diff options
Diffstat (limited to 'lib/libc/arm')
46 files changed, 5831 insertions, 0 deletions
diff --git a/lib/libc/arm/Makefile.inc b/lib/libc/arm/Makefile.inc new file mode 100644 index 0000000..4c72e56 --- /dev/null +++ b/lib/libc/arm/Makefile.inc @@ -0,0 +1,10 @@ +# $FreeBSD$ +# +# Machine dependent definitions for the arm architecture. +# + +SOFTFLOAT_BITS=32 + +# Long double is just double precision. +MDSRCS+=machdep_ldisd.c +SYM_MAPS+=${.CURDIR}/arm/Symbol.map diff --git a/lib/libc/arm/SYS.h b/lib/libc/arm/SYS.h new file mode 100644 index 0000000..584c151 --- /dev/null +++ b/lib/libc/arm/SYS.h @@ -0,0 +1,74 @@ +/* $NetBSD: SYS.h,v 1.8 2003/08/07 16:42:02 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * This code is derived from software contributed to Berkeley by + * William Jolitz. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)SYS.h 5.5 (Berkeley) 5/7/91 + * $FreeBSD$ + */ + +#include <machine/asm.h> +#include <sys/syscall.h> +#include <machine/swi.h> + +#define SYSTRAP(x) swi 0 | SYS_ ## x + +#define CERROR _C_LABEL(cerror) +#define CURBRK _C_LABEL(curbrk) + +#define _SYSCALL_NOERROR(x) \ + ENTRY(__CONCAT(__sys_, x)); \ + .weak _C_LABEL(x); \ + .set _C_LABEL(x), _C_LABEL(__CONCAT(__sys_,x)); \ + .weak _C_LABEL(__CONCAT(_,x)); \ + .set _C_LABEL(__CONCAT(_,x)),_C_LABEL(__CONCAT(__sys_,x)); \ + SYSTRAP(x) + +#define _SYSCALL(x) \ + _SYSCALL_NOERROR(x); \ + bcs PIC_SYM(CERROR, PLT) + +#define SYSCALL(x) \ + _SYSCALL(x) + +#define PSEUDO(x) \ + ENTRY(__CONCAT(__sys_, x)); \ + .weak _C_LABEL(__CONCAT(_,x)); \ + .set _C_LABEL(__CONCAT(_,x)),_C_LABEL(__CONCAT(__sys_,x)); \ + SYSTRAP(x); \ + bcs PIC_SYM(CERROR, PLT); \ + RET + +#define RSYSCALL(x) \ + _SYSCALL(x); \ + RET + + .globl CERROR diff --git a/lib/libc/arm/Symbol.map b/lib/libc/arm/Symbol.map new file mode 100644 index 0000000..6b7e6e4 --- /dev/null +++ b/lib/libc/arm/Symbol.map @@ -0,0 +1,79 @@ +/* + * $FreeBSD$ + */ + +/* + * This only needs to contain symbols that are not listed in + * symbol maps from other parts of libc (i.e., not found in + * stdlib/Symbol.map, string/Symbol.map, sys/Symbol.map, ...). + */ +FBSD_1.0 { + /* PSEUDO syscalls */ + _exit; + + __mcount; + _setjmp; + _longjmp; + alloca; + fabs; + __infinity; + __nan; + makecontext; + modf; + setjmp; + longjmp; + sigsetjmp; + siglongjmp; + htonl; + htons; + ntohl; + ntohs; + vfork; + brk; + cerror; /* XXX - Should this be .cerror (see sys/cerror.S)? */ + sbrk; +}; + +FBSDprivate_1.0 { + /* PSEUDO syscalls */ + __sys_getlogin; + _getlogin; + __sys_exit; + + _set_tp; + ___longjmp; + __umodsi3; + __modsi3; + __udivsi3; + __divsi3; + __makecontext; + __longjmp; + signalcontext; + _signalcontext; + __siglongjmp; + __sys_vfork; + _vfork; + _brk; + _end; + curbrk; + minbrk; + _sbrk; + + /* softfloat */ + __addsf3; + __adddf3; + __subsf3; + __subdf3; + __mulsf3; + __muldf3; + __divsf3; + __divdf3; + __floatsisf; + __floatsidf; + __fixsfsi; + __fixdfsi; + __fixunssfsi; + __fixunsdfsi; + __extendsfdf2; + __truncdfsf2; +}; diff --git a/lib/libc/arm/_fpmath.h b/lib/libc/arm/_fpmath.h new file mode 100644 index 0000000..b0ce8f4 --- /dev/null +++ b/lib/libc/arm/_fpmath.h @@ -0,0 +1,66 @@ +/*- + * Copyright (c) 2002, 2003 David Schultz <das@FreeBSD.ORG> + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#if defined(__VFP_FP__) +#define _IEEE_WORD_ORDER _BYTE_ORDER +#else +#define _IEEE_WORD_ORDER _BIG_ENDIAN +#endif + +union IEEEl2bits { + long double e; + struct { +#if _BYTE_ORDER == _LITTLE_ENDIAN +#if _IEEE_WORD_ORDER == _LITTLE_ENDIAN + unsigned int manl :32; +#endif + unsigned int manh :20; + unsigned int exp :11; + unsigned int sign :1; +#if _IEEE_WORD_ORDER == _BIG_ENDIAN + unsigned int manl :32; +#endif +#else /* _BYTE_ORDER == _LITTLE_ENDIAN */ + unsigned int sign :1; + unsigned int exp :11; + unsigned int manh :20; + unsigned int manl :32; +#endif + } bits; +}; + +#define LDBL_NBIT 0 +#define mask_nbit_l(u) ((void)0) + +#define LDBL_MANH_SIZE 32 +#define LDBL_MANL_SIZE 32 + +#define LDBL_TO_ARRAY32(u, a) do { \ + (a)[0] = (uint32_t)(u).bits.manl; \ + (a)[1] = (uint32_t)(u).bits.manh; \ +} while(0) diff --git a/lib/libc/arm/arith.h b/lib/libc/arm/arith.h new file mode 100644 index 0000000..be78d86 --- /dev/null +++ b/lib/libc/arm/arith.h @@ -0,0 +1,21 @@ +/* + * MD header for contrib/gdtoa + * + * $FreeBSD$ + */ + +/* + * NOTE: The definitions in this file must be correct or strtod(3) and + * floating point formats in printf(3) will break! The file can be + * generated by running contrib/gdtoa/arithchk.c on the target + * architecture. See contrib/gdtoa/gdtoaimp.h for details. + */ + +#if !defined(__ARMEB__) && defined(__VFP_FP__) +#define IEEE_8087 +#define Arith_Kind_ASL 1 +#define Sudden_Underflow +#else +#define IEEE_MC68k +#define Arith_Kind_ASL 2 +#endif diff --git a/lib/libc/arm/gd_qnan.h b/lib/libc/arm/gd_qnan.h new file mode 100644 index 0000000..8fa9edd --- /dev/null +++ b/lib/libc/arm/gd_qnan.h @@ -0,0 +1,23 @@ +/* + * MD header for contrib/gdtoa + * + * This file can be generated by compiling and running contrib/gdtoa/qnan.c + * on the target architecture after arith.h has been generated. + * + * XXX I don't have ARM hardware, so I just guessed. --das + * + * $FreeBSD$ + */ + +#define f_QNAN 0x7fc00000 +#define d_QNAN0 0x0 +#define d_QNAN1 0x7ff80000 +#define ld_QNAN0 0x0 +#define ld_QNAN1 0xc0000000 +#define ld_QNAN2 0x7fff +#define ld_QNAN3 0x0 +#define ldus_QNAN0 0x0 +#define ldus_QNAN1 0x0 +#define ldus_QNAN2 0x0 +#define ldus_QNAN3 0xc000 +#define ldus_QNAN4 0x7fff diff --git a/lib/libc/arm/gen/Makefile.inc b/lib/libc/arm/gen/Makefile.inc new file mode 100644 index 0000000..bbe2257 --- /dev/null +++ b/lib/libc/arm/gen/Makefile.inc @@ -0,0 +1,6 @@ +# @(#)Makefile.inc 8.1 (Berkeley) 6/4/93 +# $FreeBSD$ + +SRCS+= _ctx_start.S _setjmp.S _set_tp.c alloca.S fabs.c \ + infinity.c ldexp.c makecontext.c modf.c \ + setjmp.S signalcontext.c sigsetjmp.S divsi3.S diff --git a/lib/libc/arm/gen/_ctx_start.S b/lib/libc/arm/gen/_ctx_start.S new file mode 100644 index 0000000..fbde357 --- /dev/null +++ b/lib/libc/arm/gen/_ctx_start.S @@ -0,0 +1,9 @@ +#include <machine/asm.h> + +.ident "$FreeBSD$" +ENTRY(_ctx_start) + mov lr, pc + mov pc, r4 + mov r0, r5 + bl _C_LABEL(ctx_done) + bl _C_LABEL(abort) diff --git a/lib/libc/arm/gen/_set_tp.c b/lib/libc/arm/gen/_set_tp.c new file mode 100644 index 0000000..4c26dff --- /dev/null +++ b/lib/libc/arm/gen/_set_tp.c @@ -0,0 +1,35 @@ +/*- + * Copyright (c) 2004 Doug Rabson + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#include <string.h> +#include <stdint.h> + +void +_set_tp(void *tp) +{ +} diff --git a/lib/libc/arm/gen/_setjmp.S b/lib/libc/arm/gen/_setjmp.S new file mode 100644 index 0000000..6594afb --- /dev/null +++ b/lib/libc/arm/gen/_setjmp.S @@ -0,0 +1,110 @@ +/* $NetBSD: _setjmp.S,v 1.5 2003/04/05 23:08:51 bjh21 Exp $ */ + +/* + * Copyright (c) 1997 Mark Brinicombe + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +/* + * C library -- _setjmp, _longjmp + * + * _longjmp(a,v) + * will generate a "return(v)" from the last call to + * _setjmp(a) + * by restoring registers from the stack. + * The previous signal state is NOT restored. + * + * Note: r0 is the return value + * r1-r3 are scratch registers in functions + */ + +ENTRY(_setjmp) + ldr r1, .L_setjmp_magic + str r1, [r0], #4 +#ifdef __SOFTFP__ + add r0, r0, #52 +#else + /* Store fp registers */ + sfm f4, 4, [r0], #48 + /* Store fpsr */ + rfs r1 + str r1, [r0], #0x0004 +#endif /* __SOFTFP__ */ + /* Store integer registers */ + stmia r0, {r4-r14} + + mov r0, #0x00000000 + RET + +.L_setjmp_magic: + .word _JB_MAGIC__SETJMP + +WEAK_ALIAS(___longjmp, _longjmp) +ENTRY(_longjmp) + ldr r2, .L_setjmp_magic + ldr r3, [r0], #4 + teq r2, r3 + bne botch + +#ifdef __SOFTFP__ + add r0, r0, #52 +#else + /* Restore fp registers */ + lfm f4, 4, [r0], #48 + /* Restore fpsr */ + ldr r4, [r0], #0x0004 + wfs r4 +#endif /* __SOFTFP__ */ + /* Restore integer registers */ + ldmia r0, {r4-r14} + + /* Validate sp and r14 */ + teq sp, #0 + teqne r14, #0 + beq botch + + /* Set return value */ + mov r0, r1 + teq r0, #0x00000000 + moveq r0, #0x00000001 + RET + + /* validation failed, die die die. */ +botch: +#if !defined(_STANDALONE) + bl PIC_SYM(_C_LABEL(longjmperror), PLT) + bl PIC_SYM(_C_LABEL(abort), PLT) + b . - 8 /* Cannot get here */ +#else + b . +#endif diff --git a/lib/libc/arm/gen/alloca.S b/lib/libc/arm/gen/alloca.S new file mode 100644 index 0000000..9569d86 --- /dev/null +++ b/lib/libc/arm/gen/alloca.S @@ -0,0 +1,45 @@ +/* $NetBSD: alloca.S,v 1.3 2003/04/05 23:08:51 bjh21 Exp $ */ + +/* + * Copyright (c) 1995 Mark Brinicombe + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +/* like alloc, but automatic automatic free in return */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +ENTRY(alloca) + add r0, r0, #0x00000007 /* round up to next 8 byte alignment */ + bic r0, r0, #0x00000007 + sub sp, sp, r0 /* Adjust the stack pointer */ + mov r0, sp /* r0 = base of new space */ + RET diff --git a/lib/libc/arm/gen/divsi3.S b/lib/libc/arm/gen/divsi3.S new file mode 100644 index 0000000..104a958 --- /dev/null +++ b/lib/libc/arm/gen/divsi3.S @@ -0,0 +1,387 @@ +/* $NetBSD: divsi3.S,v 1.4 2003/04/05 23:27:15 bjh21 Exp $ */ + +/* + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +/* + * stack is aligned as there's a possibility of branching to L_overflow + * which makes a C call + */ + +ENTRY(__umodsi3) + stmfd sp!, {lr} + sub sp, sp, #4 /* align stack */ + bl .L_udivide + add sp, sp, #4 /* unalign stack */ + mov r0, r1 + ldmfd sp!, {pc} + +ENTRY(__modsi3) + stmfd sp!, {lr} + sub sp, sp, #4 /* align stack */ + bl .L_divide + add sp, sp, #4 /* unalign stack */ + mov r0, r1 + ldmfd sp!, {pc} + +.L_overflow: +#if !defined(_KERNEL) && !defined(_STANDALONE) + mov r0, #8 /* SIGFPE */ + bl PIC_SYM(_C_LABEL(raise), PLT) /* raise it */ + mov r0, #0 +#else + /* XXX should cause a fatal error */ + mvn r0, #0 +#endif + RET + +ENTRY(__udivsi3) +.L_udivide: /* r0 = r0 / r1; r1 = r0 % r1 */ + eor r0, r1, r0 + eor r1, r0, r1 + eor r0, r1, r0 + /* r0 = r1 / r0; r1 = r1 % r0 */ + cmp r0, #1 + bcc .L_overflow + beq .L_divide_l0 + mov ip, #0 + movs r1, r1 + bpl .L_divide_l1 + orr ip, ip, #0x20000000 /* ip bit 0x20000000 = -ve r1 */ + movs r1, r1, lsr #1 + orrcs ip, ip, #0x10000000 /* ip bit 0x10000000 = bit 0 of r1 */ + b .L_divide_l1 + +.L_divide_l0: /* r0 == 1 */ + mov r0, r1 + mov r1, #0 + RET + +ENTRY(__divsi3) +.L_divide: /* r0 = r0 / r1; r1 = r0 % r1 */ + eor r0, r1, r0 + eor r1, r0, r1 + eor r0, r1, r0 + /* r0 = r1 / r0; r1 = r1 % r0 */ + cmp r0, #1 + bcc .L_overflow + beq .L_divide_l0 + ands ip, r0, #0x80000000 + rsbmi r0, r0, #0 + ands r2, r1, #0x80000000 + eor ip, ip, r2 + rsbmi r1, r1, #0 + orr ip, r2, ip, lsr #1 /* ip bit 0x40000000 = -ve division */ + /* ip bit 0x80000000 = -ve remainder */ + +.L_divide_l1: + mov r2, #1 + mov r3, #0 + + /* + * If the highest bit of the dividend is set, we have to be + * careful when shifting the divisor. Test this. + */ + movs r1,r1 + bpl .L_old_code + + /* + * At this point, the highest bit of r1 is known to be set. + * We abuse this below in the tst instructions. + */ + tst r1, r0 /*, lsl #0 */ + bmi .L_divide_b1 + tst r1, r0, lsl #1 + bmi .L_divide_b2 + tst r1, r0, lsl #2 + bmi .L_divide_b3 + tst r1, r0, lsl #3 + bmi .L_divide_b4 + tst r1, r0, lsl #4 + bmi .L_divide_b5 + tst r1, r0, lsl #5 + bmi .L_divide_b6 + tst r1, r0, lsl #6 + bmi .L_divide_b7 + tst r1, r0, lsl #7 + bmi .L_divide_b8 + tst r1, r0, lsl #8 + bmi .L_divide_b9 + tst r1, r0, lsl #9 + bmi .L_divide_b10 + tst r1, r0, lsl #10 + bmi .L_divide_b11 + tst r1, r0, lsl #11 + bmi .L_divide_b12 + tst r1, r0, lsl #12 + bmi .L_divide_b13 + tst r1, r0, lsl #13 + bmi .L_divide_b14 + tst r1, r0, lsl #14 + bmi .L_divide_b15 + tst r1, r0, lsl #15 + bmi .L_divide_b16 + tst r1, r0, lsl #16 + bmi .L_divide_b17 + tst r1, r0, lsl #17 + bmi .L_divide_b18 + tst r1, r0, lsl #18 + bmi .L_divide_b19 + tst r1, r0, lsl #19 + bmi .L_divide_b20 + tst r1, r0, lsl #20 + bmi .L_divide_b21 + tst r1, r0, lsl #21 + bmi .L_divide_b22 + tst r1, r0, lsl #22 + bmi .L_divide_b23 + tst r1, r0, lsl #23 + bmi .L_divide_b24 + tst r1, r0, lsl #24 + bmi .L_divide_b25 + tst r1, r0, lsl #25 + bmi .L_divide_b26 + tst r1, r0, lsl #26 + bmi .L_divide_b27 + tst r1, r0, lsl #27 + bmi .L_divide_b28 + tst r1, r0, lsl #28 + bmi .L_divide_b29 + tst r1, r0, lsl #29 + bmi .L_divide_b30 + tst r1, r0, lsl #30 + bmi .L_divide_b31 +/* + * instead of: + * tst r1, r0, lsl #31 + * bmi .L_divide_b32 + */ + b .L_divide_b32 + +.L_old_code: + cmp r1, r0 + bcc .L_divide_b0 + cmp r1, r0, lsl #1 + bcc .L_divide_b1 + cmp r1, r0, lsl #2 + bcc .L_divide_b2 + cmp r1, r0, lsl #3 + bcc .L_divide_b3 + cmp r1, r0, lsl #4 + bcc .L_divide_b4 + cmp r1, r0, lsl #5 + bcc .L_divide_b5 + cmp r1, r0, lsl #6 + bcc .L_divide_b6 + cmp r1, r0, lsl #7 + bcc .L_divide_b7 + cmp r1, r0, lsl #8 + bcc .L_divide_b8 + cmp r1, r0, lsl #9 + bcc .L_divide_b9 + cmp r1, r0, lsl #10 + bcc .L_divide_b10 + cmp r1, r0, lsl #11 + bcc .L_divide_b11 + cmp r1, r0, lsl #12 + bcc .L_divide_b12 + cmp r1, r0, lsl #13 + bcc .L_divide_b13 + cmp r1, r0, lsl #14 + bcc .L_divide_b14 + cmp r1, r0, lsl #15 + bcc .L_divide_b15 + cmp r1, r0, lsl #16 + bcc .L_divide_b16 + cmp r1, r0, lsl #17 + bcc .L_divide_b17 + cmp r1, r0, lsl #18 + bcc .L_divide_b18 + cmp r1, r0, lsl #19 + bcc .L_divide_b19 + cmp r1, r0, lsl #20 + bcc .L_divide_b20 + cmp r1, r0, lsl #21 + bcc .L_divide_b21 + cmp r1, r0, lsl #22 + bcc .L_divide_b22 + cmp r1, r0, lsl #23 + bcc .L_divide_b23 + cmp r1, r0, lsl #24 + bcc .L_divide_b24 + cmp r1, r0, lsl #25 + bcc .L_divide_b25 + cmp r1, r0, lsl #26 + bcc .L_divide_b26 + cmp r1, r0, lsl #27 + bcc .L_divide_b27 + cmp r1, r0, lsl #28 + bcc .L_divide_b28 + cmp r1, r0, lsl #29 + bcc .L_divide_b29 + cmp r1, r0, lsl #30 + bcc .L_divide_b30 +.L_divide_b32: + cmp r1, r0, lsl #31 + subhs r1, r1,r0, lsl #31 + addhs r3, r3,r2, lsl #31 +.L_divide_b31: + cmp r1, r0, lsl #30 + subhs r1, r1,r0, lsl #30 + addhs r3, r3,r2, lsl #30 +.L_divide_b30: + cmp r1, r0, lsl #29 + subhs r1, r1,r0, lsl #29 + addhs r3, r3,r2, lsl #29 +.L_divide_b29: + cmp r1, r0, lsl #28 + subhs r1, r1,r0, lsl #28 + addhs r3, r3,r2, lsl #28 +.L_divide_b28: + cmp r1, r0, lsl #27 + subhs r1, r1,r0, lsl #27 + addhs r3, r3,r2, lsl #27 +.L_divide_b27: + cmp r1, r0, lsl #26 + subhs r1, r1,r0, lsl #26 + addhs r3, r3,r2, lsl #26 +.L_divide_b26: + cmp r1, r0, lsl #25 + subhs r1, r1,r0, lsl #25 + addhs r3, r3,r2, lsl #25 +.L_divide_b25: + cmp r1, r0, lsl #24 + subhs r1, r1,r0, lsl #24 + addhs r3, r3,r2, lsl #24 +.L_divide_b24: + cmp r1, r0, lsl #23 + subhs r1, r1,r0, lsl #23 + addhs r3, r3,r2, lsl #23 +.L_divide_b23: + cmp r1, r0, lsl #22 + subhs r1, r1,r0, lsl #22 + addhs r3, r3,r2, lsl #22 +.L_divide_b22: + cmp r1, r0, lsl #21 + subhs r1, r1,r0, lsl #21 + addhs r3, r3,r2, lsl #21 +.L_divide_b21: + cmp r1, r0, lsl #20 + subhs r1, r1,r0, lsl #20 + addhs r3, r3,r2, lsl #20 +.L_divide_b20: + cmp r1, r0, lsl #19 + subhs r1, r1,r0, lsl #19 + addhs r3, r3,r2, lsl #19 +.L_divide_b19: + cmp r1, r0, lsl #18 + subhs r1, r1,r0, lsl #18 + addhs r3, r3,r2, lsl #18 +.L_divide_b18: + cmp r1, r0, lsl #17 + subhs r1, r1,r0, lsl #17 + addhs r3, r3,r2, lsl #17 +.L_divide_b17: + cmp r1, r0, lsl #16 + subhs r1, r1,r0, lsl #16 + addhs r3, r3,r2, lsl #16 +.L_divide_b16: + cmp r1, r0, lsl #15 + subhs r1, r1,r0, lsl #15 + addhs r3, r3,r2, lsl #15 +.L_divide_b15: + cmp r1, r0, lsl #14 + subhs r1, r1,r0, lsl #14 + addhs r3, r3,r2, lsl #14 +.L_divide_b14: + cmp r1, r0, lsl #13 + subhs r1, r1,r0, lsl #13 + addhs r3, r3,r2, lsl #13 +.L_divide_b13: + cmp r1, r0, lsl #12 + subhs r1, r1,r0, lsl #12 + addhs r3, r3,r2, lsl #12 +.L_divide_b12: + cmp r1, r0, lsl #11 + subhs r1, r1,r0, lsl #11 + addhs r3, r3,r2, lsl #11 +.L_divide_b11: + cmp r1, r0, lsl #10 + subhs r1, r1,r0, lsl #10 + addhs r3, r3,r2, lsl #10 +.L_divide_b10: + cmp r1, r0, lsl #9 + subhs r1, r1,r0, lsl #9 + addhs r3, r3,r2, lsl #9 +.L_divide_b9: + cmp r1, r0, lsl #8 + subhs r1, r1,r0, lsl #8 + addhs r3, r3,r2, lsl #8 +.L_divide_b8: + cmp r1, r0, lsl #7 + subhs r1, r1,r0, lsl #7 + addhs r3, r3,r2, lsl #7 +.L_divide_b7: + cmp r1, r0, lsl #6 + subhs r1, r1,r0, lsl #6 + addhs r3, r3,r2, lsl #6 +.L_divide_b6: + cmp r1, r0, lsl #5 + subhs r1, r1,r0, lsl #5 + addhs r3, r3,r2, lsl #5 +.L_divide_b5: + cmp r1, r0, lsl #4 + subhs r1, r1,r0, lsl #4 + addhs r3, r3,r2, lsl #4 +.L_divide_b4: + cmp r1, r0, lsl #3 + subhs r1, r1,r0, lsl #3 + addhs r3, r3,r2, lsl #3 +.L_divide_b3: + cmp r1, r0, lsl #2 + subhs r1, r1,r0, lsl #2 + addhs r3, r3,r2, lsl #2 +.L_divide_b2: + cmp r1, r0, lsl #1 + subhs r1, r1,r0, lsl #1 + addhs r3, r3,r2, lsl #1 +.L_divide_b1: + cmp r1, r0 + subhs r1, r1, r0 + addhs r3, r3, r2 +.L_divide_b0: + + tst ip, #0x20000000 + bne .L_udivide_l1 + mov r0, r3 + cmp ip, #0 + rsbmi r1, r1, #0 + movs ip, ip, lsl #1 + bicmi r0, r0, #0x80000000 /* Fix incase we divided 0x80000000 */ + rsbmi r0, r0, #0 + RET + +.L_udivide_l1: + tst ip, #0x10000000 + mov r1, r1, lsl #1 + orrne r1, r1, #1 + mov r3, r3, lsl #1 + cmp r1, r0 + subhs r1, r1, r0 + addhs r3, r3, r2 + mov r0, r3 + RET diff --git a/lib/libc/arm/gen/fabs.c b/lib/libc/arm/gen/fabs.c new file mode 100644 index 0000000..8bb1502 --- /dev/null +++ b/lib/libc/arm/gen/fabs.c @@ -0,0 +1,46 @@ +/* $NetBSD: fabs.c,v 1.2 2002/05/26 11:48:01 wiz Exp $ */ + +/* + * Copyright (c) 1996 Mark Brinicombe + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +/* + * fabs(x) returns the absolute value of x. + */ +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +double +fabs(double x) +{ + if (x < 0) + x = -x; + return(x); +} diff --git a/lib/libc/arm/gen/infinity.c b/lib/libc/arm/gen/infinity.c new file mode 100644 index 0000000..60faf42 --- /dev/null +++ b/lib/libc/arm/gen/infinity.c @@ -0,0 +1,26 @@ +/* + * infinity.c + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <math.h> + +/* bytes for +Infinity on a 387 */ +const union __infinity_un __infinity = { +#if BYTE_ORDER == BIG_ENDIAN + { 0x7f, 0xf0, 0, 0, 0, 0, 0, 0 } +#else + { 0, 0, 0, 0, 0, 0, 0xf0, 0x7f } +#endif +}; + +/* bytes for NaN */ +const union __nan_un __nan = { +#if BYTE_ORDER == BIG_ENDIAN + {0xff, 0xc0, 0, 0} +#else + { 0, 0, 0xc0, 0xff } +#endif +}; diff --git a/lib/libc/arm/gen/makecontext.c b/lib/libc/arm/gen/makecontext.c new file mode 100644 index 0000000..4b0f92d --- /dev/null +++ b/lib/libc/arm/gen/makecontext.c @@ -0,0 +1,89 @@ +/* $NetBSD: makecontext.c,v 1.2 2003/01/18 11:06:24 thorpej Exp $ */ + +/*- + * Copyright (c) 2001 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Klaus Klein. