+/*
+ * Copyright (c) 1994 by Xerox Corporation. All rights reserved.
+ * Copyright (c) 1996 by Silicon Graphics. All rights reserved.
+ * Copyright (c) 1998 by Fergus Henderson. All rights reserved.
+ * Copyright (c) 2000-2008 by Hewlett-Packard Development Company.
+ * All rights reserved.
+ *
+ * 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.
+ */
+
+#include "private/gc_priv.h"
+
+#if defined(GC_WIN32_THREADS)
+
+#ifndef WIN32_LEAN_AND_MEAN
+# define WIN32_LEAN_AND_MEAN 1
+#endif
+#define NOSERVICE
+#include <windows.h>
+
+#ifdef THREAD_LOCAL_ALLOC
+# include "private/thread_local_alloc.h"
+#endif /* THREAD_LOCAL_ALLOC */
+
+/* Allocation lock declarations. */
+#if !defined(USE_PTHREAD_LOCKS)
+ GC_INNER CRITICAL_SECTION GC_allocate_ml;
+ GC_INNER DWORD GC_lock_holder = NO_THREAD;
+ /* Thread id for current holder of allocation lock */
+#else
+ GC_INNER pthread_mutex_t GC_allocate_ml = PTHREAD_MUTEX_INITIALIZER;
+ GC_INNER unsigned long GC_lock_holder = NO_THREAD;
+#endif
+
+#ifdef GC_PTHREADS
+# include <errno.h> /* for EAGAIN */
+
+ /* Cygwin-specific forward decls */
+# undef pthread_create
+# undef pthread_join
+# undef pthread_detach
+
+# ifndef GC_NO_PTHREAD_SIGMASK
+# undef pthread_sigmask
+# endif
+
+ STATIC void * GC_pthread_start(void * arg);
+ STATIC void GC_thread_exit_proc(void *arg);
+
+# include <pthread.h>
+# ifdef CAN_HANDLE_FORK
+# include <unistd.h>
+# endif
+
+#else
+
+# undef CreateThread
+# undef ExitThread
+# undef _beginthreadex
+# undef _endthreadex
+
+# ifdef MSWINCE
+ /* Force DONT_USE_SIGNALANDWAIT implementation of PARALLEL_MARK */
+ /* for WinCE (since Win32 SignalObjectAndWait() is missing). */
+# ifndef DONT_USE_SIGNALANDWAIT
+# define DONT_USE_SIGNALANDWAIT
+# endif
+# else
+# include <process.h> /* For _beginthreadex, _endthreadex */
+# include <errno.h> /* for errno, EAGAIN */
+# endif
+
+#endif
+
+/* DllMain-based thread registration is currently incompatible */
+/* with thread-local allocation, pthreads and WinCE. */
+#if defined(GC_DLL) && !defined(GC_NO_THREADS_DISCOVERY) && !defined(MSWINCE) \
+ && !defined(THREAD_LOCAL_ALLOC) && !defined(GC_PTHREADS)
+# include "atomic_ops.h"
+
+ /* This code operates in two distinct modes, depending on */
+ /* the setting of GC_win32_dll_threads. */
+ /* If GC_win32_dll_threads is set, all threads in the process */
+ /* are implicitly registered with the GC by DllMain. */
+ /* No explicit registration is required, and attempts at */
+ /* explicit registration are ignored. This mode is */
+ /* very different from the Posix operation of the collector. */
+ /* In this mode access to the thread table is lock-free. */
+ /* Hence there is a static limit on the number of threads. */
+
+# ifdef GC_DISCOVER_TASK_THREADS
+ /* GC_DISCOVER_TASK_THREADS should be used if DllMain-based */
+ /* thread registration is required but it is impossible to */
+ /* call GC_use_threads_discovery before other GC routines. */
+# define GC_win32_dll_threads TRUE
+# else
+ STATIC GC_bool GC_win32_dll_threads = FALSE;
+ /* GC_win32_dll_threads must be set (if needed) at the */
+ /* application initialization time, i.e. before any */
+ /* collector or thread calls. We make it a "dynamic" */
+ /* option only to avoid multiple library versions. */
+# endif
+
+#else
+ /* If GC_win32_dll_threads is FALSE (or the collector is */
+ /* built without GC_DLL defined), things operate in a way */
+ /* that is very similar to Posix platforms, and new threads */
+ /* must be registered with the collector, e.g. by using */
+ /* preprocessor-based interception of the thread primitives. */
+ /* In this case, we use a real data structure for the thread */
+ /* table. Note that there is no equivalent of linker-based */
+ /* call interception, since we don't have ELF-like */
+ /* facilities. The Windows analog appears to be "API */
+ /* hooking", which really seems to be a standard way to */
+ /* do minor binary rewriting (?). I'd prefer not to have */
+ /* the basic collector rely on such facilities, but an */
+ /* optional package that intercepts thread calls this way */
+ /* would probably be nice. */
+# ifndef GC_NO_THREADS_DISCOVERY
+# define GC_NO_THREADS_DISCOVERY
+# endif
+# define GC_win32_dll_threads FALSE
+# undef MAX_THREADS
+# define MAX_THREADS 1 /* dll_thread_table[] is always empty. */
+#endif /* GC_NO_THREADS_DISCOVERY */
+
+/* We have two versions of the thread table. Which one */
+/* we us depends on whether or not GC_win32_dll_threads */
+/* is set. Note that before initialization, we don't */
+/* add any entries to either table, even if DllMain is */
+/* called. The main thread will be added on */
+/* initialization. */
+
+/* The type of the first argument to InterlockedExchange. */
+/* Documented to be LONG volatile *, but at least gcc likes */
+/* this better. */
+typedef LONG * IE_t;
+
+STATIC GC_bool GC_thr_initialized = FALSE;
+
+GC_INNER GC_bool GC_need_to_lock = FALSE;
+
+static GC_bool parallel_initialized = FALSE;
+
+/* GC_use_threads_discovery() is currently incompatible with pthreads */
+/* and WinCE. It might be possible to get DllMain-based thread */
+/* registration to work with Cygwin, but if you try it then you are on */
+/* your own. */
+GC_API void GC_CALL GC_use_threads_discovery(void)
+{
+# ifdef GC_NO_THREADS_DISCOVERY
+ ABORT("GC DllMain-based thread registration unsupported");
+# else
+ /* Turn on GC_win32_dll_threads. */
+ GC_ASSERT(!parallel_initialized);
+# ifndef GC_DISCOVER_TASK_THREADS
+ GC_win32_dll_threads = TRUE;
+# endif
+ GC_init_parallel();
+# endif
+}
+
+STATIC DWORD GC_main_thread = 0;
+
+#define ADDR_LIMIT ((ptr_t)(word)-1)
+
+struct GC_Thread_Rep {
+ union {
+# ifndef GC_NO_THREADS_DISCOVERY
+ AO_t in_use; /* Updated without lock. */
+ /* We assert that unused */
+ /* entries have invalid ids of */
+ /* zero and zero stack fields. */
+ /* Used only with GC_win32_dll_threads. */
+# endif
+ struct GC_Thread_Rep * next;
+ /* Hash table link without */
+ /* GC_win32_dll_threads. */
+ /* More recently allocated threads */
+ /* with a given pthread id come */
+ /* first. (All but the first are */
+ /* guaranteed to be dead, but we may */
+ /* not yet have registered the join.) */
+ } tm; /* table_management */
+ DWORD id;
+
+# ifdef MSWINCE
+ /* According to MSDN specs for WinCE targets: */
+ /* - DuplicateHandle() is not applicable to thread handles; and */
+ /* - the value returned by GetCurrentThreadId() could be used as */
+ /* a "real" thread handle (for SuspendThread(), ResumeThread() and */
+ /* GetThreadContext()). */
+# define THREAD_HANDLE(t) (HANDLE)(word)(t)->id
+# else
+ HANDLE handle;
+# define THREAD_HANDLE(t) (t)->handle
+# endif
+
+ ptr_t stack_base; /* The cold end of the stack. */
+ /* 0 ==> entry not valid. */
+ /* !in_use ==> stack_base == 0 */
+ ptr_t last_stack_min; /* Last known minimum (hottest) address */
+ /* in stack or ADDR_LIMIT if unset */
+# ifdef IA64
+ ptr_t backing_store_end;
+ ptr_t backing_store_ptr;
+# endif
+
+ ptr_t thread_blocked_sp; /* Protected by GC lock. */
+ /* NULL value means thread unblocked. */
+ /* If set to non-NULL, thread will */
+ /* acquire GC lock before doing any */
+ /* pointer manipulations. Thus it does */
+ /* not need to stop this thread. */
+
+ struct GC_traced_stack_sect_s *traced_stack_sect;
+ /* Points to the "stack section" data */
+ /* held in stack by the innermost */
+ /* GC_call_with_gc_active() of this */
+ /* thread. May be NULL. */
+
+ unsigned short finalizer_skipped;
+ unsigned char finalizer_nested;
+ /* Used by GC_check_finalizer_nested() */
+ /* to minimize the level of recursion */
+ /* when a client finalizer allocates */
+ /* memory (initially both are 0). */
+
+ unsigned char suspended; /* really of GC_bool type */
+
+# ifdef GC_PTHREADS
+ unsigned char flags; /* Protected by GC lock. */
+# define FINISHED 1 /* Thread has exited. */
+# define DETACHED 2 /* Thread is intended to be detached. */
+# define KNOWN_FINISHED(t) (((t) -> flags) & FINISHED)
+ pthread_t pthread_id;
+ void *status; /* hold exit value until join in case it's a pointer */
+# else
+# define KNOWN_FINISHED(t) 0
+# endif
+
+# ifdef THREAD_LOCAL_ALLOC
+ struct thread_local_freelists tlfs;
+# endif
+};
+
+typedef struct GC_Thread_Rep * GC_thread;
+typedef volatile struct GC_Thread_Rep * GC_vthread;
+
+#ifndef GC_NO_THREADS_DISCOVERY
+ /* We assumed that volatile ==> memory ordering, at least among */
+ /* volatiles. This code should consistently use atomic_ops. */
+ STATIC volatile GC_bool GC_please_stop = FALSE;
+#elif defined(GC_ASSERTIONS)
+ STATIC GC_bool GC_please_stop = FALSE;
+#endif
+
+/*
+ * We track thread attachments while the world is supposed to be stopped.
+ * Unfortunately, we can't stop them from starting, since blocking in
+ * DllMain seems to cause the world to deadlock. Thus we have to recover
+ * If we notice this in the middle of marking.
