1*4882a593Smuzhiyun/* SPDX-License-Identifier: GPL-2.0 */ 2*4882a593Smuzhiyun/* 3*4882a593Smuzhiyun * arch/alpha/lib/ev67-strlen.S 4*4882a593Smuzhiyun * 21264 version by Rick Gorton <rick.gorton@alpha-processor.com> 5*4882a593Smuzhiyun * 6*4882a593Smuzhiyun * Finds length of a 0-terminated string. Optimized for the 7*4882a593Smuzhiyun * Alpha architecture: 8*4882a593Smuzhiyun * 9*4882a593Smuzhiyun * - memory accessed as aligned quadwords only 10*4882a593Smuzhiyun * - uses bcmpge to compare 8 bytes in parallel 11*4882a593Smuzhiyun * 12*4882a593Smuzhiyun * Much of the information about 21264 scheduling/coding comes from: 13*4882a593Smuzhiyun * Compiler Writer's Guide for the Alpha 21264 14*4882a593Smuzhiyun * abbreviated as 'CWG' in other comments here 15*4882a593Smuzhiyun * ftp.digital.com/pub/Digital/info/semiconductor/literature/dsc-library.html 16*4882a593Smuzhiyun * Scheduling notation: 17*4882a593Smuzhiyun * E - either cluster 18*4882a593Smuzhiyun * U - upper subcluster; U0 - subcluster U0; U1 - subcluster U1 19*4882a593Smuzhiyun * L - lower subcluster; L0 - subcluster L0; L1 - subcluster L1 20*4882a593Smuzhiyun */ 21*4882a593Smuzhiyun#include <asm/export.h> 22*4882a593Smuzhiyun .set noreorder 23*4882a593Smuzhiyun .set noat 24*4882a593Smuzhiyun 25*4882a593Smuzhiyun .globl strlen 26*4882a593Smuzhiyun .ent strlen 27*4882a593Smuzhiyun .align 4 28*4882a593Smuzhiyunstrlen: 29*4882a593Smuzhiyun ldq_u $1, 0($16) # L : load first quadword ($16 may be misaligned) 30*4882a593Smuzhiyun lda $2, -1($31) # E : 31*4882a593Smuzhiyun insqh $2, $16, $2 # U : 32*4882a593Smuzhiyun andnot $16, 7, $0 # E : 33*4882a593Smuzhiyun 34*4882a593Smuzhiyun or $2, $1, $1 # E : 35*4882a593Smuzhiyun cmpbge $31, $1, $2 # E : $2 <- bitmask: bit i == 1 <==> i-th byte == 0 36*4882a593Smuzhiyun nop # E : 37*4882a593Smuzhiyun bne $2, $found # U : 38*4882a593Smuzhiyun 39*4882a593Smuzhiyun$loop: ldq $1, 8($0) # L : 40*4882a593Smuzhiyun addq $0, 8, $0 # E : addr += 8 41*4882a593Smuzhiyun cmpbge $31, $1, $2 # E : 42*4882a593Smuzhiyun beq $2, $loop # U : 43*4882a593Smuzhiyun 44*4882a593Smuzhiyun$found: 45*4882a593Smuzhiyun cttz $2, $3 # U0 : 46*4882a593Smuzhiyun addq $0, $3, $0 # E : 47*4882a593Smuzhiyun subq $0, $16, $0 # E : 48*4882a593Smuzhiyun ret $31, ($26) # L0 : 49*4882a593Smuzhiyun 50*4882a593Smuzhiyun .end strlen 51*4882a593Smuzhiyun EXPORT_SYMBOL(strlen) 52