erts: Introduce thread suspend functions

These functions allow any thread to suspend any other thread
immediately and then resume all threads. This is useful when
doing a crash dump in order to get a more accurate picture
of what state the system is in.

3 files changed, 85 insertions, 4 deletions

diff --git a/erts/emulator/beam/erl_threads.h b/erts/emulator/beam/erl_threads.h
index 1ab5418741..463271252d 100644
--- a/erts/emulator/beam/erl_threads.h
+++ b/erts/emulator/beam/erl_threads.h
@@ -621,11 +621,17 @@ ERTS_GLB_INLINE void erts_thr_set_main_status(int, int);
 ERTS_GLB_INLINE int erts_thr_get_main_status(void);
 ERTS_GLB_INLINE void erts_thr_yield(void);
 
+
 #ifdef ETHR_HAVE_ETHR_SIG_FUNCS
 #define ERTS_THR_HAVE_SIG_FUNCS 1
 ERTS_GLB_INLINE void erts_thr_sigmask(int how, const sigset_t *set,
 				      sigset_t *oset);
 ERTS_GLB_INLINE void erts_thr_sigwait(const sigset_t *set, int *sig);
+
+#ifdef USE_THREADS
+ERTS_GLB_INLINE void erts_thr_kill(erts_tid_t tid, int sig);
+#endif
+
 #endif /* #ifdef HAVE_ETHR_SIG_FUNCS */
 
 #ifdef USE_THREADS
@@ -2849,6 +2855,15 @@ ERTS_GLB_INLINE void erts_thr_yield(void)
 #ifdef ETHR_HAVE_ETHR_SIG_FUNCS
 
 ERTS_GLB_INLINE void
+erts_thr_kill(erts_tid_t tid, int sig) {
+#ifdef USE_THREADS
+  int res = ethr_kill((ethr_tid)tid, sig);
+  if (res)
+    erts_thr_fatal_error(res, "killing thread");
+#endif
+}
+
+ERTS_GLB_INLINE void
 erts_thr_sigmask(int how, const sigset_t *set, sigset_t *oset)
 {
 #ifdef USE_THREADS
diff --git a/erts/emulator/beam/sys.h b/erts/emulator/beam/sys.h
index c29d4b3777..bdc5237815 100644
--- a/erts/emulator/beam/sys.h
+++ b/erts/emulator/beam/sys.h
@@ -784,6 +784,11 @@ int erts_sys_unsetenv(char *key);
 char *erts_read_env(char *key);
 void erts_free_read_env(void *value);
 
+#if defined(ERTS_THR_HAVE_SIG_FUNCS) && !defined(ETHR_UNUSABLE_SIGUSRX)
+extern void sys_thr_resume(erts_tid_t tid);
+extern void sys_thr_suspend(erts_tid_t tid);
+#endif
+
 /* utils.c */
 
 /* Options to sys_alloc_opt */
diff --git a/erts/emulator/sys/unix/sys.c b/erts/emulator/sys/unix/sys.c
index 81ed5626c2..841f0c7575 100644
--- a/erts/emulator/sys/unix/sys.c
+++ b/erts/emulator/sys/unix/sys.c
@@ -216,6 +216,9 @@ static erts_smp_atomic_t sys_misc_mem_sz;
 #if defined(ERTS_SMP)
 static void smp_sig_notify(char c);
 static int sig_notify_fds[2] = {-1, -1};
+
+static int sig_suspend_fds[2] = {-1, -1};
+#define ERTS_SYS_SUSPEND_SIGNAL SIGUSR2
 #elif defined(USE_THREADS)
 static int async_fd[2];
 #endif
@@ -789,6 +792,9 @@ prepare_crash_dump(int secs)
 	/* We don't want to close the signal notification pipe... */
 	if (i == sig_notify_fds[0] || i == sig_notify_fds[1])
 	    continue;
+	/* We don't want to close the signal syspend pipe... */
+	if (i == sig_suspend_fds[0] || i == sig_suspend_fds[1])
+	    continue;
 #elif defined(USE_THREADS)
 	/* We don't want to close the async notification pipe... */
 	if (i == async_fd[0] || i == async_fd[1])
@@ -877,9 +883,23 @@ sigusr1_exit(void)
 
 #ifdef ETHR_UNUSABLE_SIGUSRX
 #warning "Unusable SIGUSR1 & SIGUSR2. Disabling use of these signals"
-#endif
 