+ */
+
+#ifndef GC_NO_THREADS_DISCOVERY
+ STATIC AO_t GC_attached_thread = FALSE;
+#endif
+
+#if !defined(__GNUC__)
+ /* Return TRUE if an thread was attached since we last asked or */
+ /* since GC_attached_thread was explicitly reset. */
+ GC_bool GC_started_thread_while_stopped(void)
+ {
+# ifndef GC_NO_THREADS_DISCOVERY
+ AO_t result;
+
+ if (GC_win32_dll_threads) {
+ AO_nop_full(); /* Prior heap reads need to complete earlier. */
+ result = AO_load(&GC_attached_thread);
+ if (result) {
+ AO_store(&GC_attached_thread, FALSE);
+ return TRUE;
+ }
+ }
+# endif
+ return FALSE;
+ }
+#endif /* !__GNUC__ */
+
+/* Thread table used if GC_win32_dll_threads is set. */
+/* This is a fixed size array. */
+/* Since we use runtime conditionals, both versions */
+/* are always defined. */
+# ifndef MAX_THREADS
+# define MAX_THREADS 512
+# endif
+
+/* Things may get quite slow for large numbers of threads, */
+/* since we look them up with sequential search. */
+volatile struct GC_Thread_Rep dll_thread_table[MAX_THREADS];
+
+STATIC volatile LONG GC_max_thread_index = 0;
+ /* Largest index in dll_thread_table */
+ /* that was ever used. */
+
+/* And now the version used if GC_win32_dll_threads is not set. */
+/* This is a chained hash table, with much of the code borrowed */
+/* From the Posix implementation. */
+#ifndef THREAD_TABLE_SZ
+# define THREAD_TABLE_SZ 256 /* Power of 2 (for speed). */
+#endif
+#define THREAD_TABLE_INDEX(id) (((word)(id) >> 2) % THREAD_TABLE_SZ)
+STATIC GC_thread GC_threads[THREAD_TABLE_SZ];
+
+/* It may not be safe to allocate when we register the first thread. */
+/* Thus we allocated one statically. */
+static struct GC_Thread_Rep first_thread;
+static GC_bool first_thread_used = FALSE;
+
+/* Add a thread to GC_threads. We assume it wasn't already there. */
+/* Caller holds allocation lock. */
+/* Unlike the pthreads version, the id field is set by the caller. */
+STATIC GC_thread GC_new_thread(DWORD id)
+{
+ word hv = THREAD_TABLE_INDEX(id);
+ GC_thread result;
+
+ GC_ASSERT(I_HOLD_LOCK());
+ if (!first_thread_used) {
+ result = &first_thread;
+ first_thread_used = TRUE;
+ } else {
+ GC_ASSERT(!GC_win32_dll_threads);
+ result = (struct GC_Thread_Rep *)
+ GC_INTERNAL_MALLOC(sizeof(struct GC_Thread_Rep), NORMAL);
+ /* result can be NULL */
+ if (result == 0) return(0);
+ }
+ /* result -> id = id; Done by caller. */
+ result -> tm.next = GC_threads[hv];
+ GC_threads[hv] = result;
+# ifdef GC_PTHREADS
+ GC_ASSERT(result -> flags == 0);
+# endif
+ GC_ASSERT(result -> thread_blocked_sp == NULL);
+ return(result);
+}
+
+STATIC GC_bool GC_in_thread_creation = FALSE;
+ /* Protected by allocation lock. */
+
+GC_INLINE void GC_record_stack_base(GC_vthread me,
+ const struct GC_stack_base *sb)
+{
+ me -> stack_base = sb -> mem_base;
+# ifdef IA64
+ me -> backing_store_end = sb -> reg_base;
+# endif
+ if (me -> stack_base == NULL)
+ ABORT("Bad stack base in GC_register_my_thread");
+}
+
+/* This may be called from DllMain, and hence operates under unusual */
+/* constraints. In particular, it must be lock-free if */
+/* GC_win32_dll_threads is set. Always called from the thread being */
+/* added. If GC_win32_dll_threads is not set, we already hold the */
+/* allocation lock except possibly during single-threaded startup code. */
+STATIC GC_thread GC_register_my_thread_inner(const struct GC_stack_base *sb,
+ DWORD thread_id)
+{
+ GC_vthread me;
+
+ /* The following should be a no-op according to the win32 */
+ /* documentation. There is empirical evidence that it */
+ /* isn't. - HB */
+# if defined(MPROTECT_VDB)
+ if (GC_incremental
+# ifdef GWW_VDB
+ && !GC_gww_dirty_init()
+# endif
+ )
+ GC_set_write_fault_handler();
+# endif
+
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ int i;
+ /* It appears to be unsafe to acquire a lock here, since this */
+ /* code is apparently not preemptible on some systems. */
+ /* (This is based on complaints, not on Microsoft's official */
+ /* documentation, which says this should perform "only simple */
+ /* initialization tasks".) */
+ /* Hence we make do with nonblocking synchronization. */
+ /* It has been claimed that DllMain is really only executed with */
+ /* a particular system lock held, and thus careful use of locking */
+ /* around code that doesn't call back into the system libraries */
+ /* might be OK. But this hasn't been tested across all win32 */
+ /* variants. */
+ /* cast away volatile qualifier */
+ for (i = 0;
+ InterlockedExchange((void*)&dll_thread_table[i].tm.in_use, 1) != 0;
+ i++) {
+ /* Compare-and-swap would make this cleaner, but that's not */
+ /* supported before Windows 98 and NT 4.0. In Windows 2000, */
+ /* InterlockedExchange is supposed to be replaced by */
+ /* InterlockedExchangePointer, but that's not really what I */
+ /* want here. */
+ /* FIXME: We should eventually declare Win95 dead and use AO_ */
+ /* primitives here. */
+ if (i == MAX_THREADS - 1)
+ ABORT("Too many threads");
+ }
+ /* Update GC_max_thread_index if necessary. The following is */
+ /* safe, and unlike CompareExchange-based solutions seems to work */
+ /* on all Windows95 and later platforms. */
+ /* Unfortunately, GC_max_thread_index may be temporarily out of */
+ /* bounds, so readers have to compensate. */
+ while (i > GC_max_thread_index) {
+ InterlockedIncrement((IE_t)&GC_max_thread_index);
+ }
+ if (GC_max_thread_index >= MAX_THREADS) {
+ /* We overshot due to simultaneous increments. */
+ /* Setting it to MAX_THREADS-1 is always safe. */
+ GC_max_thread_index = MAX_THREADS - 1;
+ }
+ me = dll_thread_table + i;
+ } else
+# endif
+ /* else */ /* Not using DllMain */ {
+ GC_ASSERT(I_HOLD_LOCK());
+ GC_in_thread_creation = TRUE; /* OK to collect from unknown thread. */
+ me = GC_new_thread(thread_id);
+ GC_in_thread_creation = FALSE;
+ if (me == 0)
+ ABORT("Failed to allocate memory for thread registering");
+ }
+# ifdef GC_PTHREADS
+ /* me can be NULL -> segfault */
+ me -> pthread_id = pthread_self();
+# endif
+# ifndef MSWINCE
+ /* GetCurrentThread() returns a pseudohandle (a const value). */
+ if (!DuplicateHandle(GetCurrentProcess(), GetCurrentThread(),
+ GetCurrentProcess(),
+ (HANDLE*)&(me -> handle),
+ 0 /* dwDesiredAccess */, FALSE /* bInheritHandle */,
+ DUPLICATE_SAME_ACCESS)) {
+ if (GC_print_stats)
+ GC_log_printf("DuplicateHandle failed with error code: %d\n",
+ (int)GetLastError());
+ ABORT("DuplicateHandle failed");
+ }
+# endif
+ me -> last_stack_min = ADDR_LIMIT;
+ GC_record_stack_base(me, sb);
+ /* Up until this point, GC_push_all_stacks considers this thread */
+ /* invalid. */
+ /* Up until this point, this entry is viewed as reserved but invalid */
+ /* by GC_delete_thread. */
+ me -> id = thread_id;
+# if defined(THREAD_LOCAL_ALLOC)
+ GC_init_thread_local((GC_tlfs)(&(me->tlfs)));
+# endif
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ if (GC_please_stop) {
+ AO_store(&GC_attached_thread, TRUE);
+ AO_nop_full(); /* Later updates must become visible after this. */
+ }
+ /* We'd like to wait here, but can't, since waiting in DllMain */
+ /* provokes deadlocks. */
+ /* Thus we force marking to be restarted instead. */
+ } else
+# endif
+ /* else */ {
+ GC_ASSERT(!GC_please_stop);
+ /* Otherwise both we and the thread stopping code would be */
+ /* holding the allocation lock. */
+ }
+ return (GC_thread)(me);
+}
+
+/*
+ * GC_max_thread_index may temporarily be larger than MAX_THREADS.
+ * To avoid subscript errors, we check on access.
+ */
+GC_INLINE LONG GC_get_max_thread_index(void)
+{
+ LONG my_max = GC_max_thread_index;
+ if (my_max >= MAX_THREADS) return MAX_THREADS - 1;
+ return my_max;
+}
+
+/* Return the GC_thread corresponding to a thread id. May be called */
+/* without a lock, but should be called in contexts in which the */
+/* requested thread cannot be asynchronously deleted, e.g. from the */
+/* thread itself. */
+/* This version assumes that either GC_win32_dll_threads is set, or */
+/* we hold the allocator lock. */
+/* Also used (for assertion checking only) from thread_local_alloc.c. */
+STATIC GC_thread GC_lookup_thread_inner(DWORD thread_id)
+{
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ int i;
+ LONG my_max = GC_get_max_thread_index();
+ for (i = 0; i <= my_max &&
+ (!AO_load_acquire(&dll_thread_table[i].tm.in_use)
+ || dll_thread_table[i].id != thread_id);
+ /* Must still be in_use, since nobody else can store our */
+ /* thread_id. */
+ i++) {
+ /* empty */
+ }
+ return i <= my_max ? (GC_thread)(dll_thread_table + i) : NULL;
+ } else
+# endif
+ /* else */ {
+ word hv = THREAD_TABLE_INDEX(thread_id);
+ register GC_thread p = GC_threads[hv];
+
+ GC_ASSERT(I_HOLD_LOCK());
+ while (p != 0 && p -> id != thread_id) p = p -> tm.next;
+ return(p);
+ }
+}
+
+#ifdef LINT2
+# define CHECK_LOOKUP_MY_THREAD(me) \
+ if (!(me)) ABORT("GC_lookup_thread_inner(GetCurrentThreadId) failed")
+#else
+# define CHECK_LOOKUP_MY_THREAD(me) /* empty */
+#endif
+
+/* Called by GC_finalize() (in case of an allocation failure observed). */
+/* GC_reset_finalizer_nested() is the same as in pthread_support.c. */
+GC_INNER void GC_reset_finalizer_nested(void)
+{
+ GC_thread me = GC_lookup_thread_inner(GetCurrentThreadId());
+ CHECK_LOOKUP_MY_THREAD(me);
+ me->finalizer_nested = 0;
+}
+
+/* Checks and updates the thread-local level of finalizers recursion. */
+/* Returns NULL if GC_invoke_finalizers() should not be called by the */
+/* collector (to minimize the risk of a deep finalizers recursion), */
+/* otherwise returns a pointer to the thread-local finalizer_nested. */
+/* Called by GC_notify_or_invoke_finalizers() only (the lock is held). */
+/* GC_check_finalizer_nested() is the same as in pthread_support.c. */
+GC_INNER unsigned char *GC_check_finalizer_nested(void)
+{
+ GC_thread me = GC_lookup_thread_inner(GetCurrentThreadId());
+ unsigned nesting_level;
+ CHECK_LOOKUP_MY_THREAD(me);
+ nesting_level = me->finalizer_nested;
+ if (nesting_level) {
+ /* We are inside another GC_invoke_finalizers(). */
+ /* Skip some implicitly-called GC_invoke_finalizers() */
+ /* depending on the nesting (recursion) level. */
+ if (++me->finalizer_skipped < (1U << nesting_level)) return NULL;
+ me->finalizer_skipped = 0;
+ }
+ me->finalizer_nested = (unsigned char)(nesting_level + 1);
+ return &me->finalizer_nested;
+}
+
+#if defined(GC_ASSERTIONS) && defined(THREAD_LOCAL_ALLOC)
+ /* This is called from thread-local GC_malloc(). */
+ GC_bool GC_is_thread_tsd_valid(void *tsd)
+ {
+ GC_thread me;
+ DCL_LOCK_STATE;
+
+ LOCK();
+ me = GC_lookup_thread_inner(GetCurrentThreadId());
+ UNLOCK();
+ return (char *)tsd >= (char *)&me->tlfs
+ && (char *)tsd < (char *)&me->tlfs + sizeof(me->tlfs);
+ }
+#endif /* GC_ASSERTIONS && THREAD_LOCAL_ALLOC */
+
+/* Make sure thread descriptor t is not protected by the VDB */
+/* implementation. */
+/* Used to prevent write faults when the world is (partially) stopped, */
+/* since it may have been stopped with a system lock held, and that */
+/* lock may be required for fault handling. */
+#if defined(MPROTECT_VDB)
+# define UNPROTECT_THREAD(t) \
+ if (!GC_win32_dll_threads && GC_dirty_maintained \
+ && t != &first_thread) { \
+ GC_ASSERT(SMALL_OBJ(GC_size(t))); \
+ GC_remove_protection(HBLKPTR(t), 1, FALSE); \
+ }
+#else
+# define UNPROTECT_THREAD(t)
+#endif
+
+#ifdef CYGWIN32
+# define GC_PTHREAD_PTRVAL(pthread_id) pthread_id
+#elif defined(GC_WIN32_PTHREADS)
+# define GC_PTHREAD_PTRVAL(pthread_id) pthread_id.p
+#endif
+
+/* If a thread has been joined, but we have not yet */
+/* been notified, then there may be more than one thread */
+/* in the table with the same win32 id. */
