7a966f2ded
This provides a 30% reduction in system time and a 6% reduction in wallclock time for a make buildworld on my xp1000 (one 21264). FWIW, I've been running this for nearly 2 months without problems. Portions submitted by: ticso, jhb Tested by: jhb (ds20 dual 21264)
524 lines
14 KiB
C
524 lines
14 KiB
C
/*-
|
|
* Copyright (c) 1998 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.
|
|
*
|
|
* $FreeBSD$
|
|
*/
|
|
|
|
#ifndef _MACHINE_ATOMIC_H_
|
|
#define _MACHINE_ATOMIC_H_
|
|
|
|
#include <machine/alpha_cpu.h>
|
|
|
|
/*
|
|
* Various simple arithmetic on memory which is atomic in the presence
|
|
* of interrupts and SMP safe.
|
|
*/
|
|
|
|
void atomic_set_8(volatile u_int8_t *, u_int8_t);
|
|
void atomic_clear_8(volatile u_int8_t *, u_int8_t);
|
|
void atomic_add_8(volatile u_int8_t *, u_int8_t);
|
|
void atomic_subtract_8(volatile u_int8_t *, u_int8_t);
|
|
|
|
void atomic_set_16(volatile u_int16_t *, u_int16_t);
|
|
void atomic_clear_16(volatile u_int16_t *, u_int16_t);
|
|
void atomic_add_16(volatile u_int16_t *, u_int16_t);
|
|
void atomic_subtract_16(volatile u_int16_t *, u_int16_t);
|
|
|
|
static __inline void atomic_set_32(volatile u_int32_t *p, u_int32_t v)
|
|
{
|
|
u_int32_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldl_l %0, %2\n\t" /* load old value */
|
|
"bis %0, %3, %0\n\t" /* calculate new value */
|
|
"stl_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "=m" (*p)
|
|
: "m" (*p), "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_clear_32(volatile u_int32_t *p, u_int32_t v)
|
|
{
|
|
u_int32_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldl_l %0, %1\n\t" /* load old value */
|
|
"bic %0, %2, %0\n\t" /* calculate new value */
|
|
"stl_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_add_32(volatile u_int32_t *p, u_int32_t v)
|
|
{
|
|
u_int32_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldl_l %0, %1\n\t" /* load old value */
|
|
"addl %0, %2, %0\n\t" /* calculate new value */
|
|
"stl_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_subtract_32(volatile u_int32_t *p, u_int32_t v)
|
|
{
|
|
u_int32_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldl_l %0, %1\n\t" /* load old value */
|
|
"subl %0, %2, %0\n\t" /* calculate new value */
|
|
"stl_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline u_int32_t atomic_readandclear_32(volatile u_int32_t *addr)
|
|
{
|
|
u_int32_t result,temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"wmb\n" /* ensure pending writes have drained */
|
|
"1:\tldl_l %0,%2\n\t" /* load current value, asserting lock */
|
|
"ldiq %1,0\n\t" /* value to store */
|
|
"stl_c %1,%2\n\t" /* attempt to store */
|
|
"beq %1,2f\n\t" /* if the store failed, spin */
|
|
"br 3f\n" /* it worked, exit */
|
|
"2:\tbr 1b\n" /* *addr not updated, loop */
|
|
"3:\n" /* it worked */
|
|
: "=&r"(result), "=&r"(temp), "+m" (*addr)
|
|
:
|
|
: "memory");
|
|
#endif
|
|
|
|
return result;
|
|
}
|
|
|
|
static __inline void atomic_set_64(volatile u_int64_t *p, u_int64_t v)
|
|
{
|
|
u_int64_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldq_l %0, %1\n\t" /* load old value */
|
|
"bis %0, %2, %0\n\t" /* calculate new value */
|
|
"stq_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_clear_64(volatile u_int64_t *p, u_int64_t v)
|
|
{
|
|
u_int64_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldq_l %0, %1\n\t" /* load old value */
|
|
"bic %0, %2, %0\n\t" /* calculate new value */
|
|
"stq_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_add_64(volatile u_int64_t *p, u_int64_t v)
|
|
{
|
|
u_int64_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldq_l %0, %1\n\t" /* load old value */
