From d71ff82e7019e91ae9f76af391e7f677d421e205 Mon Sep 17 00:00:00 2001 From: Mike Frysinger <vapier@gentoo.org> Date: Fri, 26 Aug 2005 00:01:54 +0000 Subject: import sparc-optimized string functions from glibc --- libc/string/Makefile | 2 +- libc/string/sparc/Makefile | 31 + libc/string/sparc/_glibc_inc.h | 30 + libc/string/sparc/bcopy.c | 1 + libc/string/sparc/bzero.c | 1 + libc/string/sparc/memchr.S | 13 + libc/string/sparc/memcpy.S | 19 + libc/string/sparc/memmove.c | 1 + libc/string/sparc/memset.S | 13 + libc/string/sparc/sparc32/memchr.S | 144 +++++ libc/string/sparc/sparc32/memcpy.S | 966 ++++++++++++++++++++++++++++ libc/string/sparc/sparc32/memset.S | 152 +++++ libc/string/sparc/sparc32/sparcv9b/memchr.S | 4 + libc/string/sparc/sparc32/sparcv9b/memcpy.S | 4 + libc/string/sparc/sparc32/sparcv9b/memset.S | 4 + libc/string/sparc/sparc32/sparcv9b/stpcpy.S | 4 + libc/string/sparc/sparc32/sparcv9b/strcat.S | 4 + libc/string/sparc/sparc32/sparcv9b/strchr.S | 4 + libc/string/sparc/sparc32/sparcv9b/strcmp.S | 4 + libc/string/sparc/sparc32/sparcv9b/strcpy.S | 4 + libc/string/sparc/sparc32/sparcv9b/strlen.S | 4 + libc/string/sparc/sparc32/stpcpy.S | 161 +++++ libc/string/sparc/sparc32/strcat.S | 346 ++++++++++ libc/string/sparc/sparc32/strchr.S | 280 ++++++++ libc/string/sparc/sparc32/strcmp.S | 255 ++++++++ libc/string/sparc/sparc32/strcpy.S | 272 ++++++++ libc/string/sparc/sparc32/strlen.S | 101 +++ libc/string/sparc/sparc64/memchr.S | 261 ++++++++ libc/string/sparc/sparc64/memcpy.S | 918 ++++++++++++++++++++++++++ libc/string/sparc/sparc64/memset.S | 313 +++++++++ libc/string/sparc/sparc64/sparcv9b/memcpy.S | 606 +++++++++++++++++ libc/string/sparc/sparc64/stpcpy.S | 272 ++++++++ libc/string/sparc/sparc64/strcat.S | 338 ++++++++++ libc/string/sparc/sparc64/strchr.S | 480 ++++++++++++++ libc/string/sparc/sparc64/strcmp.S | 278 ++++++++ libc/string/sparc/sparc64/strcpy.S | 243 +++++++ libc/string/sparc/sparc64/strlen.S | 172 +++++ libc/string/sparc/stpcpy.S | 13 + libc/string/sparc/strcat.S | 13 + libc/string/sparc/strchr.S | 13 + libc/string/sparc/strcmp.S | 13 + libc/string/sparc/strcpy.S | 13 + libc/string/sparc/strlen.S | 13 + libc/string/sparc/strrchr.c | 1 + 44 files changed, 6783 insertions(+), 1 deletion(-) create mode 100644 libc/string/sparc/Makefile create mode 100644 libc/string/sparc/_glibc_inc.h create mode 100644 libc/string/sparc/bcopy.c create mode 100644 libc/string/sparc/bzero.c create mode 100644 libc/string/sparc/memchr.S create mode 100644 libc/string/sparc/memcpy.S create mode 100644 libc/string/sparc/memmove.c create mode 100644 libc/string/sparc/memset.S create mode 100644 libc/string/sparc/sparc32/memchr.S create mode 100644 libc/string/sparc/sparc32/memcpy.S create mode 100644 libc/string/sparc/sparc32/memset.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/memchr.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/memcpy.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/memset.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/stpcpy.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/strcat.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/strchr.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/strcmp.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/strcpy.S create mode 100644 libc/string/sparc/sparc32/sparcv9b/strlen.S create mode 100644 libc/string/sparc/sparc32/stpcpy.S create mode 100644 libc/string/sparc/sparc32/strcat.S create mode 100644 libc/string/sparc/sparc32/strchr.S create mode 100644 libc/string/sparc/sparc32/strcmp.S create mode 100644 libc/string/sparc/sparc32/strcpy.S create mode 100644 libc/string/sparc/sparc32/strlen.S create mode 100644 libc/string/sparc/sparc64/memchr.S create mode 100644 libc/string/sparc/sparc64/memcpy.S create mode 100644 libc/string/sparc/sparc64/memset.S create mode 100644 libc/string/sparc/sparc64/sparcv9b/memcpy.S create mode 100644 libc/string/sparc/sparc64/stpcpy.S create mode 100644 libc/string/sparc/sparc64/strcat.S create mode 100644 libc/string/sparc/sparc64/strchr.S create mode 100644 libc/string/sparc/sparc64/strcmp.S create mode 100644 libc/string/sparc/sparc64/strcpy.S create mode 100644 libc/string/sparc/sparc64/strlen.S create mode 100644 libc/string/sparc/stpcpy.S create mode 100644 libc/string/sparc/strcat.S create mode 100644 libc/string/sparc/strchr.S create mode 100644 libc/string/sparc/strcmp.S create mode 100644 libc/string/sparc/strcpy.S create mode 100644 libc/string/sparc/strlen.S create mode 100644 libc/string/sparc/strrchr.c diff --git a/libc/string/Makefile b/libc/string/Makefile index 6b516c1c5..600f3b81d 100644 --- a/libc/string/Makefile +++ b/libc/string/Makefile @@ -31,7 +31,7 @@ DIRS += $(TARGET_ARCH) endif endif -ALL_SUBDIRS = generic arm frv i386 sh64 powerpc mips +ALL_SUBDIRS = generic arm frv i386 mips powerpc sh64 sparc MSRC= wstring.c MOBJ= basename.o bcopy.o bzero.o dirname.o ffs.o memccpy.o memchr.o memcmp.o \ diff --git a/libc/string/sparc/Makefile b/libc/string/sparc/Makefile new file mode 100644 index 000000000..2215a6025 --- /dev/null +++ b/libc/string/sparc/Makefile @@ -0,0 +1,31 @@ +# Makefile for uClibc +# +# Copyright (C) 2000-2005 Erik Andersen <andersen@uclibc.org> +# +# Licensed under the GNU Library General Public License version 2 or later. +# See the COPYING.LIB file in the toplevel for more information. + +TOPDIR=../../../ +include $(TOPDIR)Rules.mak + +CSRCS = $(wildcard *.c) +COBJS = $(patsubst %.c,%.o,$(CSRCS)) + +SSRCS = $(wildcard *.S) +SOBJS = $(patsubst %.S,%.o,$(SSRCS)) + +OBJS = $(COBJS) $(SOBJS) + +OBJ_LIST = ../../obj.string.$(TARGET_ARCH) + +all: $(OBJ_LIST) + +$(OBJ_LIST): $(OBJS) + echo $(patsubst %, string/$(TARGET_ARCH)/%, $(OBJS)) > $(OBJ_LIST) + +$(COBJS): %.o : %.c + $(CC) $(CFLAGS) -c $< -o $@ + $(STRIPTOOL) -x -R .note -R .comment $*.o + +clean: + $(RM) *.[oa] *~ core diff --git a/libc/string/sparc/_glibc_inc.h b/libc/string/sparc/_glibc_inc.h new file mode 100644 index 000000000..7840ba1bf --- /dev/null +++ b/libc/string/sparc/_glibc_inc.h @@ -0,0 +1,30 @@ +/* + * Setup some glibc defines so we can just drop in the + * asm files from glibc without any modification. + */ + +#include <features.h> +#include <bits/wordsize.h> + +#if __WORDSIZE == 32 +# define ENTRY_ALIGN 4 +#else +# define ENTRY_ALIGN 2 +#endif + +#define ENTRY(sym) \ + .global sym; \ + .align ENTRY_ALIGN; \ + .type sym,%function; \ + sym: + +#define LOC(sym) \ + .L ## sym + +#define END(sym) \ + .size sym,.