+/* This is OK, but we need a way to delete a specific one. */
+/* Assumes we hold the allocation lock unless */
+/* GC_win32_dll_threads is set. */
+/* If GC_win32_dll_threads is set it should be called from the */
+/* thread being deleted. */
+STATIC void GC_delete_gc_thread(GC_vthread t)
+{
+# ifndef MSWINCE
+ CloseHandle(t->handle);
+# endif
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ /* This is intended to be lock-free. */
+ /* It is either called synchronously from the thread being */
+ /* deleted, or by the joining thread. */
+ /* In this branch asynchronous changes to (*t) are possible. */
+ /* It's not allowed to call GC_printf (and the friends) here, */
+ /* see GC_stop_world() for the information. */
+ t -> stack_base = 0;
+ t -> id = 0;
+# ifdef GC_PTHREADS
+ GC_PTHREAD_PTRVAL(t->pthread_id) = 0;
+# endif
+ AO_store_release(&t->tm.in_use, FALSE);
+ } else
+# endif
+ /* else */ {
+ DWORD id = ((GC_thread)t) -> id;
+ /* Cast away volatile qualifier, since we have lock. */
+ word hv = THREAD_TABLE_INDEX(id);
+ register GC_thread p = GC_threads[hv];
+ register GC_thread prev = 0;
+
+ GC_ASSERT(I_HOLD_LOCK());
+ while (p != (GC_thread)t) {
+ prev = p;
+ p = p -> tm.next;
+ }
+ if (prev == 0) {
+ GC_threads[hv] = p -> tm.next;
+ } else {
+ prev -> tm.next = p -> tm.next;
+ }
+ GC_INTERNAL_FREE(p);
+ }
+}
+
+/* Delete a thread from GC_threads. We assume it is there. */
+/* (The code intentionally traps if it wasn't.) Assumes we */
+/* hold the allocation lock unless GC_win32_dll_threads is set. */
+/* If GC_win32_dll_threads is set then it should be called from */
+/* the thread being deleted. It is also safe to delete the */
+/* main thread (unless GC_win32_dll_threads). */
+STATIC void GC_delete_thread(DWORD id)
+{
+ if (GC_win32_dll_threads) {
+ GC_thread t = GC_lookup_thread_inner(id);
+
+ if (0 == t) {
+ WARN("Removing nonexistent thread, id = %" GC_PRIdPTR "\n", id);
+ } else {
+ GC_delete_gc_thread(t);
+ }
+ } else {
+ word hv = THREAD_TABLE_INDEX(id);
+ register GC_thread p = GC_threads[hv];
+ register GC_thread prev = 0;
+
+ GC_ASSERT(I_HOLD_LOCK());
+ while (p -> id != id) {
+ prev = p;
+ p = p -> tm.next;
+ }
+# ifndef MSWINCE
+ CloseHandle(p->handle);
+# endif
+ if (prev == 0) {
+ GC_threads[hv] = p -> tm.next;
+ } else {
+ prev -> tm.next = p -> tm.next;
+ }
+ if (p != &first_thread) {
+ GC_INTERNAL_FREE(p);
+ }
+ }
+}
+
+GC_API void GC_CALL GC_allow_register_threads(void)
+{
+ /* Check GC is initialized and the current thread is registered. */
+ GC_ASSERT(GC_lookup_thread_inner(GetCurrentThreadId()) != 0);
+
+# if !defined(GC_NO_THREADS_DISCOVERY) && !defined(PARALLEL_MARK)
+ /* GC_init() doesn't call GC_init_parallel() in this case. */
+ parallel_initialized = TRUE;
+# endif
+ GC_need_to_lock = TRUE; /* We are multi-threaded now. */
+}
+
+GC_API int GC_CALL GC_register_my_thread(const struct GC_stack_base *sb)
+{
+ GC_thread me;
+ DWORD thread_id = GetCurrentThreadId();
+ DCL_LOCK_STATE;
+
+ if (GC_need_to_lock == FALSE)
+ ABORT("Threads explicit registering is not previously enabled");
+
+ /* We lock here, since we want to wait for an ongoing GC. */
+ LOCK();
+ me = GC_lookup_thread_inner(thread_id);
+ if (me == 0) {
+# ifdef GC_PTHREADS
+ me = GC_register_my_thread_inner(sb, thread_id);
+ me -> flags |= DETACHED;
+ /* Treat as detached, since we do not need to worry about */
+ /* pointer results. */
+# else
+ GC_register_my_thread_inner(sb, thread_id);
+# endif
+ UNLOCK();
+ return GC_SUCCESS;
+ } else
+# ifdef GC_PTHREADS
+ /* else */ if ((me -> flags & FINISHED) != 0) {
+ GC_record_stack_base(me, sb);
+ me -> flags &= ~FINISHED; /* but not DETACHED */
+# ifdef THREAD_LOCAL_ALLOC
+ GC_init_thread_local((GC_tlfs)(&me->tlfs));
+# endif
+ UNLOCK();
+ return GC_SUCCESS;
+ } else
+# endif
+ /* else */ {
+ UNLOCK();
+ return GC_DUPLICATE;
+ }
+}
+
+/* Similar to that in pthread_support.c. */
+STATIC void GC_wait_for_gc_completion(GC_bool wait_for_all)
+{
+ GC_ASSERT(I_HOLD_LOCK());
+ if (GC_incremental && GC_collection_in_progress()) {
+ word old_gc_no = GC_gc_no;
+
+ /* Make sure that no part of our stack is still on the mark stack, */
+ /* since it's about to be unmapped. */
+ do {
+ ENTER_GC();
+ GC_in_thread_creation = TRUE;
+ GC_collect_a_little_inner(1);
+ GC_in_thread_creation = FALSE;
+ EXIT_GC();
+
+ UNLOCK();
+ Sleep(0); /* yield */
+ LOCK();
+ } while (GC_incremental && GC_collection_in_progress()
+ && (wait_for_all || old_gc_no == GC_gc_no));
+ }
+}
+
+GC_API int GC_CALL GC_unregister_my_thread(void)
+{
+ DCL_LOCK_STATE;
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("Unregistering thread 0x%lx\n", (long)GetCurrentThreadId());
+# endif
+
+ if (GC_win32_dll_threads) {
+# if defined(THREAD_LOCAL_ALLOC)
+ /* Can't happen: see GC_use_threads_discovery(). */
+ GC_ASSERT(FALSE);
+# else
+# ifdef GC_PTHREADS
+ /* FIXME: If not DETACHED then just set FINISHED. */
+# endif
+ /* FIXME: Should we just ignore this? */
+ GC_delete_thread(GetCurrentThreadId());
+# endif
+ } else {
+# if defined(THREAD_LOCAL_ALLOC) || defined(GC_PTHREADS)
+ GC_thread me;
+# endif
+ DWORD thread_id = GetCurrentThreadId();
+
+ LOCK();
+ GC_wait_for_gc_completion(FALSE);
+# if defined(THREAD_LOCAL_ALLOC) || defined(GC_PTHREADS)
+ me = GC_lookup_thread_inner(thread_id);
+ CHECK_LOOKUP_MY_THREAD(me);
+ GC_ASSERT(!KNOWN_FINISHED(me));
+# endif
+# if defined(THREAD_LOCAL_ALLOC)
+ GC_destroy_thread_local(&(me->tlfs));
+# endif
+# ifdef GC_PTHREADS
+ if ((me -> flags & DETACHED) == 0) {
+ me -> flags |= FINISHED;
+ } else
+# endif
+ /* else */ {
+ GC_delete_thread(thread_id);
+ }
+ UNLOCK();
+ }
+ return GC_SUCCESS;
+}
+
+/* Wrapper for functions that are likely to block for an appreciable */
+/* length of time. */
+
+/* GC_do_blocking_inner() is nearly the same as in pthread_support.c */
+/*ARGSUSED*/
+GC_INNER void GC_do_blocking_inner(ptr_t data, void * context)
+{
+ struct blocking_data * d = (struct blocking_data *) data;
+ DWORD thread_id = GetCurrentThreadId();
+ GC_thread me;
+# ifdef IA64
+ ptr_t stack_ptr = GC_save_regs_in_stack();
+# endif
+ DCL_LOCK_STATE;
+
+ LOCK();
+ me = GC_lookup_thread_inner(thread_id);
+ CHECK_LOOKUP_MY_THREAD(me);
+ GC_ASSERT(me -> thread_blocked_sp == NULL);
+# ifdef IA64
+ me -> backing_store_ptr = stack_ptr;
+# endif
+ me -> thread_blocked_sp = (ptr_t) &d; /* save approx. sp */
+ /* Save context here if we want to support precise stack marking */
+ UNLOCK();
+ d -> client_data = (d -> fn)(d -> client_data);
+ LOCK(); /* This will block if the world is stopped. */
+ me -> thread_blocked_sp = NULL;
+ UNLOCK();
+}
+
+/* GC_call_with_gc_active() has the opposite to GC_do_blocking() */
+/* functionality. It might be called from a user function invoked by */
+/* GC_do_blocking() to temporarily back allow calling any GC function */
+/* and/or manipulating pointers to the garbage collected heap. */
+GC_API void * GC_CALL GC_call_with_gc_active(GC_fn_type fn,
+ void * client_data)
+{
+ struct GC_traced_stack_sect_s stacksect;
+ GC_thread me;
+ DCL_LOCK_STATE;
+
+ LOCK(); /* This will block if the world is stopped. */
+ me = GC_lookup_thread_inner(GetCurrentThreadId());
+ CHECK_LOOKUP_MY_THREAD(me);
+ /* Adjust our stack base value (this could happen unless */
+ /* GC_get_stack_base() was used which returned GC_SUCCESS). */
+ GC_ASSERT(me -> stack_base != NULL);
+ if (me -> stack_base < (ptr_t)(&stacksect))
+ me -> stack_base = (ptr_t)(&stacksect);
+
+ if (me -> thread_blocked_sp == NULL) {
+ /* We are not inside GC_do_blocking() - do nothing more. */
+ UNLOCK();
+ return fn(client_data);
+ }
+
+ /* Setup new "stack section". */
+ stacksect.saved_stack_ptr = me -> thread_blocked_sp;
+# ifdef IA64
+ /* This is the same as in GC_call_with_stack_base(). */
+ stacksect.backing_store_end = GC_save_regs_in_stack();
+ /* Unnecessarily flushes register stack, */
+ /* but that probably doesn't hurt. */
+ stacksect.saved_backing_store_ptr = me -> backing_store_ptr;
+# endif
+ stacksect.prev = me -> traced_stack_sect;
+ me -> thread_blocked_sp = NULL;
+ me -> traced_stack_sect = &stacksect;
+
+ UNLOCK();
+ client_data = fn(client_data);
+ GC_ASSERT(me -> thread_blocked_sp == NULL);
+ GC_ASSERT(me -> traced_stack_sect == &stacksect);
+
+ /* Restore original "stack section". */
+ LOCK();
+ me -> traced_stack_sect = stacksect.prev;
+# ifdef IA64
+ me -> backing_store_ptr = stacksect.saved_backing_store_ptr;
+# endif
+ me -> thread_blocked_sp = stacksect.saved_stack_ptr;
+ UNLOCK();
+
+ return client_data; /* result */
+}
+
+#ifdef GC_PTHREADS
+
+ /* A quick-and-dirty cache of the mapping between pthread_t */
+ /* and win32 thread id. */
+# define PTHREAD_MAP_SIZE 512
+ DWORD GC_pthread_map_cache[PTHREAD_MAP_SIZE] = {0};
+# define PTHREAD_MAP_INDEX(pthread_id) \
+ ((NUMERIC_THREAD_ID(pthread_id) >> 5) % PTHREAD_MAP_SIZE)
+ /* It appears pthread_t is really a pointer type ... */
+# define SET_PTHREAD_MAP_CACHE(pthread_id, win32_id) \
+ (GC_pthread_map_cache[PTHREAD_MAP_INDEX(pthread_id)] = (win32_id))
+# define GET_PTHREAD_MAP_CACHE(pthread_id) \
+ GC_pthread_map_cache[PTHREAD_MAP_INDEX(pthread_id)]
+
+ /* Return a GC_thread corresponding to a given pthread_t. */
+ /* Returns 0 if it's not there. */
+ /* We assume that this is only called for pthread ids that */
+ /* have not yet terminated or are still joinable, and */
+ /* cannot be concurrently terminated. */
+ /* Assumes we do NOT hold the allocation lock. */
+ STATIC GC_thread GC_lookup_pthread(pthread_t id)
+ {
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ int i;
+ LONG my_max = GC_get_max_thread_index();
+
+ for (i = 0; i <= my_max &&
+ (!AO_load_acquire(&dll_thread_table[i].tm.in_use)
+ || THREAD_EQUAL(dll_thread_table[i].pthread_id, id));
+ /* Must still be in_use, since nobody else can */
+ /* store our thread_id. */
+ i++) {
+ /* empty */
+ }
+ return i <= my_max ? (GC_thread)(dll_thread_table + i) : NULL;
+ } else
+# endif
+ /* else */ {
+ /* We first try the cache. If that fails, we use a very slow */
+ /* approach. */
+ word hv_guess = THREAD_TABLE_INDEX(GET_PTHREAD_MAP_CACHE(id));
+ int hv;
+ GC_thread p;
+ DCL_LOCK_STATE;
+
+ LOCK();
+ for (p = GC_threads[hv_guess]; 0 != p; p = p -> tm.next) {
+ if (THREAD_EQUAL(p -> pthread_id, id))
+ goto foundit;
+ }
+ for (hv = 0; hv < THREAD_TABLE_SZ; ++hv) {
+ for (p = GC_threads[hv]; 0 != p; p = p -> tm.next) {
+ if (THREAD_EQUAL(p -> pthread_id, id))
+ goto foundit;
+ }
+ }
+ p = 0;
+ foundit:
+ UNLOCK();
+ return p;
+ }
+ }
+
+#endif /* GC_PTHREADS */
+
+#ifdef CAN_HANDLE_FORK
+ /* Similar to that in pthread_support.c but also rehashes the table */
+ /* since hash map key (thread_id) differs from that in the parent. */
+ STATIC void GC_remove_all_threads_but_me(void)
+ {
+ int hv;
+ GC_thread p, next, me = NULL;
+ DWORD thread_id;
+ pthread_t pthread_id = pthread_self(); /* same as in parent */
+
+ GC_ASSERT(!GC_win32_dll_threads);
+ for (hv = 0; hv < THREAD_TABLE_SZ; ++hv) {
+ for (p = GC_threads[hv]; 0 != p; p = next) {
+ next = p -> tm.next;
+ if (THREAD_EQUAL(p -> pthread_id, pthread_id)) {
+ GC_ASSERT(me == NULL);
+ me = p;
+ p -> tm.next = 0;
+ } else {
+# ifdef THREAD_LOCAL_ALLOC
+ if ((p -> flags & FINISHED) == 0) {
+ GC_destroy_thread_local(&p->tlfs);
+ }
+# endif
+ if (&first_thread != p)
+ GC_INTERNAL_FREE(p);
+ }
+ }
+ GC_threads[hv] = NULL;
+ }
+
+ /* Put "me" back to GC_threads. */
+ GC_ASSERT(me != NULL);
+ thread_id = GetCurrentThreadId(); /* differs from that in parent */
+ GC_threads[THREAD_TABLE_INDEX(thread_id)] = me;
+
+ /* Update Win32 thread Id and handle. */
+ me -> id = thread_id;
+# ifndef MSWINCE
+ if (!DuplicateHandle(GetCurrentProcess(), GetCurrentThread(),
