From 8760d94e0ef6d52260765a9246aaac073613055e Mon Sep 17 00:00:00 2001 From: Paolo Bonzini Date: Mon, 1 Mar 2010 13:57:00 -0500 Subject: [PATCH 1/1] add uatomic_generic.h, use it for common definitions uatomic_generic.h can be included by uatomic_arch_*.h to provide useful default definitions. uatomic_arch_*.h can define whatever builtins it wants to override, then uatomic_generic.h will provide what is not already defined, as follows: - uatomic_cmpxchg will use __sync_val_compare_and_swap builtins; - uatomic_add_return will use __sync_fetch_and_add if uatomic_arch_*.h did not provide a definition of uatomic_cmpxchg. If it did, we assume __sync builtins are buggy or otherwise undesirable on this platform, so uatomic_generic.h will implement uatomic_add_return in terms of uatomic_cmpxchg too. - uatomic_xchg is like uatomic_add_return. However, since GCC does not provide an adequate builtin, it needs to use a compare-and-swap loop using __sync_bool_compare_and_swap if uatomic_cmpxchg is not provided. - uatomic_sub_return/uatomic_add/uatomic_sub will be implemented in terms of uatomic_add_return; - uatomic_inc/uatomic_dec will be implemented in terms of uatomic_add. After this patch, uatomic_generic.h is already used for the latter two categories. The next patch will use uatomic_generic.h whenever there is no assembly code involved, or otherwise it makes no difference to use uatomic_generic.h. We keep custom per-arch macros to provide support for compilers back to early GCC 3.x versions; however future ports may not have this limitation. The hunk in tests/test_uatomic.c is only needed for bisectability and will be removed later. [edit by Mathieu Desnoyers] Fixed Makefile.am conflict. Signed-off-by: Paolo Bonzini Signed-off-by: Mathieu Desnoyers --- Makefile.am | 3 +- tests/test_uatomic.c | 2 + urcu/uatomic_arch_ppc.h | 15 +- urcu/uatomic_arch_s390.h | 15 +- urcu/uatomic_arch_sparc64.h | 15 +- urcu/uatomic_arch_x86.h | 24 +--- urcu/uatomic_generic.h | 263 ++++++++++++++++++++++++++++++++++++ 7 files changed, 280 insertions(+), 57 deletions(-) create mode 100644 urcu/uatomic_generic.h diff --git a/Makefile.am b/Makefile.am index 2d6c65b..73a3015 100644 --- a/Makefile.am +++ b/Makefile.am @@ -6,7 +6,8 @@ SUBDIRS = . tests include_HEADERS = urcu.h $(top_srcdir)/urcu-*.h nobase_dist_include_HEADERS = urcu/compiler.h urcu/hlist.h urcu/list.h \ - urcu/rculist.h urcu/rcuhlist.h urcu/system.h urcu/urcu-futex.h + urcu/rculist.h urcu/rcuhlist.h urcu/system.h urcu/urcu-futex.h \ + urcu/uatomic_generic.h nobase_nodist_include_HEADERS = urcu/arch.h urcu/uatomic_arch.h urcu/config.h EXTRA_DIST = $(top_srcdir)/urcu/arch_*.h $(top_srcdir)/urcu/uatomic_arch_*.h \ diff --git a/tests/test_uatomic.c b/tests/test_uatomic.c index 68cb6df..c0f36fe 100644 --- a/tests/test_uatomic.c +++ b/tests/test_uatomic.c @@ -1,5 +1,7 @@ #include #include + +#define UATOMIC_NO_LINK_ERROR #include #if (defined(__i386__) || defined(__x86_64__)) diff --git a/urcu/uatomic_arch_ppc.h b/urcu/uatomic_arch_ppc.h