core: use raw_clone instead of fork in signal handler

fork() is not async-signal-safe and calling it from the signal handler
could result in a deadlock when at_fork() handlers are called. Using
the raw clone() syscall sidesteps that problem.

The tricky part is that raise() does not work, since getpid() does not
work. Add raw_getpid() to get the real pid, and use kill() instead of
raise().

https://bugs.freedesktop.org/show_bug.cgi?id=86604

3 files changed, 42 insertions, 13 deletions

diff --git a/src/core/main.c b/src/core/main.c
index 77980e392..300567a92 100644
--- a/src/core/main.c
+++ b/src/core/main.c
@@ -142,7 +142,7 @@ noreturn static void crash(int sig) {
                 /* We want to wait for the core process, hence let's enable SIGCHLD */
                 sigaction(SIGCHLD, &sa, NULL);
 
-                pid = fork();
+                pid = raw_clone(SIGCHLD, NULL);
                 if (pid < 0)
                         log_emergency_errno(errno, "Caught <%s>, cannot fork for core dump: %m", signal_to_string(sig));
 
@@ -163,11 +163,11 @@ noreturn static void crash(int sig) {
                         chdir("/");
 
                         /* Raise the signal again */
-                        raise(sig);
+                        pid = raw_getpid();
+                        kill(pid, sig); /* raise() would kill the parent */
 
                         assert_not_reached("We shouldn't be here...");
                         _exit(1);
-
                 } else {
                         siginfo_t status;
                         int r;
@@ -177,7 +177,13 @@ noreturn static void crash(int sig) {
                         if (r < 0)
                                 log_emergency_errno(r, "Caught <%s>, waitpid() failed: %m", signal_to_string(sig));
                         else if (status.si_code != CLD_DUMPED)
-                                log_emergency("Caught <%s>, core dump failed.", signal_to_string(sig));
+                                log_emergency("Caught <%s>, core dump failed (child "PID_FMT", code=%s, status=%i/%s).",
+                                              signal_to_string(sig),
+                                              pid, sigchld_code_to_string(status.si_code),
+                                              status.si_status,
+                                              strna(status.si_code == CLD_EXITED
+                                                    ? exit_status_to_string(status.si_status, EXIT_STATUS_FULL)
+                                                    : signal_to_string(status.si_status)));
                         else
                                 log_emergency("Caught <%s>, dumped core as pid "PID_FMT".", signal_to_string(sig), pid);
                 }
@@ -199,18 +205,17 @@ noreturn static void crash(int sig) {
                 /* Let the kernel reap children for us */
                 assert_se(sigaction(SIGCHLD, &sa, NULL) == 0);
 
-                pid = fork();
+                pid = raw_clone(SIGCHLD, NULL);
                 if (pid < 0)
                         log_emergency_errno(errno, "Failed to fork off crash shell: %m");
                 else if (pid == 0) {
                         make_console_stdio();
-                        execl("/bin/sh", "/bin/sh", NULL);
+                        execle("/bin/sh", "/bin/sh", NULL, environ);
 
-                        log_emergency_errno(errno, "execl() failed: %m");
+                        log_emergency_errno(errno, "execle() failed: %m");
                         _exit(1);
-                }
-
-                log_info("Successfully spawned crash shell as pid "PID_FMT".", pid);
+                } else
+                        log_info("Successfully spawned crash shell as PID "PID_FMT".", pid);
         }
 
         log_emergency("Freezing execution.");
diff --git a/src/shared/missing.h b/src/shared/missing.h
index bea125436..91a621522 100644
--- a/src/shared/missing.h
+++ b/src/shared/missing.h
@@ -636,12 +636,16 @@ static inline int setns(int fd, int nstype) {
 #define CAP_AUDIT_READ 37
 #endif
 
-static inline long raw_clone(unsigned long flags, void *child_stack) {
+static inline int raw_clone(unsigned long flags, void *child_stack) {
 #if defined(__s390__) || defined(__CRIS__)
         /* On s390 and cris the order of the first and second arguments
          * of the raw clone() system call is reversed. */
-        return syscall(__NR_clone, child_stack, flags);
+        return (int) syscall(__NR_clone, child_stack, flags);
 #else
-        return syscall(__NR_clone, flags, child_stack);
+        return (int) syscall(__NR_clone, flags, child_stack);
 #endif
 }
+
+static inline pid_t raw_getpid(void) {
+        return (pid_t) syscall(__NR_getpid);
+}
diff --git a/src/test/test-util.c b/src/test/test-util.c
index 6c7d77b19..bbf751283 100644
--- a/src/test/test-util.c
+++ b/src/test/test-util.c
@@ -1312,6 +1312,25 @@ static void test_parse_proc_cmdline(void) {
         assert_se(parse_proc_cmdline(parse_item) >= 0);
 }
 
+static void test_raw_clone(void) {
+        pid_t parent, pid, pid2;
+
+        parent = getpid();
+        log_info("before clone: getpid()→"PID_FMT, parent);
+        assert_se(raw_getpid() == parent);
+
+        pid = raw_clone(0, NULL);
+        assert(pid >= 0);
+
+        pid2 = raw_getpid();
+        log_info("raw_clone: "PID_FMT" getpid()→"PID_FMT" raw_getpid()→"PID_FMT,
+                 pid, getpid(), pid2);
+        if (pid == 0)
+                assert(pid2 != parent);
+        else
+                assert(pid2 == parent);
+}
+
 int main(int argc, char *argv[]) {
         log_parse_environment();
         log_open();
@@ -1384,6 +1403,7 @@ int main(int argc, char *argv[]) {
         test_unquote_first_word();
         test_unquote_many_words();
         test_parse_proc_cmdline();
+        test_raw_clone();
 
         return 0;
 }