+ GetCurrentProcess(), (HANDLE *)&me->handle,
+ 0 /* dwDesiredAccess */, FALSE /* bInheritHandle */,
+ DUPLICATE_SAME_ACCESS))
+ ABORT("DuplicateHandle failed");
+# endif
+
+# if defined(THREAD_LOCAL_ALLOC) && !defined(USE_CUSTOM_SPECIFIC)
+ /* For Cygwin, we need to re-assign thread-local pointer to */
+ /* 'tlfs' (it is ok to call GC_destroy_thread_local and */
+ /* GC_free_internal before this action). */
+ if (GC_setspecific(GC_thread_key, &me->tlfs) != 0)
+ ABORT("GC_setspecific failed (in child)");
+# endif
+ }
+
+ STATIC void GC_fork_prepare_proc(void)
+ {
+ LOCK();
+# ifdef PARALLEL_MARK
+ if (GC_parallel)
+ GC_wait_for_reclaim();
+# endif
+ GC_wait_for_gc_completion(TRUE);
+# ifdef PARALLEL_MARK
+ if (GC_parallel)
+ GC_acquire_mark_lock();
+# endif
+ }
+
+ STATIC void GC_fork_parent_proc(void)
+ {
+# ifdef PARALLEL_MARK
+ if (GC_parallel)
+ GC_release_mark_lock();
+# endif
+ UNLOCK();
+ }
+
+ STATIC void GC_fork_child_proc(void)
+ {
+# ifdef PARALLEL_MARK
+ if (GC_parallel) {
+ GC_release_mark_lock();
+ GC_markers = 1;
+ GC_parallel = FALSE;
+ /* Turn off parallel marking in the child, since we are */
+ /* probably just going to exec, and we would have to */
+ /* restart mark threads. */
+ }
+# endif
+ GC_remove_all_threads_but_me();
+ UNLOCK();
+ }
+#endif /* CAN_HANDLE_FORK */
+
+void GC_push_thread_structures(void)
+{
+ GC_ASSERT(I_HOLD_LOCK());
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ /* Unlike the other threads implementations, the thread table here */
+ /* contains no pointers to the collectable heap. Thus we have */
+ /* no private structures we need to preserve. */
+# ifdef GC_PTHREADS
+ int i; /* pthreads may keep a pointer in the thread exit value */
+ LONG my_max = GC_get_max_thread_index();
+
+ for (i = 0; i <= my_max; i++)
+ if (dll_thread_table[i].tm.in_use)
+ GC_push_all((ptr_t)&(dll_thread_table[i].status),
+ (ptr_t)(&(dll_thread_table[i].status)+1));
+# endif
+ } else
+# endif
+ /* else */ {
+ GC_push_all((ptr_t)(GC_threads), (ptr_t)(GC_threads)+sizeof(GC_threads));
+ }
+# if defined(THREAD_LOCAL_ALLOC)
+ GC_push_all((ptr_t)(&GC_thread_key),
+ (ptr_t)(&GC_thread_key)+sizeof(&GC_thread_key));
+ /* Just in case we ever use our own TLS implementation. */
+# endif
+}
+
+/* Suspend the given thread, if it's still active. */
+STATIC void GC_suspend(GC_thread t)
+{
+# ifndef MSWINCE
+ /* Apparently the Windows 95 GetOpenFileName call creates */
+ /* a thread that does not properly get cleaned up, and */
+ /* SuspendThread on its descriptor may provoke a crash. */
+ /* This reduces the probability of that event, though it still */
+ /* appears there's a race here. */
+ DWORD exitCode;
+# endif
+ UNPROTECT_THREAD(t);
+# ifndef MSWINCE
+ if (GetExitCodeThread(t -> handle, &exitCode) &&
+ exitCode != STILL_ACTIVE) {
+# ifdef GC_PTHREADS
+ t -> stack_base = 0; /* prevent stack from being pushed */
+# else
+ /* this breaks pthread_join on Cygwin, which is guaranteed to */
+ /* only see user pthreads */
+ GC_ASSERT(GC_win32_dll_threads);
+ GC_delete_gc_thread(t);
+# endif
+ return;
+ }
+# endif
+# if defined(MPROTECT_VDB)
+ /* Acquire the spin lock we use to update dirty bits. */
+ /* Threads shouldn't get stopped holding it. But we may */
+ /* acquire and release it in the UNPROTECT_THREAD call. */
+ while (AO_test_and_set_acquire(&GC_fault_handler_lock) == AO_TS_SET) {
+ /* empty */
+ }
+# endif
+
+# ifdef MSWINCE
+ /* SuspendThread() will fail if thread is running kernel code. */
+ while (SuspendThread(THREAD_HANDLE(t)) == (DWORD)-1)
+ Sleep(10); /* in millis */
+# else
+ if (SuspendThread(t -> handle) == (DWORD)-1)
+ ABORT("SuspendThread failed");
+# endif /* !MSWINCE */
+ t -> suspended = (unsigned char)TRUE;
+# if defined(MPROTECT_VDB)
+ AO_CLEAR(&GC_fault_handler_lock);
+# endif
+}
+
+#if defined(GC_ASSERTIONS) && !defined(CYGWIN32)
+ GC_INNER GC_bool GC_write_disabled = FALSE;
+ /* TRUE only if GC_stop_world() acquired GC_write_cs. */
+#endif
+
+GC_INNER void GC_stop_world(void)
+{
+ DWORD thread_id = GetCurrentThreadId();
+
+ if (!GC_thr_initialized)
+ ABORT("GC_stop_world() called before GC_thr_init()");
+ GC_ASSERT(I_HOLD_LOCK());
+
+ /* This code is the same as in pthread_stop_world.c */
+# ifdef PARALLEL_MARK
+ if (GC_parallel) {
+ GC_acquire_mark_lock();
+ GC_ASSERT(GC_fl_builder_count == 0);
+ /* We should have previously waited for it to become zero. */
+ }
+# endif /* PARALLEL_MARK */
+
+# if !defined(GC_NO_THREADS_DISCOVERY) || defined(GC_ASSERTIONS)
+ GC_please_stop = TRUE;
+# endif
+# ifndef CYGWIN32
+ GC_ASSERT(!GC_write_disabled);
+ EnterCriticalSection(&GC_write_cs);
+ /* It's not allowed to call GC_printf() (and friends) here down to */
+ /* LeaveCriticalSection (same applies recursively to */
+ /* GC_get_max_thread_index(), GC_suspend(), GC_delete_gc_thread() */
+ /* (only if GC_win32_dll_threads), GC_size() and */
+ /* GC_remove_protection()). */
+# ifdef GC_ASSERTIONS
+ GC_write_disabled = TRUE;
+# endif
+# endif
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ int i;
+ int my_max;
+ /* Any threads being created during this loop will end up setting */
+ /* GC_attached_thread when they start. This will force marking */
+ /* to restart. This is not ideal, but hopefully correct. */
+ GC_attached_thread = FALSE;
+ my_max = (int)GC_get_max_thread_index();
+ for (i = 0; i <= my_max; i++) {
+ GC_vthread t = dll_thread_table + i;
+ if (t -> stack_base != 0 && t -> thread_blocked_sp == NULL
+ && t -> id != thread_id) {
+ GC_suspend((GC_thread)t);
+ }
+ }
+ } else
+# endif
+ /* else */ {
+ GC_thread t;
+ int i;
+
+ for (i = 0; i < THREAD_TABLE_SZ; i++) {
+ for (t = GC_threads[i]; t != 0; t = t -> tm.next) {
+ if (t -> stack_base != 0 && t -> thread_blocked_sp == NULL
+ && !KNOWN_FINISHED(t) && t -> id != thread_id) {
+ GC_suspend(t);
+ }
+ }
+ }
+ }
+# ifndef CYGWIN32
+# ifdef GC_ASSERTIONS
+ GC_write_disabled = FALSE;
+# endif
+ LeaveCriticalSection(&GC_write_cs);
+# endif
+# ifdef PARALLEL_MARK
+ if (GC_parallel)
+ GC_release_mark_lock();
+# endif
+}
+
+GC_INNER void GC_start_world(void)
+{
+# ifdef GC_ASSERTIONS
+ DWORD thread_id = GetCurrentThreadId();
+# endif
+ int i;
+
+ GC_ASSERT(I_HOLD_LOCK());
+ if (GC_win32_dll_threads) {
+ LONG my_max = GC_get_max_thread_index();
+ for (i = 0; i <= my_max; i++) {
+ GC_thread t = (GC_thread)(dll_thread_table + i);
+ if (t -> suspended) {
+ GC_ASSERT(t -> stack_base != 0 && t -> id != thread_id);
+ if (ResumeThread(THREAD_HANDLE(t)) == (DWORD)-1)
+ ABORT("ResumeThread failed");
+ t -> suspended = FALSE;
+ }
+ }
+ } else {
+ GC_thread t;
+ int i;
+
+ for (i = 0; i < THREAD_TABLE_SZ; i++) {
+ for (t = GC_threads[i]; t != 0; t = t -> tm.next) {
+ if (t -> suspended) {
+ GC_ASSERT(t -> stack_base != 0 && t -> id != thread_id);
+ if (ResumeThread(THREAD_HANDLE(t)) == (DWORD)-1)
+ ABORT("ResumeThread failed");
+ UNPROTECT_THREAD(t);
+ t -> suspended = FALSE;
+ }
+ }
+ }
+ }
+# if !defined(GC_NO_THREADS_DISCOVERY) || defined(GC_ASSERTIONS)
+ GC_please_stop = FALSE;
+# endif
+}
+
+#ifdef MSWINCE
+ /* The VirtualQuery calls below won't work properly on some old WinCE */
+ /* versions, but since each stack is restricted to an aligned 64 KiB */
+ /* region of virtual memory we can just take the next lowest multiple */
+ /* of 64 KiB. The result of this macro must not be used as its */
+ /* argument later and must not be used as the lower bound for sp */
+ /* check (since the stack may be bigger than 64 KiB). */
+# define GC_wince_evaluate_stack_min(s) \
+ (ptr_t)(((word)(s) - 1) & ~(word)0xFFFF)
+#elif defined(GC_ASSERTIONS)
+# define GC_dont_query_stack_min FALSE
+#endif
+
+/* A cache holding the results of the recent VirtualQuery call. */
+/* Protected by the allocation lock. */
+static ptr_t last_address = 0;
+static MEMORY_BASIC_INFORMATION last_info;
+
+/* Probe stack memory region (starting at "s") to find out its */
+/* lowest address (i.e. stack top). */
+/* S must be a mapped address inside the region, NOT the first */
+/* unmapped address. */
+STATIC ptr_t GC_get_stack_min(ptr_t s)
+{
+ ptr_t bottom;
+
+ GC_ASSERT(I_HOLD_LOCK());
+ if (s != last_address) {
+ VirtualQuery(s, &last_info, sizeof(last_info));
+ last_address = s;
+ }
+ do {
+ bottom = last_info.BaseAddress;
+ VirtualQuery(bottom - 1, &last_info, sizeof(last_info));
+ last_address = bottom - 1;
+ } while ((last_info.Protect & PAGE_READWRITE)
+ && !(last_info.Protect & PAGE_GUARD));
+ return(bottom);
+}
+
+/* Return true if the page at s has protections appropriate */
+/* for a stack page. */
+static GC_bool may_be_in_stack(ptr_t s)
+{
+ GC_ASSERT(I_HOLD_LOCK());
+ if (s != last_address) {
+ VirtualQuery(s, &last_info, sizeof(last_info));
+ last_address = s;
+ }
+ return (last_info.Protect & PAGE_READWRITE)
+ && !(last_info.Protect & PAGE_GUARD);
+}
+
+STATIC word GC_push_stack_for(GC_thread thread, DWORD me)
+{
+ ptr_t sp, stack_min;
+
+ struct GC_traced_stack_sect_s *traced_stack_sect =
+ thread -> traced_stack_sect;
+ if (thread -> id == me) {
+ GC_ASSERT(thread -> thread_blocked_sp == NULL);
+ sp = GC_approx_sp();
+ } else if ((sp = thread -> thread_blocked_sp) == NULL) {
+ /* Use saved sp value for blocked threads. */
+ /* For unblocked threads call GetThreadContext(). */
+ CONTEXT context;
+ context.ContextFlags = CONTEXT_INTEGER|CONTEXT_CONTROL;
+ if (!GetThreadContext(THREAD_HANDLE(thread), &context))
+ ABORT("GetThreadContext failed");
+
+ /* Push all registers that might point into the heap. Frame */
+ /* pointer registers are included in case client code was */
+ /* compiled with the 'omit frame pointer' optimisation. */
+# define PUSH1(reg) GC_push_one((word)context.reg)
+# define PUSH2(r1,r2) PUSH1(r1), PUSH1(r2)
+# define PUSH4(r1,r2,r3,r4) PUSH2(r1,r2), PUSH2(r3,r4)
+# if defined(I386)
+ PUSH4(Edi,Esi,Ebx,Edx), PUSH2(Ecx,Eax), PUSH1(Ebp);
+ sp = (ptr_t)context.Esp;
+# elif defined(X86_64)
+ PUSH4(Rax,Rcx,Rdx,Rbx); PUSH2(Rbp, Rsi); PUSH1(Rdi);
+ PUSH4(R8, R9, R10, R11); PUSH4(R12, R13, R14, R15);
+ sp = (ptr_t)context.Rsp;
+# elif defined(ARM32)
+ PUSH4(R0,R1,R2,R3),PUSH4(R4,R5,R6,R7),PUSH4(R8,R9,R10,R11);
+ PUSH1(R12);
+ sp = (ptr_t)context.Sp;
+# elif defined(SHx)
+ PUSH4(R0,R1,R2,R3), PUSH4(R4,R5,R6,R7), PUSH4(R8,R9,R10,R11);
+ PUSH2(R12,R13), PUSH1(R14);
+ sp = (ptr_t)context.R15;
+# elif defined(MIPS)
+ PUSH4(IntAt,IntV0,IntV1,IntA0), PUSH4(IntA1,IntA2,IntA3,IntT0);
+ PUSH4(IntT1,IntT2,IntT3,IntT4), PUSH4(IntT5,IntT6,IntT7,IntS0);
+ PUSH4(IntS1,IntS2,IntS3,IntS4), PUSH4(IntS5,IntS6,IntS7,IntT8);
+ PUSH4(IntT9,IntK0,IntK1,IntS8);
+ sp = (ptr_t)context.IntSp;
+# elif defined(PPC)
+ PUSH4(Gpr0, Gpr3, Gpr4, Gpr5), PUSH4(Gpr6, Gpr7, Gpr8, Gpr9);
+ PUSH4(Gpr10,Gpr11,Gpr12,Gpr14), PUSH4(Gpr15,Gpr16,Gpr17,Gpr18);
+ PUSH4(Gpr19,Gpr20,Gpr21,Gpr22), PUSH4(Gpr23,Gpr24,Gpr25,Gpr26);
+ PUSH4(Gpr27,Gpr28,Gpr29,Gpr30), PUSH1(Gpr31);
+ sp = (ptr_t)context.Gpr1;
+# elif defined(ALPHA)
+ PUSH4(IntV0,IntT0,IntT1,IntT2), PUSH4(IntT3,IntT4,IntT5,IntT6);
+ PUSH4(IntT7,IntS0,IntS1,IntS2), PUSH4(IntS3,IntS4,IntS5,IntFp);
+ PUSH4(IntA0,IntA1,IntA2,IntA3), PUSH4(IntA4,IntA5,IntT8,IntT9);
+ PUSH4(IntT10,IntT11,IntT12,IntAt);
+ sp = (ptr_t)context.IntSp;
+# else
+# error "architecture is not supported"
+# endif
+ } /* ! current thread */
+
+ /* Set stack_min to the lowest address in the thread stack, */
+ /* or to an address in the thread stack no larger than sp, */
+ /* taking advantage of the old value to avoid slow traversals */
+ /* of large stacks. */
+ if (thread -> last_stack_min == ADDR_LIMIT) {
+# ifdef MSWINCE
+ if (GC_dont_query_stack_min) {
+ stack_min = GC_wince_evaluate_stack_min(traced_stack_sect != NULL ?