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <stdlib.h> +#include <stddef.h> +#include <inttypes.h> +#include <ucontext.h> + +#include <stdarg.h> + +extern void _ctx_start(void); + +void +ctx_done(ucontext_t *ucp) +{ + + if (ucp->uc_link == NULL) + exit(0); + else { + setcontext((const ucontext_t *)ucp->uc_link); + abort(); + } +} + +__weak_reference(__makecontext, makecontext); + +void +__makecontext(ucontext_t *ucp, void (*func)(void), int argc, ...) +{ + __greg_t *gr = ucp->uc_mcontext.__gregs; + int i; + unsigned int *sp; + va_list ap; + + /* Compute and align stack pointer. */ + sp = (unsigned int *) + (((uintptr_t)ucp->uc_stack.ss_sp + ucp->uc_stack.ss_size - + sizeof(double)) & ~7); + /* Allocate necessary stack space for arguments exceeding r0-3. */ + if (argc > 4) + sp -= argc - 4; + gr[_REG_SP] = (__greg_t)sp; + /* Wipe out frame pointer. */ + gr[_REG_FP] = 0; + /* Arrange for return via the trampoline code. */ + gr[_REG_PC] = (__greg_t)_ctx_start; + gr[_REG_R4] = (__greg_t)func; + gr[_REG_R5] = (__greg_t)ucp; + + va_start(ap, argc); + /* Pass up to four arguments in r0-3. */ + for (i = 0; i < argc && i < 4; i++) + gr[_REG_R0 + i] = va_arg(ap, int); + /* Pass any additional arguments on the stack. */ + for (argc -= i; argc > 0; argc--) + *sp++ = va_arg(ap, int); + va_end(ap); +} diff --git a/lib/libc/arm/gen/modf.c b/lib/libc/arm/gen/modf.c new file mode 100644 index 0000000..347090c --- /dev/null +++ b/lib/libc/arm/gen/modf.c @@ -0,0 +1,107 @@ +/* + * Copyright (c) 1994, 1995 Carnegie-Mellon University. + * All rights reserved. + * + * Author: Chris G. Demetriou + * + * Permission to use, copy, modify and distribute this software and + * its documentation is hereby granted, provided that both the copyright + * notice and this permission notice appear in all copies of the + * software, derivative works or modified versions, and any portions + * thereof, and that both notices appear in supporting documentation. + * + * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" + * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND + * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. + * + * Carnegie Mellon requests users of this software to return to + * + * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU + * School of Computer Science + * Carnegie Mellon University + * Pittsburgh PA 15213-3890 + * + * any improvements or extensions that they make and grant Carnegie the + * rights to redistribute these changes. + * + * $NetBSD: modf.c,v 1.1 1995/02/10 17:50:25 cgd Exp $ + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <sys/types.h> +#include <errno.h> +#include <math.h> +#include <machine/ieee.h> + +/* + * double modf(double val, double *iptr) + * returns: f and i such that |f| < 1.0, (f + i) = val, and + * sign(f) == sign(i) == sign(val). + * + * Beware signedness when doing subtraction, and also operand size! + */ +double +modf(val, iptr) + double val, *iptr; +{ + union doub { + double v; + struct ieee_double s; + } u, v; + u_int64_t frac; + + /* + * If input is Inf or NaN, return it and leave i alone. + */ + u.v = val; + if (u.s.dbl_exp == DBL_EXP_INFNAN) + return (u.v); + + /* + * If input can't have a fractional part, return + * (appropriately signed) zero, and make i be the input. + */ + if ((int)u.s.dbl_exp - DBL_EXP_BIAS > DBL_FRACBITS - 1) { + *iptr = u.v; + v.v = 0.0; + v.s.dbl_sign = u.s.dbl_sign; + return (v.v); + } + + /* + * If |input| < 1.0, return it, and set i to the appropriately + * signed zero. + */ + if (u.s.dbl_exp < DBL_EXP_BIAS) { + v.v = 0.0; + v.s.dbl_sign = u.s.dbl_sign; + *iptr = v.v; + return (u.v); + } + + /* + * There can be a fractional part of the input. + * If you look at the math involved for a few seconds, it's + * plain to see that the integral part is the input, with the + * low (DBL_FRACBITS - (exponent - DBL_EXP_BIAS)) bits zeroed, + * the the fractional part is the part with the rest of the + * bits zeroed. Just zeroing the high bits to get the + * fractional part would yield a fraction in need of + * normalization. Therefore, we take the easy way out, and + * just use subtraction to get the fractional part. + */ + v.v = u.v; + /* Zero the low bits of the fraction, the sleazy way. */ + frac = ((u_int64_t)v.s.dbl_frach << 32) + v.s.dbl_fracl; + frac >>= DBL_FRACBITS - (u.s.dbl_exp - DBL_EXP_BIAS); + frac <<= DBL_FRACBITS - (u.s.dbl_exp - DBL_EXP_BIAS); + v.s.dbl_fracl = frac & 0xffffffff; + v.s.dbl_frach = frac >> 32; + *iptr = v.v; + + u.v -= v.v; + u.s.dbl_sign = v.s.dbl_sign; + return (u.v); +} diff --git a/lib/libc/arm/gen/setjmp.S b/lib/libc/arm/gen/setjmp.S new file mode 100644 index 0000000..e5f581f --- /dev/null +++ b/lib/libc/arm/gen/setjmp.S @@ -0,0 +1,129 @@ +/* $NetBSD: setjmp.S,v 1.5 2003/04/05 23:08:51 bjh21 Exp $ */ + +/* + * Copyright (c) 1997 Mark Brinicombe + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +/* + * C library -- setjmp, longjmp + * + * longjmp(a,v) + * will generate a "return(v)" from the last call to + * setjmp(a) + * by restoring registers from the stack. + * The previous signal state is restored. + */ + +ENTRY(setjmp) + /* Block all signals and retrieve the old signal mask */ + stmfd sp!, {r0, r14} + add r2, r0, #(25 * 4) /* oset */ + mov r0, #0x00000001 /* SIG_BLOCK */ + mov r1, #0 /* set */ + + bl PIC_SYM(_C_LABEL(sigprocmask), PLT) + + ldmfd sp!, {r0, r14} + + ldr r1, .Lsetjmp_magic + str r1, [r0], #4 + +#ifdef __SOFTFP__ + add r0, r0, #52 +#else + /* Store fp registers */ + sfm f4, 4, [r0], #48 + /* Store fpsr */ + rfs r1 + str r1, [r0], #0x0004 +#endif /* __SOFTFP__ */ + /* Store integer registers */ + stmia r0, {r4-r14} + mov r0, #0x00000000 + RET + +.Lsetjmp_magic: + .word _JB_MAGIC_SETJMP + + +.weak _C_LABEL(longjmp) +.set _C_LABEL(longjmp), _C_LABEL(__longjmp) +ENTRY(__longjmp) + ldr r2, .Lsetjmp_magic + ldr r3, [r0] + teq r2, r3 + bne botch + + + /* Set signal mask */ + stmfd sp!, {r0, r1, r14} + sub sp, sp, #4 /* align the stack */ + + add r1, r0, #(25 * 4) /* Signal mask */ + mov r0, #3 /* SIG_SETMASK */ + mov r2, #0 + bl PIC_SYM(_C_LABEL(sigprocmask), PLT) + + add sp, sp, #4 /* unalign the stack */ + ldmfd sp!, {r0, r1, r14} + + add r0, r0, #4 +#ifdef __SOFTFP__ + add r0, r0, #52 +#else + /* Restore fp registers */ + lfm f4, 4, [r0], #48 + /* Restore FPSR */ + ldr r4, [r0], #0x0004 + wfs r4 +#endif /* __SOFTFP__ */ + /* Restore integer registers */ + ldmia r0, {r4-r14} + + /* Validate sp and r14 */ + teq sp, #0 + teqne r14, #0 + beq botch + + /* Set return value */ + + mov r0, r1 + teq r0, #0x00000000 + moveq r0, #0x00000001 + RET + + /* validation failed, die die die. */ +botch: + bl PIC_SYM(_C_LABEL(longjmperror), PLT) + bl PIC_SYM(_C_LABEL(abort), PLT) + b . - 8 /* Cannot get here */ diff --git a/lib/libc/arm/gen/signalcontext.c b/lib/libc/arm/gen/signalcontext.c new file mode 100644 index 0000000..bb9a54d --- /dev/null +++ b/lib/libc/arm/gen/signalcontext.c @@ -0,0 +1,78 @@ +/*- + * Copyright (c) 2004 Olivier Houchard + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <sys/param.h> +#include <sys/signal.h> +#include <sys/ucontext.h> + +#include <machine/frame.h> +#include <machine/sigframe.h> + +#include <errno.h> +#include <stdarg.h> +#include <stdlib.h> +#include <unistd.h> +#include <strings.h> +#include <signal.h> + +__weak_reference(__signalcontext, signalcontext); + +extern void _ctx_start(void); + +int +__signalcontext(ucontext_t *ucp, int sig, __sighandler_t *func) +{ + struct sigframe *sfp; + __greg_t *gr = ucp->uc_mcontext.__gregs; + unsigned int *sp; + + sp = (unsigned int *)gr[_REG_SP]; + + sfp = (struct sigframe *)sp - 1; + + bzero(sfp, sizeof(*sfp)); + bcopy(ucp, &sfp->sf_uc, sizeof(*ucp)); + sfp->sf_si.si_signo = sig; + + gr[_REG_SP] = (__greg_t)sfp; + /* Wipe out frame pointer. */ + gr[_REG_FP] = 0; + /* Arrange for return via the trampoline code. */ + gr[_REG_PC] = (__greg_t)_ctx_start; + gr[_REG_R4] = (__greg_t)func; + gr[_REG_R5] = (__greg_t)ucp; + gr[_REG_R0] = (__greg_t)sig; + gr[_REG_R1] = (__greg_t)&sfp->sf_si; + gr[_REG_R2] = (__greg_t)&sfp->sf_uc; + + ucp->uc_link = &sfp->sf_uc; + sigdelset(&ucp->uc_sigmask, sig); + + return (0); +} diff --git a/lib/libc/arm/gen/sigsetjmp.S b/lib/libc/arm/gen/sigsetjmp.S new file mode 100644 index 0000000..09ec256 --- /dev/null +++ b/lib/libc/arm/gen/sigsetjmp.S @@ -0,0 +1,62 @@ +/* $NetBSD: sigsetjmp.S,v 1.3 2002/08/17 19:54:30 thorpej Exp $ */ + +/* + * Copyright (c) 1997 Mark Brinicombe + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +/* + * C library -- sigsetjmp, siglongjmp + * + * longjmp(a,v) + * will generate a "return(v)" from the last call to + * setjmp(a, m) + * by restoring registers from the stack. + * The previous signal state is restored. + */ + +ENTRY(sigsetjmp) + teq r1, #0 + beq PIC_SYM(_C_LABEL(_setjmp), PLT) + b PIC_SYM(_C_LABEL(setjmp), PLT) + +.L_setjmp_magic: + .word _JB_MAGIC__SETJMP +WEAK_ALIAS(__siglongjmp, siglongjmp) + +ENTRY(siglongjmp) + ldr r2, .L_setjmp_magic + ldr r3, [r0] + teq r2, r3 + beq PIC_SYM(_C_LABEL(_longjmp), PLT) + b PIC_SYM(_C_LABEL(longjmp), PLT) diff --git a/lib/libc/arm/softfloat/arm-gcc.h b/lib/libc/arm/softfloat/arm-gcc.h new file mode 100644 index 0000000..1204108 --- /dev/null +++ b/lib/libc/arm/softfloat/arm-gcc.h @@ -0,0 +1,101 @@ +/* $NetBSD: arm-gcc.h,v 1.2 2001/02/21 18:09:25 bjh21 Exp $ */ +/* $FreeBSD$ */ + +/* +------------------------------------------------------------------------------- +One of the macros `BIGENDIAN' or `LITTLEENDIAN' must be defined. +------------------------------------------------------------------------------- +*/ +#ifdef __ARMEB__ +#define BIGENDIAN +#else +#define LITTLEENDIAN +#endif + +/* +------------------------------------------------------------------------------- +The macro `BITS64' can be defined to indicate that 64-bit integer types are +supported by the compiler. +------------------------------------------------------------------------------- +*/ +#define BITS64 + +/* +------------------------------------------------------------------------------- +Each of the following `typedef's defines the most convenient type that holds +integers of at least as many bits as specified. For example, `uint8' should +be the most convenient type that can hold unsigned integers of as many as +8 bits. The `flag' type must be able to hold either a 0 or 1. For most +implementations of C, `flag', `uint8', and `int8' should all be `typedef'ed +to the same as `int'. +------------------------------------------------------------------------------- +*/ +typedef int flag; +typedef int uint8; +typedef int int8; +typedef int uint16; +typedef int int16; +typedef unsigned int uint32; +typedef signed int int32; +#ifdef BITS64 +typedef unsigned long long int uint64; +typedef signed long long int int64; +#endif + +/* +------------------------------------------------------------------------------- +Each of the following `typedef's defines a type that holds integers +of _exactly_ the number of bits specified. For instance, for most +implementation of C, `bits16' and `sbits16' should be `typedef'ed to +`unsigned short int' and `signed short int' (or `short int'), respectively. +------------------------------------------------------------------------------- +*/ +typedef unsigned char bits8; +typedef signed char sbits8; +typedef unsigned short int bits16; +typedef signed short int sbits16; +typedef unsigned int bits32; +typedef signed int sbits32; +#ifdef BITS64 +typedef unsigned long long int bits64; +typedef signed long long int sbits64; +#endif + +#ifdef BITS64 +/* +------------------------------------------------------------------------------- +The `LIT64' macro takes as its argument a textual integer literal and +if necessary ``marks'' the literal as having a 64-bit integer type. +For example, the GNU C Compiler (`gcc') requires that 64-bit literals be +appended with the letters `LL' standing for `long long', which is `gcc's +name for the 64-bit integer type. Some compilers may allow `LIT64' to be +defined as the identity macro: `#define LIT64( a ) a'. +------------------------------------------------------------------------------- +*/ +#define LIT64( a ) a##LL +#endif + +/* +------------------------------------------------------------------------------- +The macro `INLINE' can be used before functions that should be inlined. If +a compiler does not support explicit inlining, this macro should be defined +to be `static'. +------------------------------------------------------------------------------- +*/ +#define INLINE static __inline + +/* +------------------------------------------------------------------------------- +The ARM FPA is odd in that it stores doubles high-order word first, no matter +what the endianness of the CPU. VFP is sane. +------------------------------------------------------------------------------- +*/ +#if defined(SOFTFLOAT_FOR_GCC) +#if defined(__VFP_FP__) || defined(__ARMEB__) +#define FLOAT64_DEMANGLE(a) (a) +#define FLOAT64_MANGLE(a) (a) +#else +#define FLOAT64_DEMANGLE(a) (((a) << 32) | ((a) >> 32)) +#define FLOAT64_MANGLE(a) FLOAT64_DEMANGLE(a) +#endif +#endif diff --git a/lib/libc/arm/softfloat/milieu.h b/lib/libc/arm/softfloat/milieu.h new file mode 100644 index 0000000..54775c7 --- /dev/null +++ b/lib/libc/arm/softfloat/milieu.h @@ -0,0 +1,49 @@ +/* $NetBSD: milieu.h,v 1.1 2000/12/29 20:13:54 bjh21 Exp $ */ +/* $FreeBSD$ */ + +/* +=============================================================================== + +This C header file is part of the SoftFloat IEC/IEEE Floating-point +Arithmetic Package, Release 2a. + +Written by John R. Hauser. This work was made possible in part by the +International Computer Science Institute, located at Suite 600, 1947 Center +Street, Berkeley, California 94704. Funding was partially provided by the +National Science Foundation under grant MIP-9311980. The original version +of this code was written as part of a project to build a fixed-point vector +processor in collaboration with the University of California at Berkeley, +overseen by Profs. Nelson