make last thread's pthread_exit give exit(0) a consistent state

[musl] / src / thread / pthread_create.c

diff --git a/src/thread/pthread_create.c b/src/thread/pthread_create.c
index e6760ab..5d5d3a6 100644 (file)
--- a/src/thread/pthread_create.c
+++ b/src/thread/pthread_create.c
@@ -12,7 +12,7 @@ weak_alias(dummy_0, __pthread_tsd_run_dtors);
 _Noreturn void pthread_exit(void *result)
 {
        pthread_t self = pthread_self();
-       int n;
+       sigset_t set;
 
        self->result = result;
 
@@ -32,15 +32,35 @@ _Noreturn void pthread_exit(void *result)
        self->dead = 1;
        __unlock(self->killlock);
 
-       do n = libc.threads_minus_1;
-       while (n && a_cas(&libc.threads_minus_1, n, n-1)!=n);
-       if (!n) exit(0);
+       /* Block all signals before decrementing the live thread count.
+        * This is important to ensure that dynamically allocated TLS
+        * is not under-allocated/over-committed, and possibly for other
+        * reasons as well. */
+       __syscall(SYS_rt_sigprocmask, SIG_BLOCK, SIGALL_SET, &set, _NSIG/8);
+
+       /* It's impossible to determine whether this is "the last thread"
+        * until performing the atomic decrement, since multiple threads
+        * could exit at the same time. For the last thread, revert the
+        * decrement and unblock signals to give the atexit handlers and
+        * stdio cleanup code a consistent state. */
+       if (a_fetch_add(&libc.threads_minus_1, -1)==0) {
+               libc.threads_minus_1 = 0;
+               __syscall(SYS_rt_sigprocmask, SIG_SETMASK, &set, 0, _NSIG/8);
+               exit(0);
+       }
 
        if (self->detached && self->map_base) {
-               if (self->detached == 2)
-                       __syscall(SYS_set_tid_address, 0);
-               __syscall(SYS_rt_sigprocmask, SIG_BLOCK,
-                       SIGALL_SET, 0, _NSIG/8);
+               /* Detached threads must avoid the kernel clear_child_tid
+                * feature, since the virtual address will have been
+                * unmapped and possibly already reused by a new mapping
+                * at the time the kernel would perform the write. In
+                * the case of threads that started out detached, the
+                * initial clone flags are correct, but if the thread was
+                * detached later (== 2), we need to clear it here. */
+               if (self->detached == 2) __syscall(SYS_set_tid_address, 0);
+
+               /* The following call unmaps the thread's stack mapping
+                * and then exits without touching the stack. */
                __unmapself(self->map_base, self->map_size);
        }