+ (ptr_t)traced_stack_sect : thread -> stack_base);
+ /* Keep last_stack_min value unmodified. */
+ } else
+# endif
+ /* else */ {
+ stack_min = GC_get_stack_min(traced_stack_sect != NULL ?
+ (ptr_t)traced_stack_sect : thread -> stack_base);
+ UNPROTECT_THREAD(thread);
+ thread -> last_stack_min = stack_min;
+ }
+ } else {
+ /* First, adjust the latest known minimum stack address if we */
+ /* are inside GC_call_with_gc_active(). */
+ if (traced_stack_sect != NULL &&
+ thread -> last_stack_min > (ptr_t)traced_stack_sect) {
+ UNPROTECT_THREAD(thread);
+ thread -> last_stack_min = (ptr_t)traced_stack_sect;
+ }
+
+ if (sp < thread -> stack_base && sp >= thread -> last_stack_min) {
+ stack_min = sp;
+ } else {
+ /* In the current thread it is always safe to use sp value. */
+ if (may_be_in_stack(thread -> id == me &&
+ sp < thread -> last_stack_min ?
+ sp : thread -> last_stack_min)) {
+ stack_min = last_info.BaseAddress;
+ /* Do not probe rest of the stack if sp is correct. */
+ if (sp < stack_min || sp >= thread->stack_base)
+ stack_min = GC_get_stack_min(thread -> last_stack_min);
+ } else {
+ /* Stack shrunk? Is this possible? */
+ stack_min = GC_get_stack_min(thread -> stack_base);
+ }
+ UNPROTECT_THREAD(thread);
+ thread -> last_stack_min = stack_min;
+ }
+ }
+
+ GC_ASSERT(GC_dont_query_stack_min
+ || stack_min == GC_get_stack_min(thread -> stack_base)
+ || (sp >= stack_min && stack_min < thread -> stack_base
+ && stack_min > GC_get_stack_min(thread -> stack_base)));
+
+ if (sp >= stack_min && sp < thread->stack_base) {
+# ifdef DEBUG_THREADS
+ GC_log_printf("Pushing stack for 0x%x from sp %p to %p from 0x%x\n",
+ (int)thread -> id, sp, thread -> stack_base, (int)me);
+# endif
+ GC_push_all_stack_sections(sp, thread->stack_base, traced_stack_sect);
+ } else {
+ /* If not current thread then it is possible for sp to point to */
+ /* the guarded (untouched yet) page just below the current */
+ /* stack_min of the thread. */
+ if (thread -> id == me || sp >= thread->stack_base
+ || sp + GC_page_size < stack_min)
+ WARN("Thread stack pointer %p out of range, pushing everything\n",
+ sp);
+# ifdef DEBUG_THREADS
+ GC_log_printf("Pushing stack for 0x%x from (min) %p to %p from 0x%x\n",
+ (int)thread->id, stack_min, thread->stack_base, (int)me);
+# endif
+ /* Push everything - ignore "traced stack section" data. */
+ GC_push_all_stack(stack_min, thread->stack_base);
+ }
+ return thread->stack_base - sp; /* stack grows down */
+}
+
+GC_INNER void GC_push_all_stacks(void)
+{
+ DWORD thread_id = GetCurrentThreadId();
+ GC_bool found_me = FALSE;
+# ifndef SMALL_CONFIG
+ unsigned nthreads = 0;
+# endif
+ word total_size = 0;
+# ifndef GC_NO_THREADS_DISCOVERY
+ if (GC_win32_dll_threads) {
+ int i;
+ LONG my_max = GC_get_max_thread_index();
+
+ for (i = 0; i <= my_max; i++) {
+ GC_thread t = (GC_thread)(dll_thread_table + i);
+ if (t -> tm.in_use && t -> stack_base) {
+# ifndef SMALL_CONFIG
+ ++nthreads;
+# endif
+ total_size += GC_push_stack_for(t, thread_id);
+ if (t -> id == thread_id) found_me = TRUE;
+ }
+ }
+ } else
+# endif
+ /* else */ {
+ int i;
+ for (i = 0; i < THREAD_TABLE_SZ; i++) {
+ GC_thread t;
+ for (t = GC_threads[i]; t != 0; t = t -> tm.next) {
+ if (!KNOWN_FINISHED(t) && t -> stack_base) {
+# ifndef SMALL_CONFIG
+ ++nthreads;
+# endif
+ total_size += GC_push_stack_for(t, thread_id);
+ if (t -> id == thread_id) found_me = TRUE;
+ }
+ }
+ }
+ }
+# ifndef SMALL_CONFIG
+ if (GC_print_stats == VERBOSE) {
+ GC_log_printf("Pushed %d thread stacks%s\n", nthreads,
+ GC_win32_dll_threads ? " based on DllMain thread tracking" : "");
+ }
+# endif
+ if (!found_me && !GC_in_thread_creation)
+ ABORT("Collecting from unknown thread");
+ GC_total_stacksize = total_size;
+}
+
+#ifdef PARALLEL_MARK
+
+# ifndef MAX_MARKERS
+# define MAX_MARKERS 16
+# endif
+
+ static ptr_t marker_sp[MAX_MARKERS - 1]; /* The cold end of the stack */
+ /* for markers. */
+# ifdef IA64
+ static ptr_t marker_bsp[MAX_MARKERS - 1];
+# endif
+
+ static ptr_t marker_last_stack_min[MAX_MARKERS - 1];
+ /* Last known minimum (hottest) address */
+ /* in stack (or ADDR_LIMIT if unset) */
+ /* for markers. */
+
+#endif /* PARALLEL_MARK */
+
+/* Find stack with the lowest address which overlaps the */
+/* interval [start, limit). */
+/* Return stack bounds in *lo and *hi. If no such stack */
+/* is found, both *hi and *lo will be set to an address */
+/* higher than limit. */
+GC_INNER void GC_get_next_stack(char *start, char *limit,
+ char **lo, char **hi)
+{
+ int i;
+ char * current_min = ADDR_LIMIT; /* Least in-range stack base */
+ ptr_t *plast_stack_min = NULL; /* Address of last_stack_min */
+ /* field for thread corresponding */
+ /* to current_min. */
+ GC_thread thread = NULL; /* Either NULL or points to the */
+ /* thread's hash table entry */
+ /* containing *plast_stack_min. */
+
+ /* First set current_min, ignoring limit. */
+ if (GC_win32_dll_threads) {
+ LONG my_max = GC_get_max_thread_index();
+
+ for (i = 0; i <= my_max; i++) {
+ ptr_t s = (ptr_t)(dll_thread_table[i].stack_base);
+
+ if (s > start && s < current_min) {
+ /* Update address of last_stack_min. */
+ plast_stack_min = (ptr_t * /* no volatile */)
+ &dll_thread_table[i].last_stack_min;
+ current_min = s;
+ }
+ }
+ } else {
+ for (i = 0; i < THREAD_TABLE_SZ; i++) {
+ GC_thread t;
+
+ for (t = GC_threads[i]; t != 0; t = t -> tm.next) {
+ ptr_t s = t -> stack_base;
+
+ if (s > start && s < current_min) {
+ /* Update address of last_stack_min. */
+ plast_stack_min = &t -> last_stack_min;
+ thread = t; /* Remember current thread to unprotect. */
+ current_min = s;
+ }
+ }
+ }
+# ifdef PARALLEL_MARK
+ for (i = 0; i < GC_markers - 1; ++i) {
+ ptr_t s = marker_sp[i];
+# ifdef IA64
+ /* FIXME: not implemented */
+# endif
+ if (s > start && s < current_min) {
+ GC_ASSERT(marker_last_stack_min[i] != NULL);
+ plast_stack_min = &marker_last_stack_min[i];
+ current_min = s;
+ thread = NULL; /* Not a thread's hash table entry. */
+ }
+ }
+# endif
+ }
+
+ *hi = current_min;
+ if (current_min == ADDR_LIMIT) {
+ *lo = ADDR_LIMIT;
+ return;
+ }
+
+ GC_ASSERT(current_min > start && plast_stack_min != NULL);
+# ifdef MSWINCE
+ if (GC_dont_query_stack_min) {
+ *lo = GC_wince_evaluate_stack_min(current_min);
+ /* Keep last_stack_min value unmodified. */
+ return;
+ }
+# endif
+
+ if (current_min > limit && !may_be_in_stack(limit)) {
+ /* Skip the rest since the memory region at limit address is */
+ /* not a stack (so the lowest address of the found stack would */
+ /* be above the limit value anyway). */
+ *lo = ADDR_LIMIT;
+ return;
+ }
+
+ /* Get the minimum address of the found stack by probing its memory */
+ /* region starting from the recent known minimum (if set). */
+ if (*plast_stack_min == ADDR_LIMIT
+ || !may_be_in_stack(*plast_stack_min)) {
+ /* Unsafe to start from last_stack_min value. */
+ *lo = GC_get_stack_min(current_min);
+ } else {
+ /* Use the recent value to optimize search for min address. */
+ *lo = GC_get_stack_min(*plast_stack_min);
+ }
+
+ /* Remember current stack_min value. */
+ if (thread != NULL) {
+ UNPROTECT_THREAD(thread);
+ }
+ *plast_stack_min = *lo;
+}
+
+#ifdef PARALLEL_MARK
+
+# if defined(GC_PTHREADS) && !defined(GC_PTHREADS_PARAMARK)
+ /* Use pthread-based parallel mark implementation. */
+# define GC_PTHREADS_PARAMARK
+# endif
+
+ /* GC_mark_thread() is the same as in pthread_support.c */
+# ifdef GC_PTHREADS_PARAMARK
+ STATIC void * GC_mark_thread(void * id)
+# else
+# ifdef MSWINCE
+ STATIC DWORD WINAPI GC_mark_thread(LPVOID id)
+# else
+ STATIC unsigned __stdcall GC_mark_thread(void * id)
+# endif
+# endif
+ {
+ word my_mark_no = 0;
+
+ if ((word)id == (word)-1) return 0; /* to make compiler happy */
+ marker_sp[(word)id] = GC_approx_sp();
+# ifdef IA64
+ marker_bsp[(word)id] = GC_save_regs_in_stack();
+# endif
+
+ for (;; ++my_mark_no) {
+ if (my_mark_no - GC_mark_no > (word)2) {
+ /* resynchronize if we get far off, e.g. because GC_mark_no */
+ /* wrapped. */
+ my_mark_no = GC_mark_no;
+ }
+# ifdef DEBUG_THREADS
+ GC_log_printf("Starting mark helper for mark number %lu\n",
+ (unsigned long)my_mark_no);
+# endif
+ GC_help_marker(my_mark_no);
+ }
+ }
+
+# ifdef GC_ASSERTIONS
+ GC_INNER unsigned long GC_mark_lock_holder = NO_THREAD;
+# endif
+
+ /* GC_mark_threads[] is unused here unlike that in pthread_support.c */
+
+# ifdef GC_PTHREADS_PARAMARK
+# include <pthread.h>
+
+# ifndef NUMERIC_THREAD_ID
+# define NUMERIC_THREAD_ID(id) (unsigned long)GC_PTHREAD_PTRVAL(id)
+# endif
+
+ /* start_mark_threads() is the same as in pthread_support.c except for: */
+ /* - GC_markers value is adjusted already; */
+ /* - thread stack is assumed to be large enough; and */
+ /* - statistics about the number of marker threads is printed outside. */
+ static void start_mark_threads(void)
+ {
+ int i;
+ pthread_attr_t attr;
+ pthread_t new_thread;
+
+ if (0 != pthread_attr_init(&attr)) ABORT("pthread_attr_init failed");
+
+ if (0 != pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED))
+ ABORT("pthread_attr_setdetachstate failed");
+
+ for (i = 0; i < GC_markers - 1; ++i) {
+ marker_last_stack_min[i] = ADDR_LIMIT;
+ if (0 != pthread_create(&new_thread, &attr,
+ GC_mark_thread, (void *)(word)i)) {
+ WARN("Marker thread creation failed.\n", 0);
+ /* Don't try to create other marker threads. */
+ GC_markers = i + 1;
+ if (i == 0) GC_parallel = FALSE;
+ break;
+ }
+ }
+ pthread_attr_destroy(&attr);
+ }
+
+ static pthread_mutex_t mark_mutex = PTHREAD_MUTEX_INITIALIZER;
+
+ static pthread_cond_t builder_cv = PTHREAD_COND_INITIALIZER;
+
+ /* GC_acquire/release_mark_lock(), GC_wait_builder/marker(), */
+ /* GC_wait_for_reclaim(), GC_notify_all_builder/marker() are the same */
+ /* as in pthread_support.c except that GC_generic_lock() is not used. */
+
+# ifdef LOCK_STATS
+ AO_t GC_block_count = 0;
+# endif
+
+ GC_INNER void GC_acquire_mark_lock(void)
+ {
+ if (pthread_mutex_lock(&mark_mutex) != 0) {
+ ABORT("pthread_mutex_lock failed");
+ }
+# ifdef LOCK_STATS
+ (void)AO_fetch_and_add1(&GC_block_count);
+# endif
+ /* GC_generic_lock(&mark_mutex); */
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NUMERIC_THREAD_ID(pthread_self());
+# endif
+ }
+
+ GC_INNER void GC_release_mark_lock(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == NUMERIC_THREAD_ID(pthread_self()));
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NO_THREAD;
+# endif
+ if (pthread_mutex_unlock(&mark_mutex) != 0) {
+ ABORT("pthread_mutex_unlock failed");
+ }
+ }
+
+ /* Collector must wait for a freelist builders for 2 reasons: */
+ /* 1) Mark bits may still be getting examined without lock. */
+ /* 2) Partial free lists referenced only by locals may not be */