|
|
"addq %0, %2, %0\n\t" /* calculate new value */
|
|
"stq_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline void atomic_subtract_64(volatile u_int64_t *p, u_int64_t v)
|
|
{
|
|
u_int64_t temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldq_l %0, %1\n\t" /* load old value */
|
|
"subq %0, %2, %0\n\t" /* calculate new value */
|
|
"stq_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 2f\n\t" /* spin if failed */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"2:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (temp), "+m" (*p)
|
|
: "r" (v)
|
|
: "memory");
|
|
#endif
|
|
}
|
|
|
|
static __inline u_int64_t atomic_readandclear_64(volatile u_int64_t *addr)
|
|
{
|
|
u_int64_t result,temp;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"wmb\n" /* ensure pending writes have drained */
|
|
"1:\tldq_l %0,%2\n\t" /* load current value, asserting lock */
|
|
"ldiq %1,0\n\t" /* value to store */
|
|
"stq_c %1,%2\n\t" /* attempt to store */
|
|
"beq %1,2f\n\t" /* if the store failed, spin */
|
|
"br 3f\n" /* it worked, exit */
|
|
"2:\tbr 1b\n" /* *addr not updated, loop */
|
|
"3:\n" /* it worked */
|
|
: "=&r"(result), "=&r"(temp), "+m" (*addr)
|
|
:
|
|
: "memory");
|
|
#endif
|
|
|
|
return result;
|
|
}
|
|
|
|
#define atomic_set_char atomic_set_8
|
|
#define atomic_clear_char atomic_clear_8
|
|
#define atomic_add_char atomic_add_8
|
|
#define atomic_subtract_char atomic_subtract_8
|
|
|
|
#define atomic_set_short atomic_set_16
|
|
#define atomic_clear_short atomic_clear_16
|
|
#define atomic_add_short atomic_add_16
|
|
#define atomic_subtract_short atomic_subtract_16
|
|
|
|
#define atomic_set_int atomic_set_32
|
|
#define atomic_clear_int atomic_clear_32
|
|
#define atomic_add_int atomic_add_32
|
|
#define atomic_subtract_int atomic_subtract_32
|
|
#define atomic_readandclear_int atomic_readandclear_32
|
|
|
|
#define atomic_set_long atomic_set_64
|
|
#define atomic_clear_long atomic_clear_64
|
|
#define atomic_add_long atomic_add_64
|
|
#define atomic_subtract_long atomic_subtract_64
|
|
#define atomic_readandclear_long atomic_readandclear_64
|
|
|
|
#define ATOMIC_ACQ_REL(NAME, WIDTH, TYPE) \
|
|
static __inline void \
|
|
atomic_##NAME##_acq_##WIDTH(volatile u_int##WIDTH##_t *p, u_int##WIDTH##_t v)\
|
|
{ \
|
|
atomic_##NAME##_##WIDTH(p, v); \
|
|
alpha_mb(); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_##NAME##_rel_##WIDTH(volatile u_int##WIDTH##_t *p, u_int##WIDTH##_t v)\
|
|
{ \
|
|
alpha_mb(); \
|
|
atomic_##NAME##_##WIDTH(p, v); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_##NAME##_acq_##TYPE(volatile u_int##WIDTH##_t *p, u_int##WIDTH##_t v)\
|
|
{ \
|
|
atomic_##NAME##_##WIDTH(p, v); \
|
|
alpha_mb(); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_##NAME##_rel_##TYPE(volatile u_int##WIDTH##_t *p, u_int##WIDTH##_t v)\
|
|
{ \
|
|
alpha_mb(); \
|
|
atomic_##NAME##_##WIDTH(p, v); \
|
|
}
|
|
|
|
ATOMIC_ACQ_REL(set, 8, char)
|
|
ATOMIC_ACQ_REL(clear, 8, char)
|
|
ATOMIC_ACQ_REL(add, 8, char)
|
|
ATOMIC_ACQ_REL(subtract, 8, char)
|
|
ATOMIC_ACQ_REL(set, 16, short)
|
|
ATOMIC_ACQ_REL(clear, 16, short)
|
|
ATOMIC_ACQ_REL(add, 16, short)
|
|
ATOMIC_ACQ_REL(subtract, 16, short)
|
|
ATOMIC_ACQ_REL(set, 32, int)
|
|
ATOMIC_ACQ_REL(clear, 32, int)
|
|
ATOMIC_ACQ_REL(add, 32, int)
|
|
ATOMIC_ACQ_REL(subtract, 32, int)
|
|
ATOMIC_ACQ_REL(set, 64, long)
|
|
ATOMIC_ACQ_REL(clear, 64, long)
|
|
ATOMIC_ACQ_REL(add, 64, long)
|
|
ATOMIC_ACQ_REL(subtract, 64, long)
|
|
|
|
#undef ATOMIC_ACQ_REL
|
|
|
|
/*
|
|
* We assume that a = b will do atomic loads and stores.