Morgan and John Wawrzynek. More information +is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/ +arithmetic/SoftFloat.html'. + +THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort +has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT +TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO +PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY +AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE. + +Derivative works are acceptable, even for commercial purposes, so long as +(1) they include prominent notice that the work is derivative, and (2) they +include prominent notice akin to these four paragraphs for those parts of +this code that are retained. + +=============================================================================== +*/ + +/* +------------------------------------------------------------------------------- +Include common integer types and flags. +------------------------------------------------------------------------------- +*/ +#include "arm-gcc.h" + +/* +------------------------------------------------------------------------------- +Symbolic Boolean literals. +------------------------------------------------------------------------------- +*/ +enum { + FALSE = 0, + TRUE = 1 +}; diff --git a/lib/libc/arm/softfloat/softfloat.h b/lib/libc/arm/softfloat/softfloat.h new file mode 100644 index 0000000..50b5fa6 --- /dev/null +++ b/lib/libc/arm/softfloat/softfloat.h @@ -0,0 +1,315 @@ +/* $NetBSD: softfloat.h,v 1.6 2002/05/12 13:12:46 bjh21 Exp $ */ +/* $FreeBSD$ */ + +/* This is a derivative work. */ + +/* +=============================================================================== + +This C header file is part of the SoftFloat IEC/IEEE Floating-point +Arithmetic Package, Release 2a. + +Written by John R. Hauser. This work was made possible in part by the +International Computer Science Institute, located at Suite 600, 1947 Center +Street, Berkeley, California 94704. Funding was partially provided by the +National Science Foundation under grant MIP-9311980. The original version +of this code was written as part of a project to build a fixed-point vector +processor in collaboration with the University of California at Berkeley, +overseen by Profs. Nelson Morgan and John Wawrzynek. More information +is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/ +arithmetic/SoftFloat.html'. + +THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort +has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT +TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO +PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY +AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE. + +Derivative works are acceptable, even for commercial purposes, so long as +(1) they include prominent notice that the work is derivative, and (2) they +include prominent notice akin to these four paragraphs for those parts of +this code that are retained. + +=============================================================================== +*/ + +/* +------------------------------------------------------------------------------- +The macro `FLOATX80' must be defined to enable the extended double-precision +floating-point format `floatx80'. If this macro is not defined, the +`floatx80' type will not be defined, and none of the functions that either +input or output the `floatx80' type will be defined. The same applies to +the `FLOAT128' macro and the quadruple-precision format `float128'. +------------------------------------------------------------------------------- +*/ +/* #define FLOATX80 */ +/* #define FLOAT128 */ + +#include <machine/ieeefp.h> + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE floating-point types. +------------------------------------------------------------------------------- +*/ +typedef unsigned int float32; +typedef unsigned long long float64; +#ifdef FLOATX80 +typedef struct { + unsigned short high; + unsigned long long low; +} floatx80; +#endif +#ifdef FLOAT128 +typedef struct { + unsigned long long high, low; +} float128; +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE floating-point underflow tininess-detection mode. +------------------------------------------------------------------------------- +*/ +#ifndef SOFTFLOAT_FOR_GCC +extern int float_detect_tininess; +#endif +enum { + float_tininess_after_rounding = 0, + float_tininess_before_rounding = 1 +}; + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE floating-point rounding mode. +------------------------------------------------------------------------------- +*/ +extern fp_rnd_t float_rounding_mode; +enum { + float_round_nearest_even = FP_RN, + float_round_to_zero = FP_RZ, + float_round_down = FP_RM, + float_round_up = FP_RP +}; + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE floating-point exception flags. +------------------------------------------------------------------------------- +*/ +extern fp_except float_exception_flags; +extern fp_except float_exception_mask; +enum { + float_flag_inexact = FP_X_IMP, + float_flag_underflow = FP_X_UFL, + float_flag_overflow = FP_X_OFL, + float_flag_divbyzero = FP_X_DZ, + float_flag_invalid = FP_X_INV +}; + +/* +------------------------------------------------------------------------------- +Routine to raise any or all of the software IEC/IEEE floating-point +exception flags. +------------------------------------------------------------------------------- +*/ +void float_raise( fp_except ); + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE integer-to-floating-point conversion routines. +------------------------------------------------------------------------------- +*/ +float32 int32_to_float32( int ); +float64 int32_to_float64( int ); +#ifdef FLOATX80 +floatx80 int32_to_floatx80( int ); +#endif +#ifdef FLOAT128 +float128 int32_to_float128( int ); +#endif +#ifndef SOFTFLOAT_FOR_GCC /* __floatdi?f is in libgcc2.c */ +float32 int64_to_float32( long long ); +float64 int64_to_float64( long long ); +#ifdef FLOATX80 +floatx80 int64_to_floatx80( long long ); +#endif +#ifdef FLOAT128 +float128 int64_to_float128( long long ); +#endif +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE single-precision conversion routines. +------------------------------------------------------------------------------- +*/ +int float32_to_int32( float32 ); +int float32_to_int32_round_to_zero( float32 ); +#if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS) +unsigned int float32_to_uint32_round_to_zero( float32 ); +#endif +#ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */ +long long float32_to_int64( float32 ); +long long float32_to_int64_round_to_zero( float32 ); +#endif +float64 float32_to_float64( float32 ); +#ifdef FLOATX80 +floatx80 float32_to_floatx80( float32 ); +#endif +#ifdef FLOAT128 +float128 float32_to_float128( float32 ); +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE single-precision operations. +------------------------------------------------------------------------------- +*/ +float32 float32_round_to_int( float32 ); +float32 float32_add( float32, float32 ); +float32 float32_sub( float32, float32 ); +float32 float32_mul( float32, float32 ); +float32 float32_div( float32, float32 ); +float32 float32_rem( float32, float32 ); +float32 float32_sqrt( float32 ); +int float32_eq( float32, float32 ); +int float32_le( float32, float32 ); +int float32_lt( float32, float32 ); +int float32_eq_signaling( float32, float32 ); +int float32_le_quiet( float32, float32 ); +int float32_lt_quiet( float32, float32 ); +#ifndef SOFTFLOAT_FOR_GCC +int float32_is_signaling_nan( float32 ); +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE double-precision conversion routines. +------------------------------------------------------------------------------- +*/ +int float64_to_int32( float64 ); +int float64_to_int32_round_to_zero( float64 ); +#if defined(SOFTFLOAT_FOR_GCC) && defined(SOFTFLOAT_NEED_FIXUNS) +unsigned int float64_to_uint32_round_to_zero( float64 ); +#endif +#ifndef SOFTFLOAT_FOR_GCC /* __fix?fdi provided by libgcc2.c */ +long long float64_to_int64( float64 ); +long long float64_to_int64_round_to_zero( float64 ); +#endif +float32 float64_to_float32( float64 ); +#ifdef FLOATX80 +floatx80 float64_to_floatx80( float64 ); +#endif +#ifdef FLOAT128 +float128 float64_to_float128( float64 ); +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE double-precision operations. +------------------------------------------------------------------------------- +*/ +float64 float64_round_to_int( float64 ); +float64 float64_add( float64, float64 ); +float64 float64_sub( float64, float64 ); +float64 float64_mul( float64, float64 ); +float64 float64_div( float64, float64 ); +float64 float64_rem( float64, float64 ); +float64 float64_sqrt( float64 ); +int float64_eq( float64, float64 ); +int float64_le( float64, float64 ); +int float64_lt( float64, float64 ); +int float64_eq_signaling( float64, float64 ); +int float64_le_quiet( float64, float64 ); +int float64_lt_quiet( float64, float64 ); +#ifndef SOFTFLOAT_FOR_GCC +int float64_is_signaling_nan( float64 ); +#endif + +#ifdef FLOATX80 + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE extended double-precision conversion routines. +------------------------------------------------------------------------------- +*/ +int floatx80_to_int32( floatx80 ); +int floatx80_to_int32_round_to_zero( floatx80 ); +long long floatx80_to_int64( floatx80 ); +long long floatx80_to_int64_round_to_zero( floatx80 ); +float32 floatx80_to_float32( floatx80 ); +float64 floatx80_to_float64( floatx80 ); +#ifdef FLOAT128 +float128 floatx80_to_float128( floatx80 ); +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE extended double-precision rounding precision. Valid +values are 32, 64, and 80. +------------------------------------------------------------------------------- +*/ +extern int floatx80_rounding_precision; + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE extended double-precision operations. +------------------------------------------------------------------------------- +*/ +floatx80 floatx80_round_to_int( floatx80 ); +floatx80 floatx80_add( floatx80, floatx80 ); +floatx80 floatx80_sub( floatx80, floatx80 ); +floatx80 floatx80_mul( floatx80, floatx80 ); +floatx80 floatx80_div( floatx80, floatx80 ); +floatx80 floatx80_rem( floatx80, floatx80 ); +floatx80 floatx80_sqrt( floatx80 ); +int floatx80_eq( floatx80, floatx80 ); +int floatx80_le( floatx80, floatx80 ); +int floatx80_lt( floatx80, floatx80 ); +int floatx80_eq_signaling( floatx80, floatx80 ); +int floatx80_le_quiet( floatx80, floatx80 ); +int floatx80_lt_quiet( floatx80, floatx80 ); +int floatx80_is_signaling_nan( floatx80 ); + +#endif + +#ifdef FLOAT128 + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE quadruple-precision conversion routines. +------------------------------------------------------------------------------- +*/ +int float128_to_int32( float128 ); +int float128_to_int32_round_to_zero( float128 ); +long long float128_to_int64( float128 ); +long long float128_to_int64_round_to_zero( float128 ); +float32 float128_to_float32( float128 ); +float64 float128_to_float64( float128 ); +#ifdef FLOATX80 +floatx80 float128_to_floatx80( float128 ); +#endif + +/* +------------------------------------------------------------------------------- +Software IEC/IEEE quadruple-precision operations. +------------------------------------------------------------------------------- +*/ +float128 float128_round_to_int( float128 ); +float128 float128_add( float128, float128 ); +float128 float128_sub( float128, float128 ); +float128 float128_mul( float128, float128 ); +float128 float128_div( float128, float128 ); +float128 float128_rem( float128, float128 ); +float128 float128_sqrt( float128 ); +int float128_eq( float128, float128 ); +int float128_le( float128, float128 ); +int float128_lt( float128, float128 ); +int float128_eq_signaling( float128, float128 ); +int float128_le_quiet( float128, float128 ); +int float128_lt_quiet( float128, float128 ); +int float128_is_signaling_nan( float128 ); + +#endif + diff --git a/lib/libc/arm/string/Makefile.inc b/lib/libc/arm/string/Makefile.inc new file mode 100644 index 0000000..6658ce7 --- /dev/null +++ b/lib/libc/arm/string/Makefile.inc @@ -0,0 +1,7 @@ +# @(#)Makefile.inc 8.1 (Berkeley) 6/4/93 +# $FreeBSD$ + +MDSRCS+=bcmp.c bcopy.S bzero.S ffs.S index.c memchr.c memcmp.S memcpy.S \ + memmove.S memset.S rindex.c strcat.c strchr.c strcmp.S strcpy.c \ + strlen.S strncmp.S strrchr.c swab.c wcschr.c wcscmp.c wcslen.c \ + wmemchr.c diff --git a/lib/libc/arm/string/bcopy.S b/lib/libc/arm/string/bcopy.S new file mode 100644 index 0000000..f2583fc --- /dev/null +++ b/lib/libc/arm/string/bcopy.S @@ -0,0 +1,6 @@ +/* $NetBSD: bcopy.S,v 1.3 2003/10/14 07:51:45 scw Exp $ */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#define _BCOPY +#include "memmove.S" diff --git a/lib/libc/arm/string/bzero.S b/lib/libc/arm/string/bzero.S new file mode 100644 index 0000000..b31358d --- /dev/null +++ b/lib/libc/arm/string/bzero.S @@ -0,0 +1,37 @@ +/* $NetBSD: bzero.S,v 1.3 2003/10/14 07:51:45 scw Exp $ */ + +/*- + * Copyright (c) 1997 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Neil A. Carson and Mark Brinicombe + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> + +__FBSDID("$FreeBSD$"); + +#define _BZERO +#include "memset.S" diff --git a/lib/libc/arm/string/ffs.S b/lib/libc/arm/string/ffs.S new file mode 100644 index 0000000..af4e118 --- /dev/null +++ b/lib/libc/arm/string/ffs.S @@ -0,0 +1,82 @@ +/* $NetBSD: ffs.S,v 1.5 2003/04/05 23:08:52 bjh21 Exp $ */ +/* + * Copyright (c) 2001 Christopher Gilbert + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the company nor the name of the author may be used to + * endorse or promote products derived from this software without specific + * prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, + * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR + * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> + +__FBSDID("$FreeBSD$"); + +/* + * ffs - find first set bit, this algorithm isolates the first set + * bit, then multiplies the number by 0x0450fbaf which leaves the top + * 6 bits as an index into the table. This algorithm should be a win + * over the checking each bit in turn as per the C compiled version. + * + * under ARMv5 there's an instruction called CLZ (count leading Zero's) that + * could be used + * + * This is the ffs algorithm devised by d.seal and posted to comp.sys.arm on + * 16 Feb 1994. + */ + +ENTRY(ffs) + /* Standard trick to isolate bottom bit in r0 or 0 if r0 = 0 on entry */ + rsb r1, r0, #0 + ands r0, r0, r1 +#ifndef _ARM_ARCH_5 + /* + * now r0 has at most one set bit, call this X + * if X = 0, all further instructions are skipped + */ + adrne r2, .L_ffs_table + orrne r0, r0, r0, lsl #4 /* r0 = X * 0x11 */ + orrne r0, r0, r0, lsl #6 /* r0 = X * 0x451 */ + rsbne r0, r0, r0, lsl #16 /* r0 = X * 0x0450fbaf */ + + /* now lookup in table indexed on top 6 bits of r0 */ + ldrneb r0, [ r2, r0, lsr #26 ] + + RET +.text; +.type .L_ffs_table, _ASM_TYPE_OBJECT; +.L_ffs_table: +/* 0 1 2 3 4 5 6 7 */ + .byte 0, 1, 2, 13, 3, 7, 0, 14 /* 0- 7 */ + .byte 4, 0, 8, 0, 0, 0, 0, 15 /* 8-15 */ + .byte 11, 5, 0, 0, 9, 0, 0, 26 /* 16-23 */ + .byte 0, 0, 0, 0, 0, 22, 28, 16 /* 24-31 */ + .byte 32, 12, 6, 0, 0, 0, 0, 0 /* 32-39 */ + .byte 10, 0, 0, 25, 0, 0, 21, 27 /* 40-47 */ + .byte 31, 0, 0, 0, 0, 24, 0, 20 /* 48-55 */ + .byte 30, 0, 23, 19, 29, 18, 17, 0 /* 56-63 */ +#else + clzne r0, r0 + rsbne r0, r0, #32 + RET +#endif diff --git a/lib/libc/arm/string/memcmp.S b/lib/libc/arm/string/memcmp.S new file mode 100644 index 0000000..a81c960 --- /dev/null +++ b/lib/libc/arm/string/memcmp.S @@ -0,0 +1,180 @@ +/* $NetBSD: memcmp.S,v 1.3 2003/10/14 07:51:45 scw Exp $ */ + +/* + * Copyright 2003 Wasabi Systems, Inc. + * All rights reserved. + * + * Written by Steve C. Woodford for Wasabi Systems, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed for the NetBSD Project by + * Wasabi Systems, Inc. + * 4. The name of Wasabi Systems, Inc. may not be used to endorse + * or promote products derived from this software without specific prior + * written permission. + * + * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ +/* + * Copyright (c) 2002 