+ /* scanned correctly, e.g. if they contain "pointer-free" objects, */
+ /* since the free-list link may be ignored. */
+ STATIC void GC_wait_builder(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == NUMERIC_THREAD_ID(pthread_self()));
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NO_THREAD;
+# endif
+ if (pthread_cond_wait(&builder_cv, &mark_mutex) != 0) {
+ ABORT("pthread_cond_wait failed");
+ }
+ GC_ASSERT(GC_mark_lock_holder == NO_THREAD);
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NUMERIC_THREAD_ID(pthread_self());
+# endif
+ }
+
+ GC_INNER void GC_wait_for_reclaim(void)
+ {
+ GC_acquire_mark_lock();
+ while (GC_fl_builder_count > 0) {
+ GC_wait_builder();
+ }
+ GC_release_mark_lock();
+ }
+
+ GC_INNER void GC_notify_all_builder(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == NUMERIC_THREAD_ID(pthread_self()));
+ if (pthread_cond_broadcast(&builder_cv) != 0) {
+ ABORT("pthread_cond_broadcast failed");
+ }
+ }
+
+ static pthread_cond_t mark_cv = PTHREAD_COND_INITIALIZER;
+
+ GC_INNER void GC_wait_marker(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == NUMERIC_THREAD_ID(pthread_self()));
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NO_THREAD;
+# endif
+ if (pthread_cond_wait(&mark_cv, &mark_mutex) != 0) {
+ ABORT("pthread_cond_wait failed");
+ }
+ GC_ASSERT(GC_mark_lock_holder == NO_THREAD);
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NUMERIC_THREAD_ID(pthread_self());
+# endif
+ }
+
+ GC_INNER void GC_notify_all_marker(void)
+ {
+ if (pthread_cond_broadcast(&mark_cv) != 0) {
+ ABORT("pthread_cond_broadcast failed");
+ }
+ }
+
+# else /* ! GC_PTHREADS_PARAMARK */
+
+# ifdef DONT_USE_SIGNALANDWAIT
+ STATIC HANDLE GC_marker_cv[MAX_MARKERS - 1] = {0};
+ /* Events with manual reset (one for each */
+ /* mark helper). */
+
+ STATIC DWORD GC_marker_Id[MAX_MARKERS - 1] = {0};
+ /* This table is used for mapping helper */
+ /* threads ID to mark helper index (linear */
+ /* search is used since the mapping contains */
+ /* only a few entries). */
+# endif
+
+# ifndef MARK_THREAD_STACK_SIZE
+# define MARK_THREAD_STACK_SIZE 0 /* default value */
+# endif
+
+ /* mark_mutex_event, builder_cv, mark_cv are initialized in GC_thr_init */
+ static HANDLE mark_mutex_event = (HANDLE)0; /* Event with auto-reset. */
+ static HANDLE builder_cv = (HANDLE)0; /* Event with manual reset. */
+ static HANDLE mark_cv = (HANDLE)0; /* Event with manual reset. */
+
+ static void start_mark_threads(void)
+ {
+ int i;
+# ifdef MSWINCE
+ HANDLE handle;
+ DWORD thread_id;
+# else
+ GC_uintptr_t handle;
+ unsigned thread_id;
+# endif
+
+# ifdef DONT_USE_SIGNALANDWAIT
+ /* Initialize GC_marker_cv[] and GC_marker_Id[] fully before */
+ /* starting the first helper thread. */
+ for (i = 0; i < GC_markers - 1; ++i) {
+ GC_marker_Id[i] = GetCurrentThreadId();
+ if ((GC_marker_cv[i] = CreateEvent(NULL /* attrs */,
+ TRUE /* isManualReset */,
+ FALSE /* initialState */,
+ NULL /* name (A/W) */)) == (HANDLE)0)
+ ABORT("CreateEvent() failed");
+ }
+# endif
+
+ for (i = 0; i < GC_markers - 1; ++i) {
+ marker_last_stack_min[i] = ADDR_LIMIT;
+# ifdef MSWINCE
+ /* There is no _beginthreadex() in WinCE. */
+ handle = CreateThread(NULL /* lpsa */,
+ MARK_THREAD_STACK_SIZE /* ignored */,
+ GC_mark_thread, (LPVOID)(word)i,
+ 0 /* fdwCreate */, &thread_id);
+ if (handle == NULL) {
+ WARN("Marker thread creation failed\n", 0);
+ /* The most probable failure reason is "not enough memory". */
+ /* Don't try to create other marker threads. */
+ break;
+ } else {
+ /* It's safe to detach the thread. */
+ CloseHandle(handle);
+ }
+# else
+ handle = _beginthreadex(NULL /* security_attr */,
+ MARK_THREAD_STACK_SIZE, GC_mark_thread,
+ (void *)(word)i, 0 /* flags */, &thread_id);
+ if (!handle || handle == (GC_uintptr_t)-1L) {
+ WARN("Marker thread creation failed\n", 0);
+ /* Don't try to create other marker threads. */
+ break;
+ } else {/* We may detach the thread (if handle is of HANDLE type) */
+ /* CloseHandle((HANDLE)handle); */
+ }
+# endif
+ }
+
+ /* Adjust GC_markers (and free unused resources) in case of failure. */
+# ifdef DONT_USE_SIGNALANDWAIT
+ while ((int)GC_markers > i + 1) {
+ GC_markers--;
+ CloseHandle(GC_marker_cv[(int)GC_markers - 1]);
+ }
+# else
+ GC_markers = i + 1;
+# endif
+ if (i == 0) {
+ GC_parallel = FALSE;
+ CloseHandle(mark_cv);
+ CloseHandle(builder_cv);
+ CloseHandle(mark_mutex_event);
+ }
+ }
+
+# ifdef DONT_USE_SIGNALANDWAIT
+ STATIC /* volatile */ LONG GC_mark_mutex_state = 0;
+ /* Mutex state: 0 - unlocked, */
+ /* 1 - locked and no other waiters, */
+ /* -1 - locked and waiters may exist. */
+ /* Accessed by InterlockedExchange(). */
+# else
+ STATIC volatile AO_t GC_mark_mutex_waitcnt = 0;
+ /* Number of waiters + 1; 0 - unlocked. */
+# endif
+
+ /* #define LOCK_STATS */
+# ifdef LOCK_STATS
+ AO_t GC_block_count = 0;
+ AO_t GC_unlocked_count = 0;
+# endif
+
+ GC_INNER void GC_acquire_mark_lock(void)
+ {
+# ifdef DONT_USE_SIGNALANDWAIT
+ if (InterlockedExchange(&GC_mark_mutex_state, 1 /* locked */) != 0)
+# else
+ if (AO_fetch_and_add1_acquire(&GC_mark_mutex_waitcnt) != 0)
+# endif
+ {
+# ifdef LOCK_STATS
+ (void)AO_fetch_and_add1(&GC_block_count);
+# endif
+# ifdef DONT_USE_SIGNALANDWAIT
+ /* Repeatedly reset the state and wait until acquire the lock. */
+ while (InterlockedExchange(&GC_mark_mutex_state,
+ -1 /* locked_and_has_waiters */) != 0)
+# endif
+ {
+ if (WaitForSingleObject(mark_mutex_event, INFINITE) == WAIT_FAILED)
+ ABORT("WaitForSingleObject() failed");
+ }
+ }
+# ifdef LOCK_STATS
+ else {
+ (void)AO_fetch_and_add1(&GC_unlocked_count);
+ }
+# endif
+
+ GC_ASSERT(GC_mark_lock_holder == NO_THREAD);
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = (unsigned long)GetCurrentThreadId();
+# endif
+ }
+
+ GC_INNER void GC_release_mark_lock(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == (unsigned long)GetCurrentThreadId());
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NO_THREAD;
+# endif
+# ifdef DONT_USE_SIGNALANDWAIT
+ if (InterlockedExchange(&GC_mark_mutex_state, 0 /* unlocked */) < 0)
+# else
+ GC_ASSERT(AO_load(&GC_mark_mutex_waitcnt) != 0);
+ if (AO_fetch_and_sub1_release(&GC_mark_mutex_waitcnt) > 1)
+# endif
+ {
+ /* wake a waiter */
+ if (SetEvent(mark_mutex_event) == FALSE)
+ ABORT("SetEvent() failed");
+ }
+ }
+
+ /* In GC_wait_for_reclaim/GC_notify_all_builder() we emulate POSIX */
+ /* cond_wait/cond_broadcast() primitives with WinAPI Event object */
+ /* (working in "manual reset" mode). This works here because */
+ /* GC_notify_all_builder() is always called holding lock on */
+ /* mark_mutex and the checked condition (GC_fl_builder_count == 0) */
+ /* is the only one for which broadcasting on builder_cv is performed. */
+
+ GC_INNER void GC_wait_for_reclaim(void)
+ {
+ GC_ASSERT(builder_cv != 0);
+ for (;;) {
+ GC_acquire_mark_lock();
+ if (GC_fl_builder_count == 0)
+ break;
+ if (ResetEvent(builder_cv) == FALSE)
+ ABORT("ResetEvent() failed");
+ GC_release_mark_lock();
+ if (WaitForSingleObject(builder_cv, INFINITE) == WAIT_FAILED)
+ ABORT("WaitForSingleObject() failed");
+ }
+ GC_release_mark_lock();
+ }
+
+ GC_INNER void GC_notify_all_builder(void)
+ {
+ GC_ASSERT(GC_mark_lock_holder == (unsigned long)GetCurrentThreadId());
+ GC_ASSERT(builder_cv != 0);
+ GC_ASSERT(GC_fl_builder_count == 0);
+ if (SetEvent(builder_cv) == FALSE)
+ ABORT("SetEvent() failed");
+ }
+
+# ifdef DONT_USE_SIGNALANDWAIT
+
+ /* mark_cv is used (for waiting) by a non-helper thread. */
+
+ GC_INNER void GC_wait_marker(void)
+ {
+ HANDLE event = mark_cv;
+ DWORD thread_id = GetCurrentThreadId();
+ int i = (int)GC_markers - 1;
+ while (i-- > 0) {
+ if (GC_marker_Id[i] == thread_id) {
+ event = GC_marker_cv[i];
+ break;
+ }
+ }
+
+ if (ResetEvent(event) == FALSE)
+ ABORT("ResetEvent() failed");
+ GC_release_mark_lock();
+ if (WaitForSingleObject(event, INFINITE) == WAIT_FAILED)
+ ABORT("WaitForSingleObject() failed");
+ GC_acquire_mark_lock();
+ }
+
+ GC_INNER void GC_notify_all_marker(void)
+ {
+ DWORD thread_id = GetCurrentThreadId();
+ int i = (int)GC_markers - 1;
+ while (i-- > 0) {
+ /* Notify every marker ignoring self (for efficiency). */
+ if (SetEvent(GC_marker_Id[i] != thread_id ? GC_marker_cv[i] :
+ mark_cv) == FALSE)
+ ABORT("SetEvent() failed");
+ }
+ }
+
+# else /* DONT_USE_SIGNALANDWAIT */
+
+ /* For GC_wait_marker/GC_notify_all_marker() the above technique */
+ /* does not work because they are used with different checked */
+ /* conditions in different places (and, in addition, notifying is */
+ /* done after leaving critical section) and this could result in */
+ /* a signal loosing between checking for a particular condition */
+ /* and calling WaitForSingleObject. So, we use PulseEvent() and */
+ /* NT SignalObjectAndWait() (which atomically sets mutex event to */
+ /* signaled state and starts waiting on condvar). A special */
+ /* case here is GC_mark_mutex_waitcnt == 1 (i.e. nobody waits for */
+ /* mark lock at this moment) - we don't change it (otherwise we */
+ /* may loose a signal sent between decrementing */
+ /* GC_mark_mutex_waitcnt and calling WaitForSingleObject()). */
+
+# ifdef MSWINCE
+ /* SignalObjectAndWait() is missing in WinCE (for now), so you */
+ /* should supply its emulation (externally) to use this code. */
+ WINBASEAPI DWORD WINAPI SignalObjectAndWait(HANDLE, HANDLE, DWORD,
+ BOOL);
+# define signalObjectAndWait_func SignalObjectAndWait
+# else
+ typedef DWORD (WINAPI * SignalObjectAndWait_type)(HANDLE, HANDLE,
+ DWORD, BOOL);
+ static SignalObjectAndWait_type signalObjectAndWait_func = 0;
+# endif
+
+ GC_INNER void GC_wait_marker(void)
+ {
+ /* Here we assume that GC_wait_marker() is always called */
+ /* from a while(check_cond) loop. */
+ AO_t waitcnt;
+ GC_ASSERT(mark_cv != 0);
+
+ /* We inline GC_release_mark_lock() to have atomic */
+ /* unlock-and-wait action here. */
+ GC_ASSERT(GC_mark_lock_holder == (unsigned long)GetCurrentThreadId());
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = NO_THREAD;
+# endif
+
+ if ((waitcnt = AO_load(&GC_mark_mutex_waitcnt)) > 1) {
+ (void)AO_fetch_and_sub1_release(&GC_mark_mutex_waitcnt);
+ } else {
+ GC_ASSERT(AO_load(&GC_mark_mutex_waitcnt) != 0);
+ }
+
+ /* The state of mark_cv is non-signaled here. */
+ if (signalObjectAndWait_func(mark_mutex_event /* hObjectToSignal */,
+ mark_cv /* hObjectToWaitOn */,
+ INFINITE /* timeout */,
+ FALSE /* isAlertable */) == WAIT_FAILED)
+ ABORT("SignalObjectAndWait() failed");
+ /* The state of mark_cv is non-signaled here again. */
+
+ if (waitcnt > 1) {
+ GC_acquire_mark_lock();
+ } else {
+ GC_ASSERT(GC_mark_mutex_waitcnt != 0);
+ /* Acquire mark lock */