|
|
*/
|
|
#define ATOMIC_STORE_LOAD(TYPE, WIDTH) \
|
|
static __inline u_##TYPE \
|
|
atomic_load_acq_##WIDTH(volatile u_##TYPE *p) \
|
|
{ \
|
|
u_##TYPE v; \
|
|
\
|
|
v = *p; \
|
|
alpha_mb(); \
|
|
return (v); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_store_rel_##WIDTH(volatile u_##TYPE *p, u_##TYPE v)\
|
|
{ \
|
|
alpha_mb(); \
|
|
*p = v; \
|
|
} \
|
|
static __inline u_##TYPE \
|
|
atomic_load_acq_##TYPE(volatile u_##TYPE *p) \
|
|
{ \
|
|
u_##TYPE v; \
|
|
\
|
|
v = *p; \
|
|
alpha_mb(); \
|
|
return (v); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_store_rel_##TYPE(volatile u_##TYPE *p, u_##TYPE v)\
|
|
{ \
|
|
alpha_mb(); \
|
|
*p = v; \
|
|
}
|
|
|
|
ATOMIC_STORE_LOAD(char, 8)
|
|
ATOMIC_STORE_LOAD(short, 16)
|
|
ATOMIC_STORE_LOAD(int, 32)
|
|
ATOMIC_STORE_LOAD(long, 64)
|
|
|
|
#undef ATOMIC_STORE_LOAD
|
|
|
|
/*
|
|
* Atomically compare the value stored at *p with cmpval and if the
|
|
* two values are equal, update the value of *p with newval. Returns
|
|
* zero if the compare failed, nonzero otherwise.
|
|
*/
|
|
static __inline u_int32_t
|
|
atomic_cmpset_32(volatile u_int32_t* p, u_int32_t cmpval, u_int32_t newval)
|
|
{
|
|
u_int32_t ret;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldl_l %0, %1\n\t" /* load old value */
|
|
"cmpeq %0, %2, %0\n\t" /* compare */
|
|
"beq %0, 2f\n\t" /* exit if not equal */
|
|
"mov %3, %0\n\t" /* value to store */
|
|
"stl_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 3f\n\t" /* if it failed, spin */
|
|
"2:\n" /* done */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"3:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (ret), "+m" (*p)
|
|
: "r" ((long)(int)cmpval), "r" (newval)
|
|
: "memory");
|
|
#endif
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Atomically compare the value stored at *p with cmpval and if the
|
|
* two values are equal, update the value of *p with newval. Returns
|
|
* zero if the compare failed, nonzero otherwise.