ARM Ltd + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the company may not be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED + * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED + * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF + * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING + * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> + +__FBSDID("$FreeBSD$"); + +ENTRY(memcmp) + mov ip, r0 +#if defined(_KERNEL) && !defined(_STANDALONE) + cmp r2, #0x06 + beq .Lmemcmp_6bytes +#endif + mov r0, #0x00 + + /* Are both addresses aligned the same way? */ + cmp r2, #0x00 + eornes r3, ip, r1 + RETeq /* len == 0, or same addresses! */ + tst r3, #0x03 + subne r2, r2, #0x01 + bne .Lmemcmp_bytewise2 /* Badly aligned. Do it the slow way */ + + /* Word-align the addresses, if necessary */ + sub r3, r1, #0x05 + ands r3, r3, #0x03 + add r3, r3, r3, lsl #1 + addne pc, pc, r3, lsl #3 + nop + + /* Compare up to 3 bytes */ + ldrb r0, [ip], #0x01 + ldrb r3, [r1], #0x01 + subs r0, r0, r3 + RETne + subs r2, r2, #0x01 + RETeq + + /* Compare up to 2 bytes */ + ldrb r0, [ip], #0x01 + ldrb r3, [r1], #0x01 + subs r0, r0, r3 + RETne + subs r2, r2, #0x01 + RETeq + + /* Compare 1 byte */ + ldrb r0, [ip], #0x01 + ldrb r3, [r1], #0x01 + subs r0, r0, r3 + RETne + subs r2, r2, #0x01 + RETeq + + /* Compare 4 bytes at a time, if possible */ + subs r2, r2, #0x04 + bcc .Lmemcmp_bytewise +.Lmemcmp_word_aligned: + ldr r0, [ip], #0x04 + ldr r3, [r1], #0x04 + subs r2, r2, #0x04 + cmpcs r0, r3 + beq .Lmemcmp_word_aligned + sub r0, r0, r3 + + /* Correct for extra subtraction, and check if done */ + adds r2, r2, #0x04 + cmpeq r0, #0x00 /* If done, did all bytes match? */ + RETeq /* Yup. Just return */ + + /* Re-do the final word byte-wise */ + sub ip, ip, #0x04 + sub r1, r1, #0x04 + +.Lmemcmp_bytewise: + add r2, r2, #0x03 +.Lmemcmp_bytewise2: + ldrb r0, [ip], #0x01 + ldrb r3, [r1], #0x01 + subs r2, r2, #0x01 + cmpcs r0, r3 + beq .Lmemcmp_bytewise2 + sub r0, r0, r3 + RET + +#if defined(_KERNEL) && !defined(_STANDALONE) + /* + * 6 byte compares are very common, thanks to the network stack. + * This code is hand-scheduled to reduce the number of stalls for + * load results. Everything else being equal, this will be ~32% + * faster than a byte-wise memcmp. + */ + .align 5 +.Lmemcmp_6bytes: + ldrb r3, [r1, #0x00] /* r3 = b2#0 */ + ldrb r0, [ip, #0x00] /* r0 = b1#0 */ + ldrb r2, [r1, #0x01] /* r2 = b2#1 */ + subs r0, r0, r3 /* r0 = b1#0 - b2#0 */ + ldreqb r3, [ip, #0x01] /* r3 = b1#1 */ + RETne /* Return if mismatch on #0 */ + subs r0, r3, r2 /* r0 = b1#1 - b2#1 */ + ldreqb r3, [r1, #0x02] /* r3 = b2#2 */ + ldreqb r0, [ip, #0x02] /* r0 = b1#2 */ + RETne /* Return if mismatch on #1 */ + ldrb r2, [r1, #0x03] /* r2 = b2#3 */ + subs r0, r0, r3 /* r0 = b1#2 - b2#2 */ + ldreqb r3, [ip, #0x03] /* r3 = b1#3 */ + RETne /* Return if mismatch on #2 */ + subs r0, r3, r2 /* r0 = b1#3 - b2#3 */ + ldreqb r3, [r1, #0x04] /* r3 = b2#4 */ + ldreqb r0, [ip, #0x04] /* r0 = b1#4 */ + RETne /* Return if mismatch on #3 */ + ldrb r2, [r1, #0x05] /* r2 = b2#5 */ + subs r0, r0, r3 /* r0 = b1#4 - b2#4 */ + ldreqb r3, [ip, #0x05] /* r3 = b1#5 */ + RETne /* Return if mismatch on #4 */ + sub r0, r3, r2 /* r0 = b1#5 - b2#5 */ + RET +#endif diff --git a/lib/libc/arm/string/memcpy.S b/lib/libc/arm/string/memcpy.S new file mode 100644 index 0000000..3628652 --- /dev/null +++ b/lib/libc/arm/string/memcpy.S @@ -0,0 +1,9 @@ +/* $NetBSD: memcpy.S,v 1.4 2003/10/14 07:51:45 scw Exp $ */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#if !defined(_ARM_ARCH_5E) || defined(_STANDALONE) +#include "memcpy_arm.S" +#else +#include "memcpy_xscale.S" +#endif diff --git a/lib/libc/arm/string/memcpy_arm.S b/lib/libc/arm/string/memcpy_arm.S new file mode 100644 index 0000000..b84a32e --- /dev/null +++ b/lib/libc/arm/string/memcpy_arm.S @@ -0,0 +1,332 @@ +/* $NetBSD: memcpy_arm.S,v 1.1 2003/10/14 07:51:45 scw Exp $ */ + +/*- + * Copyright (c) 1997 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Neil A. Carson and Mark Brinicombe + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +/* + * This is one fun bit of code ... + * Some easy listening music is suggested while trying to understand this + * code e.g. Iron Maiden + * + * For anyone attempting to understand it : + * + * The core code is implemented here with simple stubs for memcpy(). + * + * All local labels are prefixed with Lmemcpy_ + * Following the prefix a label starting f is used in the forward copy code + * while a label using b is used in the backwards copy code + * The source and destination addresses determine whether a forward or + * backward copy is performed. + * Separate bits of code are used to deal with the following situations + * for both the forward and backwards copy. + * unaligned source address + * unaligned destination address + * Separate copy routines are used to produce an optimised result for each + * of these cases. + * The copy code will use LDM/STM instructions to copy up to 32 bytes at + * a time where possible. + * + * Note: r12 (aka ip) can be trashed during the function along with + * r0-r3 although r0-r2 have defined uses i.e. src, dest, len through out. + * Additional registers are preserved prior to use i.e. r4, r5 & lr + * + * Apologies for the state of the comments ;-) + */ +/* LINTSTUB: Func: void *memcpy(void *dst, const void *src, size_t len) */ +ENTRY(memcpy) + /* save leaf functions having to store this away */ + stmdb sp!, {r0, lr} /* memcpy() returns dest addr */ + + subs r2, r2, #4 + blt .Lmemcpy_l4 /* less than 4 bytes */ + ands r12, r0, #3 + bne .Lmemcpy_destul /* oh unaligned destination addr */ + ands r12, r1, #3 + bne .Lmemcpy_srcul /* oh unaligned source addr */ + +.Lmemcpy_t8: + /* We have aligned source and destination */ + subs r2, r2, #8 + blt .Lmemcpy_l12 /* less than 12 bytes (4 from above) */ + subs r2, r2, #0x14 + blt .Lmemcpy_l32 /* less than 32 bytes (12 from above) */ + stmdb sp!, {r4} /* borrow r4 */ + + /* blat 32 bytes at a time */ + /* XXX for really big copies perhaps we should use more registers */ +.Lmemcpy_loop32: + ldmia r1!, {r3, r4, r12, lr} + stmia r0!, {r3, r4, r12, lr} + ldmia r1!, {r3, r4, r12, lr} + stmia r0!, {r3, r4, r12, lr} + subs r2, r2, #0x20 + bge .Lmemcpy_loop32 + + cmn r2, #0x10 + ldmgeia r1!, {r3, r4, r12, lr} /* blat a remaining 16 bytes */ + stmgeia r0!, {r3, r4, r12, lr} + subge r2, r2, #0x10 + ldmia sp!, {r4} /* return r4 */ + +.Lmemcpy_l32: + adds r2, r2, #0x14 + + /* blat 12 bytes at a time */ +.Lmemcpy_loop12: + ldmgeia r1!, {r3, r12, lr} + stmgeia r0!, {r3, r12, lr} + subges r2, r2, #0x0c + bge .Lmemcpy_loop12 + +.Lmemcpy_l12: + adds r2, r2, #8 + blt .Lmemcpy_l4 + + subs r2, r2, #4 + ldrlt r3, [r1], #4 + strlt r3, [r0], #4 + ldmgeia r1!, {r3, r12} + stmgeia r0!, {r3, r12} + subge r2, r2, #4 + +.Lmemcpy_l4: + /* less than 4 bytes to go */ + adds r2, r2, #4 +#ifdef __APCS_26_ + ldmeqia sp!, {r0, pc}^ /* done */ +#else + ldmeqia sp!, {r0, pc} /* done */ +#endif + /* copy the crud byte at a time */ + cmp r2, #2 + ldrb r3, [r1], #1 + strb r3, [r0], #1 + ldrgeb r3, [r1], #1 + strgeb r3, [r0], #1 + ldrgtb r3, [r1], #1 + strgtb r3, [r0], #1 + ldmia sp!, {r0, pc} + + /* erg - unaligned destination */ +.Lmemcpy_destul: + rsb r12, r12, #4 + cmp r12, #2 + + /* align destination with byte copies */ + ldrb r3, [r1], #1 + strb r3, [r0], #1 + ldrgeb r3, [r1], #1 + strgeb r3, [r0], #1 + ldrgtb r3, [r1], #1 + strgtb r3, [r0], #1 + subs r2, r2, r12 + blt .Lmemcpy_l4 /* less the 4 bytes */ + + ands r12, r1, #3 + beq .Lmemcpy_t8 /* we have an aligned source */ + + /* erg - unaligned source */ + /* This is where it gets nasty ... */ +.Lmemcpy_srcul: + bic r1, r1, #3 + ldr lr, [r1], #4 + cmp r12, #2 + bgt .Lmemcpy_srcul3 + beq .Lmemcpy_srcul2 + cmp r2, #0x0c + blt .Lmemcpy_srcul1loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemcpy_srcul1loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #8 +#else + mov r3, lr, lsr #8 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #24 + mov r4, r4, lsl #8 + orr r4, r4, r5, lsr #24 + mov r5, r5, lsl #8 + orr r5, r5, r12, lsr #24 + mov r12, r12, lsl #8 + orr r12, r12, lr, lsr #24 +#else + orr r3, r3, r4, lsl #24 + mov r4, r4, lsr #8 + orr r4, r4, r5, lsl #24 + mov r5, r5, lsr #8 + orr r5, r5, r12, lsl #24 + mov r12, r12, lsr #8 + orr r12, r12, lr, lsl #24 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemcpy_srcul1loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemcpy_srcul1l4 + +.Lmemcpy_srcul1loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #8 +#else + mov r12, lr, lsr #8 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #24 +#else + orr r12, r12, lr, lsl #24 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemcpy_srcul1loop4 + +.Lmemcpy_srcul1l4: + sub r1, r1, #3 + b .Lmemcpy_l4 + +.Lmemcpy_srcul2: + cmp r2, #0x0c + blt .Lmemcpy_srcul2loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemcpy_srcul2loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #16 +#else + mov r3, lr, lsr #16 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #16 + mov r4, r4, lsl #16 + orr r4, r4, r5, lsr #16 + mov r5, r5, lsl #16 + orr r5, r5, r12, lsr #16 + mov r12, r12, lsl #16 + orr r12, r12, lr, lsr #16 +#else + orr r3, r3, r4, lsl #16 + mov r4, r4, lsr #16 + orr r4, r4, r5, lsl #16 + mov r5, r5, lsr #16 + orr r5, r5, r12, lsl #16 + mov r12, r12, lsr #16 + orr r12, r12, lr, lsl #16 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemcpy_srcul2loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemcpy_srcul2l4 + +.Lmemcpy_srcul2loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #16 +#else + mov r12, lr, lsr #16 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #16 +#else + orr r12, r12, lr, lsl #16 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemcpy_srcul2loop4 + +.Lmemcpy_srcul2l4: + sub r1, r1, #2 + b .Lmemcpy_l4 + +.Lmemcpy_srcul3: + cmp r2, #0x0c + blt .Lmemcpy_srcul3loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemcpy_srcul3loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #24 +#else + mov r3, lr, lsr #24 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #8 + mov r4, r4, lsl #24 + orr r4, r4, r5, lsr #8 + mov r5, r5, lsl #24 + orr r5, r5, r12, lsr #8 + mov r12, r12, lsl #24 + orr r12, r12, lr, lsr #8 +#else + orr r3, r3, r4, lsl #8 + mov r4, r4, lsr #24 + orr r4, r4, r5, lsl #8 + mov r5, r5, lsr #24 + orr r5, r5, r12, lsl #8 + mov r12, r12, lsr #24 + orr r12, r12, lr, lsl #8 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemcpy_srcul3loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemcpy_srcul3l4 + +.Lmemcpy_srcul3loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #24 +#else + mov r12, lr, lsr #24 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #8 +#else + orr r12, r12, lr, lsl #8 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemcpy_srcul3loop4 + +.Lmemcpy_srcul3l4: + sub r1, r1, #1 + b .Lmemcpy_l4 diff --git a/lib/libc/arm/string/memcpy_xscale.S b/lib/libc/arm/string/memcpy_xscale.S new file mode 100644 index 0000000..02cca5e --- /dev/null +++ b/lib/libc/arm/string/memcpy_xscale.S @@ -0,0 +1,1783 @@ +/* $NetBSD: memcpy_xscale.S,v 1.1 2003/10/14 07:51:45 scw Exp $ */ + +/* + * Copyright 2003 Wasabi Systems, Inc. + * All rights reserved. + * + * Written by Steve C. Woodford for Wasabi Systems, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed for the NetBSD Project by + * Wasabi Systems, Inc. + * 4. The name of Wasabi Systems, Inc. may not be used to endorse + * or promote products derived from this software without specific prior + * written permission. + * + * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +/* LINTSTUB: Func: void *memcpy(void *dst, const void *src, size_t len) */ +ENTRY(memcpy) + pld [r1] + cmp r2, #0x0c + ble .Lmemcpy_short /* <= 12 bytes */ + mov r3, r0 /* We must not clobber r0 */ + + /* Word-align the destination buffer */ + ands ip, r3, #0x03 /* Already word aligned? */ + beq .Lmemcpy_wordaligned /* Yup */ + cmp ip, #0x02 + ldrb ip, [r1], #0x01 + sub r2, r2, #0x01 + strb ip, [r3], #0x01 + ldrleb ip, [r1], #0x01 + suble r2, r2, #0x01 + strleb ip, [r3], #0x01 + ldrltb ip, [r1], #0x01 + sublt r2, r2, #0x01 + strltb ip, [r3], #0x01 + + /* Destination buffer is now word aligned */ +.Lmemcpy_wordaligned: + ands ip, r1, #0x03 /* Is src also word-aligned? */ + bne .Lmemcpy_bad_align /* Nope. Things just got bad */ + + /* Quad-align the destination buffer */ + tst r3, #0x07 /* Already quad aligned? */ + ldrne ip, [r1], #0x04 + stmfd sp!, {r4-r9} /* Free up some registers */ + subne r2, r2, #0x04 + strne ip, [r3], #0x04 + + /* Destination buffer quad aligned, source is at least word aligned */ + subs r2, r2, #0x80 + blt .Lmemcpy_w_lessthan128 + + /* Copy 128 bytes at a time */ +.Lmemcpy_w_loop128: + ldr r4, [r1], #0x04 /* LD:00-03 */ + ldr r5, [r1], #0x04 /* LD:04-07 */ + pld [r1, #0x18] /* Prefetch 0x20 */ + ldr r6, [r1], #0x04 /* LD:08-0b */ + ldr r7, [r1], #0x04 /* LD:0c-0f */ + ldr r8, [r1], #0x04 /* LD:10-13 */ + ldr r9, [r1], #0x04 /* LD:14-17 */ + strd r4, [r3], #0x08 /* ST:00-07 */ + ldr r4, [r1], #0x04 /* LD:18-1b */ + ldr r5, [r1], #0x04 /* LD:1c-1f */ + strd r6, [r3], #0x08 /* ST:08-0f */ + ldr r6, [r1], #0x04 /* LD:20-23 */ + ldr r7, [r1], #0x04 /* LD:24-27 */ + pld [r1, #0x18] /* Prefetch 0x40 */ + strd r8, [r3], #0x08 /* ST:10-17 */ + ldr r8, [r1], #0x04 /* LD:28-2b */ + ldr r9, [r1], #0x04 /* LD:2c-2f */ + strd r4, [r3], #0x08 /* ST:18-1f */ + ldr r4, [r1], #0x04 /* LD:30-33 */ + ldr r5, [r1], #0x04 /* LD:34-37 */ + strd r6, [r3], #0x08 /* ST:20-27 */ + ldr r6, [r1], #0x04 /* LD:38-3b */ + ldr r7, [r1], #0x04 /* LD:3c-3f */ + strd r8, [r3], #0x08 /* ST:28-2f */ + ldr r8, [r1], #0x04 /* LD:40-43 */ + ldr r9, [r1], #0x04 /* LD:44-47 */ + pld [r1, #0x18] /* Prefetch 0x60 */ + strd r4, [r3], #0x08 /* ST:30-37 */ + ldr r4, [r1], #0x04 /* LD:48-4b */ + ldr r5, [r1], #0x04 /* LD:4c-4f */ + strd r6, [r3], #0x08 /* ST:38-3f */ + ldr r6, [r1], #0x04 /* LD:50-53 */ + ldr r7, [r1], #0x04 /* LD:54-57 */ + strd r8, [r3], #0x08 /* ST:40-47 */ + ldr r8, [r1], #0x04 /* LD:58-5b */ + ldr r9, [r1], #0x04 /* LD:5c-5f */ + strd r4, [r3], #0x08 /* ST:48-4f */ + ldr r4, [r1], #0x04 /* LD:60-63 */ + ldr r5, [r1], #0x04 /* LD:64-67 */ + pld [r1, #0x18] /* Prefetch 0x80 */ + strd r6, [r3], #0x08 /* ST:50-57 */ + ldr r6, [r1], #0x04 /* LD:68-6b */ + ldr r7, [r1], #0x04 /* LD:6c-6f */ + strd r8, [r3], #0x08 /* ST:58-5f */ + ldr r8, [r1], #0x04 /* LD:70-73 */ + ldr r9, [r1], #0x04 /* LD:74-77 */ + strd r4, [r3], #0x08 /* ST:60-67 */ + ldr r4, [r1], #0x04 /* LD:78-7b */ + ldr r5, [r1], #0x04 /* LD:7c-7f */ + strd r6, [r3], #0x08 /* ST:68-6f */ + strd r8, [r3], #0x08 /* ST:70-77 */ + subs r2, r2, #0x80 + strd r4, [r3], #0x08 /* ST:78-7f */ + bge .Lmemcpy_w_loop128 + +.Lmemcpy_w_lessthan128: + adds r2, r2, #0x80 /* Adjust for extra sub */ + ldmeqfd sp!, {r4-r9} + bxeq lr /* Return now if done */ + subs r2, r2, #0x20 + blt .Lmemcpy_w_lessthan32 + + /* Copy 32 bytes at a time */ +.Lmemcpy_w_loop32: + ldr r4, [r1], #0x04 + ldr r5, [r1], #0x04 + pld [r1, #0x18] + ldr r6, [r1], #0x04 + ldr r7, [r1], #0x04 + ldr r8, [r1], #0x04 + ldr r9, [r1], #0x04 + strd r4, [r3], #0x08 + ldr r4, [r1], #0x04 + ldr r5, [r1], #0x04 + strd r6, [r3], #0x08 + strd r8, [r3], #0x08 + subs r2, r2, #0x20 + strd r4, [r3], #0x08 + bge .Lmemcpy_w_loop32 + +.Lmemcpy_w_lessthan32: + adds r2, r2, #0x20 /* Adjust for extra sub */ + ldmeqfd sp!, {r4-r9} + bxeq lr /* Return now if done */ + + and r4, r2, #0x18 + rsbs r4, r4, #0x18 + addne pc, pc, r4, lsl #1 + nop + + /* At least 24 bytes remaining */ + ldr r4, [r1], #0x04 + ldr r5, [r1], #0x04 + sub r2, r2, #0x08 + strd r4, [r3], #0x08 + + /* At least 16 bytes remaining */ + ldr r4, [r1], #0x04 + ldr r5, [r1], #0x04 + sub r2, r2, #0x08 + strd r4, [r3], #0x08 + + /* At least 8 bytes remaining */ + ldr r4, [r1], #0x04 + ldr r5, [r1], #0x04 + subs r2, r2, #0x08 + strd r4, [r3], #0x08 + + /* Less than 8 bytes remaining */ + ldmfd sp!, {r4-r9} + bxeq lr /* Return now if done */ + subs r2, r2, #0x04 + ldrge ip, [r1], #0x04 + strge ip, [r3], #0x04 + bxeq lr /* Return now if done */ + addlt r2, r2, #0x04 + ldrb ip, [r1], #0x01 + cmp r2, #0x02 + ldrgeb r2, [r1], #0x01 + strb ip, [r3], #0x01 + ldrgtb ip, [r1] + strgeb r2, [r3], #0x01 + strgtb ip, [r3] + bx lr + + +/* + * At this point, it has not been possible to word align both buffers. + * The destination buffer is word aligned, but the source buffer is not. + */ +.Lmemcpy_bad_align: + stmfd sp!, {r4-r7} + bic r1, r1, #0x03 + cmp ip, #2 + ldr ip, [r1], #0x04 + bgt .Lmemcpy_bad3 + beq .Lmemcpy_bad2 + b .Lmemcpy_bad1 + +.Lmemcpy_bad1_loop16: +#ifdef __ARMEB__ + mov r4, ip, lsl #8 +#else + mov r4, ip, lsr #8 +#endif + ldr r5, [r1], #0x04 + pld [r1, #0x018] + ldr r6, [r1], #0x04 + ldr r7, [r1], #0x04 + ldr ip, [r1], #0x04 +#ifdef __ARMEB__ + orr r4, r4, r5, lsr #24 + mov r5, r5, lsl #8 + orr r5, r5, r6, lsr #24 + mov r6, r6, lsl #8 + orr r6, r6, r7, lsr #24 + mov r7, r7, lsl #8 + orr r7, r7, ip, lsr #24 +#else + orr r4, r4, r5, lsl #24 + mov r5, r5, lsr #8 + orr r5, r5, r6, lsl #24 + mov r6, r6, lsr #8 + orr r6, r6, r7, lsl #24 + mov r7, r7, lsr #8 + orr r7, r7, ip, lsl #24 +#endif + str r4, [r3], #0x04 + str r5, [r3], #0x04 + str r6, [r3], #0x04 + str r7, [r3], #0x04 +.Lmemcpy_bad1: + subs r2, r2, #0x10 + bge .Lmemcpy_bad1_loop16 + + adds r2, r2, #0x10 + ldmeqfd sp!, {r4-r7} + bxeq lr /* Return now if done */ + subs r2, r2, #0x04 + sublt r1, r1, #0x03 + blt .Lmemcpy_bad_done + +.Lmemcpy_bad1_loop4: +#ifdef __ARMEB__ + mov r4, ip, lsl #8 +#else + mov r4, ip, lsr #8 +#endif + ldr ip, [r1], #0x04 + subs r2, r2, #0x04 +#ifdef __ARMEB__ + orr r4, r4, ip, lsr #24 +#else + orr r4, r4, ip, lsl #24 +#endif + str r4, [r3], #0x04 + bge .Lmemcpy_bad1_loop4 + sub r1, r1, #0x03 + b .Lmemcpy_bad_done + +.Lmemcpy_bad2_loop16: +#ifdef __ARMEB__ + mov r4, ip, lsl #16 +#else + mov r4, ip, lsr #16 +#endif + ldr r5, [r1], #0x04 + pld [r1, #0x018] + ldr r6, [r1], #0x04 + ldr r7, [r1], #0x04 + ldr ip, [r1], #0x04 +#ifdef __ARMEB__ + orr r4, r4, r5, lsr #16 + mov r5, r5, lsl #16 + orr r5, r5, r6, lsr #16 + mov r6, r6, lsl #16 + orr r6, r6, r7, lsr #16 + mov r7, r7, lsl #16 + orr r7, r7, ip, lsr #16 +#else + orr r4, r4, r5, lsl #16 + mov r5, r5, lsr #16 + orr r5, r5, r6, lsl #16 + mov r6, r6, lsr #16 + orr r6, r6, r7, lsl #16 + mov r7, r7, lsr #16 + orr r7, r7, ip, lsl #16 +#endif + str r4, [r3], #0x04 + str r5, [r3], #0x04 + str r6, [r3], #0x04 + str r7, [r3], #0x04 +.Lmemcpy_bad2: + subs r2, r2, #0x10 + bge .Lmemcpy_bad2_loop16 + + adds r2, r2, #0x10 + ldmeqfd sp!, {r4-r7} + bxeq lr /* Return now if done */ + subs r2, r2, #0x04 + sublt r1, r1, #0x02 + blt .Lmemcpy_bad_done + +.Lmemcpy_bad2_loop4: +#ifdef __ARMEB__ + mov r4, ip, lsl #16 +#else + mov r4, ip, lsr #16 +#endif + ldr ip, [r1], #0x04 + subs r2, r2, #0x04 +#ifdef __ARMEB__ + orr r4, r4, ip, lsr #16 +#else + orr r4, r4, ip, lsl #16 +#endif + str r4, [r3], #0x04 + bge .Lmemcpy_bad2_loop4 + sub r1, r1, #0x02 + b .Lmemcpy_bad_done + +.Lmemcpy_bad3_loop16: +#ifdef __ARMEB__ + mov r4, ip, lsl #24 +#else + mov r4, ip, lsr #24 +#endif + ldr r5, [r1], #0x04 + pld [r1, #0x018] + ldr r6, [r1], #0x04 + ldr r7, [r1], #0x04 + ldr ip, [r1], #0x04 +#ifdef __ARMEB__ + orr r4, r4, r5, lsr #8 + mov r5, r5, lsl #24 + orr r5, r5, r6, lsr #8 + mov r6, r6, lsl #24 + orr r6, r6, r7, lsr #8 + mov r7, r7, lsl #24 + orr r7, r7, ip, lsr #8 +#else + orr r4, r4, r5, lsl #8 + mov r5, r5, lsr #24 + orr r5, r5, r6, lsl #8 + mov r6, r6, lsr #24 + orr r6, r6, r7, lsl #8 + mov r7, r7, lsr #24 + orr r7, r7, ip, lsl #8 +#endif + str r4, [r3], #0x04 + str r5, [r3], #0x04 + str r6, [r3], #0x04 + str r7, [r3], #0x04 +.Lmemcpy_bad3: + subs r2, r2, #0x10 + bge .Lmemcpy_bad3_loop16 + + adds r2, r2, #0x10 + ldmeqfd sp!, {r4-r7} + bxeq lr /* Return now if done */ + subs r2, r2, #0x04 + sublt r1, r1, #0x01 + blt .Lmemcpy_bad_done + +.Lmemcpy_bad3_loop4: +#ifdef __ARMEB__ + mov r4, ip, lsl #24 +#else + mov r4, ip, lsr #24 +#endif + ldr ip, [r1], #0x04 + subs r2, r2, #0x04 +#ifdef __ARMEB__ + orr r4, r4, ip, lsr #8 +#else + orr r4, r4, ip, lsl #8 +#endif + str r4, [r3], #0x04 + bge .Lmemcpy_bad3_loop4 + sub r1, r1, #0x01 + +.Lmemcpy_bad_done: + ldmfd sp!, {r4-r7} + adds r2, r2, #0x04 + bxeq lr + ldrb ip, [r1], #0x01 + cmp r2, #0x02 + ldrgeb r2, [r1], #0x01 + strb ip, [r3], #0x01 + ldrgtb ip, [r1] + strgeb r2, [r3], #0x01 + strgtb ip, [r3] + bx lr + + +/* + * Handle short copies (less than 16 bytes), possibly misaligned. + * Some of these are *very* common, thanks to the network stack, + * and so are handled specially. + */ +.Lmemcpy_short: +#ifndef _STANDALONE + add pc, pc, r2, lsl #2 + nop + bx lr /* 0x00 */ + b .Lmemcpy_bytewise /* 0x01 */ + b .Lmemcpy_bytewise /* 0x02 */ + b .Lmemcpy_bytewise /* 0x03 */ + b .Lmemcpy_4 /* 0x04 */ + b .Lmemcpy_bytewise /* 0x05 */ + b .Lmemcpy_6 /* 0x06 */ + b .Lmemcpy_bytewise /* 0x07 */ + b .Lmemcpy_8 /* 0x08 */ + b .Lmemcpy_bytewise /* 0x09 */ + b .Lmemcpy_bytewise /* 0x0a */ + b .Lmemcpy_bytewise /* 0x0b */ + b .Lmemcpy_c /* 0x0c */ +#endif +.Lmemcpy_bytewise: + mov r3, r0 /* We must not clobber r0 */ + ldrb ip, [r1], #0x01 +1: subs r2, r2, #0x01 + strb ip, [r3], #0x01 + ldrneb ip, [r1], #0x01 + bne 1b + bx lr + +#ifndef _STANDALONE +/****************************************************************************** + * Special case for 4 byte copies + */ +#define LMEMCPY_4_LOG2 6 /* 64 bytes */ +#define LMEMCPY_4_PAD .align LMEMCPY_4_LOG2 + LMEMCPY_4_PAD +.Lmemcpy_4: + and r2, r1, #0x03 + orr r2, r2, r0, lsl #2 + ands r2, r2, #0x0f + sub r3, pc, #0x14 + addne pc, r3, r2, lsl #LMEMCPY_4_LOG2 + +/* + * 0000: dst is 32-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] + str r2, [r0] + bx lr + LMEMCPY_4_PAD + +/* + * 0001: dst is 32-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #-1] /* BE:r3 = x012 LE:r3 = 210x */ + ldr r2, [r1, #3] /* BE:r2 = 3xxx LE:r2 = xxx3 */ +#ifdef __ARMEB__ + mov r3, r3, lsl #8 /* r3 = 012. */ + orr r3, r3, r2, lsr #24 /* r3 = 0123 */ +#else + mov r3, r3, lsr #8 /* r3 = .210 */ + orr r3, r3, r2, lsl #24 /* r3 = 3210 */ +#endif + str r3, [r0] + bx lr + LMEMCPY_4_PAD + +/* + * 0010: dst is 32-bit aligned, src is 16-bit aligned + */ +#ifdef __ARMEB__ + ldrh r3, [r1] + ldrh r2, [r1, #0x02] +#else + ldrh r3, [r1, #0x02] + ldrh r2, [r1] +#endif + orr r3, r2, r3, lsl #16 + str r3, [r0] + bx lr + LMEMCPY_4_PAD + +/* + * 0011: dst is 32-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #-3] /* BE:r3 = xxx0 LE:r3 = 0xxx */ + ldr r2, [r1, #1] /* BE:r2 = 123x LE:r2 = x321 */ +#ifdef __ARMEB__ + mov r3, r3, lsl #24 /* r3 = 0... */ + orr r3, r3, r2, lsr #8 /* r3 = 0123 */ +#else + mov r3, r3, lsr #24 /* r3 = ...0 */ + orr r3, r3, r2, lsl #8 /* r3 = 3210 */ +#endif + str r3, [r0] + bx lr + LMEMCPY_4_PAD + +/* + * 0100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] +#ifdef __ARMEB__ + strb r2, [r0, #0x03] + mov r3, r2, lsr #8 + mov r1, r2, lsr #24 + strb r1, [r0] +#else + strb r2, [r0] + mov r3, r2, lsr #8 + mov r1, r2, lsr #24 + strb r1, [r0, #0x03] +#endif + strh r3, [r0, #0x01] + bx lr + LMEMCPY_4_PAD + +/* + * 0101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrb r1, [r1, #0x03] + strb r2, [r0] + strh r3, [r0, #0x01] + strb r1, [r0, #0x03] + bx lr + LMEMCPY_4_PAD + +/* + * 0110: dst is 8-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldrh r3, [r1, #0x02] /* LE:r3 = ..23 LE:r3 = ..32 */ +#ifdef __ARMEB__ + mov r1, r2, lsr #8 /* r1 = ...0 */ + strb r1, [r0] + mov r2, r2, lsl #8 /* r2 = .01. */ + orr r2, r2, r3, lsr #8 /* r2 = .012 */ +#else + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = ...1 */ + orr r2, r2, r3, lsl #8 /* r2 = .321 */ + mov r3, r3, lsr #8 /* r3 = ...3 */ +#endif + strh r2, [r0, #0x01] + strb r3, [r0, #0x03] + bx lr + LMEMCPY_4_PAD + +/* + * 0111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrb r1, [r1, #0x03] + strb r2, [r0] + strh r3, [r0, #0x01] + strb r1, [r0, #0x03] + bx lr + LMEMCPY_4_PAD + +/* + * 1000: dst is 16-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] +#ifdef __ARMEB__ + strh r2, [r0, #0x02] + mov r3, r2, lsr #16 + strh r3, [r0] +#else + strh r2, [r0] + mov r3, r2, lsr #16 + strh r3, [r0, #0x02] +#endif + bx lr + LMEMCPY_4_PAD + +/* + * 1001: dst is 16-bit aligned, src is 8-bit aligned + */ + ldr r2, [r1, #-1] /* BE:r2 = x012 LE:r2 = 210x */ + ldr r3, [r1, #3] /* BE:r3 = 3xxx LE:r3 = xxx3 */ + mov r1, r2, lsr #8 /* BE:r1 = .x01 LE:r1 = .210 */ + strh r1, [r0] +#ifdef __ARMEB__ + mov r2, r2, lsl #8 /* r2 = 012. */ + orr r2, r2, r3, lsr #24 /* r2 = 0123 */ +#else + mov r2, r2, lsr #24 /* r2 = ...2 */ + orr r2, r2, r3, lsl #8 /* r2 = xx32 */ +#endif + strh r2, [r0, #0x02] + bx lr + LMEMCPY_4_PAD + +/* + * 1010: dst is 16-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] + ldrh r3, [r1, #0x02] + strh r2, [r0] + strh r3, [r0, #0x02] + bx lr + LMEMCPY_4_PAD + +/* + * 1011: dst is 16-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #1] /* BE:r3 = 123x LE:r3 = x321 */ + ldr r2, [r1, #-3] /* BE:r2 = xxx0 LE:r2 = 0xxx */ + mov r1, r3, lsr #8 /* BE:r1 = .123 LE:r1 = .x32 */ + strh r1, [r0, #0x02] +#ifdef __ARMEB__ + mov r3, r3, lsr #24 /* r3 = ...1 */ + orr r3, r3, r2, lsl #8 /* r3 = xx01 */ +#else + mov r3, r3, lsl #8 /* r3 = 321. */ + orr r3, r3, r2, lsr #24 /* r3 = 3210 */ +#endif + strh r3, [r0] + bx lr + LMEMCPY_4_PAD + +/* + * 1100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ +#ifdef __ARMEB__ + strb r2, [r0, #0x03] + mov r3, r2, lsr #8 + mov r1, r2, lsr #24 + strh r3, [r0, #0x01] + strb r1, [r0] +#else + strb r2, [r0] + mov r3, r2, lsr #8 + mov r1, r2, lsr #24 + strh r3, [r0, #0x01] + strb r1, [r0, #0x03] +#endif + bx lr + LMEMCPY_4_PAD + +/* + * 1101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrb r1, [r1, #0x03] + strb r2, [r0] + strh r3, [r0, #0x01] + strb r1, [r0, #0x03] + bx lr + LMEMCPY_4_PAD + +/* + * 1110: dst is 8-bit aligned, src is 16-bit aligned + */ +#ifdef __ARMEB__ + ldrh r3, [r1, #0x02] /* BE:r3 = ..23 LE:r3 = ..32 */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + strb r3, [r0, #0x03] + mov r3, r3, lsr #8 /* r3 = ...2 */ + orr r3, r3, r2, lsl #8 /* r3 = ..12 */ + strh r3, [r0, #0x01] + mov r2, r2, lsr #8 /* r2 = ...0 */ + strb r2, [r0] +#else + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldrh r3, [r1, #0x02] /* BE:r3 = ..23 LE:r3 = ..32 */ + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = ...1 */ + orr r2, r2, r3, lsl #8 /* r2 = .321 */ + strh r2, [r0, #0x01] + mov r3, r3, lsr #8 /* r3 = ...3 */ + strb r3, [r0, #0x03] +#endif + bx lr + LMEMCPY_4_PAD + +/* + * 1111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrb r1, [r1, #0x03] + strb r2, [r0] + strh r3, [r0, #0x01] + strb r1, [r0, #0x03] + bx lr + LMEMCPY_4_PAD + + +/****************************************************************************** + * Special case for 6 byte copies + */ +#define LMEMCPY_6_LOG2 6 /* 64 bytes */ +#define LMEMCPY_6_PAD .align LMEMCPY_6_LOG2 + LMEMCPY_6_PAD +.Lmemcpy_6: + and r2, r1, #0x03 + orr r2, r2, r0, lsl #2 + ands r2, r2, #0x0f + sub r3, pc, #0x14 + addne pc, r3, r2, lsl #LMEMCPY_6_LOG2 + +/* + * 0000: dst is 32-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] + ldrh r3, [r1, #0x04] + str r2, [r0] + strh r3, [r0, #0x04] + bx lr + LMEMCPY_6_PAD + +/* + * 0001: dst is 32-bit aligned, src is 8-bit aligned + */ + ldr r2, [r1, #-1] /* BE:r2 = x012 LE:r2 = 210x */ + ldr r3, [r1, #0x03] /* BE:r3 = 345x LE:r3 = x543 */ +#ifdef __ARMEB__ + mov r2, r2, lsl #8 /* r2 = 012. */ + orr r2, r2, r3, lsr #24 /* r2 = 0123 */ +#else + mov r2, r2, lsr #8 /* r2 = .210 */ + orr r2, r2, r3, lsl #24 /* r2 = 3210 */ +#endif + mov r3, r3, lsr #8 /* BE:r3 = .345 LE:r3 = .x54 */ + str r2, [r0] + strh r3, [r0, #0x04] + bx lr + LMEMCPY_6_PAD + +/* + * 0010: dst is 32-bit aligned, src is 16-bit aligned + */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ +#ifdef __ARMEB__ + mov r1, r3, lsr #16 /* r1 = ..23 */ + orr r1, r1, r2, lsl #16 /* r1 = 0123 */ + str r1, [r0] + strh r3, [r0, #0x04] +#else + mov r1, r3, lsr #16 /* r1 = ..54 */ + orr r2, r2, r3, lsl #16 /* r2 = 3210 */ + str r2, [r0] + strh r1, [r0, #0x04] +#endif + bx lr + LMEMCPY_6_PAD + +/* + * 0011: dst is 32-bit aligned, src is 8-bit aligned + */ + ldr r2, [r1, #-3] /* BE:r2 = xxx0 LE:r2 = 0xxx */ + ldr r3, [r1, #1] /* BE:r3 = 1234 LE:r3 = 4321 */ + ldr r1, [r1, #5] /* BE:r1 = 5xxx LE:r3 = xxx5 */ +#ifdef __ARMEB__ + mov r2, r2, lsl #24 /* r2 = 0... */ + orr r2, r2, r3, lsr #8 /* r2 = 0123 */ + mov r3, r3, lsl #8 /* r3 = 234. */ + orr r1, r3, r1, lsr #24 /* r1 = 2345 */ +#else + mov r2, r2, lsr #24 /* r2 = ...0 */ + orr r2, r2, r3, lsl #8 /* r2 = 3210 */ + mov r1, r1, lsl #8 /* r1 = xx5. */ + orr r1, r1, r3, lsr #24 /* r1 = xx54 */ +#endif + str r2, [r0] + strh r1, [r0, #0x04] + bx lr + LMEMCPY_6_PAD + +/* + * 0100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r3, [r1] /* BE:r3 = 0123 LE:r3 = 3210 */ + ldrh r2, [r1, #0x04] /* BE:r2 = ..45 LE:r2 = ..54 */ + mov r1, r3, lsr #8 /* BE:r1 = .012 LE:r1 = .321 */ + strh r1, [r0, #0x01] +#ifdef __ARMEB__ + mov r1, r3, lsr #24 /* r1 = ...0 */ + strb r1, [r0] + mov r3, r3, lsl #8 /* r3 = 123. */ + orr r3, r3, r2, lsr #8 /* r3 = 1234 */ +#else + strb r3, [r0] + mov r3, r3, lsr #24 /* r3 = ...3 */ + orr r3, r3, r2, lsl #8 /* r3 = .543 */ + mov r2, r2, lsr #8 /* r2 = ...5 */ +#endif + strh r3, [r0, #0x03] + strb r2, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 0101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrh ip, [r1, #0x03] + ldrb r1, [r1, #0x05] + strb r2, [r0] + strh r3, [r0, #0x01] + strh ip, [r0, #0x03] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 0110: dst is 8-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r1, [r1, #0x02] /* BE:r1 = 2345 LE:r1 = 5432 */ +#ifdef __ARMEB__ + mov r3, r2, lsr #8 /* r3 = ...0 */ + strb r3, [r0] + strb r1, [r0, #0x05] + mov r3, r1, lsr #8 /* r3 = .234 */ + strh r3, [r0, #0x03] + mov r3, r2, lsl #8 /* r3 = .01. */ + orr r3, r3, r1, lsr #24 /* r3 = .012 */ + strh r3, [r0, #0x01] +#else + strb r2, [r0] + mov r3, r1, lsr #24 + strb r3, [r0, #0x05] + mov r3, r1, lsr #8 /* r3 = .543 */ + strh r3, [r0, #0x03] + mov r3, r2, lsr #8 /* r3 = ...1 */ + orr r3, r3, r1, lsl #8 /* r3 = 4321 */ + strh r3, [r0, #0x01] +#endif + bx lr + LMEMCPY_6_PAD + +/* + * 0111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrh ip, [r1, #0x03] + ldrb r1, [r1, #0x05] + strb r2, [r0] + strh r3, [r0, #0x01] + strh ip, [r0, #0x03] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 1000: dst is 16-bit aligned, src is 32-bit aligned + */ +#ifdef __ARMEB__ + ldr r2, [r1] /* r2 = 0123 */ + ldrh r3, [r1, #0x04] /* r3 = ..45 */ + mov r1, r2, lsr #16 /* r1 = ..01 */ + orr r3, r3, r2, lsl#16 /* r3 = 2345 */ + strh r1, [r0] + str r3, [r0, #0x02] +#else + ldrh r2, [r1, #0x04] /* r2 = ..54 */ + ldr r3, [r1] /* r3 = 3210 */ + mov r2, r2, lsl #16 /* r2 = 54.. */ + orr r2, r2, r3, lsr #16 /* r2 = 5432 */ + strh r3, [r0] + str r2, [r0, #0x02] +#endif + bx lr + LMEMCPY_6_PAD + +/* + * 1001: dst is 16-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #-1] /* BE:r3 = x012 LE:r3 = 210x */ + ldr r2, [r1, #3] /* BE:r2 = 345x LE:r2 = x543 */ + mov r1, r3, lsr #8 /* BE:r1 = .x01 LE:r1 = .210 */ +#ifdef __ARMEB__ + mov r2, r2, lsr #8 /* r2 = .345 */ + orr r2, r2, r3, lsl #24 /* r2 = 2345 */ +#else + mov r2, r2, lsl #8 /* r2 = 543. */ + orr r2, r2, r3, lsr #24 /* r2 = 5432 */ +#endif + strh r1, [r0] + str r2, [r0, #0x02] + bx lr + LMEMCPY_6_PAD + +/* + * 1010: dst is 16-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] + ldr r3, [r1, #0x02] + strh r2, [r0] + str r3, [r0, #0x02] + bx lr + LMEMCPY_6_PAD + +/* + * 1011: dst is 16-bit aligned, src is 8-bit aligned + */ + ldrb r3, [r1] /* r3 = ...0 */ + ldr r2, [r1, #0x01] /* BE:r2 = 1234 LE:r2 = 4321 */ + ldrb r1, [r1, #0x05] /* r1 = ...5 */ +#ifdef __ARMEB__ + mov r3, r3, lsl #8 /* r3 = ..0. */ + orr r3, r3, r2, lsr #24 /* r3 = ..01 */ + orr r1, r1, r2, lsl #8 /* r1 = 2345 */ +#else + orr r3, r3, r2, lsl #8 /* r3 = 3210 */ + mov r1, r1, lsl #24 /* r1 = 5... */ + orr r1, r1, r2, lsr #8 /* r1 = 5432 */ +#endif + strh r3, [r0] + str r1, [r0, #0x02] + bx lr + LMEMCPY_6_PAD + +/* + * 1100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ + ldrh r1, [r1, #0x04] /* BE:r1 = ..45 LE:r1 = ..54 */ +#ifdef __ARMEB__ + mov r3, r2, lsr #24 /* r3 = ...0 */ + strb r3, [r0] + mov r2, r2, lsl #8 /* r2 = 123. */ + orr r2, r2, r1, lsr #8 /* r2 = 1234 */ +#else + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = .321 */ + orr r2, r2, r1, lsl #24 /* r2 = 4321 */ + mov r1, r1, lsr #8 /* r1 = ...5 */ +#endif + str r2, [r0, #0x01] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 