21, w22, [x20] // read entry values (w19 - pos, w22 - 4 bytes at pos) stp w26, w15, [x20] // write entry values (w14 - current pos, w15 - current 4 bytes) cmp w25, w24 b.ne L_try_next_match_1 // compare the 4 bytes to see if there is a match sub w19, w14, w23 // x19 - match_dist (current_pos - match_pos) cmp w19, #0x10000 ccmp w19, #0, #0xf, lo b.eq L_try_next_match_1 // verify the 0 < dist < 0x10000 b L_found_valid_match L_try_next_match_1: add x13, x13, #1 add x14, x14, #1 add x26, x14, #1 // next_match pos lsr x25, x15, #8 // next_match_first_4_bytes umull x20, w7, w25 // match_bytes * LZ4_COMPRESS_HASH_MULTIPLY lsr w20, w20, #LZ4_COMPRESS_HASH_SHIFT // use LZ4_COMPRESS_HASH_BITS MSbits as index add x20, x5, x20, lsl #3 // hash_table_entry ptr (hash + 8*index) ldp w23, w24, [x20] // read entry values (w23 - pos, w24 - 4 bytes at pos) stp w26, w25, [x20] // write entry values (w26 - next pos, w25 - next 4 bytes) cmp w15, w22 b.ne L_try_next_match_2 // compare the 4 bytes to see if there is a match sub w19, w14, w21 // x19 - match_dist (current_pos - match_pos) cmp w19, #0x10000 ccmp w19, #0, #0xf, lo b.eq L_try_next_match_2 // verify the 0 < dist < 0x10000 b L_found_valid_match L_try_next_match_2: add x13, x13, #1 add x14, x14, #1 add x26, x14, #1 // next_match pos lsr x15, x15, #16 // next_match_first_4_bytes umull x20, w7, w15 // match_bytes * LZ4_COMPRESS_HASH_MULTIPLY lsr w20, w20, #LZ4_COMPRESS_HASH_SHIFT // use LZ4_COMPRESS_HASH_BITS MSbits as index add x20, x5, x20, lsl #3 // hash_table_entry ptr (hash + 8*index) ldp w21, w22, [x20] // read entry values (w19 - pos, w22 - 4 bytes at pos) stp w26, w15, [x20] // write entry values (w14 - current pos, w15 - current 4 bytes) cmp w25, w24 b.ne L_try_next_match_3 // compare the 4 bytes to see if there is a match sub w19, w14, w23 // x19 - match_dist (current_pos - match_pos) cmp w19, #0x10000 ccmp w19, #0, #0xf, lo b.eq L_try_next_match_3 // verify the 0 < dist < 0x10000 b L_found_valid_match L_try_next_match_3: add x13, x13, #1 add x14, x14, #1 cmp w15, w22 b.ne L_try_next_matchs // compare the 4 bytes to see if there is a match sub w19, w14, w21 // x19 - match_dist (current_pos - match_pos) cmp w19, #0x10000 ccmp w19, #0, #0xf, lo b.eq L_try_next_matchs // verify the 0 < dist < 0x10000 b L_found_valid_match // this block exapnd the valid match as much as possible // first it try to expand the match forward // next it try to expand the match backword L_found_valid_match: add x20, x13, #4 // match_end = match_begin+4 (already confirmd the first 4 bytes) sub x21, x20, x19 // ref_end = match_end - dist L_found_valid_match_expand_forward_loop: ldr x22, [x20], #8 // load match_current_8_bytes (safe to load becasue of safety margin) ldr x23, [x21], #8 // load ref_current_8_bytes cmp x22, x23 b.ne L_found_valid_match_expand_forward_partial cmp x20, x12 // check if match_end reached src_end b.lo L_found_valid_match_expand_forward_loop b L_found_valid_match_expand_backward L_found_valid_match_expand_forward_partial: sub x20, x20, #8 // revert match_end by 8 and compute actual match of current 8 bytes eor x22, x22, x23 // compare the bits using xor rbit x22, x22 // revert the bits to use clz (the none equivalent bytes would have at least 1 set bit) clz x22, x22 // after the revrse for every equal prefix byte clz would count 8 add x20, x20, x22, lsr #3 // add the actual number of matching bytes is (clz result)>>3 L_found_valid_match_expand_backward: sub x15, x13, x19 // ref_begin = match_begin - dist L_found_valid_match_expand_backward_loop: cmp x13, x11 // check if match_begin reached src (previous match end) ccmp x15, x3, #0xd, gt // check if ref_begin reached src_begin b.le L_found_valid_match_emit_match ldrb w22, [x13, #-1]! // load match_current_8_bytes (safe to load becasue of safety margin) ldrb w23, [x15, #-1]! // load