index 8da192e..39c4c24 100644 --- a/urcu/uatomic_arch_ppc.h +++ b/urcu/uatomic_arch_ppc.h @@ -47,9 +47,6 @@ extern "C" { #define ILLEGAL_INSTR ".long 0xd00d00" -#define uatomic_set(addr, v) STORE_SHARED(*(addr), (v)) -#define uatomic_read(addr) LOAD_SHARED(*(addr)) - /* * Using a isync as second barrier for exchange to provide acquire semantic. * According to uatomic_ops/sysdeps/gcc/powerpc.h, the documentation is "fairly @@ -225,18 +222,10 @@ unsigned long _uatomic_add_return(void *addr, unsigned long val, (unsigned long)(v), \ sizeof(*(addr)))) -/* uatomic_sub_return, uatomic_add, uatomic_sub, uatomic_inc, uatomic_dec */ - -#define uatomic_sub_return(addr, v) uatomic_add_return((addr), -(v)) - -#define uatomic_add(addr, v) (void)uatomic_add_return((addr), (v)) -#define uatomic_sub(addr, v) (void)uatomic_sub_return((addr), (v)) - -#define uatomic_inc(addr) uatomic_add((addr), 1) -#define uatomic_dec(addr) uatomic_add((addr), -1) - #ifdef __cplusplus } #endif +#include + #endif /* _URCU_ARCH_UATOMIC_PPC_H */ diff --git a/urcu/uatomic_arch_s390.h b/urcu/uatomic_arch_s390.h index 614867f..2378ebd 100644 --- a/urcu/uatomic_arch_s390.h +++ b/urcu/uatomic_arch_s390.h @@ -78,9 +78,6 @@ struct __uatomic_dummy { }; #define __hp(x) ((struct __uatomic_dummy *)(x)) -#define uatomic_set(addr, v) STORE_SHARED(*(addr), (v)) -#define uatomic_read(addr) LOAD_SHARED(*(addr)) - /* xchg */ static inline __attribute__((always_inline)) @@ -208,18 +205,10 @@ unsigned long _uatomic_add_return(void *addr, unsigned long val, int len) (unsigned long)(v), \ sizeof(*(addr)))) -/* uatomic_sub_return, uatomic_add, uatomic_sub, uatomic_inc, uatomic_dec */ - -#define uatomic_sub_return(addr, v) uatomic_add_return((addr), -(v)) - -#define uatomic_add(addr, v) (void)uatomic_add_return((addr), (v)) -#define uatomic_sub(addr, v) (void)uatomic_sub_return((addr), (v)) - -#define uatomic_inc(addr) uatomic_add((addr), 1) -#define uatomic_dec(addr) uatomic_add((addr), -1) - #ifdef __cplusplus } #endif +#include + #endif /* _URCU_UATOMIC_ARCH_S390_H */ diff --git a/urcu/uatomic_arch_sparc64.h b/urcu/uatomic_arch_sparc64.h index d443d4f..ec9d8e8 100644 --- a/urcu/uatomic_arch_sparc64.h +++ b/urcu/uatomic_arch_sparc64.h @@ -39,9 +39,6 @@ extern "C" { #define BITS_PER_LONG (__SIZEOF_LONG__ * 8) #endif -#define uatomic_set(addr, v) STORE_SHARED(*(addr), (v)) -#define uatomic_read(addr) LOAD_SHARED(*(addr)) - /* cmpxchg */ static inline __attribute__((always_inline)) @@ -169,18 +166,10 @@ unsigned long _uatomic_add_return(void *addr, unsigned long val, int len) (unsigned long)(v), \ sizeof(*(addr)))) -/* uatomic_sub_return, uatomic_add, uatomic_sub, uatomic_inc, uatomic_dec */ - -#define uatomic_sub_return(addr, v) uatomic_add_return((addr), -(v)) - -#define uatomic_add(addr, v) (void)uatomic_add_return((addr), (v)) -#define uatomic_sub(addr, v) (void)uatomic_sub_return((addr), (v)) - -#define uatomic_inc(addr) uatomic_add((addr), 1) -#define uatomic_dec(addr) uatomic_add((addr), -1) - #ifdef __cplusplus } #endif +#include + #endif /* _URCU_ARCH_UATOMIC_PPC_H */ diff --git a/urcu/uatomic_arch_x86.h b/urcu/uatomic_arch_x86.h index 3bfe86d..e358be3 100644 --- a/urcu/uatomic_arch_x86.h +++ b/urcu/uatomic_arch_x86.h @@ -49,7 +49,6 @@ struct __uatomic_dummy { #define __hp(x) ((struct __uatomic_dummy *)(x)) #define _uatomic_set(addr, v) STORE_SHARED(*(addr), (v)) -#define _uatomic_read(addr) LOAD_SHARED(*(addr)) /* cmpxchg */ @@ -176,7 +175,7 @@ unsigned long __uatomic_exchange(void *addr, unsigned long val, int len) ((__typeof__(*(addr))) __uatomic_exchange((addr), (unsigned long)(v), \ sizeof(*(addr)))) -/* uatomic_add_return, uatomic_sub_return */ +/* uatomic_add_return */ static inline __attribute__((always_inline)) unsigned long __uatomic_add_return(void *addr, unsigned long val, @@ -241,9 +240,7 @@ unsigned long __uatomic_add_return(void *addr, unsigned long val, (unsigned long)(v), \ sizeof(*(addr)))) -#define _uatomic_sub_return(addr, v) _uatomic_add_return((addr), -(v)) - -/* uatomic_add, uatomic_sub */ +/* uatomic_add */ static inline __attribute__((always_inline)) void __uatomic_add(void *addr, unsigned long val, int len) @@ -297,8 +294,6 @@ void __uatomic_add(void *addr, unsigned long val, int len) #define _uatomic_add(addr, v) \ (__uatomic_add((addr), (unsigned long)(v), sizeof(*(addr)))) -#define _uatomic_sub(addr, v) _uatomic_add((addr), -(v)) - /* uatomic_inc */ @@ -449,36 +444,29 @@ extern unsigned long _compat_uatomic_xchg(void *addr, (unsigned long)(v), \ sizeof(*(addr)))) -#define compat_uatomic_sub_return(addr, v) \ - compat_uatomic_add_return((addr), -(v)) #define compat_uatomic_add(addr, v) \ ((void)compat_uatomic_add_return((addr), (v))) -#define compat_uatomic_sub(addr, v) \ - ((void)compat_uatomic_sub_return((addr), (v))) #define compat_uatomic_inc(addr) \ (compat_uatomic_add((addr), 1)) #define compat_uatomic_dec(addr) \ - (compat_uatomic_sub((addr), 1)) + (compat_uatomic_add((addr), -1)) #else #define UATOMIC_COMPAT(insn) (_uatomic_##insn) #endif /* Read is atomic even in compat mode */ -#define uatomic_read(addr) _uatomic_read(addr) - #define uatomic_set(addr, v) \ UATOMIC_COMPAT(set(addr, v)) + #define uatomic_cmpxchg(addr, old, _new) \ UATOMIC_COMPAT(cmpxchg(addr, old, _new)) #define uatomic_xchg(addr, v) \ UATOMIC_COMPAT(xchg(addr, v)) #define uatomic_add_return(addr, v) \ UATOMIC_COMPAT(add_return(addr, v)) -#define uatomic_sub_return(addr, v) \ - UATOMIC_COMPAT(sub_return(addr, v)) + #define uatomic_add(addr, v) UATOMIC_COMPAT(add(addr, v)) -#define uatomic_sub(addr, v) UATOMIC_COMPAT(sub(addr, v)) #define uatomic_inc(addr) UATOMIC_COMPAT(inc(addr)) #define uatomic_dec(addr) UATOMIC_COMPAT(dec(addr)) @@ -486,4 +474,6 @@ extern unsigned long _compat_uatomic_xchg(void *addr, } #endif +#include + #endif /* _URCU_ARCH_UATOMIC_X86_H */ diff --git a/urcu/uatomic_generic.h b/urcu/uatomic_generic.h new file mode 100644 index 0000000..6e2c23e --- /dev/null +++ b/urcu/uatomic_generic.h @@ -0,0 +1,263 @@ +#ifndef _URCU_UATOMIC_GENERIC_H +#define _URCU_UATOMIC_GENERIC_H + +/* + * Copyright (c) 1991-1994 by Xerox Corporation. All rights reserved. + * Copyright (c) 1996-1999 by Silicon Graphics. All rights reserved. + * Copyright (c) 1999-2004 Hewlett-Packard Development Company, L.P. + * Copyright (c) 2009 Mathieu Desnoyers + * Copyright (c) 2010 Paolo Bonzini + * + * THIS MATERIAL IS PROVIDED AS IS, WITH ABSOLUTELY NO WARRANTY EXPRESSED + * OR IMPLIED. ANY USE IS AT YOUR OWN RISK. + * + * Permission is hereby granted to use or copy this program + * for any purpose, provided the above notices are retained on all copies. + * Permission to modify the code and to distribute modified code is granted, + * provided the above notices are retained, and a notice that the code was + * modified is included with the above copyright notice. + * + * Code inspired from libuatomic_ops-1.2, inherited in part from the + * Boehm-Demers-Weiser conservative garbage collector. + */ + +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +#ifndef BITS_PER_LONG +#define BITS_PER_LONG (__SIZEOF_LONG__ * 8) +#endif + +#ifndef uatomic_set +#define uatomic_set(addr, v) STORE_SHARED(*(addr), (v)) +#endif + +#ifndef uatomic_read +#define uatomic_read(addr) LOAD_SHARED(*(addr)) +#endif + +#if !defined __OPTIMIZE__ || defined UATOMIC_NO_LINK_ERROR +static inline __attribute__((always_inline)) +void _uatomic_link_error() +{ +#ifdef ILLEGAL_INSTR + /* generate an illegal instruction. Cannot catch this with linker tricks + * when optimizations are disabled. */ + __asm__ __volatile__(ILLEGAL_INSTR); +#else + __builtin_trap (); +#endif +} + +#else /* #if !defined __OPTIMIZE__ || defined UATOMIC_NO_LINK_ERROR */ +extern void _uatomic_link_error (); +#endif /* #else #if !defined __OPTIMIZE__ || defined UATOMIC_NO_LINK_ERROR */ + +/* cmpxchg */ + +#ifndef uatomic_cmpxchg +static inline __attribute__((always_inline)) +unsigned long _uatomic_cmpxchg(void *addr, unsigned long old, + unsigned long _new, int len) +{ + switch (len) { + case 4: + return __sync_val_compare_and_swap_4(addr, old, _new); +#if (BITS_PER_LONG == 64) + case 8: + return __sync_val_compare_and_swap_8(addr, old, _new); +#endif + } + _uatomic_link_error(); + return 0; +} + + +#define uatomic_cmpxchg(addr, old, _new) \ + ((__typeof__(*(addr))) _uatomic_cmpxchg((addr), (unsigned long)(old),\ + (unsigned long)(_new), \ + sizeof(*(addr)))) + + +/* uatomic_add_return */ + +#ifndef uatomic_add_return +static inline __attribute__((always_inline)) +unsigned long _uatomic_add_return(void *addr, unsigned long val, + int len) +{ + switch (len) { + case 4: + return __sync_add_and_fetch_4(addr, val); +#if (BITS_PER_LONG == 64) + case 8: + return __sync_add_and_fetch_8(addr, val); +#endif + } + _uatomic_link_error(); + return 0; +} + + +#define uatomic_add_return(addr, v) \ + ((__typeof__(*(addr))) _uatomic_add_return((addr), \ + (unsigned long)(v), \ + sizeof(*(addr)))) +#endif /* #ifndef uatomic_add_return */ + +#ifndef uatomic_xchg +/* xchg */ + +static inline __attribute__((always_inline)) +unsigned long _uatomic_exchange(void *addr, unsigned long val, int len) +{ + switch (len) { + case 4: + { + unsigned int old; + + do { + old = uatomic_read((unsigned int *)addr); + while (!