1101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldrh ip, [r1, #0x03] + ldrb r1, [r1, #0x05] + strb r2, [r0] + strh r3, [r0, #0x01] + strh ip, [r0, #0x03] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 1110: dst is 8-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r1, [r1, #0x02] /* BE:r1 = 2345 LE:r1 = 5432 */ +#ifdef __ARMEB__ + mov r3, r2, lsr #8 /* r3 = ...0 */ + strb r3, [r0] + mov r2, r2, lsl #24 /* r2 = 1... */ + orr r2, r2, r1, lsr #8 /* r2 = 1234 */ +#else + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = ...1 */ + orr r2, r2, r1, lsl #8 /* r2 = 4321 */ + mov r1, r1, lsr #24 /* r1 = ...5 */ +#endif + str r2, [r0, #0x01] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + +/* + * 1111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldr r3, [r1, #0x01] + ldrb r1, [r1, #0x05] + strb r2, [r0] + str r3, [r0, #0x01] + strb r1, [r0, #0x05] + bx lr + LMEMCPY_6_PAD + + +/****************************************************************************** + * Special case for 8 byte copies + */ +#define LMEMCPY_8_LOG2 6 /* 64 bytes */ +#define LMEMCPY_8_PAD .align LMEMCPY_8_LOG2 + LMEMCPY_8_PAD +.Lmemcpy_8: + and r2, r1, #0x03 + orr r2, r2, r0, lsl #2 + ands r2, r2, #0x0f + sub r3, pc, #0x14 + addne pc, r3, r2, lsl #LMEMCPY_8_LOG2 + +/* + * 0000: dst is 32-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] + ldr r3, [r1, #0x04] + str r2, [r0] + str r3, [r0, #0x04] + bx lr + LMEMCPY_8_PAD + +/* + * 0001: dst is 32-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #-1] /* BE:r3 = x012 LE:r3 = 210x */ + ldr r2, [r1, #0x03] /* BE:r2 = 3456 LE:r2 = 6543 */ + ldrb r1, [r1, #0x07] /* r1 = ...7 */ +#ifdef __ARMEB__ + mov r3, r3, lsl #8 /* r3 = 012. */ + orr r3, r3, r2, lsr #24 /* r3 = 0123 */ + orr r2, r1, r2, lsl #8 /* r2 = 4567 */ +#else + mov r3, r3, lsr #8 /* r3 = .210 */ + orr r3, r3, r2, lsl #24 /* r3 = 3210 */ + mov r1, r1, lsl #24 /* r1 = 7... */ + orr r2, r1, r2, lsr #8 /* r2 = 7654 */ +#endif + str r3, [r0] + str r2, [r0, #0x04] + bx lr + LMEMCPY_8_PAD + +/* + * 0010: dst is 32-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldrh r1, [r1, #0x06] /* BE:r1 = ..67 LE:r1 = ..76 */ +#ifdef __ARMEB__ + mov r2, r2, lsl #16 /* r2 = 01.. */ + orr r2, r2, r3, lsr #16 /* r2 = 0123 */ + orr r3, r1, r3, lsl #16 /* r3 = 4567 */ +#else + orr r2, r2, r3, lsl #16 /* r2 = 3210 */ + mov r3, r3, lsr #16 /* r3 = ..54 */ + orr r3, r3, r1, lsl #16 /* r3 = 7654 */ +#endif + str r2, [r0] + str r3, [r0, #0x04] + bx lr + LMEMCPY_8_PAD + +/* + * 0011: dst is 32-bit aligned, src is 8-bit aligned + */ + ldrb r3, [r1] /* r3 = ...0 */ + ldr r2, [r1, #0x01] /* BE:r2 = 1234 LE:r2 = 4321 */ + ldr r1, [r1, #0x05] /* BE:r1 = 567x LE:r1 = x765 */ +#ifdef __ARMEB__ + mov r3, r3, lsl #24 /* r3 = 0... */ + orr r3, r3, r2, lsr #8 /* r3 = 0123 */ + mov r2, r2, lsl #24 /* r2 = 4... */ + orr r2, r2, r1, lsr #8 /* r2 = 4567 */ +#else + orr r3, r3, r2, lsl #8 /* r3 = 3210 */ + mov r2, r2, lsr #24 /* r2 = ...4 */ + orr r2, r2, r1, lsl #8 /* r2 = 7654 */ +#endif + str r3, [r0] + str r2, [r0, #0x04] + bx lr + LMEMCPY_8_PAD + +/* + * 0100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r3, [r1] /* BE:r3 = 0123 LE:r3 = 3210 */ + ldr r2, [r1, #0x04] /* BE:r2 = 4567 LE:r2 = 7654 */ +#ifdef __ARMEB__ + mov r1, r3, lsr #24 /* r1 = ...0 */ + strb r1, [r0] + mov r1, r3, lsr #8 /* r1 = .012 */ + strb r2, [r0, #0x07] + mov r3, r3, lsl #24 /* r3 = 3... */ + orr r3, r3, r2, lsr #8 /* r3 = 3456 */ +#else + strb r3, [r0] + mov r1, r2, lsr #24 /* r1 = ...7 */ + strb r1, [r0, #0x07] + mov r1, r3, lsr #8 /* r1 = .321 */ + mov r3, r3, lsr #24 /* r3 = ...3 */ + orr r3, r3, r2, lsl #8 /* r3 = 6543 */ +#endif + strh r1, [r0, #0x01] + str r3, [r0, #0x03] + bx lr + LMEMCPY_8_PAD + +/* + * 0101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldr ip, [r1, #0x03] + ldrb r1, [r1, #0x07] + strb r2, [r0] + strh r3, [r0, #0x01] + str ip, [r0, #0x03] + strb r1, [r0, #0x07] + bx lr + LMEMCPY_8_PAD + +/* + * 0110: dst is 8-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldrh r1, [r1, #0x06] /* BE:r1 = ..67 LE:r1 = ..76 */ +#ifdef __ARMEB__ + mov ip, r2, lsr #8 /* ip = ...0 */ + strb ip, [r0] + mov ip, r2, lsl #8 /* ip = .01. */ + orr ip, ip, r3, lsr #24 /* ip = .012 */ + strb r1, [r0, #0x07] + mov r3, r3, lsl #8 /* r3 = 345. */ + orr r3, r3, r1, lsr #8 /* r3 = 3456 */ +#else + strb r2, [r0] /* 0 */ + mov ip, r1, lsr #8 /* ip = ...7 */ + strb ip, [r0, #0x07] /* 7 */ + mov ip, r2, lsr #8 /* ip = ...1 */ + orr ip, ip, r3, lsl #8 /* ip = 4321 */ + mov r3, r3, lsr #8 /* r3 = .543 */ + orr r3, r3, r1, lsl #24 /* r3 = 6543 */ +#endif + strh ip, [r0, #0x01] + str r3, [r0, #0x03] + bx lr + LMEMCPY_8_PAD + +/* + * 0111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r3, [r1] /* r3 = ...0 */ + ldr ip, [r1, #0x01] /* BE:ip = 1234 LE:ip = 4321 */ + ldrh r2, [r1, #0x05] /* BE:r2 = ..56 LE:r2 = ..65 */ + ldrb r1, [r1, #0x07] /* r1 = ...7 */ + strb r3, [r0] + mov r3, ip, lsr #16 /* BE:r3 = ..12 LE:r3 = ..43 */ +#ifdef __ARMEB__ + strh r3, [r0, #0x01] + orr r2, r2, ip, lsl #16 /* r2 = 3456 */ +#else + strh ip, [r0, #0x01] + orr r2, r3, r2, lsl #16 /* r2 = 6543 */ +#endif + str r2, [r0, #0x03] + strb r1, [r0, #0x07] + bx lr + LMEMCPY_8_PAD + +/* + * 1000: dst is 16-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ + ldr r3, [r1, #0x04] /* BE:r3 = 4567 LE:r3 = 7654 */ + mov r1, r2, lsr #16 /* BE:r1 = ..01 LE:r1 = ..32 */ +#ifdef __ARMEB__ + strh r1, [r0] + mov r1, r3, lsr #16 /* r1 = ..45 */ + orr r2, r1 ,r2, lsl #16 /* r2 = 2345 */ +#else + strh r2, [r0] + orr r2, r1, r3, lsl #16 /* r2 = 5432 */ + mov r3, r3, lsr #16 /* r3 = ..76 */ +#endif + str r2, [r0, #0x02] + strh r3, [r0, #0x06] + bx lr + LMEMCPY_8_PAD + +/* + * 1001: dst is 16-bit aligned, src is 8-bit aligned + */ + ldr r2, [r1, #-1] /* BE:r2 = x012 LE:r2 = 210x */ + ldr r3, [r1, #0x03] /* BE:r3 = 3456 LE:r3 = 6543 */ + ldrb ip, [r1, #0x07] /* ip = ...7 */ + mov r1, r2, lsr #8 /* BE:r1 = .x01 LE:r1 = .210 */ + strh r1, [r0] +#ifdef __ARMEB__ + mov r1, r2, lsl #24 /* r1 = 2... */ + orr r1, r1, r3, lsr #8 /* r1 = 2345 */ + orr r3, ip, r3, lsl #8 /* r3 = 4567 */ +#else + mov r1, r2, lsr #24 /* r1 = ...2 */ + orr r1, r1, r3, lsl #8 /* r1 = 5432 */ + mov r3, r3, lsr #24 /* r3 = ...6 */ + orr r3, r3, ip, lsl #8 /* r3 = ..76 */ +#endif + str r1, [r0, #0x02] + strh r3, [r0, #0x06] + bx lr + LMEMCPY_8_PAD + +/* + * 1010: dst is 16-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] + ldr ip, [r1, #0x02] + ldrh r3, [r1, #0x06] + strh r2, [r0] + str ip, [r0, #0x02] + strh r3, [r0, #0x06] + bx lr + LMEMCPY_8_PAD + +/* + * 1011: dst is 16-bit aligned, src is 8-bit aligned + */ + ldr r3, [r1, #0x05] /* BE:r3 = 567x LE:r3 = x765 */ + ldr r2, [r1, #0x01] /* BE:r2 = 1234 LE:r2 = 4321 */ + ldrb ip, [r1] /* ip = ...0 */ + mov r1, r3, lsr #8 /* BE:r1 = .567 LE:r1 = .x76 */ + strh r1, [r0, #0x06] +#ifdef __ARMEB__ + mov r3, r3, lsr #24 /* r3 = ...5 */ + orr r3, r3, r2, lsl #8 /* r3 = 2345 */ + mov r2, r2, lsr #24 /* r2 = ...1 */ + orr r2, r2, ip, lsl #8 /* r2 = ..01 */ +#else + mov r3, r3, lsl #24 /* r3 = 5... */ + orr r3, r3, r2, lsr #8 /* r3 = 5432 */ + orr r2, ip, r2, lsl #8 /* r2 = 3210 */ +#endif + str r3, [r0, #0x02] + strh r2, [r0] + bx lr + LMEMCPY_8_PAD + +/* + * 1100: dst is 8-bit aligned, src is 32-bit aligned + */ + ldr r3, [r1, #0x04] /* BE:r3 = 4567 LE:r3 = 7654 */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ + mov r1, r3, lsr #8 /* BE:r1 = .456 LE:r1 = .765 */ + strh r1, [r0, #0x05] +#ifdef __ARMEB__ + strb r3, [r0, #0x07] + mov r1, r2, lsr #24 /* r1 = ...0 */ + strb r1, [r0] + mov r2, r2, lsl #8 /* r2 = 123. */ + orr r2, r2, r3, lsr #24 /* r2 = 1234 */ + str r2, [r0, #0x01] +#else + strb r2, [r0] + mov r1, r3, lsr #24 /* r1 = ...7 */ + strb r1, [r0, #0x07] + mov r2, r2, lsr #8 /* r2 = .321 */ + orr r2, r2, r3, lsl #24 /* r2 = 4321 */ + str r2, [r0, #0x01] +#endif + bx lr + LMEMCPY_8_PAD + +/* + * 1101: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r3, [r1] /* r3 = ...0 */ + ldrh r2, [r1, #0x01] /* BE:r2 = ..12 LE:r2 = ..21 */ + ldr ip, [r1, #0x03] /* BE:ip = 3456 LE:ip = 6543 */ + ldrb r1, [r1, #0x07] /* r1 = ...7 */ + strb r3, [r0] + mov r3, ip, lsr #16 /* BE:r3 = ..34 LE:r3 = ..65 */ +#ifdef __ARMEB__ + strh ip, [r0, #0x05] + orr r2, r3, r2, lsl #16 /* r2 = 1234 */ +#else + strh r3, [r0, #0x05] + orr r2, r2, ip, lsl #16 /* r2 = 4321 */ +#endif + str r2, [r0, #0x01] + strb r1, [r0, #0x07] + bx lr + LMEMCPY_8_PAD + +/* + * 1110: dst is 8-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldrh r1, [r1, #0x06] /* BE:r1 = ..67 LE:r1 = ..76 */ +#ifdef __ARMEB__ + mov ip, r2, lsr #8 /* ip = ...0 */ + strb ip, [r0] + mov ip, r2, lsl #24 /* ip = 1... */ + orr ip, ip, r3, lsr #8 /* ip = 1234 */ + strb r1, [r0, #0x07] + mov r1, r1, lsr #8 /* r1 = ...6 */ + orr r1, r1, r3, lsl #8 /* r1 = 3456 */ +#else + strb r2, [r0] + mov ip, r2, lsr #8 /* ip = ...1 */ + orr ip, ip, r3, lsl #8 /* ip = 4321 */ + mov r2, r1, lsr #8 /* r2 = ...7 */ + strb r2, [r0, #0x07] + mov r1, r1, lsl #8 /* r1 = .76. */ + orr r1, r1, r3, lsr #24 /* r1 = .765 */ +#endif + str ip, [r0, #0x01] + strh r1, [r0, #0x05] + bx lr + LMEMCPY_8_PAD + +/* + * 1111: dst is 8-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] + ldr ip, [r1, #0x01] + ldrh r3, [r1, #0x05] + ldrb r1, [r1, #0x07] + strb r2, [r0] + str ip, [r0, #0x01] + strh r3, [r0, #0x05] + strb r1, [r0, #0x07] + bx lr + LMEMCPY_8_PAD + +/****************************************************************************** + * Special case for 12 byte copies + */ +#define LMEMCPY_C_LOG2 7 /* 128 bytes */ +#define LMEMCPY_C_PAD .align LMEMCPY_C_LOG2 + LMEMCPY_C_PAD +.Lmemcpy_c: + and r2, r1, #0x03 + orr r2, r2, r0, lsl #2 + ands r2, r2, #0x0f + sub r3, pc, #0x14 + addne pc, r3, r2, lsl #LMEMCPY_C_LOG2 + +/* + * 0000: dst is 32-bit aligned, src is 32-bit aligned + */ + ldr r2, [r1] + ldr r3, [r1, #0x04] + ldr r1, [r1, #0x08] + str r2, [r0] + str r3, [r0, #0x04] + str r1, [r0, #0x08] + bx lr + LMEMCPY_C_PAD + +/* + * 0001: dst is 32-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1, #0xb] /* r2 = ...B */ + ldr ip, [r1, #0x07] /* BE:ip = 789A LE:ip = A987 */ + ldr r3, [r1, #0x03] /* BE:r3 = 3456 LE:r3 = 6543 */ + ldr r1, [r1, #-1] /* BE:r1 = x012 LE:r1 = 210x */ +#ifdef __ARMEB__ + orr r2, r2, ip, lsl #8 /* r2 = 89AB */ + str r2, [r0, #0x08] + mov r2, ip, lsr #24 /* r2 = ...7 */ + orr r2, r2, r3, lsl #8 /* r2 = 4567 */ + mov r1, r1, lsl #8 /* r1 = 012. */ + orr r1, r1, r3, lsr #24 /* r1 = 0123 */ +#else + mov r2, r2, lsl #24 /* r2 = B... */ + orr r2, r2, ip, lsr #8 /* r2 = BA98 */ + str r2, [r0, #0x08] + mov r2, ip, lsl #24 /* r2 = 7... */ + orr r2, r2, r3, lsr #8 /* r2 = 7654 */ + mov r1, r1, lsr #8 /* r1 = .210 */ + orr r1, r1, r3, lsl #24 /* r1 = 3210 */ +#endif + str r2, [r0, #0x04] + str r1, [r0] + bx lr + LMEMCPY_C_PAD + +/* + * 0010: dst is 32-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldr ip, [r1, #0x06] /* BE:ip = 6789 LE:ip = 9876 */ + ldrh r1, [r1, #0x0a] /* BE:r1 = ..AB LE:r1 = ..BA */ +#ifdef __ARMEB__ + mov r2, r2, lsl #16 /* r2 = 01.. */ + orr r2, r2, r3, lsr #16 /* r2 = 0123 */ + str r2, [r0] + mov r3, r3, lsl #16 /* r3 = 45.. */ + orr r3, r3, ip, lsr #16 /* r3 = 4567 */ + orr r1, r1, ip, lsl #16 /* r1 = 89AB */ +#else + orr r2, r2, r3, lsl #16 /* r2 = 3210 */ + str r2, [r0] + mov r3, r3, lsr #16 /* r3 = ..54 */ + orr r3, r3, ip, lsl #16 /* r3 = 7654 */ + mov r1, r1, lsl #16 /* r1 = BA.. */ + orr r1, r1, ip, lsr #16 /* r1 = BA98 */ +#endif + str r3, [r0, #0x04] + str r1, [r0, #0x08] + bx lr + LMEMCPY_C_PAD + +/* + * 0011: dst is 32-bit aligned, src is 8-bit aligned + */ + ldrb r2, [r1] /* r2 = ...0 */ + ldr r3, [r1, #0x01] /* BE:r3 = 1234 LE:r3 = 4321 */ + ldr ip, [r1, #0x05] /* BE:ip = 5678 LE:ip = 8765 */ + ldr r1, [r1, #0x09] /* BE:r1 = 9ABx LE:r1 = xBA9 */ +#ifdef __ARMEB__ + mov r2, r2, lsl #24 /* r2 = 0... */ + orr r2, r2, r3, lsr #8 /* r2 = 0123 */ + str r2, [r0] + mov r3, r3, lsl #24 /* r3 = 4... */ + orr r3, r3, ip, lsr #8 /* r3 = 4567 */ + mov r1, r1, lsr #8 /* r1 = .9AB */ + orr r1, r1, ip, lsl #24 /* r1 = 89AB */ +#else + orr r2, r2, r3, lsl #8 /* r2 = 3210 */ + str r2, [r0] + mov r3, r3, lsr #24 /* r3 = ...4 */ + orr r3, r3, ip, lsl #8 /* r3 = 7654 */ + mov r1, r1, lsl #8 /* r1 = BA9. */ + orr r1, r1, ip, lsr #24 /* r1 = BA98 */ +#endif + str r3, [r0, #0x04] + str r1, [r0, #0x08] + bx lr + LMEMCPY_C_PAD + +/* + * 0100: dst is 8-bit aligned (byte 1), src is 32-bit aligned + */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ + ldr r3, [r1, #0x04] /* BE:r3 = 4567 LE:r3 = 7654 */ + ldr ip, [r1, #0x08] /* BE:ip = 89AB LE:ip = BA98 */ + mov r1, r2, lsr #8 /* BE:r1 = .012 LE:r1 = .321 */ + strh r1, [r0, #0x01] +#ifdef __ARMEB__ + mov r1, r2, lsr #24 /* r1 = ...0 */ + strb r1, [r0] + mov r1, r2, lsl #24 /* r1 = 3... */ + orr r2, r1, r3, lsr #8 /* r1 = 3456 */ + mov r1, r3, lsl #24 /* r1 = 7... */ + orr r1, r1, ip, lsr #8 /* r1 = 789A */ +#else + strb r2, [r0] + mov r1, r2, lsr #24 /* r1 = ...3 */ + orr r2, r1, r3, lsl #8 /* r1 = 6543 */ + mov r1, r3, lsr #24 /* r1 = ...7 */ + orr r1, r1, ip, lsl #8 /* r1 = A987 */ + mov ip, ip, lsr #24 /* ip = ...B */ +#endif + str r2, [r0, #0x03] + str r1, [r0, #0x07] + strb ip, [r0, #0x0b] + bx lr + LMEMCPY_C_PAD + +/* + * 0101: dst is 8-bit aligned (byte 1), src is 8-bit aligned (byte 1) + */ + ldrb r2, [r1] + ldrh r3, [r1, #0x01] + ldr ip, [r1, #0x03] + strb r2, [r0] + ldr r2, [r1, #0x07] + ldrb r1, [r1, #0x0b] + strh r3, [r0, #0x01] + str ip, [r0, #0x03] + str r2, [r0, #0x07] + strb r1, [r0, #0x0b] + bx lr + LMEMCPY_C_PAD + +/* + * 0110: dst is 8-bit aligned (byte 1), src is 16-bit aligned + */ + ldrh r2, [r1] /* BE:r2 = ..01 LE:r2 = ..10 */ + ldr r3, [r1, #0x02] /* BE:r3 = 2345 LE:r3 = 5432 */ + ldr ip, [r1, #0x06] /* BE:ip = 6789 LE:ip = 9876 */ + ldrh r1, [r1, #0x0a] /* BE:r1 = ..AB LE:r1 = ..BA */ +#ifdef __ARMEB__ + mov r2, r2, ror #8 /* r2 = 1..0 */ + strb r2, [r0] + mov r2, r2, lsr #16 /* r2 = ..1. */ + orr r2, r2, r3, lsr #24 /* r2 = ..12 */ + strh r2, [r0, #0x01] + mov r2, r3, lsl #8 /* r2 = 345. */ + orr r3, r2, ip, lsr #24 /* r3 = 3456 */ + mov r2, ip, lsl #8 /* r2 = 789. */ + orr r2, r2, r1, lsr #8 /* r2 = 789A */ +#else + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = ...1 */ + orr r2, r2, r3, lsl #8 /* r2 = 4321 */ + strh r2, [r0, #0x01] + mov r2, r3, lsr #8 /* r2 = .543 */ + orr r3, r2, ip, lsl #24 /* r3 = 6543 */ + mov r2, ip, lsr #8 /* r2 = .987 */ + orr r2, r2, r1, lsl #24 /* r2 = A987 */ + mov r1, r1, lsr #8 /* r1 = ...B */ +#endif + str r3, [r0, #0x03] + str r2, [r0, #0x07] + strb r1, [r0, #0x0b] + bx lr + LMEMCPY_C_PAD + +/* + * 0111: dst is 8-bit aligned (byte 1), src is 8-bit aligned (byte 3) + */ + ldrb r2, [r1] + ldr r3, [r1, #0x01] /* BE:r3 = 1234 LE:r3 = 4321 */ + ldr ip, [r1, #0x05] /* BE:ip = 5678 LE:ip = 8765 */ + ldr r1, [r1, #0x09] /* BE:r1 = 9ABx LE:r1 = xBA9 */ + strb r2, [r0] +#ifdef __ARMEB__ + mov r2, r3, lsr #16 /* r2 = ..12 */ + strh r2, [r0, #0x01] + mov r3, r3, lsl #16 /* r3 = 34.. */ + orr r3, r3, ip, lsr #16 /* r3 = 3456 */ + mov ip, ip, lsl #16 /* ip = 78.. */ + orr ip, ip, r1, lsr #16 /* ip = 789A */ + mov r1, r1, lsr #8 /* r1 = .9AB */ +#else + strh r3, [r0, #0x01] + mov r3, r3, lsr #16 /* r3 = ..43 */ + orr r3, r3, ip, lsl #16 /* r3 = 6543 */ + mov ip, ip, lsr #16 /* ip = ..87 */ + orr ip, ip, r1, lsl #16 /* ip = A987 */ + mov r1, r1, lsr #16 /* r1 = ..xB */ +#endif + str r3, [r0, #0x03] + str ip, [r0, #0x07] + strb r1, [r0, #0x0b] + bx lr + LMEMCPY_C_PAD + +/* + * 1000: dst is 16-bit aligned, src is 32-bit aligned + */ + ldr ip, [r1] /* BE:ip = 0123 LE:ip = 3210 */ + ldr r3, [r1, #0x04] /* BE:r3 = 4567 LE:r3 = 7654 */ + ldr r2, [r1, #0x08] /* BE:r2 = 89AB LE:r2 = BA98 */ + mov r1, ip, lsr #16 /* BE:r1 = ..01 LE:r1 = ..32 */ +#ifdef __ARMEB__ + strh r1, [r0] + mov r1, ip, lsl #16 /* r1 = 23.. */ + orr r1, r1, r3, lsr #16 /* r1 = 2345 */ + mov r3, r3, lsl #16 /* r3 = 67.. */ + orr r3, r3, r2, lsr #16 /* r3 = 6789 */ +#else + strh ip, [r0] + orr r1, r1, r3, lsl #16 /* r1 = 5432 */ + mov r3, r3, lsr #16 /* r3 = ..76 */ + orr r3, r3, r2, lsl #16 /* r3 = 9876 */ + mov r2, r2, lsr #16 /* r2 = ..BA */ +#endif + str r1, [r0, #0x02] + str r3, [r0, #0x06] + strh r2, [r0, #0x0a] + bx lr + LMEMCPY_C_PAD + +/* + * 1001: dst is 16-bit aligned, src is 8-bit aligned (byte 1) + */ + ldr r2, [r1, #-1] /* BE:r2 = x012 LE:r2 = 210x */ + ldr r3, [r1, #0x03] /* BE:r3 = 3456 LE:r3 = 6543 */ + mov ip, r2, lsr #8 /* BE:ip = .x01 LE:ip = .210 */ + strh ip, [r0] + ldr ip, [r1, #0x07] /* BE:ip = 789A LE:ip = A987 */ + ldrb r1, [r1, #0x0b] /* r1 = ...B */ +#ifdef __ARMEB__ + mov r2, r2, lsl #24 /* r2 = 2... */ + orr r2, r2, r3, lsr #8 /* r2 = 2345 */ + mov r3, r3, lsl #24 /* r3 = 6... */ + orr r3, r3, ip, lsr #8 /* r3 = 6789 */ + orr r1, r1, ip, lsl #8 /* r1 = 89AB */ +#else + mov r2, r2, lsr #24 /* r2 = ...2 */ + orr r2, r2, r3, lsl #8 /* r2 = 5432 */ + mov r3, r3, lsr #24 /* r3 = ...6 */ + orr r3, r3, ip, lsl #8 /* r3 = 9876 */ + mov r1, r1, lsl #8 /* r1 = ..B. */ + orr r1, r1, ip, lsr #24 /* r1 = ..BA */ +#endif + str r2, [r0, #0x02] + str r3, [r0, #0x06] + strh r1, [r0, #0x0a] + bx lr + LMEMCPY_C_PAD + +/* + * 1010: dst is 16-bit aligned, src is 16-bit aligned + */ + ldrh r2, [r1] + ldr r3, [r1, #0x02] + ldr ip, [r1, #0x06] + ldrh r1, [r1, #0x0a] + strh r2, [r0] + str r3, [r0, #0x02] + str ip, [r0, #0x06] + strh r1, [r0, #0x0a] + bx lr + LMEMCPY_C_PAD + +/* + * 1011: dst is 16-bit aligned, src is 8-bit aligned (byte 3) + */ + ldr r2, [r1, #0x09] /* BE:r2 = 9ABx LE:r2 = xBA9 */ + ldr r3, [r1, #0x05] /* BE:r3 = 5678 LE:r3 = 8765 */ + mov ip, r2, lsr #8 /* BE:ip = .9AB LE:ip = .xBA */ + strh ip, [r0, #0x0a] + ldr ip, [r1, #0x01] /* BE:ip = 1234 LE:ip = 4321 */ + ldrb r1, [r1] /* r1 = ...0 */ +#ifdef __ARMEB__ + mov r2, r2, lsr #24 /* r2 = ...9 */ + orr r2, r2, r3, lsl #8 /* r2 = 6789 */ + mov r3, r3, lsr #24 /* r3 = ...5 */ + orr r3, r3, ip, lsl #8 /* r3 = 2345 */ + mov r1, r1, lsl #8 /* r1 = ..0. */ + orr r1, r1, ip, lsr #24 /* r1 = ..01 */ +#else + mov r2, r2, lsl #24 /* r2 = 9... */ + orr r2, r2, r3, lsr #8 /* r2 = 9876 */ + mov r3, r3, lsl #24 /* r3 = 5... */ + orr r3, r3, ip, lsr #8 /* r3 = 5432 */ + orr r1, r1, ip, lsl #8 /* r1 = 3210 */ +#endif + str r2, [r0, #0x06] + str r3, [r0, #0x02] + strh r1, [r0] + bx lr + LMEMCPY_C_PAD + +/* + * 1100: dst is 8-bit aligned (byte 3), src is 32-bit aligned + */ + ldr r2, [r1] /* BE:r2 = 0123 LE:r2 = 3210 */ + ldr ip, [r1, #0x04] /* BE:ip = 4567 LE:ip = 7654 */ + ldr r1, [r1, #0x08] /* BE:r1 = 89AB LE:r1 = BA98 */ +#ifdef __ARMEB__ + mov r3, r2, lsr #24 /* r3 = ...0 */ + strb r3, [r0] + mov r2, r2, lsl #8 /* r2 = 123. */ + orr r2, r2, ip, lsr #24 /* r2 = 1234 */ + str r2, [r0, #0x01] + mov r2, ip, lsl #8 /* r2 = 567. */ + orr r2, r2, r1, lsr #24 /* r2 = 5678 */ + str r2, [r0, #0x05] + mov r2, r1, lsr #8 /* r2 = ..9A */ + strh r2, [r0, #0x09] + strb r1, [r0, #0x0b] +#else + strb r2, [r0] + mov r3, r2, lsr #8 /* r3 = .321 */ + orr r3, r3, ip, lsl #24 /* r3 = 4321 */ + str r3, [r0, #0x01] + mov r3, ip, lsr #8 /* r3 = .765 */ + orr r3, r3, r1, lsl #24 /* r3 = 8765 */ + str r3, [r0, #0x05] + mov r1, r1, lsr #8 /* r1 = .BA9 */ + strh r1, [r0, #0x09] + mov r1, r1, lsr #16 /* r1 = ...B */ + strb r1, [r0, #0x0b] +#endif + bx lr + LMEMCPY_C_PAD + +/* + * 1101: dst is 8-bit aligned (byte 3), src is 8-bit aligned (byte 1) + */ + ldrb r2, [r1, #0x0b] /* r2 = ...B */ + ldr r3, [r1, #0x07] /* BE:r3 = 789A LE:r3 = A987 */ + ldr ip, [r1, #0x03] /* BE:ip = 3456 LE:ip = 6543 */ + ldr r1, [r1, #-1] /* BE:r1 = x012 LE:r1 = 210x */ + strb r2, [r0, #0x0b] +#ifdef __ARMEB__ + strh r3, [r0, #0x09] + mov r3, r3, lsr #16 /* r3 = ..78 */ + orr r3, r3, ip, lsl #16 /* r3 = 5678 */ + mov ip, ip, lsr #16 /* ip = ..34 */ + orr ip, ip, r1, lsl #16 /* ip = 1234 */ + mov r1, r1, lsr #16 /* r1 = ..x0 */ +#else + mov r2, r3, lsr #16 /* r2 = ..A9 */ + strh r2, [r0, #0x09] + mov r3, r3, lsl #16 /* r3 = 87.. */ + orr r3, r3, ip, lsr #16 /* r3 = 8765 */ + mov ip, ip, lsl #16 /* ip = 43.. */ + orr ip, ip, r1, lsr #16 /* ip = 4321 */ + mov r1, r1, lsr #8 /* r1 = .210 */ +#endif + str r3, [r0, #0x05] + str ip, [r0, #0x01] + strb r1, [r0] + bx lr + LMEMCPY_C_PAD + +/* + * 1110: dst is 8-bit aligned (byte 3), src is 16-bit aligned + */ +#ifdef __ARMEB__ + ldrh r2, [r1, #0x0a] /* r2 = ..AB */ + ldr ip, [r1, #0x06] /* ip = 6789 */ + ldr r3, [r1, #0x02] /* r3 = 2345 */ + ldrh r1, [r1] /* r1 = ..01 */ + strb r2, [r0, #0x0b] + mov r2, r2, lsr #8 /* r2 = ...A */ + orr r2, r2, ip, lsl #8 /* r2 = 789A */ + mov ip, ip, lsr #8 /* ip = .678 */ + orr ip, ip, r3, lsl #24 /* ip = 5678 */ + mov r3, r3, lsr #8 /* r3 = .234 */ + orr r3, r3, r1, lsl #24 /* r3 = 1234 */ + mov r1, r1, lsr #8 /* r1 = ...0 */ + strb r1, [r0] + str r3, [r0, #0x01] + str ip, [r0, #0x05] + strh r2, [r0, #0x09] +#else + ldrh r2, [r1] /* r2 = ..10 */ + ldr r3, [r1, #0x02] /* r3 = 5432 */ + ldr ip, [r1, #0x06] /* ip = 9876 */ + ldrh r1, [r1, #0x0a] /* r1 = ..BA */ + strb r2, [r0] + mov r2, r2, lsr #8 /* r2 = ...1 */ + orr r2, r2, r3, lsl #8 /* r2 = 4321 */ + mov r3, r3, lsr #24 /* r3 = ...5 */ + orr r3, r3, ip, lsl #8 /* r3 = 8765 */ + mov ip, ip, lsr #24 /* ip = ...9 */ + orr ip, ip, r1, lsl #8 /* ip = .BA9 */ + mov r1, r1, lsr #8 /* r1 = ...B */ + str r2, [r0, #0x01] + str r3, [r0, #0x05] + strh ip, [r0, #0x09] + strb r1, [r0, #0x0b] +#endif + bx lr + LMEMCPY_C_PAD + +/* + * 1111: dst is 8-bit aligned (byte 3), src is 8-bit aligned (byte 3) + */ + ldrb r2, [r1] + ldr r3, [r1, #0x01] + ldr ip, [r1, #0x05] + strb r2, [r0] + ldrh r2, [r1, #0x09] + ldrb r1, [r1, #0x0b] + str r3, [r0, #0x01] + str ip, [r0, #0x05] + strh r2, [r0, #0x09] + strb r1, [r0, #0x0b] + bx lr +#endif /* !_STANDALONE */ diff --git a/lib/libc/arm/string/memmove.S b/lib/libc/arm/string/memmove.S new file mode 100644 index 0000000..8b8baaf --- /dev/null +++ b/lib/libc/arm/string/memmove.S @@ -0,0 +1,582 @@ +/* $NetBSD: memmove.S,v 1.4 2003/10/14 07:51:45 scw Exp $ */ + +/*- + * Copyright (c) 1997 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Neil A. Carson and Mark Brinicombe + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +#ifndef _BCOPY +/* LINTSTUB: Func: void *memmove(void *, const void *, size_t) */ +ENTRY(memmove) +#else +/* bcopy = memcpy/memmove with arguments reversed. */ +/* LINTSTUB: Func: void bcopy(void *, void *, size_t) */ +ENTRY(bcopy) + /* switch the source and destination registers */ + eor r0, r1, r0 + eor r1, r0, r1 + eor r0, r1, r0 +#endif + /* Do the buffers overlap? */ + cmp r0, r1 + RETeq /* Bail now if src/dst are the same */ + subcc r3, r0, r1 /* if (dst > src) r3 = dst - src */ + subcs r3, r1, r0 /* if (src > dsr) r3 = src - dst */ + cmp r3, r2 /* if (r3 < len) we have an overlap */ + bcc PIC_SYM(_C_LABEL(memcpy), PLT) + + /* Determine copy direction */ + cmp r1, r0 + bcc .Lmemmove_backwards + + moveq r0, #0 /* Quick abort for len=0 */ + RETeq + + stmdb sp!, {r0, lr} /* memmove() returns dest addr */ + subs r2, r2, #4 + blt .Lmemmove_fl4 /* less than 4 bytes */ + ands r12, r0, #3 + bne .Lmemmove_fdestul /* oh unaligned destination addr */ + ands r12, r1, #3 + bne .Lmemmove_fsrcul /* oh unaligned source addr */ + +.Lmemmove_ft8: + /* We have aligned source and destination */ + subs r2, r2, #8 + blt .Lmemmove_fl12 /* less than 12 bytes (4 from above) */ + subs r2, r2, #0x14 + blt .Lmemmove_fl32 /* less than 32 bytes (12 from above) */ + stmdb sp!, {r4} /* borrow r4 */ + + /* blat 32 bytes at a time */ + /* XXX for really big copies perhaps we should use more registers */ +.Lmemmove_floop32: + ldmia r1!, {r3, r4, r12, lr} + stmia r0!, {r3, r4, r12, lr} + ldmia r1!, {r3, r4, r12, lr} + stmia r0!, {r3, r4, r12, lr} + subs r2, r2, #0x20 + bge .Lmemmove_floop32 + + cmn r2, #0x10 + ldmgeia r1!, {r3, r4, r12, lr} /* blat a remaining 16 bytes */ + stmgeia r0!, {r3, r4, r12, lr} + subge r2, r2, #0x10 + ldmia sp!, {r4} /* return r4 */ + +.Lmemmove_fl32: + adds r2, r2, #0x14 + + /* blat 12 bytes at a time */ +.Lmemmove_floop12: + ldmgeia r1!, {r3, r12, lr} + stmgeia r0!, {r3, r12, lr} + subges r2, r2, #0x0c + bge .Lmemmove_floop12 + +.Lmemmove_fl12: + adds r2, r2, #8 + blt .Lmemmove_fl4 + + subs r2, r2, #4 + ldrlt r3, [r1], #4 + strlt r3, [r0], #4 + ldmgeia r1!, {r3, r12} + stmgeia r0!, {r3, r12} + subge r2, r2, #4 + +.Lmemmove_fl4: + /* less than 4 bytes to go */ + adds r2, r2, #4 + ldmeqia sp!, {r0, pc} /* done */ + + /* copy the crud byte at a time */ + cmp r2, #2 + ldrb r3, [r1], #1 + strb r3, [r0], #1 + ldrgeb r3, [r1], #1 + strgeb r3, [r0], #1 + ldrgtb r3, [r1], #1 + strgtb r3, [r0], #1 + ldmia sp!, {r0, pc} + + /* erg - unaligned destination */ +.Lmemmove_fdestul: + rsb r12, r12, #4 + cmp r12, #2 + + /* align destination with byte copies */ + ldrb r3, [r1], #1 + strb r3, [r0], #1 + ldrgeb r3, [r1], #1 + strgeb r3, [r0], #1 + ldrgtb r3, [r1], #1 + strgtb r3, [r0], #1 + subs r2, r2, r12 + blt .Lmemmove_fl4 /* less the 4 bytes */ + + ands r12, r1, #3 + beq .Lmemmove_ft8 /* we have an aligned source */ + + /* erg - unaligned source */ + /* This is where it gets nasty ... */ +.Lmemmove_fsrcul: + bic r1, r1, #3 + ldr lr, [r1], #4 + cmp r12, #2 + bgt .Lmemmove_fsrcul3 + beq .Lmemmove_fsrcul2 + cmp r2, #0x0c + blt .Lmemmove_fsrcul1loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemmove_fsrcul1loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #8 +#else + mov r3, lr, lsr #8 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #24 + mov r4, r4, lsl #8 + orr r4, r4, r5, lsr #24 + mov r5, r5, lsl #8 + orr r5, r5, r12, lsr #24 + mov r12, r12, lsl #8 + orr r12, r12, lr, lsr #24 +#else + orr r3, r3, r4, lsl #24 + mov r4, r4, lsr #8 + orr r4, r4, r5, lsl #24 + mov r5, r5, lsr #8 + orr r5, r5, r12, lsl #24 + mov r12, r12, lsr #8 + orr r12, r12, lr, lsl #24 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemmove_fsrcul1loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemmove_fsrcul1l4 + +.Lmemmove_fsrcul1loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #8 +#else + mov r12, lr, lsr #8 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #24 +#else + orr r12, r12, lr, lsl #24 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemmove_fsrcul1loop4 + +.Lmemmove_fsrcul1l4: + sub r1, r1, #3 + b .Lmemmove_fl4 + +.Lmemmove_fsrcul2: + cmp r2, #0x0c + blt .Lmemmove_fsrcul2loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemmove_fsrcul2loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #16 +#else + mov r3, lr, lsr #16 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #16 + mov r4, r4, lsl #16 + orr r4, r4, r5, lsr #16 + mov r5, r5, lsl #16 + orr r5, r5, r12, lsr #16 + mov r12, r12, lsl #16 + orr r12, r12, lr, lsr #16 +#else + orr r3, r3, r4, lsl #16 + mov r4, r4, lsr #16 + orr r4, r4, r5, lsl #16 + mov r5, r5, lsr #16 + orr r5, r5, r12, lsl #16 + mov r12, r12, lsr #16 + orr r12, r12, lr, lsl #16 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemmove_fsrcul2loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemmove_fsrcul2l4 + +.Lmemmove_fsrcul2loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #16 +#else + mov r12, lr, lsr #16 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #16 +#else + orr r12, r12, lr, lsl #16 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemmove_fsrcul2loop4 + +.Lmemmove_fsrcul2l4: + sub r1, r1, #2 + b .Lmemmove_fl4 + +.Lmemmove_fsrcul3: + cmp r2, #0x0c + blt .Lmemmove_fsrcul3loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5} + +.Lmemmove_fsrcul3loop16: +#ifdef __ARMEB__ + mov r3, lr, lsl #24 +#else + mov r3, lr, lsr #24 +#endif + ldmia r1!, {r4, r5, r12, lr} +#ifdef __ARMEB__ + orr r3, r3, r4, lsr #8 + mov r4, r4, lsl #24 + orr r4, r4, r5, lsr #8 + mov r5, r5, lsl #24 + orr r5, r5, r12, lsr #8 + mov r12, r12, lsl #24 + orr r12, r12, lr, lsr #8 +#else + orr r3, r3, r4, lsl #8 + mov r4, r4, lsr #24 + orr r4, r4, r5, lsl #8 + mov r5, r5, lsr #24 + orr r5, r5, r12, lsl #8 + mov r12, r12, lsr #24 + orr r12, r12, lr, lsl #8 +#endif + stmia r0!, {r3-r5, r12} + subs r2, r2, #0x10 + bge .Lmemmove_fsrcul3loop16 + ldmia sp!, {r4, r5} + adds r2, r2, #0x0c + blt .Lmemmove_fsrcul3l4 + +.Lmemmove_fsrcul3loop4: +#ifdef __ARMEB__ + mov r12, lr, lsl #24 +#else + mov r12, lr, lsr #24 +#endif + ldr lr, [r1], #4 +#ifdef __ARMEB__ + orr r12, r12, lr, lsr #8 +#else + orr r12, r12, lr, lsl #8 +#endif + str r12, [r0], #4 + subs r2, r2, #4 + bge .Lmemmove_fsrcul3loop4 + +.Lmemmove_fsrcul3l4: + sub r1, r1, #1 + b .Lmemmove_fl4 + +.Lmemmove_backwards: + add r1, r1, r2 + add r0, r0, r2 + subs r2, r2, #4 + blt .Lmemmove_bl4 /* less than 4 bytes */ + ands r12, r0, #3 + bne .Lmemmove_bdestul /* oh unaligned destination addr */ + ands r12, r1, #3 + bne .Lmemmove_bsrcul /* oh unaligned source addr */ + +.Lmemmove_bt8: + /* We have aligned source and destination */ + subs r2, r2, #8 + blt .Lmemmove_bl12 /* less than 12 bytes (4 from above) */ + stmdb sp!, {r4, lr} + subs r2, r2, #0x14 /* less than 32 bytes (12 from above) */ + blt .Lmemmove_bl32 + + /* blat 32 bytes at a time */ + /* XXX for really big copies perhaps we should use more registers */ +.Lmemmove_bloop32: + ldmdb r1!, {r3, r4, r12, lr} + stmdb r0!, {r3, r4, r12, lr} + ldmdb r1!, {r3, r4, r12, lr} + stmdb r0!, {r3, r4, r12, lr} + subs r2, r2, #0x20 + bge .Lmemmove_bloop32 + +.Lmemmove_bl32: + cmn r2, #0x10 + ldmgedb r1!, {r3, r4, r12, lr} /* blat a remaining 16 bytes */ + stmgedb r0!, {r3, r4, r12, lr} + subge r2, r2, #0x10 + adds r2, r2, #0x14 + ldmgedb r1!, {r3, r12, lr} /* blat a remaining 12 bytes */ + stmgedb r0!, {r3, r12, lr} + subge r2, r2, #0x0c + ldmia sp!, {r4, lr} + +.Lmemmove_bl12: + adds r2, r2, #8 + blt .Lmemmove_bl4 + subs r2, r2, #4 + ldrlt r3, [r1, #-4]! + strlt r3, [r0, #-4]! + ldmgedb r1!, {r3, r12} + stmgedb r0!, {r3, r12} + subge r2, r2, #4 + +.Lmemmove_bl4: + /* less than 4 bytes to go */ + adds r2, r2, #4 + RETeq /* done */ + + /* copy the crud byte at a time */ + cmp r2, #2 + ldrb r3, [r1, #-1]! + strb r3, [r0, #-1]! + ldrgeb r3, [r1, #-1]! + strgeb r3, [r0, #-1]! + ldrgtb r3, [r1, #-1]! + strgtb r3, [r0, #-1]! + RET + + /* erg - unaligned destination */ +.Lmemmove_bdestul: + cmp r12, #2 + + /* align destination with byte copies */ + ldrb r3, [r1, #-1]! + strb r3, [r0, #-1]! + ldrgeb r3, [r1, #-1]! + strgeb r3, [r0, #-1]! + ldrgtb r3, [r1, #-1]! + strgtb r3, [r0, #-1]! + subs r2, r2, r12 + blt .Lmemmove_bl4 /* less than 4 bytes to go */ + ands r12, r1, #3 + beq .Lmemmove_bt8 /* we have an aligned source */ + + /* erg - unaligned source */ + /* This is where it gets nasty ... */ +.Lmemmove_bsrcul: + bic r1, r1, #3 + ldr r3, [r1, #0] + cmp r12, #2 + blt .Lmemmove_bsrcul1 + beq .Lmemmove_bsrcul2 + cmp r2, #0x0c + blt .Lmemmove_bsrcul3loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5, lr} + +.Lmemmove_bsrcul3loop16: +#ifdef __ARMEB__ + mov lr, r3, lsr #8 +#else + mov lr, r3, lsl #8 +#endif + ldmdb r1!, {r3-r5, r12} +#ifdef __ARMEB__ + orr lr, lr, r12, lsl #24 + mov r12, r12, lsr #8 + orr r12, r12, r5, lsl #24 + mov r5, r5, lsr #8 + orr r5, r5, r4, lsl #24 + mov r4, r4, lsr #8 + orr r4, r4, r3, lsl #24 +#else + orr lr, lr, r12, lsr #24 + mov r12, r12, lsl #8 + orr r12, r12, r5, lsr #24 + mov r5, r5, lsl #8 + orr r5, r5, r4, lsr #24 + mov r4, r4, lsl #8 + orr r4, r4, r3, lsr #24 +#endif + stmdb r0!, {r4, r5, r12, lr} + subs r2, r2, #0x10 + bge .Lmemmove_bsrcul3loop16 + ldmia sp!, {r4, r5, lr} + adds r2, r2, #0x0c + blt .Lmemmove_bsrcul3l4 + +.Lmemmove_bsrcul3loop4: +#ifdef __ARMEB__ + mov r12, r3, lsr #8 +#else + mov r12, r3, lsl #8 +#endif + ldr r3, [r1, #-4]! +#ifdef __ARMEB__ + orr r12, r12, r3, lsl #24 +#else + orr r12, r12, r3, lsr #24 +#endif + str r12, [r0, #-4]! + subs r2, r2, #4 + bge .Lmemmove_bsrcul3loop4 + +.Lmemmove_bsrcul3l4: + add r1, r1, #3 + b .Lmemmove_bl4 + +.Lmemmove_bsrcul2: + cmp r2, #0x0c + blt .Lmemmove_bsrcul2loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5, lr} + +.Lmemmove_bsrcul2loop16: +#ifdef __ARMEB__ + mov lr, r3, lsr #16 +#else + mov lr, r3, lsl #16 +#endif + ldmdb r1!, {r3-r5, r12} +#ifdef __ARMEB__ + orr lr, lr, r12, lsl #16 + mov r12, r12, lsr #16 + orr r12, r12, r5, lsl #16 + mov r5, r5, lsr #16 + orr r5, r5, r4, lsl #16 + mov r4, r4, lsr #16 + orr r4, r4, r3, lsl #16 +#else + orr lr, lr, r12, lsr #16 + mov r12, r12, lsl #16 + orr r12, r12, r5, lsr #16 + mov r5, r5, lsl #16 + orr r5, r5, r4, lsr #16 + mov r4, r4, lsl #16 + orr r4, r4, r3, lsr #16 +#endif + stmdb r0!, {r4, r5, r12, lr} + subs r2, r2, #0x10 + bge .Lmemmove_bsrcul2loop16 + ldmia sp!, {r4, r5, lr} + adds r2, r2, #0x0c + blt .Lmemmove_bsrcul2l4 + +.Lmemmove_bsrcul2loop4: +#ifdef __ARMEB__ + mov r12, r3, lsr #16 +#else + mov r12, r3, lsl #16 +#endif + ldr r3, [r1, #-4]! +#ifdef __ARMEB__ + orr r12, r12, r3, lsl #16 +#else + orr r12, r12, r3, lsr #16 +#endif + str r12, [r0, #-4]! + subs r2, r2, #4 + bge .Lmemmove_bsrcul2loop4 + +.Lmemmove_bsrcul2l4: + add r1, r1, #2 + b .Lmemmove_bl4 + +.Lmemmove_bsrcul1: + cmp r2, #0x0c + blt .Lmemmove_bsrcul1loop4 + sub r2, r2, #0x0c + stmdb sp!, {r4, r5, lr} + +.Lmemmove_bsrcul1loop32: +#ifdef __ARMEB__ + mov lr, r3, lsr #24 +#else + mov lr, r3, lsl #24 +#endif + ldmdb r1!, {r3-r5, r12} +#ifdef __ARMEB__ + orr lr, lr, r12, lsl #8 + mov r12, r12, lsr #24 + orr r12, r12, r5, lsl #8 + mov r5, r5, lsr #24 + orr r5, r5, r4, lsl #8 + mov r4, r4, lsr #24 + orr r4, r4, r3, lsl #8 +#else + orr lr, lr, r12, lsr #8 + mov r12, r12, lsl #24 + orr r12, r12, r5, lsr #8 + mov r5, r5, lsl #24 + orr r5, r5, r4, lsr #8 + mov r4, r4, lsl #24 + orr r4, r4, r3, lsr #8 +#endif + stmdb r0!, {r4, r5, r12, lr} + subs r2, r2, #0x10 + bge .Lmemmove_bsrcul1loop32 + ldmia sp!, {r4, r5, lr} + adds r2, r2, #0x0c + blt .Lmemmove_bsrcul1l4 + +.Lmemmove_bsrcul1loop4: +#ifdef __ARMEB__ + mov r12, r3, lsr #24 +#else + mov r12, r3, lsl #24 +#endif + ldr r3, [r1, #-4]! +#ifdef __ARMEB__ + orr r12, r12, r3, lsl #8 +#else + orr r12, r12, r3, lsr #8 +#endif + str r12, [r0, #-4]! + subs r2, r2, #4 + bge .Lmemmove_bsrcul1loop4 + +.Lmemmove_bsrcul1l4: + add r1, r1, #1 + b .Lmemmove_bl4 diff --git a/lib/libc/arm/string/memset.S b/lib/libc/arm/string/memset.S new file mode 100644 index 0000000..5387aab --- /dev/null +++ b/lib/libc/arm/string/memset.S @@ -0,0 +1,236 @@ +/* $NetBSD: memset.S,v 1.4 2003/10/14 07:51:45 scw Exp $ */ + +/* + * Copyright 2003 Wasabi Systems, Inc. + * All rights reserved. + * + * Written by Steve C. Woodford for Wasabi Systems, Inc. