| /* Optimized memset for PowerPC405,440,464 (32-byte cacheline). |
| Copyright (C) 2012-2014 Free Software Foundation, Inc. |
| This file is part of the GNU C Library. |
| |
| The GNU C Library is free software; you can redistribute it and/or |
| modify it under the terms of the GNU Lesser General Public |
| License as published by the Free Software Foundation; either |
| version 2.1 of the License, or (at your option) any later version. |
| |
| The GNU C Library is distributed in the hope that it will be useful, |
| but WITHOUT ANY WARRANTY; without even the implied warranty of |
| MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| Lesser General Public License for more details. |
| |
| You should have received a copy of the GNU Lesser General Public |
| License along with the GNU C Library. If not, see |
| <http://www.gnu.org/licenses/>. */ |
| |
| #include <sysdep.h> |
| |
| /* memset |
| |
| r3:destination address and return address |
| r4:source integer to copy |
| r5:byte count |
| r11:sources integer to copy in all 32 bits of reg |
| r12:temp return address |
| |
| Save return address in r12 |
| If destinationn is unaligned and count is greater tha 255 bytes |
| set 0-3 bytes to make destination aligned |
| If count is greater tha 255 bytes and setting zero to memory |
| use dbcz to set memeory when we can |
| otherwsie do the follwoing |
| If 16 or more words to set we use 16 word copy loop. |
| Finaly we set 0-15 extra bytes with string store. */ |
| |
| EALIGN (memset, 5, 0) |
| rlwinm r11,r4,0,24,31 |
| rlwimi r11,r4,8,16,23 |
| rlwimi r11,r11,16,0,15 |
| addi r12,r3,0 |
| cmpwi r5,0x00FF |
| ble L(preword8_count_loop) |
| cmpwi r4,0x00 |
| beq L(use_dcbz) |
| neg r6,r3 |
| clrlwi. r6,r6,30 |
| beq L(preword8_count_loop) |
| addi r8,0,1 |
| mtctr r6 |
| subi r3,r3,1 |
| |
| L(unaligned_bytecopy_loop): |
| stbu r11,0x1(r3) |
| subf. r5,r8,r5 |
| beq L(end_memset) |
| bdnz L(unaligned_bytecopy_loop) |
| addi r3,r3,1 |
| |
| L(preword8_count_loop): |
| srwi. r6,r5,4 |
| beq L(preword2_count_loop) |
| mtctr r6 |
| addi r3,r3,-4 |
| mr r8,r11 |
| mr r9,r11 |
| mr r10,r11 |
| |
| L(word8_count_loop_no_dcbt): |
| stwu r8,4(r3) |
| stwu r9,4(r3) |
| subi r5,r5,0x10 |
| stwu r10,4(r3) |
| stwu r11,4(r3) |
| bdnz L(word8_count_loop_no_dcbt) |
| addi r3,r3,4 |
| |
| L(preword2_count_loop): |
| clrlwi. r7,r5,28 |
| beq L(end_memset) |
| mr r8,r11 |
| mr r9,r11 |
| mr r10,r11 |
| mtxer r7 |
| stswx r8,0,r3 |
| |
| L(end_memset): |
| addi r3,r12,0 |
| blr |
| |
| L(use_dcbz): |
| neg r6,r3 |
| clrlwi. r7,r6,28 |
| beq L(skip_string_loop) |
| mr r8,r11 |
| mr r9,r11 |
| mr r10,r11 |
| subf r5,r7,r5 |
| mtxer r7 |
| stswx r8,0,r3 |
| add r3,r3,r7 |
| |
| L(skip_string_loop): |
| clrlwi r8,r6,27 |
| srwi. r8,r8,4 |
| beq L(dcbz_pre_loop) |
| mtctr r8 |
| |
| L(word_loop): |
| stw r11,0(r3) |
| subi r5,r5,0x10 |
| stw r11,4(r3) |
| stw r11,8(r3) |
| stw r11,12(r3) |
| addi r3,r3,0x10 |
| bdnz L(word_loop) |
| |
| L(dcbz_pre_loop): |
| srwi r6,r5,5 |
| mtctr r6 |
| addi r7,0,0 |
| |
| L(dcbz_loop): |
| dcbz r3,r7 |
| addi r3,r3,0x20 |
| subi r5,r5,0x20 |
| bdnz L(dcbz_loop) |
| srwi. r6,r5,4 |
| beq L(postword2_count_loop) |
| mtctr r6 |
| |
| L(postword8_count_loop): |
| stw r11,0(r3) |
| subi r5,r5,0x10 |
| stw r11,4(r3) |
| stw r11,8(r3) |
| stw r11,12(r3) |
| addi r3,r3,0x10 |
| bdnz L(postword8_count_loop) |
| |
| L(postword2_count_loop): |
| clrlwi. r7,r5,28 |
| beq L(end_memset) |
| mr r8,r11 |
| mr r9,r11 |
| mr r10,r11 |
| mtxer r7 |
| stswx r8,0,r3 |
| b L(end_memset) |
| END (memset) |
| libc_hidden_builtin_def (memset) |