-sym; + +#undef weak_alias +#define weak_alias(sym, alias) \ + .weak alias; \ + alias = sym; diff --git a/libc/string/sparc/bcopy.c b/libc/string/sparc/bcopy.c new file mode 100644 index 000000000..9a455f33c --- /dev/null +++ b/libc/string/sparc/bcopy.c @@ -0,0 +1 @@ +/* bcopy is in memcpy.S */ diff --git a/libc/string/sparc/bzero.c b/libc/string/sparc/bzero.c new file mode 100644 index 000000000..37f0f6f99 --- /dev/null +++ b/libc/string/sparc/bzero.c @@ -0,0 +1 @@ +/* bzero is in memset.S */ diff --git a/libc/string/sparc/memchr.S b/libc/string/sparc/memchr.S new file mode 100644 index 000000000..73e5101fd --- /dev/null +++ b/libc/string/sparc/memchr.S @@ -0,0 +1,13 @@ +#include "_glibc_inc.h" + +#if __WORDSIZE == 32 + +# if defined(__CONFIG_SPARC_V9B__) +# include "sparc32/sparcv9b/memchr.S" +# else +# include "sparc32/memchr.S" +# endif + +#else +# include "sparc64/memchr.S" +#endif diff --git a/libc/string/sparc/memcpy.S b/libc/string/sparc/memcpy.S new file mode 100644 index 000000000..478cb106f --- /dev/null +++ b/libc/string/sparc/memcpy.S @@ -0,0 +1,19 @@ +#include "_glibc_inc.h" + +#if __WORDSIZE == 32 + +# if defined(__CONFIG_SPARC_V9B__) +# include "sparc32/sparcv9b/memcpy.S" +# else +# include "sparc32/memcpy.S" +# endif + +#else + +# if defined(__CONFIG_SPARC_V9B__) +# include "sparc64/sparcv9b/memcpy.S" +# else +# include "sparc64/memcpy.S" +#endif + +#endif diff --git a/libc/string/sparc/memmove.c b/libc/string/sparc/memmove.c new file mode 100644 index 000000000..a8d2d4994 --- /dev/null +++ b/libc/string/sparc/memmove.c @@ -0,0 +1 @@ +/* memmove is in memcpy.S */ diff --git a/libc/string/sparc/memset.S b/libc/string/sparc/memset.S new file mode 100644 index 000000000..4aed77421 --- /dev/null +++ b/libc/string/sparc/memset.S @@ -0,0 +1,13 @@ +#include "_glibc_inc.h" + +#if __WORDSIZE == 32 + +# if defined(__CONFIG_SPARC_V9B__) +# include "sparc32/sparcv9b/memset.S" +# else +# include "sparc32/memset.S" +# endif + +#else +# include "sparc64/memset.S" +#endif diff --git a/libc/string/sparc/sparc32/memchr.S b/libc/string/sparc/sparc32/memchr.S new file mode 100644 index 000000000..c00df2427 --- /dev/null +++ b/libc/string/sparc/sparc32/memchr.S @@ -0,0 +1,144 @@ +/* memchr (str, ch, n) -- Return pointer to first occurrence of CH in STR less + than N. + For SPARC v7. + Copyright (C) 1996, 1999, 2000, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Jakub Jelinek <jj@ultra.linux.cz> and + David S. Miller <davem@caip.rutgers.edu>. + This version is developed using the same algorithm as the fast C + version which carries the following introduction: + Based on strlen implementation by Torbjorn Granlund (tege@sics.se), + with help from Dan Sahlin (dan@sics.se) and + commentary by Jim Blandy (jimb@ai.mit.edu); + adaptation to memchr suggested by Dick Karpinski (dick@cca.ucsf.edu), + and implemented by Roland McGrath (roland@ai.mit.edu). + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + .text + .align 4 +0: cmp %o2, 0 + be 9f + sethi %hi(0x80808080), %o4 + ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + subcc %o2, 1, %o2 + be 9f + andcc %o0, 3, %g0 + be 4f + or %o4, %lo(0x80808080), %o3 + ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + subcc %o2, 1, %o2 + be 9f + andcc %o0, 3, %g0 + be 5f + sethi %hi(0x01010101), %o5 + ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + subcc %o2, 1, %o2 + bne,a 7f + and %o2, 3, %g1 + retl + clr %o0 +1: retl + sub %o0, 1, %o0 + +ENTRY(__memchr) + andcc %o1, 0xff, %o1 + sll %o1, 8, %g6 + andcc %o0, 3, %g0 + or %o1, %g6, %g6 + sll %g6, 16, %o3 + bne 0b + or %o3, %g6, %g2 + sethi %hi(0x80808080), %o4 + or %o4, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %o5 +5: and %o2, 3, %g1 +7: andcc %o2, 0xfffffffc, %o2 + be 0f + or %o5, %lo(0x01010101), %g6 + ld [%o0], %g4 +6: xor %g4, %g2, %g5 + add %o0, 4, %o0 + sub %g5, %g6, %g5 + andcc %g5, %o3, %g0 + bne 8f + subcc %o2, 4, %o2 + bne,a 6b + ld [%o0], %g4 +0: cmp %g1, 0 +1: be 9f + add %o0, 4, %o0 + ldub [%o0 - 4], %g4 + cmp %g4, %o1 + be 4f + cmp %g1, 1 + be 9f + ldub [%o0 - 3], %g4 + cmp %g4, %o1 + be 3f + cmp %g1, 2 + be 9f + ldub [%o0 - 2], %g4 + cmp %g4, %o1 + be 2f + nop +9: retl + clr %o0 + + /* Check every byte. */ +8: srl %g4, 24, %g5 + and %g5, 0xff, %g5 + cmp %g5, %o1 + be 4f + srl %g4, 16, %g5 + and %g5, 0xff, %g5 + cmp %g5, %o1 + be 3f + srl %g4, 8, %g5 + and %g5, 0xff, %g5 + cmp %g5, %o1 + be 2f + and %g4, 0xff, %g5 + cmp %g5, %o1 + be 1f + cmp %o2, 0 + bne,a 6b + ld [%o0], %g4 + b 1b + cmp %g1, 0 +1: retl + sub %o0, 1, %o0 +2: retl + sub %o0, 2, %o0 +3: retl + sub %o0, 3, %o0 +4: retl + sub %o0, 4, %o0 +END(__memchr) + +weak_alias (__memchr, memchr) +#if !__BOUNDED_POINTERS__ +weak_alias (__memchr, __ubp_memchr) +#endif diff --git a/libc/string/sparc/sparc32/memcpy.S b/libc/string/sparc/sparc32/memcpy.S new file mode 100644 index 000000000..d0a80887b --- /dev/null +++ b/libc/string/sparc/sparc32/memcpy.S @@ -0,0 +1,966 @@ +/* Copy SIZE bytes from SRC to DEST. + For SPARC v7. + Copyright (C) 1996, 1999, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by David S. Miller <davem@caip.rutgers.edu>, + Eddie C. Dost <ecd@skynet.be> and + Jakub Jelinek <jj@ultra.linux.cz>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +/* Both these macros have to start with exactly the same insn */ +#define MOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \ + ldd [%src + offset + 0x00], %t0; \ + ldd [%src + offset + 0x08], %t2; \ + ldd [%src + offset + 0x10], %t4; \ + ldd [%src + offset + 0x18], %t6; \ + st %t0, [%dst + offset + 0x00]; \ + st %t1, [%dst + offset + 0x04]; \ + st %t2, [%dst + offset + 0x08]; \ + st %t3, [%dst + offset + 0x0c]; \ + st %t4, [%dst + offset + 0x10]; \ + st %t5, [%dst + offset + 0x14]; \ + st %t6, [%dst + offset + 0x18]; \ + st %t7, [%dst + offset + 0x1c]; + +#define MOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \ + ldd [%src + offset + 0x00], %t0; \ + ldd [%src + offset + 0x08], %t2; \ + ldd [%src + offset + 0x10], %t4; \ + ldd [%src + offset + 0x18], %t6; \ + std %t0, [%dst + offset + 0x00]; \ + std %t2, [%dst + offset + 0x08]; \ + std %t4, [%dst + offset + 0x10]; \ + std %t6, [%dst + offset + 0x18]; + +#define MOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \ + ldd [%src - offset - 0x10], %t0; \ + ldd [%src - offset - 0x08], %t2; \ + st %t0, [%dst - offset - 0x10]; \ + st %t1, [%dst - offset - 0x0c]; \ + st %t2, [%dst - offset - 0x08]; \ + st %t3, [%dst - offset - 0x04]; + +#define MOVE_LASTALIGNCHUNK(src, dst, offset, t0, t1, t2, t3) \ + ldd [%src - offset - 0x10], %t0; \ + ldd [%src - offset - 0x08], %t2; \ + std %t0, [%dst - offset - 0x10]; \ + std %t2, [%dst - offset - 0x08]; + +#define MOVE_SHORTCHUNK(src, dst, offset, t0, t1) \ + ldub [%src - offset - 0x02], %t0; \ + ldub [%src - offset - 0x01], %t1; \ + stb %t0, [%dst - offset - 0x02]; \ + stb %t1, [%dst - offset - 0x01]; + +/* Both these macros have to start with exactly the same insn */ +#define RMOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \ + ldd [%src - offset - 0x20], %t0; \ + ldd [%src - offset - 0x18], %t2; \ + ldd [%src - offset - 0x10], %t4; \ + ldd [%src - offset - 0x08], %t6; \ + st %t0, [%dst - offset - 0x20]; \ + st %t1, [%dst - offset - 0x1c]; \ + st %t2, [%dst - offset - 0x18]; \ + st %t3, [%dst - offset - 0x14]; \ + st %t4, [%dst - offset - 0x10]; \ + st %t5, [%dst - offset - 0x0c]; \ + st %t6, [%dst - offset - 0x08]; \ + st %t7, [%dst - offset - 0x04]; + +#define RMOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \ + ldd [%src - offset - 0x20], %t0; \ + ldd [%src - offset - 0x18], %t2; \ + ldd [%src - offset - 0x10], %t4; \ + ldd [%src - offset - 0x08], %t6; \ + std %t0, [%dst - offset - 0x20]; \ + std %t2, [%dst - offset - 0x18]; \ + std %t4, [%dst - offset - 0x10]; \ + std %t6, [%dst - offset - 0x08]; + +#define RMOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \ + ldd [%src + offset + 0x00], %t0; \ + ldd [%src + offset + 0x08], %t2; \ + st %t0, [%dst + offset + 0x00]; \ + st %t1, [%dst + offset + 0x04]; \ + st %t2, [%dst + offset + 0x08]; \ + st %t3, [%dst + offset + 0x0c]; + +#define RMOVE_SHORTCHUNK(src, dst, offset, t0, t1) \ + ldub [%src + offset + 0x00], %t0; \ + ldub [%src + offset + 0x01], %t1; \ + stb %t0, [%dst + offset + 0x00]; \ + stb %t1, [%dst + offset + 0x01]; + +#define SMOVE_CHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \ + ldd [%src + offset + 0x00], %t0; \ + ldd [%src + offset + 0x08], %t2; \ + srl %t0, shir, %t5; \ + srl %t1, shir, %t6; \ + sll %t0, shil, %t0; \ + or %t5, %prev, %t5; \ + sll %t1, shil, %prev; \ + or %t6, %t0, %t0; \ + srl %t2, shir, %t1; \ + srl %t3, shir, %t6; \ + sll %t2, shil, %t2; \ + or %t1, %prev, %t1; \ + std %t4, [%dst + offset + offset2 - 0x04]; \ + std %t0, [%dst + offset + offset2 + 0x04]; \ + sll %t3, shil, %prev; \ + or %t6, %t2, %t4; + +#define SMOVE_ALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \ + ldd [%src + offset + 0x00], %t0; \ + ldd [%src + offset + 0x08], %t2; \ + srl %t0, shir, %t4; \ + srl %t1, shir, %t5; \ + sll %t0, shil, %t6; \ + or %t4, %prev, %t0; \ + sll %t1, shil, %prev; \ + or %t5, %t6, %t1; \ + srl %t2, shir, %t4; \ + srl %t3, shir, %t5; \ + sll %t2, shil, %t6; \ + or %t4, %prev, %t2; \ + sll %t3, shil, %prev; \ + or %t5, %t6, %t3; \ + std %t0, [%dst + offset + offset2 + 0x00]; \ + std %t2, [%dst + offset + offset2 + 0x08]; + + .text + .align 4 + +70: andcc %o1, 1, %g0 + be 4f + andcc %o1, 2, %g0 + + ldub [%o1 - 1], %g2 + sub %o1, 1, %o1 + stb %g2, [%o0 - 1] + sub %o2, 1, %o2 + be 3f + sub %o0, 1, %o0 +4: lduh [%o1 - 2], %g2 + sub %o1, 2, %o1 + sth %g2, [%o0 - 2] + sub %o2, 2, %o2 + b 3f + sub %o0, 2, %o0 + +ENTRY(bcopy) + mov %o0, %o3 + mov %o1, %o0 + mov %o3, %o1 +END(bcopy) +ENTRY(memmove) + cmp %o0, %o1 + st %o0, [%sp + 64] + bleu 9f + sub %o0, %o1, %o4 + + add %o1, %o2, %o3 + cmp %o3, %o0 + bleu 0f + andcc %o4, 3, %o5 + + add %o1, %o2, %o1 + add %o0, %o2, %o0 + bne 77f + cmp %o2, 15 + bleu 91f + andcc %o1, 3, %g0 + bne 70b +3: andcc %o1, 4, %g0 + + be 2f + mov %o2, %g1 + + ld [%o1 - 4], %o4 + sub %g1, 4, %g1 + st %o4, [%o0 - 4] + sub %o1, 4, %o1 + sub %o0, 4, %o0 +2: andcc %g1, 0xffffff80, %g6 + be 3f + andcc %o0, 4, %g0 + + be 74f + 4 +5: RMOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5) + subcc %g6, 128, %g6 + sub %o1, 128, %o1 + bne 5b + sub %o0, 128, %o0 + +3: andcc %g1, 0x70, %g6 + be 72f + andcc %g1, 8, %g0 + + srl %g6, 1, %o4 + mov %o7, %g2 + add %g6, %o4, %o4 +101: call 100f + sub %o1, %g6, %o1 + mov %g2, %o7 + jmpl %o5 + (72f - 101b), %g0 + sub %o0, %g6, %o0 + +71: RMOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5) + RMOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5) +72: be 73f + andcc %g1, 4, %g0 + + ldd [%o1 - 0x08], %g2 + sub %o0, 8, %o0 + sub %o1, 8, %o1 + st %g2, [%o0] + st %g3, [%o0 + 0x04] +73: be 1f + andcc %g1, 2, %g0 + + ld [%o1 - 4], %g2 + sub %o1, 4, %o1 + st %g2, [%o0 - 4] + sub %o0, 4, %o0 +1: be 1f + andcc %g1, 1, %g0 + + lduh [%o1 - 2], %g2 + sub %o1, 2, %o1 + sth %g2, [%o0 - 2] + sub %o0, 2, %o0 +1: be 1f + nop + + ldub [%o1 - 1], %g2 + stb %g2, [%o0 - 1] +1: retl + ld [%sp + 64], %o0 + +74: RMOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5) + RMOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5) + subcc %g6, 128, %g6 + sub %o1, 128, %o1 + bne 74b + sub %o0, 128, %o0 + + andcc %g1, 0x70, %g6 + be 72b + andcc %g1, 8, %g0 + + srl %g6, 1, %o4 + mov %o7, %g2 + add %g6, %o4, %o4 +102: call 100f + sub %o1, %g6, %o1 + mov %g2, %o7 + jmpl %o5 + (72b - 102b), %g0 + sub %o0, %g6, %o0 + +75: and %o2, 0xe, %o3 + mov %o7, %g2 + sll %o3, 3, %o4 + sub %o0, %o3, %o0 +103: call 100f + sub %o1, %o3, %o1 + mov %g2, %o7 + jmpl %o5 + (76f - 103b), %g0 + andcc %o2, 1, %g0 + + RMOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3) + RMOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3) + +76: be 1f + nop + ldub [%o1 - 1], %g2 + stb %g2, [%o0 - 1] +1: retl + ld [%sp + 64], %o0 + +91: bne 75b + andcc %o2, 8, %g0 + + be 1f + andcc %o2, 4, %g0 + + ld [%o1 - 0x08], %g2 + ld [%o1 - 0x04], %g3 + sub %o1, 8, %o1 + st %g2, [%o0 - 0x08] + st %g3, [%o0 - 0x04] + sub %o0, 8, %o0 +1: b 73b + mov %o2, %g1 + +77: cmp %o2, 15 + bleu 75b + andcc %o0, 3, %g0 + be 64f + andcc %o0, 1, %g0 + be 63f + andcc %o0, 2, %g0 + ldub [%o1 - 1], %g5 + sub %o1, 1, %o1 + stb %g5, [%o0 - 1] + sub %o0, 1, %o0 + be 64f + sub %o2, 1, %o2 + +63: ldub [%o1 - 1], %g5 + sub %o1, 2, %o1 + stb %g5, [%o0 - 1] + sub %o0, 2, %o0 + ldub [%o1], %g5 + sub %o2, 2, %o2 + stb %g5, [%o0] +64: and %o1, 3, %g2 + and %o1, -4, %o1 + and %o2, 0xc, %g3 + add %o1, 4, %o1 + cmp %g3, 4 + sll %g2, 3, %g4 + mov 32, %g2 + be 4f + sub %g2, %g4, %g6 + + blu 3f + cmp %g3, 8 + + be 2f + srl %o2, 2, %g3 + + ld [%o1 - 4], %o3 + add %o0, -8, %o0 + ld [%o1 - 8], %o4 + add %o1, -16, %o1 + b 7f + add %g3, 1, %g3 +2: ld [%o1 - 4], %o4 + add %o0, -4, %o0 + ld [%o1 - 8], %g1 + add %o1, -12, %o1 + b 8f + add %g3, 2, %g3 +3: ld [%o1 - 4], %o5 + add %o0, -12, %o0 + ld [%o1 - 8], %o3 + add %o1, -20, %o1 + b 6f + srl %o2, 2, %g3 +4: ld [%o1 - 4], %g1 + srl %o2, 2, %g3 + ld [%o1 - 8], %o5 + add %o1, -24, %o1 + add %o0, -16, %o0 + add %g3, -1, %g3 + + ld [%o1 + 12], %o3 +5: sll %o5, %g4, %g2 + srl %g1, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0 + 12] +6: ld [%o1 + 8], %o4 + sll %o3, %g4, %g2 + srl %o5, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0 + 8] +7: ld [%o1 + 4], %g1 + sll %o4, %g4, %g2 + srl %o3, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0 + 4] +8: ld [%o1], %o5 + sll %g1, %g4, %g2 + srl %o4, %g6, %g5 + addcc %g3, -4, %g3 + or %g2, %g5, %g2 + add %o1, -16, %o1 + st %g2, [%o0] + add %o0, -16, %o0 + bne,a 5b + ld [%o1 + 12], %o3 + sll %o5, %g4, %g2 + srl %g1, %g6, %g5 + srl %g4, 3, %g3 + or %g2, %g5, %g2 + add %o1, %g3, %o1 + andcc %o2, 2, %g0 + st %g2, [%o0 + 12] + be 1f + andcc %o2, 1, %g0 + + ldub [%o1 + 15], %g5 + add %o1, -2, %o1 + stb %g5, [%o0 + 11] + add %o0, -2, %o0 + ldub [%o1 + 16], %g5 + stb %g5, [%o0 + 12] +1: be 1f + nop + ldub [%o1 + 15], %g5 + stb %g5, [%o0 + 11] +1: retl + ld [%sp + 64], %o0 + +78: andcc %o1, 1, %g0 + be 4f + andcc %o1, 2, %g0 + + ldub [%o1], %g2 + add %o1, 1, %o1 + stb %g2, [%o0] + sub %o2, 1, %o2 + bne 3f + add %o0, 1, %o0 +4: lduh [%o1], %g2 + add %o1, 2, %o1 + sth %g2, [%o0] + sub %o2, 2, %o2 + b 3f + add %o0, 2, %o0 +END(memmove) + +ENTRY(memcpy) /* %o0=dst %o1=src %o2=len */ + sub %o0, %o1, %o4 + st %o0, [%sp + 64] +9: andcc %o4, 3, %o5 +0: bne 86f + cmp %o2, 15 + + bleu 90f + andcc %o1, 3, %g0 + + bne 78b +3: andcc %o1, 4, %g0 + + be 2f + mov %o2, %g1 + + ld [%o1], %o4 + sub %g1, 4, %g1 + st %o4, [%o0] + add %o1, 4, %o1 + add %o0, 4, %o0 +2: andcc %g1, 0xffffff80, %g6 + be 3f + andcc %o0, 4, %g0 + + be 82f + 4 +5: MOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5) + subcc %g6, 128, %g6 + add %o1, 128, %o1 + bne 5b + add %o0, 128, %o0 +3: andcc %g1, 0x70, %g6 + be 80f + andcc %g1, 8, %g0 + + srl %g6, 1, %o4 + mov %o7, %g2 + add %g6, %o4, %o4 + add %o1, %g6, %o1 +104: call 100f + add %o0, %g6, %o0 + jmpl %o5 + (80f - 104b), %g0 + mov %g2, %o7 + +79: MOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5) + MOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5) + +80: be 81f + andcc %g1, 4, %g0 + + ldd [%o1], %g2 + add %o0, 8, %o0 + st %g2, [%o0 - 0x08] + add %o1, 8, %o1 + st %g3, [%o0 - 0x04] + +81: be 1f + andcc %g1, 2, %g0 + + ld [%o1], %g2 + add %o1, 4, %o1 + st %g2, [%o0] + add %o0, 4, %o0 +1: be 1f + andcc %g1, 1, %g0 + + lduh [%o1], %g2 + add %o1, 2, %o1 + sth %g2, [%o0] + add %o0, 2, %o0 +1: be 1f + nop + + ldub [%o1], %g2 + stb %g2, [%o0] +1: retl + ld [%sp + 64], %o0 + +82: /* ldd_std */ + MOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5) + MOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5) + subcc %g6, 128, %g6 + add %o1, 128, %o1 + bne 82b + add %o0, 128, %o0 + + andcc %g1, 0x70, %g6 + be 84f + andcc %g1, 8, %g0 + + mov %o7, %g2 +111: call 110f + add %o1, %g6, %o1 + mov %g2, %o7 + jmpl %o5 + (84f - 111b), %g0 + add %o0, %g6, %o0 + +83: MOVE_LASTALIGNCHUNK(o1, o0, 0x60, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x50, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x40, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5) + MOVE_LASTALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5) + +84: be 85f + andcc %g1, 4, %g0 + + ldd [%o1], %g2 + add %o0, 8, %o0 + std %g2, [%o0 - 0x08] + add %o1, 8, %o1 +85: be 1f + andcc %g1, 2, %g0 + + ld [%o1], %g2 + add %o1, 4, %o1 + st %g2, [%o0] + add %o0, 4, %o0 +1: be 1f + andcc %g1, 1, %g0 + + lduh [%o1], %g2 + add %o1, 2, %o1 + sth %g2, [%o0] + add %o0, 2, %o0 +1: be 1f + nop + + ldub [%o1], %g2 + stb %g2, [%o0] +1: retl + ld [%sp + 64], %o0 + +86: cmp %o2, 6 + bleu 88f + + cmp %o2, 256 + bcc 87f + + andcc %o0, 3, %g0 + be 61f + andcc %o0, 1, %g0 + be 60f + andcc %o0, 2, %g0 + + ldub [%o1], %g5 + add %o1, 1, %o1 + stb %g5, [%o0] + sub %o2, 1, %o2 + bne 61f + add %o0, 1, %o0 +60: ldub [%o1], %g3 + add %o1, 2, %o1 + stb %g3, [%o0] + sub %o2, 2, %o2 + ldub [%o1 - 1], %g3 + add %o0, 2, %o0 + stb %g3, [%o0 - 1] +61: and %o1, 3, %g2 + and %o2, 0xc, %g3 + and %o1, -4, %o1 + cmp %g3, 4 + sll %g2, 3, %g4 + mov 32, %g2 + be 4f + sub %g2, %g4, %g6 + + blu 3f + cmp %g3, 0x8 + + be 2f + srl %o2, 2, %g3 + + ld [%o1], %o3 + add %o0, -8, %o0 + ld [%o1 + 4], %o4 + b 8f + add %g3, 1, %g3 +2: ld [%o1], %o4 + add %o0, -12, %o0 + ld [%o1 + 4], %o5 + add %g3, 2, %g3 + b 9f + add %o1, -4, %o1 +3: ld [%o1], %g1 + add %o0, -4, %o0 + ld [%o1 + 4], %o3 + srl %o2, 2, %g3 + b 7f + add %o1, 4, %o1 +4: ld [%o1], %o5 + cmp %o2, 7 + ld [%o1 + 4], %g1 + srl %o2, 2, %g3 + bleu 10f + add %o1, 8, %o1 + + ld [%o1], %o3 + add %g3, -1, %g3 +5: sll %o5, %g4, %g2 + srl %g1, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0] +7: ld [%o1 + 4], %o4 + sll %g1, %g4, %g2 + srl %o3, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0 + 4] +8: ld [%o1 + 8], %o5 + sll %o3, %g4, %g2 + srl %o4, %g6, %g5 + or %g2, %g5, %g2 + st %g2, [%o0 + 8] +9: ld [%o1 + 12], %g1 + sll %o4, %g4, %g2 + srl %o5, %g6, %g5 + addcc %g3, -4, %g3 + or %g2, %g5, %g2 + add %o1, 16, %o1 + st %g2, [%o0 + 12] + add %o0, 16, %o0 + bne,a 5b + ld [%o1], %o3 +10: sll %o5, %g4, %g2 + srl %g1, %g6, %g5 + srl %g6, 3, %g3 + or %g2, %g5, %g2 + sub %o1, %g3, %o1 + andcc %o2, 2, %g0 + st %g2, [%o0] + be 1f + andcc %o2, 1, %g0 + + ldub [%o1], %g2 + add %o1, 2, %o1 + stb %g2, [%o0 + 4] + add %o0, 2, %o0 + ldub [%o1 - 1], %g2 + stb %g2, [%o0 + 3] +1: be 1f + nop + ldub [%o1], %g2 + stb %g2, [%o0 + 4] +1: retl + ld [%sp + 64], %o0 + +87: andcc %o1, 3, %g0 + be 3f + andcc %o1, 1, %g0 + + be 4f + andcc %o1, 2, %g0 + + ldub [%o1], %g2 + add %o1, 1, %o1 + stb %g2, [%o0] + sub %o2, 1, %o2 + bne 3f + add %o0, 1, %o0 +4: lduh [%o1], %g2 + add %o1, 2, %o1 + srl %g2, 8, %g3 + sub %o2, 2, %o2 + stb %g3, [%o0] + add %o0, 2, %o0 + stb %g2, [%o0 - 1] +3: andcc %o1, 4, %g0 + + bne 2f + cmp %o5, 1 + + ld [%o1], %o4 + srl %o4, 24, %g2 + stb %g2, [%o0] + srl %o4, 16, %g3 + stb %g3, [%o0 + 1] + srl %o4, 8, %g2 + stb %g2, [%o0 + 2] + sub %o2, 4, %o2 + stb %o4, [%o0 + 3] + add %o1, 4, %o1 + add %o0, 4, %o0 +2: be 33f + cmp %o5, 2 + be 32f + sub %o2, 4, %o2 +31: ld [%o1], %g2 + add %o1, 4, %o1 + srl %g2, 24, %g3 + and %o0, 7, %g5 + stb %g3, [%o0] + cmp %g5, 7 + sll %g2, 8, %g1 + add %o0, 4, %o0 + be 41f + and %o2, 0xffffffc0, %o3 + ld [%o0 - 7], %o4 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 4b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 16, %g2 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 16, %g2 +1: st %o4, [%o0 - 7] + sth %g2, [%o0 - 3] + srl %g1, 8, %g4 + b 88f + stb %g4, [%o0 - 1] +32: ld [%o1], %g2 + add %o1, 4, %o1 + srl %g2, 16, %g3 + and %o0, 7, %g5 + sth %g3, [%o0] + cmp %g5, 6 + sll %g2, 16, %g1 + add %o0, 4, %o0 + be 42f + and %o2, 0xffffffc0, %o3 + ld [%o0 - 6], %o4 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 4b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 16, %g2 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 16, %g2 +1: st %o4, [%o0 - 6] + b 88f + sth %g2, [%o0 - 2] +33: ld [%o1], %g2 + sub %o2, 4, %o2 + srl %g2, 24, %g3 + and %o0, 7, %g5 + stb %g3, [%o0] + cmp %g5, 5 + srl %g2, 8, %g4 + sll %g2, 24, %g1 + sth %g4, [%o0 + 1] + add %o1, 4, %o1 + be 43f + and %o2, 0xffffffc0, %o3 + + ld [%o0 - 1], %o4 + add %o0, 4, %o0 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, -1) + SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, -1) + SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, -1) + SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, -1) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 4b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 24, %g2 +4: SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, -1) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 24, %g2 +1: st %o4, [%o0 - 5] + b 88f + stb %g2, [%o0 - 1] +41: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 41b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 16, %g2 +4: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 8, 24, -3) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 16, %g2 +1: sth %g2, [%o0 - 3] + srl %g1, 8, %g4 + b 88f + stb %g4, [%o0 - 1] +43: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, 3) + SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, 3) + SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, 3) + SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, 3) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 43b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 24, %g2 +4: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 24, 8, 3) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 24, %g2 +1: stb %g2, [%o0 + 3] + b 88f + add %o0, 4, %o0 +42: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + subcc %o3, 64, %o3 + add %o1, 64, %o1 + bne 42b + add %o0, 64, %o0 + + andcc %o2, 0x30, %o3 + be,a 1f + srl %g1, 16, %g2 +4: SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g6, g1, 16, 16, -2) + subcc %o3, 16, %o3 + add %o1, 16, %o1 + bne 4b + add %o0, 16, %o0 + + srl %g1, 16, %g2 +1: sth %g2, [%o0 - 2] + + /* Fall through */ + +88: and %o2, 0xe, %o3 + mov %o7, %g2 + sll %o3, 3, %o4 + add %o0, %o3, %o0 +106: call 100f + add %o1, %o3, %o1 + mov %g2, %o7 + jmpl %o5 + (89f - 106b), %g0 + andcc %o2, 1, %g0 + + MOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3) + MOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3) + +89: be 1f + nop + + ldub [%o1], %g2 + stb %g2, [%o0] +1: retl + ld [%sp + 64], %o0 + +90: bne 88b + andcc %o2, 8, %g0 + + be 1f + andcc %o2, 4, %g0 + + ld [%o1 + 0x00], %g2 + ld [%o1 + 0x04], %g3 + add %o1, 8, %o1 + st %g2, [%o0 + 0x00] + st %g3, [%o0 + 0x04] + add %o0, 8, %o0 +1: b 81b + mov %o2, %g1 + +100: retl + sub %o7, %o4, %o5 +110: retl + sub %o7, %g6, %o5 +END(memcpy) diff --git a/libc/string/sparc/sparc32/memset.S b/libc/string/sparc/sparc32/memset.S new file mode 100644 index 000000000..02cebb791 --- /dev/null +++ b/libc/string/sparc/sparc32/memset.S @@ -0,0 +1,152 @@ +/* Set a block of memory to some byte value. + For SPARC v7. + Copyright (C) 1996, 1999, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by David S. Miller <davem@caip.rutgers.edu> and + Jakub Jelinek <jj@ultra.linux.cz>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Store 64 bytes at (BASE + OFFSET) using value SOURCE. */ +#define ZERO_BIG_BLOCK(base, offset, source) \ + std source, [base + offset + 0x00]; \ + std source, [base + offset + 0x08]; \ + std source, [base + offset + 0x10]; \ + std source, [base + offset + 0x18]; \ + std source, [base + offset + 0x20]; \ + std source, [base + offset + 0x28]; \ + std source, [base + offset + 0x30]; \ + std source, [base + offset + 0x38]; + +#define ZERO_LAST_BLOCKS(base, offset, source) \ + std source, [base - offset - 0x38]; \ + std source, [base - offset - 0x30]; \ + std source, [base - offset - 0x28]; \ + std source, [base - offset - 0x20]; \ + std source, [base - offset - 0x18]; \ + std source, [base - offset - 0x10]; \ + std source, [base - offset - 0x08]; \ + std source, [base - offset - 0x00]; + + .text + .align 4 +ENTRY(__bzero) + b 1f + mov %g0, %g3 + +3: cmp %o2, 3 + be 2f + stb %g3, [%o0] + + cmp %o2, 2 + be 2f + stb %g3, [%o0 + 0x01] + + stb %g3, [%o0 + 0x02] +2: sub %o2, 4, %o2 + add %o1, %o2, %o1 + b 4f + sub %o0, %o2, %o0 +END(__bzero) + +ENTRY(memset) + and %o1, 0xff, %g3 + sll %g3, 8, %g2 + or %g3, %g2, %g3 + sll %g3, 16, %g2 + or %g3, %g2, %g3 + orcc %o2, %g0, %o1 +1: cmp %o1, 7 + bleu 7f + mov %o0, %g1 + + andcc %o0, 3, %o2 + bne 3b +4: andcc %o0, 4, %g0 + + be 2f + mov %g3, %g2 + + st %g3, [%o0] + sub %o1, 4, %o1 + add %o0, 4, %o0 +2: andcc %o1, 0xffffff80, %o3 + be 9f + andcc %o1, 0x78, %o2 +4: ZERO_BIG_BLOCK (%o0, 0x00, %g2) + subcc %o3, 128, %o3 + ZERO_BIG_BLOCK (%o0, 0x40, %g2) + bne 4b + add %o0, 128, %o0 + + orcc %o2, %g0, %g0 +9: be 6f + andcc %o1, 7, %o1 + + mov %o7, %g4 +101: call 100f + srl %o2, 1, %o3 + mov %g4, %o7 + jmpl %o4 + (20f + 64 - 101b), %g0 + add %o0, %o2, %o0 + +100: retl + sub %o7, %o3, %o4 + +20: ZERO_LAST_BLOCKS(%o0, 0x48, %g2) + ZERO_LAST_BLOCKS(%o0, 0x08, %g2) + +6: be 8f + andcc %o1, 4, %g0 + be 1f + andcc %o1, 2, %g0 + st %g3, [%o0] + add %o0, 4, %o0 +1: be 1f + andcc %o1, 1, %g0 + sth %g3, [%o0] + add %o0, 2, %o0 +1: bne,a 8f + stb %g3, [%o0] +8: retl + mov %g1, %o0 +7: orcc %o1, 0, %g0 + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 1] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 2] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 3] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 4] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 5] + be 0f + subcc %o1, 1, %o1 + stb %g3, [%o0 + 6] +0: retl + nop +END(memset) + +weak_alias (__bzero, bzero) diff --git a/libc/string/sparc/sparc32/sparcv9b/memchr.S b/libc/string/sparc/sparc32/sparcv9b/memchr.S new file mode 100644 index 000000000..7e86a2972 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/memchr.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include "sparc64/memchr.S" diff --git a/libc/string/sparc/sparc32/sparcv9b/memcpy.S b/libc/string/sparc/sparc32/sparcv9b/memcpy.S new file mode 100644 index 000000000..7f697542e --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/memcpy.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include "sparc64/sparcv9b/memcpy.S" diff --git a/libc/string/sparc/sparc32/sparcv9b/memset.S b/libc/string/sparc/sparc32/sparcv9b/memset.S new file mode 100644 index 000000000..ac67b7ab7 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/memset.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/memset.