+ if (WaitForSingleObject(mark_mutex_event, INFINITE) == WAIT_FAILED)
+ ABORT("WaitForSingleObject() failed");
+ GC_ASSERT(GC_mark_lock_holder == NO_THREAD);
+# ifdef GC_ASSERTIONS
+ GC_mark_lock_holder = (unsigned long)GetCurrentThreadId();
+# endif
+ }
+ }
+
+ GC_INNER void GC_notify_all_marker(void)
+ {
+ GC_ASSERT(mark_cv != 0);
+ if (PulseEvent(mark_cv) == FALSE)
+ ABORT("PulseEvent() failed");
+ }
+
+# endif /* !DONT_USE_SIGNALANDWAIT */
+
+# endif /* ! GC_PTHREADS_PARAMARK */
+
+#endif /* PARALLEL_MARK */
+
+#ifndef GC_PTHREADS
+
+ /* We have no DllMain to take care of new threads. Thus we */
+ /* must properly intercept thread creation. */
+
+ typedef struct {
+ LPTHREAD_START_ROUTINE start;
+ LPVOID param;
+ } thread_args;
+
+ STATIC void * GC_CALLBACK GC_win32_start_inner(struct GC_stack_base *sb,
+ void *arg)
+ {
+ void * ret;
+ LPTHREAD_START_ROUTINE start = ((thread_args *)arg)->start;
+ LPVOID param = ((thread_args *)arg)->param;
+
+ GC_register_my_thread(sb); /* This waits for an in-progress GC. */
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread 0x%lx starting...\n", (long)GetCurrentThreadId());
+# endif
+
+ GC_free(arg);
+
+ /* Clear the thread entry even if we exit with an exception. */
+ /* This is probably pointless, since an uncaught exception is */
+ /* supposed to result in the process being killed. */
+# ifndef __GNUC__
+ __try
+# endif
+ {
+ ret = (void *)(word)(*start)(param);
+ }
+# ifndef __GNUC__
+ __finally
+# endif
+ {
+ GC_unregister_my_thread();
+ }
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread 0x%lx returned from start routine\n",
+ (long)GetCurrentThreadId());
+# endif
+ return ret;
+ }
+
+ STATIC DWORD WINAPI GC_win32_start(LPVOID arg)
+ {
+ return (DWORD)(word)GC_call_with_stack_base(GC_win32_start_inner, arg);
+ }
+
+ GC_API HANDLE WINAPI GC_CreateThread(
+ LPSECURITY_ATTRIBUTES lpThreadAttributes,
+ DWORD dwStackSize,
+ LPTHREAD_START_ROUTINE lpStartAddress,
+ LPVOID lpParameter, DWORD dwCreationFlags,
+ LPDWORD lpThreadId)
+ {
+ HANDLE thread_h;
+ thread_args *args;
+
+ if (!parallel_initialized) GC_init_parallel();
+ /* make sure GC is initialized (i.e. main thread is */
+ /* attached, tls initialized). */
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("About to create a thread from 0x%lx\n",
+ (long)GetCurrentThreadId());
+# endif
+ if (GC_win32_dll_threads) {
+ return CreateThread(lpThreadAttributes, dwStackSize, lpStartAddress,
+ lpParameter, dwCreationFlags, lpThreadId);
+ } else {
+ args = GC_malloc_uncollectable(sizeof(thread_args));
+ /* Handed off to and deallocated by child thread. */
+ if (0 == args) {
+ SetLastError(ERROR_NOT_ENOUGH_MEMORY);
+ return NULL;
+ }
+
+ /* set up thread arguments */
+ args -> start = lpStartAddress;
+ args -> param = lpParameter;
+
+ GC_need_to_lock = TRUE;
+ thread_h = CreateThread(lpThreadAttributes, dwStackSize, GC_win32_start,
+ args, dwCreationFlags, lpThreadId);
+ if (thread_h == 0) GC_free(args);
+ return thread_h;
+ }
+ }
+
+ GC_API DECLSPEC_NORETURN void WINAPI GC_ExitThread(DWORD dwExitCode)
+ {
+ GC_unregister_my_thread();
+ ExitThread(dwExitCode);
+ }
+
+# ifndef MSWINCE
+
+ GC_API GC_uintptr_t GC_CALL GC_beginthreadex(
+ void *security, unsigned stack_size,
+ unsigned (__stdcall *start_address)(void *),
+ void *arglist, unsigned initflag,
+ unsigned *thrdaddr)
+ {
+ GC_uintptr_t thread_h;
+ thread_args *args;
+
+ if (!parallel_initialized) GC_init_parallel();
+ /* make sure GC is initialized (i.e. main thread is */
+ /* attached, tls initialized). */
+# ifdef DEBUG_THREADS
+ GC_log_printf("About to create a thread from 0x%lx\n",
+ (long)GetCurrentThreadId());
+# endif
+
+ if (GC_win32_dll_threads) {
+ return _beginthreadex(security, stack_size, start_address,
+ arglist, initflag, thrdaddr);
+ } else {
+ args = GC_malloc_uncollectable(sizeof(thread_args));
+ /* Handed off to and deallocated by child thread. */
+ if (0 == args) {
+ /* MSDN docs say _beginthreadex() returns 0 on error and sets */
+ /* errno to either EAGAIN (too many threads) or EINVAL (the */
+ /* argument is invalid or the stack size is incorrect), so we */
+ /* set errno to EAGAIN on "not enough memory". */
+ errno = EAGAIN;
+ return 0;
+ }
+
+ /* set up thread arguments */
+ args -> start = (LPTHREAD_START_ROUTINE)start_address;
+ args -> param = arglist;
+
+ GC_need_to_lock = TRUE;
+ thread_h = _beginthreadex(security, stack_size,
+ (unsigned (__stdcall *)(void *))GC_win32_start,
+ args, initflag, thrdaddr);
+ if (thread_h == 0) GC_free(args);
+ return thread_h;
+ }
+ }
+
+ GC_API void GC_CALL GC_endthreadex(unsigned retval)
+ {
+ GC_unregister_my_thread();
+ _endthreadex(retval);
+ }
+
+# endif /* !MSWINCE */
+
+#endif /* !GC_PTHREADS */
+
+#ifdef GC_WINMAIN_REDIRECT
+ /* This might be useful on WinCE. Shouldn't be used with GC_DLL. */
+
+# if defined(MSWINCE) && defined(UNDER_CE)
+# define WINMAIN_LPTSTR LPWSTR
+# else
+# define WINMAIN_LPTSTR LPSTR
+# endif
+
+ /* This is defined in gc.h. */
+# undef WinMain
+
+ /* Defined outside GC by an application. */
+ int WINAPI GC_WinMain(HINSTANCE, HINSTANCE, WINMAIN_LPTSTR, int);
+
+ typedef struct {
+ HINSTANCE hInstance;
+ HINSTANCE hPrevInstance;
+ WINMAIN_LPTSTR lpCmdLine;
+ int nShowCmd;
+ } main_thread_args;
+
+ static DWORD WINAPI main_thread_start(LPVOID arg)
+ {
+ main_thread_args * args = (main_thread_args *) arg;
+ return (DWORD)GC_WinMain(args->hInstance, args->hPrevInstance,
+ args->lpCmdLine, args->nShowCmd);
+ }
+
+ STATIC void * GC_waitForSingleObjectInfinite(void * handle)
+ {
+ return (void *)(word)WaitForSingleObject((HANDLE)handle, INFINITE);
+ }
+
+# ifndef WINMAIN_THREAD_STACK_SIZE
+# define WINMAIN_THREAD_STACK_SIZE 0 /* default value */
+# endif
+
+ int WINAPI WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance,
+ WINMAIN_LPTSTR lpCmdLine, int nShowCmd)
+ {
+ DWORD exit_code = 1;
+
+ main_thread_args args = {
+ hInstance, hPrevInstance, lpCmdLine, nShowCmd
+ };
+ HANDLE thread_h;
+ DWORD thread_id;
+
+ /* initialize everything */
+ GC_INIT();
+
+ /* start the main thread */
+ thread_h = GC_CreateThread(NULL /* lpsa */,
+ WINMAIN_THREAD_STACK_SIZE /* ignored on WinCE */,
+ main_thread_start, &args, 0 /* fdwCreate */,
+ &thread_id);
+
+ if (thread_h != NULL) {
+ if ((DWORD)(word)GC_do_blocking(GC_waitForSingleObjectInfinite,
+ (void *)thread_h) == WAIT_FAILED)
+ ABORT("WaitForSingleObject(main_thread) failed");
+ GetExitCodeThread (thread_h, &exit_code);
+ CloseHandle (thread_h);
+ } else {
+ ABORT("GC_CreateThread(main_thread) failed");
+ }
+
+# ifdef MSWINCE
+ GC_deinit();
+ DeleteCriticalSection(&GC_allocate_ml);
+# endif
+ return (int) exit_code;
+ }
+
+#endif /* GC_WINMAIN_REDIRECT */
+
+/* Called by GC_init() - we hold the allocation lock. */
+GC_INNER void GC_thr_init(void)
+{
+ struct GC_stack_base sb;
+# ifdef GC_ASSERTIONS
+ int sb_result;
+# endif
+
+ GC_ASSERT(I_HOLD_LOCK());
+ if (GC_thr_initialized) return;
+ GC_main_thread = GetCurrentThreadId();
+ GC_thr_initialized = TRUE;
+
+# ifdef CAN_HANDLE_FORK
+ /* Prepare for forks if requested. */
+ if (GC_handle_fork
+ && pthread_atfork(GC_fork_prepare_proc, GC_fork_parent_proc,
+ GC_fork_child_proc) != 0)
+ ABORT("pthread_atfork failed");
+# endif
+
+ /* Add the initial thread, so we can stop it. */
+# ifdef GC_ASSERTIONS
+ sb_result =
+# endif
+ GC_get_stack_base(&sb);
+ GC_ASSERT(sb_result == GC_SUCCESS);
+
+# if defined(PARALLEL_MARK)
+ /* Set GC_markers. */
+ {
+ char * markers_string = GETENV("GC_MARKERS");
+ if (markers_string != NULL) {
+ GC_markers = atoi(markers_string);
+ if (GC_markers > MAX_MARKERS) {
+ WARN("Limiting number of mark threads\n", 0);
+ GC_markers = MAX_MARKERS;
+ }
+ } else {
+# ifdef MSWINCE
+ /* There is no GetProcessAffinityMask() in WinCE. */
+ /* GC_sysinfo is already initialized. */
+ GC_markers = GC_sysinfo.dwNumberOfProcessors;
+# else
+# ifdef _WIN64
+ DWORD_PTR procMask = 0;
+ DWORD_PTR sysMask;
+# else
+ DWORD procMask = 0;
+ DWORD sysMask;
+# endif
+ int ncpu = 0;
+ if (GetProcessAffinityMask(GetCurrentProcess(),
+ (void *)&procMask, (void *)&sysMask)
+ && procMask) {
+ do {
+ ncpu++;
+ } while ((procMask &= procMask - 1) != 0);
+ }
+ GC_markers = ncpu;
+# endif
+# ifdef GC_MIN_MARKERS
+ /* This is primarily for testing on systems without getenv(). */
+ if (GC_markers < GC_MIN_MARKERS)
+ GC_markers = GC_MIN_MARKERS;
+# endif
+ if (GC_markers >= MAX_MARKERS)
+ GC_markers = MAX_MARKERS; /* silently limit GC_markers value */
+ }
+ }
+
+ /* Set GC_parallel. */
+ {
+# if !defined(GC_PTHREADS_PARAMARK) && !defined(MSWINCE) \
+ && !defined(DONT_USE_SIGNALANDWAIT)
+ HMODULE hK32;
+ /* SignalObjectAndWait() API call works only under NT. */
+# endif
+ if (GC_win32_dll_threads || GC_markers <= 1
+# if !defined(GC_PTHREADS_PARAMARK) && !defined(MSWINCE) \
+ && !defined(DONT_USE_SIGNALANDWAIT)
+ || GC_wnt == FALSE
+ || (hK32 = GetModuleHandle(TEXT("kernel32.dll"))) == (HMODULE)0
+ || (signalObjectAndWait_func = (SignalObjectAndWait_type)
+ GetProcAddress(hK32, "SignalObjectAndWait")) == 0
+# endif
+ ) {
+ /* Disable parallel marking. */
+ GC_parallel = FALSE;
+ GC_markers = 1;
+ } else {
+# ifndef GC_PTHREADS_PARAMARK
+ /* Initialize Win32 event objects for parallel marking. */
+ mark_mutex_event = CreateEvent(NULL /* attrs */,
+ FALSE /* isManualReset */,
+ FALSE /* initialState */, NULL /* name */);
+ builder_cv = CreateEvent(NULL /* attrs */,
+ TRUE /* isManualReset */,
+ FALSE /* initialState */, NULL /* name */);
+ mark_cv = CreateEvent(NULL /* attrs */, TRUE /* isManualReset */,
+ FALSE /* initialState */, NULL /* name */);
+ if (mark_mutex_event == (HANDLE)0 || builder_cv == (HANDLE)0
+ || mark_cv == (HANDLE)0)
+ ABORT("CreateEvent() failed");
+# endif
+ GC_parallel = TRUE;
+ /* Disable true incremental collection, but generational is OK. */
+ GC_time_limit = GC_TIME_UNLIMITED;
+ }
+ }
+# endif /* PARALLEL_MARK */
+
+ GC_ASSERT(0 == GC_lookup_thread_inner(GC_main_thread));
+ GC_register_my_thread_inner(&sb, GC_main_thread);
+
+# ifdef PARALLEL_MARK
+ /* If we are using a parallel marker, actually start helper threads. */
+ if (GC_parallel) start_mark_threads();
+ if (GC_print_stats) {
+ GC_log_printf("Started %ld mark helper threads\n", GC_markers - 1);
+ }
+# endif
+}
+
+#ifdef GC_PTHREADS
+
+ struct start_info {
+ void *(*start_routine)(void *);
+ void *arg;
+ GC_bool detached;
+ };
+
+ GC_API int GC_pthread_join(pthread_t pthread_id, void **retval)
+ {
+ int result;
+ GC_thread t;
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread %p(0x%lx) is joining thread %p\n",
+ GC_PTHREAD_PTRVAL(pthread_self()),
+ (long)GetCurrentThreadId(), GC_PTHREAD_PTRVAL(pthread_id));
+# endif
+
+ if (!parallel_initialized) GC_init_parallel();