ref_current_8_bytes cmp w22, w23 b.eq L_found_valid_match_expand_backward_loop add x13, x13, #1 // revert x13, last compare didn't match // this block write the match into dst // it write the ML token [extra L tokens] [literals] <2byte dist> [extar M tokens] // it update src & dst positions and progress to L_search_next_available_match L_found_valid_match_emit_match: sub x21, x20, x13 // match_length - match_end - match_begin sub x21, x21, #4 // match_length - 4 (first 4 bytes are guaranteed) sub x22, x13, x11 // literals_length = match_begin - src // compute sub x26, x10, x9 // dst_remaining_space = dst_end - dst sub x26, x26, x22 // dst_remaining_space -= literals_length subs x26, x26, #3 // dst_remaining_space -= 2_dist_bytes + L/M_token b.lo L_done // exit if dst isn't sufficent and x23, x21, #0xf // store M 4 LSbits add x23, x23, x22, lsl #4 // add L 4 LSbits add x15, x9, #1 // tmp_dst = dst + 1 cmp x22, #15 // if L >= 15 need to write more L tokens b.lo L_found_valid_match_copy_literals orr x23, x23, #0xf0 // update L/M token to be 0xfM sub x24, x22, #15 // reduce 15 from number_of_literals sub x26, x26, #1 // check if there is space for the extra L token b.lo L_done cmp x24, #255 // check if need to compute number of 255 tokens b.lo L_found_valid_match_skip_L_255_tokens umull x25, w24, w28 // x25 - (literals_to_token * 1_DIV_255_magic_number) lsr x25, x25, #39 // x25 - number_of_255_tokens = (literals_to_token * 1_DIV_255_magic_number)>>39 subs x26, x26, x25 // check if there is sufficent space for the 255_tokens b.lo L_done mov x13, #255 umsubl x24, w25, w13, x24 // x24 - value_of_remainder_token = literals_to_token - (number_of_255_tokens*255) L_found_valid_match_L_255_tokens_loop: str q1, [x15], #16 // store 16 255 tokens into dst_tmp. safe to store because dst has safety_margin subs x25, x25, #16 // check if there are any 255 token left after current 16 b.hi L_found_valid_match_L_255_tokens_loop add x15, x15, x25 // revert tmp_dst if written too many 255 tokens. L_found_valid_match_skip_L_255_tokens: strb w24, [x15], #1 // write last L token L_found_valid_match_copy_literals: ldr q0, [x11], #16 // load current 16 literals. (safe becasue src_end has safety margin) str q0, [x15], #16 // store current 16 literals. (safe becasue dst_end has safety margin) subs x22, x22, #16 b.gt L_found_valid_match_copy_literals add x15, x15, x22 // revert tmp_dst if written too many literals strh w19, [x15], #2 // store dist bytes cmp x21, #15 // if M >= 15 need to write more M tokens b.lo L_found_valid_match_finish_writing_match orr x23, x23, #0xf // update L/M token to be 0xLf sub x24, x21, #15 // reduce 15 from match_length sub x26, x26, #1 // check if there is space for the extra M token b.lo L_done cmp x24, #255 // check if need to compute number of 255 tokens b.lo L_found_valid_match_skip_M_255_tokens umull x25, w24, w28 // x25 - (match_length * 1_DIV_255_magic_number) lsr x25, x25, #39 // x25 - number_of_255_tokens = (match_length * 1_DIV_255_magic_number)>>39 subs x26, x26, x25 // check if there is sufficent space for the 255_tokens b.lo L_done mov x13, #255 umsubl x24, w25, w13, x24 // x24 - value_of_remainder_token = literals_to_token - (match_length*255) L_found_valid_match_M_255_tokens_loop: str q1, [x15], #16 // store 16 255 tokens into dst_tmp. safe to store because dst has safety_margin subs x25, x25, #16 // check if there are any 255 token left after current 16 b.hi L_found_valid_match_M_255_tokens_loop add x15, x15, x25 // revert tmp_dst if written too many 255 tokens. L_found_valid_match_skip_M_255_tokens: strb w24, [x15], #1 // write last M token L_found_valid_match_finish_writing_match: strb w23, [x9] // store first token of match in dst mov x9, x15 // update dst to last postion written mov x11, x20 // update src to match_end (last byte that was encoded) cmp x11, x12 // check if src reached src_end ccmp x9, x10, #9, lt // check if dst reached dst_end b.ge L_trailing_literals b L_search_next_available_match // !