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