|
|
*/
|
|
static __inline u_int64_t
|
|
atomic_cmpset_64(volatile u_int64_t* p, u_int64_t cmpval, u_int64_t newval)
|
|
{
|
|
u_int64_t ret;
|
|
|
|
#ifdef __GNUC__
|
|
__asm __volatile (
|
|
"1:\tldq_l %0, %1\n\t" /* load old value */
|
|
"cmpeq %0, %2, %0\n\t" /* compare */
|
|
"beq %0, 2f\n\t" /* exit if not equal */
|
|
"mov %3, %0\n\t" /* value to store */
|
|
"stq_c %0, %1\n\t" /* attempt to store */
|
|
"beq %0, 3f\n\t" /* if it failed, spin */
|
|
"2:\n" /* done */
|
|
".section .text3,\"ax\"\n" /* improve branch prediction */
|
|
"3:\tbr 1b\n" /* try again */
|
|
".previous\n"
|
|
: "=&r" (ret), "+m" (*p)
|
|
: "r" (cmpval), "r" (newval)
|
|
: "memory");
|
|
#endif
|
|
|
|
return ret;
|
|
}
|
|
|
|
#define atomic_cmpset_int atomic_cmpset_32
|
|
#define atomic_cmpset_long atomic_cmpset_64
|
|
|
|
static __inline int
|
|
atomic_cmpset_ptr(volatile void *dst, void *exp, void *src)
|
|
{
|
|
|
|
return (atomic_cmpset_long((volatile u_long *)dst, (u_long)exp,
|
|
(u_long)src));
|
|
}
|
|
|
|
static __inline u_int32_t
|
|
atomic_cmpset_acq_32(volatile u_int32_t *p, u_int32_t cmpval, u_int32_t newval)
|
|
{
|
|
int retval;
|
|
|
|
retval = atomic_cmpset_32(p, cmpval, newval);
|
|
alpha_mb();
|
|
return (retval);
|
|
}
|
|
|
|
static __inline u_int32_t
|
|
atomic_cmpset_rel_32(volatile u_int32_t *p, u_int32_t cmpval, u_int32_t newval)
|
|
{
|
|
alpha_mb();
|
|
return (atomic_cmpset_32(p, cmpval, newval));
|
|
}
|
|
|
|
static __inline u_int64_t
|
|
atomic_cmpset_acq_64(volatile u_int64_t *p, u_int64_t cmpval, u_int64_t newval)
|
|
{
|
|
int retval;
|
|
|
|
retval = atomic_cmpset_64(p, cmpval, newval);
|
|
alpha_mb();
|
|
return (retval);
|
|
}
|
|
|
|
static __inline u_int64_t
|
|
atomic_cmpset_rel_64(volatile u_int64_t *p, u_int64_t cmpval, u_int64_t newval)
|
|
{
|
|
alpha_mb();
|
|
return (atomic_cmpset_64(p, cmpval, newval));
|
|
}
|
|
|
|
#define atomic_cmpset_acq_int atomic_cmpset_acq_32
|
|
#define atomic_cmpset_rel_int atomic_cmpset_rel_32
|
|
#define atomic_cmpset_acq_long atomic_cmpset_acq_64
|
|
#define atomic_cmpset_rel_long atomic_cmpset_rel_64
|
|
|
|
static __inline int
|
|
atomic_cmpset_acq_ptr(volatile void *dst, void *exp, void *src)
|
|
{
|
|
|
|
return (atomic_cmpset_acq_long((volatile u_long *)dst, (u_long)exp,
|
|
(u_long)src));
|
|
}
|
|
|
|
static __inline int
|
|
atomic_cmpset_rel_ptr(volatile void *dst, void *exp, void *src)
|
|
{
|
|
|
|
return (atomic_cmpset_rel_long((volatile u_long *)dst, (u_long)exp,
|
|
(u_long)src));
|
|
}
|
|
|
|
static __inline void *
|
|
atomic_load_acq_ptr(volatile void *p)
|
|
{
|
|
return (void *)atomic_load_acq_long((volatile u_long *)p);
|
|
}
|
|
|
|
static __inline void
|
|
atomic_store_rel_ptr(volatile void *p, void *v)
|
|
{
|
|
atomic_store_rel_long((volatile u_long *)p, (u_long)v);
|
|
}
|
|
|
|
#define ATOMIC_PTR(NAME) \
|
|
static __inline void \
|
|
atomic_##NAME##_ptr(volatile void *p, uintptr_t v) \
|
|
{ \
|
|
atomic_##NAME##_long((volatile u_long *)p, v); \
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_##NAME##_acq_ptr(volatile void *p, uintptr_t v) \
|
|
{ \
|
|
atomic_##NAME##_acq_long((volatile u_long *)p, v);\
|
|
} \
|
|
\
|
|
static __inline void \
|
|
atomic_##NAME##_rel_ptr(volatile void *p, uintptr_t v) \
|
|
{ \
|
|
atomic_##NAME##_rel_long((volatile u_long *)p, v);\
|
|
}
|
|
|
|
ATOMIC_PTR(set)
|
|
ATOMIC_PTR(clear)
|
|
ATOMIC_PTR(add)
|
|
ATOMIC_PTR(subtract)
|
|
|
|
#undef ATOMIC_PTR
|
|
|
|
#endif /* ! _MACHINE_ATOMIC_H_ */
|