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed for the NetBSD Project by + * Wasabi Systems, Inc. + * 4. The name of Wasabi Systems, Inc. may not be used to endorse + * or promote products derived from this software without specific prior + * written permission. + * + * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ +/* + * Copyright (c) 1995 Mark Brinicombe. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by Mark Brinicombe. + * 4. The name of the company nor the name of the author may be used to + * endorse or promote products derived from this software without specific + * prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED + * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, + * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES + * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR + * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +/* + * memset: Sets a block of memory to the specified value + * + * On entry: + * r0 - dest address + * r1 - byte to write + * r2 - number of bytes to write + * + * On exit: + * r0 - dest address + */ +#ifdef _BZERO +/* LINTSTUB: Func: void bzero(void *, size_t) */ +ENTRY(bzero) + mov r3, #0x00 +#else +/* LINTSTUB: Func: void *memset(void *, int, size_t) */ +ENTRY(memset) + and r3, r1, #0xff /* We deal with bytes */ + mov r1, r2 +#endif + cmp r1, #0x04 /* Do we have less than 4 bytes */ + mov ip, r0 + blt .Lmemset_lessthanfour + + /* Ok first we will word align the address */ + ands r2, ip, #0x03 /* Get the bottom two bits */ + bne .Lmemset_wordunaligned /* The address is not word aligned */ + + /* We are now word aligned */ +.Lmemset_wordaligned: +#ifndef _BZERO + orr r3, r3, r3, lsl #8 /* Extend value to 16-bits */ +#endif +#ifdef _ARM_ARCH_5E + tst ip, #0x04 /* Quad-align for armv5e */ +#else + cmp r1, #0x10 +#endif +#ifndef _BZERO + orr r3, r3, r3, lsl #16 /* Extend value to 32-bits */ +#endif +#ifdef _ARM_ARCH_5E + subne r1, r1, #0x04 /* Quad-align if necessary */ + strne r3, [ip], #0x04 + cmp r1, #0x10 +#endif + blt .Lmemset_loop4 /* If less than 16 then use words */ + mov r2, r3 /* Duplicate data */ + cmp r1, #0x80 /* If < 128 then skip the big loop */ + blt .Lmemset_loop32 + + /* Do 128 bytes at a time */ +.Lmemset_loop128: + subs r1, r1, #0x80 +#ifdef _ARM_ARCH_5E + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 +#else + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} +#endif + bgt .Lmemset_loop128 + RETeq /* Zero length so just exit */ + + add r1, r1, #0x80 /* Adjust for extra sub */ + + /* Do 32 bytes at a time */ +.Lmemset_loop32: + subs r1, r1, #0x20 +#ifdef _ARM_ARCH_5E + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 +#else + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} +#endif + bgt .Lmemset_loop32 + RETeq /* Zero length so just exit */ + + adds r1, r1, #0x10 /* Partially adjust for extra sub */ + + /* Deal with 16 bytes or more */ +#ifdef _ARM_ARCH_5E + strged r2, [ip], #0x08 + strged r2, [ip], #0x08 +#else + stmgeia ip!, {r2-r3} + stmgeia ip!, {r2-r3} +#endif + RETeq /* Zero length so just exit */ + + addlt r1, r1, #0x10 /* Possibly adjust for extra sub */ + + /* We have at least 4 bytes so copy as words */ +.Lmemset_loop4: + subs r1, r1, #0x04 + strge r3, [ip], #0x04 + bgt .Lmemset_loop4 + RETeq /* Zero length so just exit */ + +#ifdef _ARM_ARCH_5E + /* Compensate for 64-bit alignment check */ + adds r1, r1, #0x04 + RETeq + cmp r1, #2 +#else + cmp r1, #-2 +#endif + + strb r3, [ip], #0x01 /* Set 1 byte */ + strgeb r3, [ip], #0x01 /* Set another byte */ + strgtb r3, [ip] /* and a third */ + RET /* Exit */ + +.Lmemset_wordunaligned: + rsb r2, r2, #0x004 + strb r3, [ip], #0x01 /* Set 1 byte */ + cmp r2, #0x02 + strgeb r3, [ip], #0x01 /* Set another byte */ + sub r1, r1, r2 + strgtb r3, [ip], #0x01 /* and a third */ + cmp r1, #0x04 /* More than 4 bytes left? */ + bge .Lmemset_wordaligned /* Yup */ + +.Lmemset_lessthanfour: + cmp r1, #0x00 + RETeq /* Zero length so exit */ + strb r3, [ip], #0x01 /* Set 1 byte */ + cmp r1, #0x02 + strgeb r3, [ip], #0x01 /* Set another byte */ + strgtb r3, [ip] /* and a third */ + RET /* Exit */ diff --git a/lib/libc/arm/string/strcmp.S b/lib/libc/arm/string/strcmp.S new file mode 100644 index 0000000..e5cba7d --- /dev/null +++ b/lib/libc/arm/string/strcmp.S @@ -0,0 +1,43 @@ +/* $NetBSD: strcmp.S,v 1.3 2003/04/05 23:08:52 bjh21 Exp $ */ + +/* + * Copyright (c) 2002 ARM Ltd + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the company may not be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED + * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED + * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF + * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING + * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> + +__FBSDID("$FreeBSD$"); + +ENTRY(strcmp) +1: + ldrb r2, [r0], #1 + ldrb r3, [r1], #1 + cmp r2, #1 + cmpcs r2, r3 + beq 1b + sub r0, r2, r3 + RET diff --git a/lib/libc/arm/string/strlen.S b/lib/libc/arm/string/strlen.S new file mode 100644 index 0000000..378257d --- /dev/null +++ b/lib/libc/arm/string/strlen.S @@ -0,0 +1,78 @@ +/*- + * Copyright (c) 2005 Olivier Houchard + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); + +ENTRY(strlen) + mov r1, #0 + /* Check that the pointer is aligned on 32 bits. */ + ands r3, r0, #3 + beq .Loop + sub r0, r0, r3 + ldr r2, [r0] + add r0, r0, #4 + cmp r3, #2 + blt .Ldo_3 + bgt .Ldo_1 + /* So that the N bit is set. */ + cmp r3, #0 + b .Ldo_2 + +.Loop: + ldr r2, [r0] + add r0, r0, #4 +#ifndef __ARMEB__ + ands r3, r2, #0x000000ff +#else + ands r3, r2, #0xff000000 +#endif + addne r1, r1, #1 +.Ldo_3: +#ifndef __ARMEB__ + andnes r3, r2, #0x0000ff00 +#else + andnes r3, r2, #0x00ff0000 +#endif + addne r1, r1, #1 +.Ldo_2: +#ifndef __ARMEB__ + andnes r3, r2, #0x00ff0000 +#else + andnes r3, r2, #0x0000ff00 +#endif + addne r1, r1, #1 +.Ldo_1: +#ifndef __ARMEB__ + andnes r3, r2, #0xff000000 +#else + andnes r3, r2, #0x000000ff +#endif + addne r1, r1, #1 + bne .Loop +.Lexit: + mov r0, r1 + RET diff --git a/lib/libc/arm/string/strncmp.S b/lib/libc/arm/string/strncmp.S new file mode 100644 index 0000000..fce0159 --- /dev/null +++ b/lib/libc/arm/string/strncmp.S @@ -0,0 +1,54 @@ +/* $NetBSD: strncmp.S,v 1.2 2003/04/05 23:08:52 bjh21 Exp $ */ + +/* + * Copyright (c) 2002 ARM Ltd + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. The name of the company may not be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED + * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED + * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR + * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF + * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING + * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS + * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include <machine/asm.h> + +__FBSDID("$FreeBSD$"); + +ENTRY(strncmp) +/* if (len == 0) return 0 */ + cmp r2, #0 + moveq r0, #0 + moveq pc, lr + +/* ip == last src address to compare */ + adds ip, r0, r2 +/* Use last possible address on overflow. */ + movcs ip, #0 + sub ip, ip, #1 +1: + ldrb r2, [r0], #1 + ldrb r3, [r1], #1 + cmp ip, r0 + cmpcs r2, #1 + cmpcs r2, r3 + beq 1b + sub r0, r2, r3 + RET diff --git a/lib/libc/arm/sys/Makefile.inc b/lib/libc/arm/sys/Makefile.inc new file mode 100644 index 0000000..1a58eae --- /dev/null +++ b/lib/libc/arm/sys/Makefile.inc @@ -0,0 +1,11 @@ +# $FreeBSD$ + +MDASM= Ovfork.S brk.S cerror.S pipe.S ptrace.S sbrk.S shmat.S sigreturn.S syscall.S + +# Don't generate default code for these syscalls: +NOASM= break.o exit.o getlogin.o openbsd_poll.o sstk.o vfork.o yield.o + +PSEUDO= _exit.o _getlogin.o +.if !defined(WITHOUT_SYSCALL_COMPAT) +PSEUDO+= _pread.o _pwrite.o _lseek.o _mmap.o _ftruncate.o _truncate.o +.endif diff --git a/lib/libc/arm/sys/Ovfork.S b/lib/libc/arm/sys/Ovfork.S new file mode 100644 index 0000000..286347e --- /dev/null +++ b/lib/libc/arm/sys/Ovfork.S @@ -0,0 +1,54 @@ +/* $NetBSD: Ovfork.S,v 1.6 2003/08/07 16:42:03 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)Ovfork.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +/* + * pid = vfork(); + * + * On return from the SWI: + * r1 == 0 in parent process, r1 == 1 in child process. + * r0 == pid of child in parent, r0 == pid of parent in child. + */ + .text + .align 0 + +ENTRY(vfork) + mov r2, r14 + SYSTRAP(vfork) + bcs PIC_SYM(CERROR, PLT) + sub r1, r1, #1 /* r1 == 0xffffffff if parent, 0 if child */ + and r0, r0, r1 /* r0 == 0 if child, else unchanged */ + mov r15, r2 diff --git a/lib/libc/arm/sys/brk.S b/lib/libc/arm/sys/brk.S new file mode 100644 index 0000000..5fdf90c --- /dev/null +++ b/lib/libc/arm/sys/brk.S @@ -0,0 +1,100 @@ +/* $NetBSD: brk.S,v 1.6 2003/08/07 16:42:04 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)brk.s 5.2 (Berkeley) 12/17/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + + .globl _C_LABEL(_end) + .globl CURBRK + +#ifdef WEAK_ALIAS +WEAK_ALIAS(brk, _brk) +#endif + + .data + .align 0 + .globl _C_LABEL(minbrk) + .type _C_LABEL(minbrk),#object +_C_LABEL(minbrk): + .word _C_LABEL(_end) + +/* + * Change the data segment size + */ +ENTRY(_brk) +#ifdef PIC + /* Setup the GOT */ + ldr r3, .Lgot + add r3, pc, r3 +.L1: + ldr r1, .Lminbrk + ldr r1, [r3, r1] +#else + ldr r1, .Lminbrk +#endif + /* Get the minimum allowable brk address */ + ldr r1, [r1] + + /* + * Valid the address specified and set to the minimum + * if the address is below minbrk. + */ + cmp r0, r1 + movlt r0, r1 + mov r2, r0 + SYSTRAP(break) + bcs PIC_SYM(CERROR, PLT) + +#ifdef PIC + ldr r1, .Lcurbrk + ldr r1, [r3, r1] +#else + ldr r1, .Lcurbrk +#endif + /* Store the new address in curbrk */ + str r2, [r1] + + /* Return 0 for success */ + mov r0, #0x00000000 + RET + + .align 2 +#ifdef PIC +.Lgot: + .word _GLOBAL_OFFSET_TABLE_ - (.L1+4) +#endif +.Lminbrk: + .word PIC_SYM(_C_LABEL(minbrk), GOT) +.Lcurbrk: + .word PIC_SYM(CURBRK, GOT) diff --git a/lib/libc/arm/sys/cerror.S b/lib/libc/arm/sys/cerror.S new file mode 100644 index 0000000..e807285 --- /dev/null +++ b/lib/libc/arm/sys/cerror.S @@ -0,0 +1,48 @@ +/* $NetBSD: cerror.S,v 1.5 2003/08/07 16:42:04 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)cerror.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +.globl _C_LABEL(__error) +.type _C_LABEL(__error),%function + +ASENTRY(CERROR) + stmfd sp!, {r4, lr} + mov r4, r0 + bl PIC_SYM(_C_LABEL(__error), PLT) + str r4, [r0] + mvn r0, #0x00000000 + mvn r1, #0x00000000 + ldmfd sp!, {r4, pc} diff --git a/lib/libc/arm/sys/fork.S b/lib/libc/arm/sys/fork.S new file mode 100644 index 0000000..a5ae1f0 --- /dev/null +++ b/lib/libc/arm/sys/fork.S @@ -0,0 +1,49 @@ +/* $NetBSD: fork.S,v 1.5 2003/08/07 16:42:04 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)fork.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +/* + * pid = fork(); + * + * On return from the SWI: + * r1 == 0 in parent process, r1 == 1 in child process. + * r0 == pid of child in parent, r0 == pid of parent in child. + */ + +_SYSCALL(fork) + sub r1, r1, #1 /* r1 == 0xffffffff if parent, 0 if child */ + and r0, r0, r1 /* r0 == 0 if child, else unchanged */ + RET diff --git a/lib/libc/arm/sys/pipe.S b/lib/libc/arm/sys/pipe.S new file mode 100644 index 0000000..83518fc2 --- /dev/null +++ b/lib/libc/arm/sys/pipe.S @@ -0,0 +1,50 @@ +/* $NetBSD: pipe.S,v 1.5 2003/08/07 16:42:04 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)pipe.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +#ifdef WEAK_ALIAS +WEAK_ALIAS(pipe, _pipe) +WEAK_ALIAS(__sys_pipe, _pipe) +#endif + +ENTRY(_pipe) + mov r2, r0 + SYSTRAP(pipe) + bcs PIC_SYM(CERROR, PLT) + str r0, [r2, #0x0000] + str r1, [r2, #0x0004] + mov r0, #0x00000000 + RET diff --git a/lib/libc/arm/sys/ptrace.S b/lib/libc/arm/sys/ptrace.S new file mode 100644 index 0000000..3cc13f3 --- /dev/null +++ b/lib/libc/arm/sys/ptrace.S @@ -0,0 +1,48 @@ +/* $NetBSD: ptrace.S,v 1.7 2003/08/07 16:42:04 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)ptrace.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +ENTRY(ptrace) + stmfd sp!, {r0-r3, lr} + sub sp, sp, #4 /* align stack */ + bl PIC_SYM(_C_LABEL(__error), PLT) + add sp, sp, #4 /* unalign stack */ + mov r1, #0x00000000 + str r1, [r0] + ldmfd sp!, {r0-r3, lr} + SYSTRAP(ptrace) + bcs PIC_SYM(CERROR, PLT) + RET diff --git a/lib/libc/arm/sys/sbrk.S b/lib/libc/arm/sys/sbrk.S new file mode 100644 index 0000000..d76e85a --- /dev/null +++ b/lib/libc/arm/sys/sbrk.S @@ -0,0 +1,88 @@ +/* $NetBSD: sbrk.S,v 1.7 2003/08/07 16:42:05 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)sbrk.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + + .globl _C_LABEL(_end) + +#ifdef WEAK_ALIAS +WEAK_ALIAS(sbrk, _sbrk) +#endif + + .data + .align 0 + .globl CURBRK + .type CURBRK,#object +CURBRK: + .word _C_LABEL(_end) + +/* + * Change the data segment size + */ +ENTRY(_sbrk) +#ifdef PIC + /* Setup the GOT */ + ldr r3, .Lgot + add r3, pc, r3 +.L1: + ldr r2, .Lcurbrk + ldr r2, [r3, r2] +#else + ldr r2, .Lcurbrk +#endif + /* Get the current brk address */ + ldr r1, [r2] + + /* Calculate new value */ + mov r3, r0 + add r0, r0, r1 + SYSTRAP(break) + bcs PIC_SYM(CERROR, PLT) + + /* Store new curbrk value */ + ldr r0, [r2] + add r1, r0, r3 + str r1, [r2] + + /* Return old curbrk value */ + RET + + .align 0 +#ifdef PIC +.Lgot: + .word _GLOBAL_OFFSET_TABLE_ - (.L1+4) +#endif +.Lcurbrk: + .word PIC_SYM(CURBRK, GOT) diff --git a/lib/libc/arm/sys/shmat.S b/lib/libc/arm/sys/shmat.S new file mode 100644 index 0000000..3fc3d02 --- /dev/null +++ b/lib/libc/arm/sys/shmat.S @@ -0,0 +1,7 @@ +/* $NetBSD: shmat.S,v 1.1 2000/12/29 20:14:04 bjh21 Exp $ */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +RSYSCALL(shmat) diff --git a/lib/libc/arm/sys/sigreturn.S b/lib/libc/arm/sys/sigreturn.S new file mode 100644 index 0000000..1e0f245 --- /dev/null +++ b/lib/libc/arm/sys/sigreturn.S @@ -0,0 +1,42 @@ +/* $NetBSD: __sigreturn14.S,v 1.3 2003/08/07 16:42:03 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)sigreturn.s 5.2 (Berkeley) 12/17/90" + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +/* + * We must preserve the state of the registers as the user has set them up. + */ + +RSYSCALL(sigreturn) diff --git a/lib/libc/arm/sys/syscall.S b/lib/libc/arm/sys/syscall.S new file mode 100644 index 0000000..73e6b83 --- /dev/null +++ b/lib/libc/arm/sys/syscall.S @@ -0,0 +1,38 @@ +/* $NetBSD: syscall.S,v 1.4 2003/08/07 16:42:05 agc Exp $ */ + +/*- + * Copyright (c) 1990 The Regents of the University of California. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * from: @(#)syscall.s 5.1 (Berkeley) 4/23/90 + */ + +#include <machine/asm.h> +__FBSDID("$FreeBSD$"); +#include "SYS.h" + +RSYSCALL(syscall) |