libexec/rtld-elf/i386/lockdflt.c

   1 /*-
   2  * Copyright 1999, 2000 John D. Polstra.
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
  15  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  16  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  17  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
  18  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  19  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  20  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  21  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  23  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  24  *
  25  * $FreeBSD: src/libexec/rtld-elf/i386/lockdflt.c,v 1.5.2.4 2002/07/11 23:52:32 jdp Exp $
  26  * $DragonFly: src/libexec/rtld-elf/i386/lockdflt.c,v 1.3 2007/11/09 19:38:50 hasso Exp $
  27  */
  28
  29 /*
  30  * Thread locking implementation for the dynamic linker.
  31  *
  32  * We use the "simple, non-scalable reader-preference lock" from:
  33  *
  34  *   J. M. Mellor-Crummey and M. L. Scott. "Scalable Reader-Writer
  35  *   Synchronization for Shared-Memory Multiprocessors." 3rd ACM Symp. on
  36  *   Principles and Practice of Parallel Programming, April 1991.
  37  *
  38  * In this algorithm the lock is a single word.  Its low-order bit is
  39  * set when a writer holds the lock.  The remaining high-order bits
  40  * contain a count of readers desiring the lock.  The algorithm requires
  41  * atomic "compare_and_store" and "add" operations.
  42  *
  43  */
  44
  45 #include <setjmp.h>
  46 #include <signal.h>
  47 #include <stdlib.h>
  48 #include <time.h>
  49
  50 #include "debug.h"
  51 #include "rtld.h"
  52
  53 #define CACHE_LINE_SIZE         32
  54
  55 #define WAFLAG          0x1     /* A writer holds the lock */
  56 #define RC_INCR         0x2     /* Adjusts count of readers desiring lock */
  57
  58 typedef struct Struct_Lock {
  59         volatile int lock;
  60         void *base;
  61 } Lock;
  62
  63 static sigset_t fullsigmask, oldsigmask;
  64
  65 static inline int
  66 cmpxchgl(int old, int new, volatile int *m)
  67 {
  68         int result;
  69
  70         __asm __volatile ("lock; cmpxchgl %2, %0"
  71             : "+m"(*m), "=a"(result)
  72             : "r"(new), "1"(old)
  73             : "cc");
  74
  75         return result;
  76 }
  77
  78 static void *
  79 lock_create(void *context)
  80 {
  81     void *base;
  82     char *p;
  83     uintptr_t r;
  84     Lock *l;
  85
  86     /*
  87      * Arrange for the lock to occupy its own cache line.  First, we
  88      * optimistically allocate just a cache line, hoping that malloc
  89      * will give us a well-aligned block of memory.  If that doesn't
  90      * work, we allocate a larger block and take a well-aligned cache
  91      * line from it.
  92      */
  93     base = xmalloc(CACHE_LINE_SIZE);
  94     p = (char *)base;
  95     if ((uintptr_t)p % CACHE_LINE_SIZE != 0) {
  96         free(base);
  97         base = xmalloc(2 * CACHE_LINE_SIZE);
  98         p = (char *)base;
  99         if ((r = (uintptr_t)p % CACHE_LINE_SIZE) != 0)
 100             p += CACHE_LINE_SIZE - r;
 101     }
 102     l = (Lock *)p;
 103     l->base = base;
 104     l->lock = 0;
 105     return l;
 106 }
 107
 108 static void
 109 lock_destroy(void *lock)
 110 {
 111     Lock *l = (Lock *)lock;
 112
 113     free(l->base);
 114 }
 115
 116 /*
 117  * Reader/writer locks.
 118  */
 119 static void
 120 rlock_acquire(void *lock)
 121 {
 122     Lock *l = (Lock *)lock;
 123
 124     atomic_add_int(&l->lock, RC_INCR);
 125     while (l->lock & WAFLAG)
 126             ;   /* Spin */
 127 }
 128
 129 static void
 130 wlock_acquire(void *lock)
 131 {
 132     Lock *l = (Lock *)lock;
 133     sigset_t tmp_oldsigmask;
 134
 135     for ( ; ; ) {
 136         sigprocmask(SIG_BLOCK, &fullsigmask, &tmp_oldsigmask);
 137         if (cmpxchgl(0, WAFLAG, &l->lock) == 0)
 138             break;
 139         sigprocmask(SIG_SETMASK, &tmp_oldsigmask, NULL);
 140     }
 141     oldsigmask = tmp_oldsigmask;
 142 }
 143
 144 static void
 145 rlock_release(void *lock)
 146 {
 147     Lock *l = (Lock *)lock;
 148
 149     atomic_add_int(&l->lock, -RC_INCR);
 150 }
 151
 152 static void
 153 wlock_release(void *lock)
 154 {
 155     Lock *l = (Lock *)lock;
 156
 157     atomic_add_int(&l->lock, -WAFLAG);
 158     sigprocmask(SIG_SETMASK, &oldsigmask, NULL);
 159 }
 160
 161 void
 162 lockdflt_init(LockInfo *li)
 163 {
 164     li->context = NULL;
 165     li->context_destroy = NULL;
 166     li->lock_create = lock_create;
 167     li->lock_destroy = lock_destroy;
 168     li->rlock_acquire = rlock_acquire;
 169     li->wlock_acquire = wlock_acquire;
 170     li->rlock_release = rlock_release;
 171     li->wlock_release = wlock_release;
 172     /*
 173      * Construct a mask to block all signals except traps which might
 174      * conceivably be generated within the dynamic linker itself.
 175      */
 176     sigfillset(&fullsigmask);
 177     sigdelset(&fullsigmask, SIGILL);
 178     sigdelset(&fullsigmask, SIGTRAP);
 179     sigdelset(&fullsigmask, SIGABRT);
 180     sigdelset(&fullsigmask, SIGEMT);
 181     sigdelset(&fullsigmask, SIGFPE);
 182     sigdelset(&fullsigmask, SIGBUS);
 183     sigdelset(&fullsigmask, SIGSEGV);
 184     sigdelset(&fullsigmask, SIGSYS);
 185 }