__sync_bool_compare_and_swap_4(addr, old, val)); + + } return old; + } +#if (BITS_PER_LONG == 64) + case 8: + { + unsigned long old; + + do { + old = uatomic_read((unsigned long *)addr); + } while (!__sync_bool_compare_and_swap_8(addr, old, val)); + + return old; + } +#endif + } + _uatomic_link_error(); + return 0; +} + +#define uatomic_xchg(addr, v) \ + ((__typeof__(*(addr))) _uatomic_exchange((addr), (unsigned long)(v), \ + sizeof(*(addr)))) +#endif /* #ifndef uatomic_xchg */ + +#else /* #ifndef uatomic_cmpxchg */ + +#ifndef uatomic_add_return +/* uatomic_add_return */ + +static inline __attribute__((always_inline)) +unsigned long _uatomic_add_return(void *addr, unsigned long val, int len) +{ + switch (len) { + case 4: + { + unsigned int old, oldt; + + oldt = uatomic_read((unsigned int *)addr); + do { + old = oldt; + oldt = _uatomic_cmpxchg(addr, old, old + val, 4); + } while (oldt != old); + + return old + val; + } +#if (BITS_PER_LONG == 64) + case 8: + { + unsigned long old, oldt; + + oldt = uatomic_read((unsigned long *)addr); + do { + old = oldt; + oldt = _uatomic_cmpxchg(addr, old, old + val, 8); + } while (oldt != old); + + return old + val; + } +#endif + } + _uatomic_link_error(); + return 0; +} + +#define uatomic_add_return(addr, v) \ + ((__typeof__(*(addr))) _uatomic_add_return((addr), \ + (unsigned long)(v), \ + sizeof(*(addr)))) +#endif /* #ifndef uatomic_add_return */ + +#ifndef uatomic_xchg +/* xchg */ + +static inline __attribute__((always_inline)) +unsigned long _uatomic_exchange(void *addr, unsigned long val, int len) +{ + switch (len) { + case 4: + { + unsigned int old, oldt; + + oldt = uatomic_read((unsigned int *)addr); + do { + old = oldt; + oldt = _uatomic_cmpxchg(addr, old, val, 4); + } while (oldt != old); + + return old; + } +#if (BITS_PER_LONG == 64) + case 8: + { + unsigned long old, oldt; + + oldt = uatomic_read((unsigned long *)addr); + do { + old = oldt; + oldt = _uatomic_cmpxchg(addr, old, val, 8); + } while (oldt != old); + + return old; + } +#endif + } + _uatomic_link_error(); + return 0; +} + +#define uatomic_xchg(addr, v) \ + ((__typeof__(*(addr))) _uatomic_exchange((addr), (unsigned long)(v), \ + sizeof(*(addr)))) +#endif /* #ifndef uatomic_xchg */ + +#endif /* #else #ifndef uatomic_cmpxchg */ + +/* uatomic_sub_return, uatomic_add, uatomic_sub, uatomic_inc, uatomic_dec */ + +#ifndef uatomic_add +#define uatomic_add(addr, v) (void)uatomic_add_return((addr), (v)) +#endif + +#define uatomic_sub_return(addr, v) uatomic_add_return((addr), -(v)) +#define uatomic_sub(addr, v) uatomic_add((addr), -(v)) + +#ifndef uatomic_inc +#define uatomic_inc(addr) uatomic_add((addr), 1) +#endif + +#ifndef uatomic_dec +#define uatomic_dec(addr) uatomic_add((addr), -1) +#endif + +#ifdef __cplusplus +} +#endif + +#endif /* _URCU_UATOMIC_GENERIC_H */ -- 2.34.1