-#ifndef ETHR_UNUSABLE_SIGUSRX
+#else
+
+#ifdef ERTS_SMP
+void
+sys_thr_suspend(erts_tid_t tid) {
+    erts_thr_kill(tid, ERTS_SYS_SUSPEND_SIGNAL);
+}
+
+void
+sys_thr_resume(erts_tid_t tid) {
+    int i = 0, res;
+    do {
+        res = write(sig_suspend_fds[1],&i,sizeof(i));
+    } while (res < 0 && errno == EAGAIN);
+}
+#endif
 
 #if (defined(SIG_SIGSET) || defined(SIG_SIGNAL))
 static RETSIGTYPE user_signal1(void)
@@ -894,6 +914,21 @@ static RETSIGTYPE user_signal1(int signum)
 #endif
 }
 
+#ifdef ERTS_SMP
+#if (defined(SIG_SIGSET) || defined(SIG_SIGNAL))
+static RETSIGTYPE suspend_signal(void)
+#else
+static RETSIGTYPE suspend_signal(int signum)
+#endif
+{
+   int res;
+   int buf[1];
+   do {
+     res = read(sig_suspend_fds[0], buf, sizeof(int));
+   } while (res < 0 && errno == EINTR);
+}
+#endif /* #ifdef ERTS_SMP */
+
 #endif /* #ifndef ETHR_UNUSABLE_SIGUSRX */
 
 static void
@@ -944,7 +979,10 @@ void init_break_handler(void)
 {
    sys_sigset(SIGINT, request_break);
 #ifndef ETHR_UNUSABLE_SIGUSRX
-   sys_sigset(SIGUSR1, user_signal1);
+   sys_signal(SIGUSR1, user_signal1);
+#ifdef ERTS_SMP
+   sys_signal(ERTS_SYS_SUSPEND_SIGNAL, suspend_signal);
+#endif /* #ifdef ERTS_SMP */
 #endif /* #ifndef ETHR_UNUSABLE_SIGUSRX */
    sys_sigset(SIGQUIT, do_quit);
 }
@@ -963,8 +1001,13 @@ static void block_signals(void)
    sys_sigblock(SIGINT);
 #ifndef ETHR_UNUSABLE_SIGUSRX
    sys_sigblock(SIGUSR1);
+#endif /* #ifndef ETHR_UNUSABLE_SIGUSRX */
+#endif /* #ifndef ERTS_SMP */
+
+#if defined(ERTS_SMP) && !defined(ETHR_UNUSABLE_SIGUSRX)
+   sys_sigblock(ERTS_SYS_SUSPEND_SIGNAL);
 #endif
-#endif
+
 }
 
 static void unblock_signals(void)
@@ -978,8 +1021,14 @@ static void unblock_signals(void)
 #ifndef ETHR_UNUSABLE_SIGUSRX
    sys_sigrelease(SIGUSR1);
 #endif /* #ifndef ETHR_UNUSABLE_SIGUSRX */
+#endif /* #ifndef ERTS_SMP */
+
+#if defined(ERTS_SMP) && !defined(ETHR_UNUSABLE_SIGUSRX)
+   sys_sigrelease(ERTS_SYS_SUSPEND_SIGNAL);
 #endif
+
 }
+
 /************************** Time stuff **************************/
 #ifdef HAVE_GETHRTIME
 #ifdef GETHRTIME_WITH_CLOCK_GETTIME
@@ -3221,6 +3270,17 @@ init_smp_sig_notify(void)
 			NULL,
 			&thr_opts);
 }
+
+static void
+init_smp_sig_suspend(void) {
+  if (pipe(sig_suspend_fds) < 0) {
+    erl_exit(ERTS_ABORT_EXIT,
+	     "Failed to create sig_suspend pipe: %s (%d)\n",
+	     erl_errno_id(errno),
+	     errno);
+  }
+}
+
 #ifdef __DARWIN__
 
 int erts_darwin_main_thread_pipe[2];
@@ -3371,6 +3431,7 @@ erl_sys_args(int* argc, char** argv)
 
 #ifdef ERTS_SMP
     init_smp_sig_notify();
+    init_smp_sig_suspend();
 #endif
 
     /* Handled arguments have been marked with NULL. Slide arguments