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/stpcpy.S b/libc/string/sparc/sparc32/sparcv9b/stpcpy.S new file mode 100644 index 000000000..440ad7e21 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/stpcpy.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/stpcpy.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/strcat.S b/libc/string/sparc/sparc32/sparcv9b/strcat.S new file mode 100644 index 000000000..7a2223570 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/strcat.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/strcat.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/strchr.S b/libc/string/sparc/sparc32/sparcv9b/strchr.S new file mode 100644 index 000000000..ddd32120d --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/strchr.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/strchr.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/strcmp.S b/libc/string/sparc/sparc32/sparcv9b/strcmp.S new file mode 100644 index 000000000..5330f4359 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/strcmp.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/strcmp.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/strcpy.S b/libc/string/sparc/sparc32/sparcv9b/strcpy.S new file mode 100644 index 000000000..0b35c9be0 --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/strcpy.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/strcpy.S> diff --git a/libc/string/sparc/sparc32/sparcv9b/strlen.S b/libc/string/sparc/sparc32/sparcv9b/strlen.S new file mode 100644 index 000000000..b8f4dba4f --- /dev/null +++ b/libc/string/sparc/sparc32/sparcv9b/strlen.S @@ -0,0 +1,4 @@ +#define ASI_PNF 0x82 +#define ASI_BLK_P 0xf0 +#define XCC icc +#include <sparc64/strlen.S> diff --git a/libc/string/sparc/sparc32/stpcpy.S b/libc/string/sparc/sparc32/stpcpy.S new file mode 100644 index 000000000..b7e7a00f8 --- /dev/null +++ b/libc/string/sparc/sparc32/stpcpy.S @@ -0,0 +1,161 @@ +/* Copy SRC to DEST returning the address of the terminating '\0' in DEST. + For SPARC v7. + Copyright (C) 1996, 1999, 2002, 2004 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Jakub Jelinek <jj@ultra.linux.cz>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Normally, this uses ((xword - 0x01010101) & 0x80808080) test + to find out if any byte in xword could be zero. This is fast, but + also gives false alarm for any byte in range 0x81-0xff. It does + not matter for correctness, as if this test tells us there could + be some zero byte, we check it byte by byte, but if bytes with + high bits set are common in the strings, then this will give poor + performance. You can #define EIGHTBIT_NOT_RARE and the algorithm + will use one tick slower, but more precise test + ((xword - 0x01010101) & (~xword) & 0x80808080), + which does not give any false alarms (but if some bits are set, + one cannot assume from it which bytes are zero and which are not). + It is yet to be measured, what is the correct default for glibc + in these days for an average user. + */ + + .text + .align 4 +10: ldub [%o1], %o5 + stb %o5, [%o0] + cmp %o5, 0 + add %o0, 1, %o0 + be 1f + add %o1, 1, %o1 + andcc %o1, 3, %g0 + be 4f + or %o4, %lo(0x80808080), %o3 + ldub [%o1], %o5 + stb %o5, [%o0] + cmp %o5, 0 + add %o0, 1, %o0 + be 1f + add %o1, 1, %o1 + andcc %o1, 3, %g0 + be 5f + sethi %hi(0x01010101), %o4 + ldub [%o1], %o5 + stb %o5, [%o0] + cmp %o5, 0 + add %o0, 1, %o0 + be 1f + add %o1, 1, %o1 + b 6f + or %o4, %lo(0x01010101), %o2 +1: retl + add %o0, -1, %o0 + +ENTRY(__stpcpy) + andcc %o1, 3, %g0 + bne 10b + sethi %hi(0x80808080), %o4 + or %o4, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %o4 +5: or %o4, %lo(0x01010101), %o2 +6: andcc %o0, 3, %g0 + bne 16f + sub %g0, 4, %g1 + +11: add %g1, 4, %g1 + ld [%o1 + %g1], %o5 + sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + andcc %o4, %o3, %g0 + be,a 11b + st %o5, [%o0 + %g1] + + /* Check every byte. */ + srl %o5, 24, %g5 + andcc %g5, 0xff, %g0 + be 14f + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + be 13f + srl %o5, 8, %g5 + andcc %g5, 0xff, %g0 + be 12f + andcc %o5, 0xff, %g0 + bne 11b + st %o5, [%o0 + %g1] + add %o0, %g1, %o0 + retl + add %o0, 3, %o0 +12: srl %o5, 16, %o5 + sth %o5, [%o0 + %g1] + add %g1, 2, %g1 + stb %g0, [%o0 + %g1] + retl + add %o0, %g1, %o0 +13: srl %o5, 16, %o5 + sth %o5, [%o0 + %g1] + add %g1, 1, %g1 + retl + add %o0, %g1, %o0 +14: stb %g0, [%o0 + %g1] + retl + add %o0, %g1, %o0 + +15: srl %o5, 24, %o4 + srl %o5, 16, %g1 + stb %o4, [%o0] + srl %o5, 8, %g4 + stb %g1, [%o0 + 1] + stb %g4, [%o0 + 2] + stb %o5, [%o0 + 3] + add %o0, 4, %o0 +16: ld [%o1], %o5 + sub %o5, %o2, %o4 + andcc %o4, %o3, %g0 + be 15b + add %o1, 4, %o1 + + /* Check every byte. */ + srl %o5, 24, %g5 + andcc %g5, 0xff, %g4 + be 19f + stb %g4, [%o0] + srl %o5, 16, %g5 + andcc %g5, 0xff, %g4 + be 18f + stb %g4, [%o0 + 1] + srl %o5, 8, %g5 + andcc %g5, 0xff, %g4 + be 17f + stb %g4, [%o0 + 2] + andcc %o5, 0xff, %g4 + stb %g4, [%o0 + 3] + bne 16b + add %o0, 4, %o0 + retl + sub %o0, 1, %o0 +17: retl + add %o0, 2, %o0 +18: retl + add %o0, 1, %o0 +19: retl + nop +END(__stpcpy) + +weak_alias (__stpcpy, stpcpy) diff --git a/libc/string/sparc/sparc32/strcat.S b/libc/string/sparc/sparc32/strcat.S new file mode 100644 index 000000000..e4ff21d01 --- /dev/null +++ b/libc/string/sparc/sparc32/strcat.S @@ -0,0 +1,346 @@ +/* strcat (dest, src) -- Append SRC on the end of DEST. + For SPARC v7. + Copyright (C) 1996, 1999, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Jakub Jelinek <jj@ultra.linux.cz>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Normally, this uses ((xword - 0x01010101) & 0x80808080) test + to find out if any byte in xword could be zero. This is fast, but + also gives false alarm for any byte in range 0x81-0xff. It does + not matter for correctness, as if this test tells us there could + be some zero byte, we check it byte by byte, but if bytes with + high bits set are common in the strings, then this will give poor + performance. You can #define EIGHTBIT_NOT_RARE and the algorithm + will use one tick slower, but more precise test + ((xword - 0x01010101) & (~xword) & 0x80808080), + which does not give any false alarms (but if some bits are set, + one cannot assume from it which bytes are zero and which are not). + It is yet to be measured, what is the correct default for glibc + in these days for an average user. + */ + + .text + .align 4 +10: cmp %o4, 2 + be 1f + cmp %o4, 3 + ldub [%o1], %o5 + add %o1, 1, %o1 + stb %o5, [%o0] + be 3f + cmp %o5, 0 + be 0f + add %o0, 1, %o0 +1: lduh [%o1], %o5 + add %o1, 2, %o1 + srl %o5, 8, %o4 + cmp %o4, 0 + stb %o4, [%o0] + bne,a 2f + stb %o5, [%o0 + 1] + retl + mov %g2, %o0 +2: andcc %o5, 0xff, %o5 + bne 4f + add %o0, 2, %o0 + retl + mov %g2, %o0 +3: bne 4f + add %o0, 1, %o0 + retl + mov %g2, %o0 + +11: ldub [%o0], %o5 + cmp %o5, 0 + be 1f + add %o0, 1, %o0 + andcc %o0, 3, %g0 + be 7f + or %o4, %lo(0x80808080), %o3 + ldub [%o0], %o5 + cmp %o5, 0 + be 2f + add %o0, 1, %o0 + andcc %o0, 3, %g0 + be 8f + sethi %hi(0x01010101), %o4 + ldub [%o0], %o5 + cmp %o5, 0 + be 3f + add %o0, 1, %o0 + b 9f + or %o4, %lo(0x01010101), %o2 +1: or %o4, %lo(0x80808080), %o3 +2: sethi %hi(0x01010101), %o4 +3: or %o4, %lo(0x01010101), %o2 + b 3f + sub %o0, 1, %o0 + +ENTRY(strcat) + mov %o0, %g2 + andcc %o0, 3, %g0 + bne 11b + sethi %hi(0x80808080), %o4 + or %o4, %lo(0x80808080), %o3 +7: sethi %hi(0x01010101), %o4 +8: or %o4, %lo(0x01010101), %o2 +9: ld [%o0], %o5 +7: sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + andcc %o4, %o3, %g0 + be 9b + add %o0, 4, %o0 + + srl %o5, 24, %g5 + andcc %g5, 0xff, %g0 + be 3f + add %o0, -4, %o0 + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + be 3f + add %o0, 1, %o0 + srl %o5, 8, %g5 + andcc %g5, 0xff, %g0 + be 3f + add %o0, 1, %o0 + andcc %o5, 0xff, %g0 + add %o0, 2, %o0 + bne,a 7b + ld [%o0], %o5 + sub %o0, 1, %o0 +3: andcc %o1, 3, %o4 + bne 10b +4: andcc %o0, 3, %g3 + bne 12f +1: ld [%o1], %o5 + add %o1, 4, %o1 + sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + add %o0, 4, %o0 + andcc %o4, %o3, %g0 + be,a 1b + st %o5, [%o0 - 4] + + srl %o5, 24, %g5 + andcc %g5, 0xff, %g0 + be 1f + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + be 2f + srl %o5, 8, %g5 + andcc %g5, 0xff, %g0 + be 3f + andcc %o5, 0xff, %g0 + bne 1b + st %o5, [%o0 - 4] + retl + mov %g2, %o0 +3: srl %o5, 16, %o5 + sth %o5, [%o0 - 4] + stb %g0, [%o0 - 2] + retl + mov %g2, %o0 +2: srl %o5, 16, %o5 + sth %o5, [%o0 - 4] + retl + mov %g2, %o0 +1: stb %g0, [%o0 - 4] + retl + mov %g2, %o0 + +12: add %o1, 4, %o1 + sub %o5, %o2, %o4 + cmp %g3, 2 + be 2f + cmp %g3, 3 + be 3f + andcc %o4, %o3, %g0 + bne 5f + srl %o5, 24, %g5 + stb %g5, [%o0] + sub %o0, 1, %o0 + srl %o5, 8, %g5 + sth %g5, [%o0 + 2] +1: add %o0, 4, %o0 +4: sll %o5, 24, %g6 + ld [%o1], %o5 + add %o1, 4, %o1 + srl %o5, 8, %g5 + sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + or %g5, %g6, %g5 + andcc %o4, %o3, %g0 + be,a 1b + st %g5, [%o0] + srl %o5, 24, %o4 + andcc %o4, 0xff, %g0 + be 6f + srl %o5, 16, %o4 + andcc %o4, 0xff, %g0 + be 7f + srl %o5, 8, %o4 + st %g5, [%o0] + andcc %o4, 0xff, %g0 + be 0f + andcc %o5, 0xff, %g0 +1: bne 4b + add %o0, 4, %o0 +9: stb %g0, [%o0] +0: retl + mov %g2, %o0 + +6: srl %g5, 16, %g5 + sth %g5, [%o0] + retl + mov %g2, %o0 + +7: srl %g5, 16, %g5 + sth %g5, [%o0] + stb %g0, [%o0 + 2] + retl + mov %g2, %o0 + +5: andcc %g5, 0xff, %g4 + be 9b + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + be 7f + srl %o5, 8, %g5 + andcc %g5, 0xff, %g0 + stb %g4, [%o0] + sth %g5, [%o0 + 1] + sub %o0, 1, %o0 + bne 1b + andcc %o5, 0xff, %g0 + retl + mov %g2, %o0 + +7: stb %g4, [%o0] + stb %g0, [%o0 + 1] + retl + mov %g2, %o0 + +2: andcc %o4, %o3, %g0 + bne 5f + srl %o5, 16, %g5 + sth %g5, [%o0] + sub %o0, 2, %o0 +1: add %o0, 4, %o0 +4: sll %o5, 16, %g6 + ld [%o1], %o5 + add %o1, 4, %o1 + srl %o5, 16, %g5 + sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + or %g5, %g6, %g5 + andcc %o4, %o3, %g0 + be,a 1b + st %g5, [%o0] + srl %o5, 24, %o4 + andcc %o4, 0xff, %g0 + be 7f + srl %o5, 16, %o4 + st %g5, [%o0] + andcc %o4, 0xff, %g0 + be 0b + srl %o5, 8, %o4 +1: andcc %o4, 0xff, %g0 + be 8f + andcc %o5, 0xff, %g0 + bne 4b + add %o0, 4, %o0 + sth %o5, [%o0] + retl + mov %g2, %o0 + +7: srl %g5, 16, %g5 + sth %g5, [%o0] + stb %g0, [%o0 + 2] + retl + mov %g2, %o0 + +8: stb %g0, [%o0 + 4] + retl + mov %g2, %o0 + +5: srl %o5, 24, %g5 + andcc %g5, 0xff, %g0 + be 9b + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + sth %g5, [%o0] + sub %o0, 2, %o0 + bne 1b + srl %o5, 8, %o4 + retl + mov %g2, %o0 + +3: bne 5f + srl %o5, 24, %g5 + stb %g5, [%o0] + sub %o0, 3, %o0 +1: add %o0, 4, %o0 +4: sll %o5, 8, %g6 + ld [%o1], %o5 + add %o1, 4, %o1 + srl %o5, 24, %g5 + sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + or %g5, %g6, %g5 + andcc %o4, %o3, %g0 + be 1b + st %g5, [%o0] + srl %o5, 24, %o4 + andcc %o4, 0xff, %g0 + be 0b + srl %o5, 16, %o4 +1: andcc %o4, 0xff, %g0 + be 8b + srl %o5, 8, %o4 + andcc %o4, 0xff, %g0 + be 9f + andcc %o5, 0xff, %g0 + bne 4b + add %o0, 4, %o0 + srl %o5, 8, %o5 + sth %o5, [%o0] + stb %g0, [%o0 + 2] + retl + mov %g2, %o0 +9: srl %o5, 8, %o5 + sth %o5, [%o0 + 4] + retl + mov %g2, %o0 +5: andcc %g5, 0xff, %g0 + stb %g5, [%o0] + sub %o0, 3, %o0 + bne 1b + srl %o5, 16, %o4 + retl + mov %g2, %o0 +END(strcat) diff --git a/libc/string/sparc/sparc32/strchr.S b/libc/string/sparc/sparc32/strchr.S new file mode 100644 index 000000000..f18a8abc7 --- /dev/null +++ b/libc/string/sparc/sparc32/strchr.S @@ -0,0 +1,280 @@ +/* strchr (str, ch) -- Return pointer to first occurrence of CH in STR. + For SPARC v7. + Copyright (C) 1996, 1999, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Jakub Jelinek <jj@ultra.linux.cz> and + David S. Miller <davem@caip.rutgers.edu>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Normally, this uses ((xword - 0x01010101) & 0x80808080) test + to find out if any byte in xword could be zero. This is fast, but + also gives false alarm for any byte in range 0x81-0xff. It does + not matter for correctness, as if this test tells us there could + be some zero byte, we check it byte by byte, but if bytes with + high bits set are common in the strings, then this will give poor + performance. You can #define EIGHTBIT_NOT_RARE and the algorithm + will use one tick slower, but more precise test + ((xword - 0x01010101) & (~xword) & 0x80808080), + which does not give any false alarms (but if some bits are set, + one cannot assume from it which bytes are zero and which are not). + It is yet to be measured, what is the correct default for glibc + in these days for an average user. + */ + + .text + .align 4 +10: ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + cmp %g4, 0 + be 9f + andcc %o0, 3, %g0 + be 4f + or %o4, %lo(0x80808080), %o3 + ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + cmp %g4, 0 + be 9f + andcc %o0, 3, %g0 + be 5f + sethi %hi(0x01010101), %o5 + ldub [%o0], %g4 + cmp %g4, %o1 + be 1f + add %o0, 1, %o0 + cmp %g4, 0 + be 9f + or %o5, %lo(0x01010101), %o2 + b 6f + ld [%o0], %g4 +1: retl + sub %o0, 1, %o0 + +ENTRY(strchr) + andcc %o1, 0xff, %o1 + be 12f + sll %o1, 8, %o2 + andcc %o0, 3, %g0 + or %o1, %o2, %o2 + sethi %hi(0x80808080), %o4 + sll %o2, 16, %o3 + bne 10b + or %o3, %o2, %g2 + or %o4, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %o5 +5: or %o5, %lo(0x01010101), %o2 +7: ld [%o0], %g4 +6: xor %g4, %g2, %g5 + sub %g4, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + sub %g5, %o2, %g6 + andn %o4, %g4, %o4 + andn %g6, %g5, %g5 +#else + sub %g5, %o2, %g5 +#endif + or %g5, %o4, %o4 + andcc %o4, %o3, %g0 + be 7b + add %o0, 4, %o0 + + /* Check every byte. */ +8: srl %g4, 24, %g5 +7: andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be 4f + srl %g4, 16, %g5 + andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be 3f + srl %g4, 8, %g5 + andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be 2f + andcc %g4, 0xff, %g5 + be 9f + cmp %g5, %o1 + bne,a 6b + ld [%o0], %g4 + retl + sub %o0, 1, %o0 +2: retl + sub %o0, 2, %o0 +3: retl + sub %o0, 3, %o0 +4: retl + sub %o0, 4, %o0 +9: retl + clr %o0 + +11: ldub [%o0], %o5 + cmp %o5, 0 + be 1f + add %o0, 1, %o0 + andcc %o0, 3, %g0 + be 4f + or %o4, %lo(0x80808080), %o3 + ldub [%o0], %o5 + cmp %o5, 0 + be 1f + add %o0, 1, %o0 + andcc %o0, 3, %g0 + be 5f + sethi %hi(0x01010101), %o4 + ldub [%o0], %o5 + cmp %o5, 0 + be 1f + add %o0, 1, %o0 + b 6f + or %o4, %lo(0x01010101), %o2 +1: retl + sub %o0, 1, %o0 + +12: andcc %o0, 3, %g0 + bne 11b + sethi %hi(0x80808080), %o4 + or %o4, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %o4 +5: or %o4, %lo(0x01010101), %o2 +6: ld [%o0], %o5 +7: sub %o5, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + andn %o4, %o5, %o4 +#endif + andcc %o4, %o3, %g0 + be 6b + add %o0, 4, %o0 + + /* Check every byte. */ + srl %o5, 24, %g5 + andcc %g5, 0xff, %g0 + be 8f + add %o0, -4, %o4 + srl %o5, 16, %g5 + andcc %g5, 0xff, %g0 + be 8f + add %o4, 1, %o4 + srl %o5, 8, %g5 + andcc %g5, 0xff, %g0 + be 8f + add %o4, 1, %o4 + andcc %o5, 0xff, %g0 + bne,a 7b + ld [%o0], %o5 + add %o4, 1, %o4 +8: retl + mov %o4, %o0 + +13: ldub [%o0], %g4 + cmp %g4, %o1 + add %o0, 1, %o0 + be,a 1f + sub %o0, 1, %o5 + cmp %g4, 0 + be 9f +1: andcc %o0, 3, %g0 + be 4f + or %o4, %lo(0x80808080), %o3 + ldub [%o0], %g4 + cmp %g4, %o1 + add %o0, 1, %o0 + be,a 1f + sub %o0, 1, %o5 + cmp %g4, 0 + be 9f +1: andcc %o0, 3, %g0 + be 5f + sethi %hi(0x01010101), %o4 + ldub [%o0], %g4 + cmp %g4, %o1 + add %o0, 1, %o0 + be,a 1f + sub %o0, 1, %o5 + cmp %g4, 0 + be 9f +1: or %o4, %lo(0x01010101), %o2 + b 7f + ld [%o0], %g4 +END(strchr) + +ENTRY(strrchr) + andcc %o1, 0xff, %o1 + clr %o5 + be 12b + sll %o1, 8, %o2 + andcc %o0, 3, %g0 + or %o1, %o2, %o2 + sethi %hi(0x80808080), %o4 + sll %o2, 16, %o3 + bne 13b + or %o3, %o2, %g2 + or %o4, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %o4 +5: or %o4, %lo(0x01010101), %o2 +6: ld [%o0], %g4 +7: xor %g4, %g2, %g5 + sub %g4, %o2, %o4 +#ifdef EIGHTBIT_NOT_RARE + sub %g5, %o2, %g6 + andn %o4, %g4, %o4 + andn %g6, %g5, %g5 +#else + sub %g5, %o2, %g5 +#endif + or %g5, %o4, %o4 + andcc %o4, %o3, %g0 + be 6b + add %o0, 4, %o0 + + /* Check every byte. */ +3: srl %g4, 24, %g5 +8: andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be,a 1f + sub %o0, 4, %o5 +1: srl %g4, 16, %g5 + andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be,a 1f + sub %o0, 3, %o5 +1: srl %g4, 8, %g5 + andcc %g5, 0xff, %g5 + be 9f + cmp %g5, %o1 + be,a 1f + sub %o0, 2, %o5 +1: andcc %g4, 0xff, %g5 + be 9f + cmp %g5, %o1 + be,a 1f + sub %o0, 1, %o5 +1: b 7b + ld [%o0], %g4 +9: retl + mov %o5, %o0 +END(strrchr) + +weak_alias (strchr, index) +weak_alias (strrchr, rindex) diff --git a/libc/string/sparc/sparc32/strcmp.S b/libc/string/sparc/sparc32/strcmp.S new file mode 100644 index 000000000..d6cf9e7d3 --- /dev/null +++ b/libc/string/sparc/sparc32/strcmp.S @@ -0,0 +1,255 @@ +/* Compare two strings for differences. + For SPARC v7. + Copyright (C) 1996, 97, 99, 2003 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Jakub Jelinek <jj@ultra.linux.cz>. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Normally, this uses ((xword - 0x01010101) & 0x80808080) test + to find out if any byte in xword could be zero. This is fast, but + also gives false alarm for any byte in range 0x81-0xff. It does + not matter for correctness, as if this test tells us there could + be some zero byte, we check it byte by byte, but if bytes with + high bits set are common in the strings, then this will give poor + performance. You can #define EIGHTBIT_NOT_RARE and the algorithm + will use one tick slower, but more precise test + ((xword - 0x01010101) & (~xword) & 0x80808080), + which does not give any false alarms (but if some bits are set, + one cannot assume from it which bytes are zero and which are not). + It is yet to be measured, what is the correct default for glibc + in these days for an average user. + */ + + .text + .align 4 +10: ldub [%o0], %o4 + add %o0, 1, %o0 + ldub [%o1], %o5 + cmp %o4, 0 + add %o1, 1, %o1 + be 2f + subcc %o4, %o5, %o4 + bne 2f + andcc %o0, 3, %g0 + be 4f + or %g1, %lo(0x80808080), %o3 + ldub [%o0], %o4 + add %o0, 1, %o0 + ldub [%o1], %o5 + cmp %o4, 0 + add %o1, 1, %o1 + be 2f + subcc %o4, %o5, %o4 + bne 2f + andcc %o0, 3, %g0 + be 5f + sethi %hi(0x01010101), %g1 + ldub [%o0], %o4 + add %o0, 1, %o0 + ldub [%o1], %o5 + cmp %o4, 0 + add %o1, 1, %o1 + be 2f + subcc %o4, %o5, %o4 + bne 2f + andcc %o1, 3, %g2 + bne 12f + or %g1, %lo(0x01010101), %o2 + b 1f + ld [%o0], %o4 +2: retl + mov %o4, %o0 + +ENTRY(strcmp) + andcc %o0, 3, %g0 + bne 10b + sethi %hi(0x80808080), %g1 + or %g1, %lo(0x80808080), %o3 +4: sethi %hi(0x01010101), %g1 +5: andcc %o1, 3, %g2 + bne 12f + or %g1, %lo(0x01010101), %o2 + +0: ld [%o0], %o4 +1: ld [%o1], %o5 + sub %o4, %o2, %g1 + add %o0, 4, %o0 + cmp %o4, %o5 +#ifdef EIGHTBIT_NOT_RARE + andn %g1, %o4, %g1 +#endif + bne 11f + andcc %g1, %o3, %g0 + be 0b + add %o1, 4, %o1 + + srl %o4, 24, %g4 + andcc %g4, 0xff, %g0 + be 2f + srl %o4, 16, %g4 + andcc %g4, 0xff, %g0 + be 2f + srl %o4, 8, %g4 + andcc %g4, 0xff, %g0 + be 2f + andcc %o4, 0xff, %g0 + bne,a 1b + ld [%o0], %o4 +2: retl + clr %o0 + +11: srl %o4, 24, %g4 + srl %o5, 24, %g5 + andcc %g4, 0xff, %g0 + be 3f + subcc %g4, %g5, %g4 + bne 3f + srl %o5, 16, %g5 + srl %o4, 16, %g4 + andcc %g4, 0xff, %g0 + be 3f + subcc %g4, %g5, %g4 + bne 3f + srl %o5, 8, %g5 + srl %o4, 8, %g4 + andcc %g4, 0xff, %g0 + be 3f + subcc %g4, %g5, %g4 + bne 3f + subcc %o4, %o5, %o4 + retl + mov %o4, %o0 +3: retl + mov %g4, %o0 + +12: save %sp, -64, %sp + ld [%i0], %i4 + sll %g2, 3, %g3 + andn %i1, 3, %i1 + mov 32, %l1 + ld [%i1], %l2 + mov -1, %g6 + add %i1, 4, %i1 + sub %l1, %g3, %l1 + sll %g6, %g3, %g6 + +1: sll %l2, %g3, %g5 + and %i4, %g6, %l3 + sub %i4, %i2, %g1 +#ifdef EIGHTBIT_NOT_RARE + andn %g1, %i4, %g1 +#endif + andcc %g1, %i3, %g1 + bne 3f + cmp %g5, %l3 + bne 2f + add %i0, 4, %i0 + ld [%i1], %l