1*4882a593Smuzhiyun/* 2*4882a593Smuzhiyun * Copyright (C) 2004, 2007-2010, 2011-2014 Synopsys, Inc. All rights reserved. 3*4882a593Smuzhiyun * 4*4882a593Smuzhiyun * SPDX-License-Identifier: GPL-2.0+ 5*4882a593Smuzhiyun */ 6*4882a593Smuzhiyun 7*4882a593Smuzhiyun/* 8*4882a593Smuzhiyun * If dst and src are 4 byte aligned, copy 8 bytes at a time. 9*4882a593Smuzhiyun * If the src is 4, but not 8 byte aligned, we first read 4 bytes to get 10*4882a593Smuzhiyun * it 8 byte aligned. Thus, we can do a little read-ahead, without 11*4882a593Smuzhiyun * dereferencing a cache line that we should not touch. 12*4882a593Smuzhiyun * Note that short and long instructions have been scheduled to avoid 13*4882a593Smuzhiyun * branch stalls. 14*4882a593Smuzhiyun * The beq_s to r3z could be made unaligned & long to avoid a stall 15*4882a593Smuzhiyun * there, but it is not likely to be taken often, and it would also be likely 16*4882a593Smuzhiyun * to cost an unaligned mispredict at the next call. 17*4882a593Smuzhiyun */ 18*4882a593Smuzhiyun 19*4882a593Smuzhiyun.global strcpy 20*4882a593Smuzhiyun.align 4 21*4882a593Smuzhiyunstrcpy: 22*4882a593Smuzhiyun or %r2, %r0, %r1 23*4882a593Smuzhiyun bmsk_s %r2, %r2, 1 24*4882a593Smuzhiyun brne.d %r2, 0, charloop 25*4882a593Smuzhiyun mov_s %r10, %r0 26*4882a593Smuzhiyun ld_s %r3, [%r1, 0] 27*4882a593Smuzhiyun mov %r8, 0x01010101 28*4882a593Smuzhiyun bbit0.d %r1, 2, loop_start 29*4882a593Smuzhiyun ror %r12, %r8 30*4882a593Smuzhiyun sub %r2, %r3, %r8 31*4882a593Smuzhiyun bic_s %r2, %r2, %r3 32*4882a593Smuzhiyun tst_s %r2,%r12 33*4882a593Smuzhiyun bne r3z 34*4882a593Smuzhiyun mov_s %r4,%r3 35*4882a593Smuzhiyun .balign 4 36*4882a593Smuzhiyunloop: 37*4882a593Smuzhiyun ld.a %r3, [%r1, 4] 38*4882a593Smuzhiyun st.ab %r4, [%r10, 4] 39*4882a593Smuzhiyunloop_start: 40*4882a593Smuzhiyun ld.a %r4, [%r1, 4] 41*4882a593Smuzhiyun sub %r2, %r3, %r8 42*4882a593Smuzhiyun bic_s %r2, %r2, %r3 43*4882a593Smuzhiyun tst_s %r2, %r12 44*4882a593Smuzhiyun bne_s r3z 45*4882a593Smuzhiyun st.ab %r3, [%r10, 4] 46*4882a593Smuzhiyun sub %r2, %r4, %r8 47*4882a593Smuzhiyun bic %r2, %r2, %r4 48*4882a593Smuzhiyun tst %r2, %r12 49*4882a593Smuzhiyun beq loop 50*4882a593Smuzhiyun mov_s %r3, %r4 51*4882a593Smuzhiyun#ifdef __LITTLE_ENDIAN__ 52*4882a593Smuzhiyunr3z: bmsk.f %r1, %r3, 7 53*4882a593Smuzhiyun lsr_s %r3, %r3, 8 54*4882a593Smuzhiyun#else /* __BIG_ENDIAN__ */ 55*4882a593Smuzhiyunr3z: lsr.f %r1, %r3, 24 56*4882a593Smuzhiyun asl_s %r3, %r3, 8 57*4882a593Smuzhiyun#endif /* _ENDIAN__ */ 58*4882a593Smuzhiyun bne.d r3z 59*4882a593Smuzhiyun stb.ab %r1, [%r10, 1] 60*4882a593Smuzhiyun j_s [%blink] 61*4882a593Smuzhiyun 62*4882a593Smuzhiyun .balign 4 63*4882a593Smuzhiyuncharloop: 64*4882a593Smuzhiyun ldb.ab %r3, [%r1, 1] 65*4882a593Smuzhiyun brne.d %r3, 0, charloop 66*4882a593Smuzhiyun stb.ab %r3, [%r10, 1] 67*4882a593Smuzhiyun j [%blink] 68