+
+ /* Thread being joined might not have registered itself yet. */
+ /* After the join,thread id may have been recycled. */
+ /* FIXME: It would be better if this worked more like */
+ /* pthread_support.c. */
+# ifndef GC_WIN32_PTHREADS
+ while ((t = GC_lookup_pthread(pthread_id)) == 0)
+ Sleep(10);
+# endif
+
+ result = pthread_join(pthread_id, retval);
+
+# ifdef GC_WIN32_PTHREADS
+ /* win32_pthreads id are unique */
+ t = GC_lookup_pthread(pthread_id);
+# endif
+
+ if (!GC_win32_dll_threads) {
+ DCL_LOCK_STATE;
+
+ LOCK();
+ GC_delete_gc_thread(t);
+ UNLOCK();
+ } /* otherwise DllMain handles it. */
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread %p(0x%lx) completed join with thread %p\n",
+ GC_PTHREAD_PTRVAL(pthread_self()),
+ (long)GetCurrentThreadId(), GC_PTHREAD_PTRVAL(pthread_id));
+# endif
+ return result;
+ }
+
+ /* Cygwin-pthreads calls CreateThread internally, but it's not easily */
+ /* interceptible by us..., so intercept pthread_create instead. */
+ GC_API int GC_pthread_create(pthread_t *new_thread,
+ GC_PTHREAD_CREATE_CONST pthread_attr_t *attr,
+ void *(*start_routine)(void *), void *arg)
+ {
+ if (!parallel_initialized) GC_init_parallel();
+ /* make sure GC is initialized (i.e. main thread is attached) */
+ if (GC_win32_dll_threads) {
+ return pthread_create(new_thread, attr, start_routine, arg);
+ } else {
+ int result;
+ struct start_info * si;
+
+ /* This is otherwise saved only in an area mmapped by the thread */
+ /* library, which isn't visible to the collector. */
+ si = GC_malloc_uncollectable(sizeof(struct start_info));
+ if (0 == si) return(EAGAIN);
+
+ si -> start_routine = start_routine;
+ si -> arg = arg;
+ if (attr != 0 &&
+ pthread_attr_getdetachstate(attr, &si->detached)
+ == PTHREAD_CREATE_DETACHED) {
+ si->detached = TRUE;
+ }
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("About to create a thread from %p(0x%lx)\n",
+ GC_PTHREAD_PTRVAL(pthread_self()),
+ (long)GetCurrentThreadId());
+# endif
+ GC_need_to_lock = TRUE;
+ result = pthread_create(new_thread, attr, GC_pthread_start, si);
+
+ if (result) { /* failure */
+ GC_free(si);
+ }
+ return(result);
+ }
+ }
+
+ STATIC void * GC_CALLBACK GC_pthread_start_inner(struct GC_stack_base *sb,
+ void * arg)
+ {
+ struct start_info * si = arg;
+ void * result;
+ void *(*start)(void *);
+ void *start_arg;
+ DWORD thread_id = GetCurrentThreadId();
+ pthread_t pthread_id = pthread_self();
+ GC_thread me;
+ DCL_LOCK_STATE;
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread %p(0x%x) starting...\n",
+ GC_PTHREAD_PTRVAL(pthread_id), (int)thread_id);
+# endif
+
+ GC_ASSERT(!GC_win32_dll_threads);
+ /* If a GC occurs before the thread is registered, that GC will */
+ /* ignore this thread. That's fine, since it will block trying to */
+ /* acquire the allocation lock, and won't yet hold interesting */
+ /* pointers. */
+ LOCK();
+ /* We register the thread here instead of in the parent, so that */
+ /* we don't need to hold the allocation lock during pthread_create. */
+ me = GC_register_my_thread_inner(sb, thread_id);
+ SET_PTHREAD_MAP_CACHE(pthread_id, thread_id);
+ me -> pthread_id = pthread_id;
+ if (si->detached) me -> flags |= DETACHED;
+ UNLOCK();
+
+ start = si -> start_routine;
+ start_arg = si -> arg;
+
+ GC_free(si); /* was allocated uncollectable */
+
+ pthread_cleanup_push(GC_thread_exit_proc, (void *)me);
+ result = (*start)(start_arg);
+ me -> status = result;
+ pthread_cleanup_pop(1);
+
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread %p(0x%x) returned from start routine\n",
+ GC_PTHREAD_PTRVAL(pthread_id), (int)thread_id);
+# endif
+ return(result);
+ }
+
+ STATIC void * GC_pthread_start(void * arg)
+ {
+ return GC_call_with_stack_base(GC_pthread_start_inner, arg);
+ }
+
+ STATIC void GC_thread_exit_proc(void *arg)
+ {
+ GC_thread me = (GC_thread)arg;
+ DCL_LOCK_STATE;
+
+ GC_ASSERT(!GC_win32_dll_threads);
+# ifdef DEBUG_THREADS
+ GC_log_printf("thread %p(0x%lx) called pthread_exit()\n",
+ GC_PTHREAD_PTRVAL(pthread_self()),
+ (long)GetCurrentThreadId());
+# endif
+
+ LOCK();
+ GC_wait_for_gc_completion(FALSE);
+# if defined(THREAD_LOCAL_ALLOC)
+ GC_destroy_thread_local(&(me->tlfs));
+# endif
+ if (me -> flags & DETACHED) {
+ GC_delete_thread(GetCurrentThreadId());
+ } else {
+ /* deallocate it as part of join */
+ me -> flags |= FINISHED;
+ }
+ UNLOCK();
+ }
+
+# ifndef GC_NO_PTHREAD_SIGMASK
+ /* Win32 pthread does not support sigmask. */
+ /* So, nothing required here... */
+ GC_API int GC_pthread_sigmask(int how, const sigset_t *set,
+ sigset_t *oset)
+ {
+ if (!parallel_initialized) GC_init_parallel();
+ return pthread_sigmask(how, set, oset);
+ }
+# endif /* !GC_NO_PTHREAD_SIGMASK */
+
+ GC_API int GC_pthread_detach(pthread_t thread)
+ {
+ int result;
+ GC_thread t;
+ DCL_LOCK_STATE;
+
+ if (!parallel_initialized) GC_init_parallel();
+ LOCK();
+ t = GC_lookup_pthread(thread);
+ UNLOCK();
+ result = pthread_detach(thread);
+ if (result == 0) {
+ LOCK();
+ t -> flags |= DETACHED;
+ /* Here the pthread thread id may have been recycled. */
+ if ((t -> flags & FINISHED) != 0) {
+ GC_delete_gc_thread(t);
+ }
+ UNLOCK();
+ }
+ return result;
+ }
+
+#else /* !GC_PTHREADS */
+
+# ifndef GC_NO_THREADS_DISCOVERY
+ /* We avoid acquiring locks here, since this doesn't seem to be */
+ /* preemptible. This may run with an uninitialized collector, in */
+ /* which case we don't do much. This implies that no threads other */
+ /* than the main one should be created with an uninitialized */
+ /* collector. (The alternative of initializing the collector here */
+ /* seems dangerous, since DllMain is limited in what it can do.) */
+
+ /*ARGSUSED*/
+ BOOL WINAPI DllMain(HINSTANCE inst, ULONG reason, LPVOID reserved)
+ {
+ struct GC_stack_base sb;
+ DWORD thread_id;
+# ifdef GC_ASSERTIONS
+ int sb_result;
+# endif
+ static int entry_count = 0;
+
+ if (!GC_win32_dll_threads && parallel_initialized) return TRUE;
+
+ switch (reason) {
+ case DLL_THREAD_ATTACH:
+# ifdef PARALLEL_MARK
+ /* Don't register marker threads. */
+ if (GC_parallel) {
+ /* We could reach here only if parallel_initialized == FALSE. */
+ break;
+ }
+# endif
+ GC_ASSERT(entry_count == 0 || parallel_initialized);
+ ++entry_count; /* and fall through: */
+ case DLL_PROCESS_ATTACH:
+ /* This may run with the collector uninitialized. */
+ thread_id = GetCurrentThreadId();
+ if (parallel_initialized && GC_main_thread != thread_id) {
+ /* Don't lock here. */
+# ifdef GC_ASSERTIONS
+ sb_result =
+# endif
+ GC_get_stack_base(&sb);
+ GC_ASSERT(sb_result == GC_SUCCESS);
+# if defined(THREAD_LOCAL_ALLOC) || defined(PARALLEL_MARK)
+ ABORT("Cannot initialize thread local cache from DllMain");
+# endif
+ GC_register_my_thread_inner(&sb, thread_id);
+ } /* o.w. we already did it during GC_thr_init, called by GC_init */
+ break;
+
+ case DLL_THREAD_DETACH:
+ /* We are hopefully running in the context of the exiting thread. */
+ GC_ASSERT(parallel_initialized);
+ if (GC_win32_dll_threads) {
+ GC_delete_thread(GetCurrentThreadId());
+ }
+ break;
+
+ case DLL_PROCESS_DETACH:
+ if (GC_win32_dll_threads) {
+ int i;
+ int my_max = (int)GC_get_max_thread_index();
+
+ for (i = 0; i <= my_max; ++i) {
+ if (AO_load(&(dll_thread_table[i].tm.in_use)))
+ GC_delete_gc_thread(dll_thread_table + i);
+ }
+ GC_deinit();
+ DeleteCriticalSection(&GC_allocate_ml);
+ }
+ break;
+ }
+ return TRUE;
+ }
+# endif /* !GC_NO_THREADS_DISCOVERY */
+
+#endif /* !GC_PTHREADS */
+
+/* Perform all initializations, including those that */
+/* may require allocation. */
+/* Called without allocation lock. */
+/* Must be called before a second thread is created. */
+GC_INNER void GC_init_parallel(void)
+{
+# if defined(THREAD_LOCAL_ALLOC)
+ GC_thread me;
+ DCL_LOCK_STATE;
+# endif
+
+ if (parallel_initialized) return;
+ parallel_initialized = TRUE;
+ /* GC_init() calls us back, so set flag first. */
+
+ if (!GC_is_initialized) GC_init();
+ if (GC_win32_dll_threads) {
+ GC_need_to_lock = TRUE;
+ /* Cannot intercept thread creation. Hence we don't know if */
+ /* other threads exist. However, client is not allowed to */
+ /* create other threads before collector initialization. */
+ /* Thus it's OK not to lock before this. */
+ }
+ /* Initialize thread local free lists if used. */
+# if defined(THREAD_LOCAL_ALLOC)
+ LOCK();
+ me = GC_lookup_thread_inner(GetCurrentThreadId());
+ CHECK_LOOKUP_MY_THREAD(me);
+ GC_init_thread_local(&me->tlfs);
+ UNLOCK();
+# endif
+}
+
+#if defined(USE_PTHREAD_LOCKS)
+ /* Support for pthread locking code. */
+ /* Pthread_mutex_try_lock may not win here, */
+ /* due to builtin support for spinning first? */
+
+ GC_INNER volatile GC_bool GC_collecting = 0;
+ /* A hint that we're in the collector and */
+ /* holding the allocation lock for an */
+ /* extended period. */
+
+ GC_INNER void GC_lock(void)
+ {
+ pthread_mutex_lock(&GC_allocate_ml);
+ }
+#endif /* USE_PTHREAD_LOCKS */
+
+#if defined(THREAD_LOCAL_ALLOC)
+
+ /* Add thread-local allocation support. VC++ uses __declspec(thread). */
+
+ /* We must explicitly mark ptrfree and gcj free lists, since the free */
+ /* list links wouldn't otherwise be found. We also set them in the */
+ /* normal free lists, since that involves touching less memory than if */
+ /* we scanned them normally. */
+ GC_INNER void GC_mark_thread_local_free_lists(void)
+ {
+ int i;
+ GC_thread p;
+
+ for (i = 0; i < THREAD_TABLE_SZ; ++i) {
+ for (p = GC_threads[i]; 0 != p; p = p -> tm.next) {
+ if (!KNOWN_FINISHED(p)) {
+# ifdef DEBUG_THREADS
+ GC_log_printf("Marking thread locals for 0x%x\n", (int)p -> id);
+# endif
+ GC_mark_thread_local_fls_for(&(p->tlfs));
+ }
+ }
+ }
+ }
+
+# if defined(GC_ASSERTIONS)
+ void GC_check_tls_for(GC_tlfs p);
+# if defined(USE_CUSTOM_SPECIFIC)
+ void GC_check_tsd_marks(tsd *key);
+# endif
+ /* Check that all thread-local free-lists are completely marked. */
+ /* also check that thread-specific-data structures are marked. */
+ void GC_check_tls(void)
+ {
+ int i;
+ GC_thread p;
+
+ for (i = 0; i < THREAD_TABLE_SZ; ++i) {
+ for (p = GC_threads[i]; 0 != p; p = p -> tm.next) {
+ if (!KNOWN_FINISHED(p))
+ GC_check_tls_for(&(p->tlfs));
+ }
+ }
+# if defined(USE_CUSTOM_SPECIFIC)
+ if (GC_thread_key != 0)
+ GC_check_tsd_marks(GC_thread_key);
+# endif
+ }
+# endif /* GC_ASSERTIONS */
+
+#endif /* THREAD_LOCAL_ALLOC ... */
+
+# ifndef GC_NO_THREAD_REDIRECTS
+ /* Restore thread calls redirection. */
+# define CreateThread GC_CreateThread
+# define ExitThread GC_ExitThread
+# undef _beginthreadex
+# define _beginthreadex GC_beginthreadex
+# undef _endthreadex
+# define _endthreadex GC_endthreadex
+# endif /* !GC_NO_THREAD_REDIRECTS */
+
+#endif /* GC_WIN32_THREADS */