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! // attempted to hash three quad values from the end of each emited match // this eneded up being slower and less compression (???) // this block set match_begin and pos for next hash search and // compute the hash values for the last 3 bytes of currently emited match // only need to comute these hash becasue other "quads" were hashed when the original // data was read. L_try_next_matchs: add x13, x13, #1 // move to next match add x14, x14, #1 // update next match pos cmp x13, x12 // check match_begin didn't reach src_end b.lo L_hash_match L_trailing_literals: // unless skip_final_literals is set // write the trailing bytes as literals // traliing bytes include the whole src (with the safty margin) // need to verify whole dst (withthe safty margin) has sufficent space tst x6, x6 b.ne L_done // if skip_final_literals is set skip writing them add x12, x12, #LZ4_GOFAST_SAFETY_MARGIN // add safety_margin subs x13, x12, x11 // remaining_src b.eq L_done // finish if there are 0 trailing literals add x10, x10, #LZ4_GOFAST_SAFETY_MARGIN // add safety_margin sub x14, x10, x9 // remaining dst (dst_end - dst) sub x14, x14, #1 // 1 byte is needed at least to write literals token subs x14, x14, x13 // finish if dst can't contain all remaining literals + 1 literals token b.le L_done // (need to verify that it has room for literals tokens cmp x13, #15 b.lt L_trailing_literals_store_less_than_15_literals subs x14, x14, #1 // 1-extra byte is needed for literals tokens b.mi L_done mov w15, #0xf0 strb w15, [x9], #1 // write literals first token (Important !!! if 255 tokens exist but dst isn't sufficent need to revert dst by 1) sub x15, x13, #15 cmp x15, #255 b.lo L_trailing_literals_no_255_tokens umull x19, w15, w28 // x19 - (literals_to_token * 1_DIV_255_magic_number) lsr x19, x19, #39 // x19 - number_of_255_tokens = (literals_to_token * 1_DIV_255_magic_number)>>39 subs x14, x14, x19 b.mi L_revert_x9_and_done mov x26, #255 umsubl x15, w26, w19, x15 // x15 - value_of_remainder_token = literals_to_token - (number_of_255_tokens*255) L_tariling_literals_write_16_255_tokens: str q1, [x9], #16 // store 16 255 tokens each iteration (this is safe becasue there is space for 15 or more literals + remainder token) subs x19, x19, #16 b.gt L_tariling_literals_write_16_255_tokens add x9, x9, x19 // fixes dst to actual number of tokens (x19 might not be a mulitple of 16) L_trailing_literals_no_255_tokens: strb w15, [x9], #1 // store remainder_token lsr x14, x13, #4 // check if there are more than 16 literals left to be written tst x14, x14 b.eq L_trailing_literals_copy_less_than_16_literals L_trailing_literals_copy_16_literals: ldr q0, [x11], #16 // load current_16_literals str q0, [ x9], #16 // *dst16++ = current_16_literals subs x14, x14, #1 b.gt L_trailing_literals_copy_16_literals cmp x11, x12 b.lo L_trailing_literals_copy_less_than_16_literals b L_done L_trailing_literals_store_less_than_15_literals: lsl x14, x13, #4 // literals_only_token is 0xL0 (where L is 4 bits) strb w14, [x9], #1 // *dst++ = literals_only_token L_trailing_literals_copy_less_than_16_literals: ldrb w13, [x11], #1 // load current_literal strb w13, [ x9], #1 // *dst++ = current_literal cmp x11, x12 b.lo L_trailing_literals_copy_less_than_16_literals // this block upadte dst & src pointers and remove frame L_done: str x9, [x0] str x11, [x2] ldp x27, x28, [sp], #16 ldp x25, x26, [sp], #16 ldp x23, x24, [sp], #16 ldp x21, x22, [sp], #16 ldp x19, x20, [sp], #16 // clear frame ldp fp, lr, [sp], #16 ARM64_STACK_EPILOG L_revert_x9_and_done: sub x9, x9, #1 b L_done .p2align 2 L_constant: .long LZ4_COMPRESS_HASH_MULTIPLY .long 0x80808081 #endif