linux: improve diagnosis on exception handling

Under some circumstances, the diagnostic message in the exception signal
handler was not printed. This could happen due to a dead lock in the
console library if the console code itself produces the exception while
possessing the mutex, e.g., by exhausting a undersized stack. Now, we
directly write to the log session via the stdout_write() hook or use
raw_write_str() in core.

repos/base-linux/src/base/thread/thread_env.cc

@@ -31,6 +31,19 @@ char **lx_environ;
  */
 int main_thread_futex_counter __attribute__((aligned(sizeof(addr_t))));
 
+/**
+ * Genode console hook
+ */
+extern "C" int stdout_write(char const *);
+
+/*
+ * Core lacks the hook, so provide a base-linux specific weak implementation
+ */
+extern "C" __attribute__((weak)) int stdout_write(char const *s)
+{
+	return raw_write_str(s);
+}
+
 /**
  * Signal handler for exceptions like segmentation faults
  */
@@ -46,7 +59,18 @@ static void exception_signal_handler(int signum)
 
 	default: /* unexpected signal */ return;
 	}
-	PERR("%s (signum=%d), see Linux kernel log for details", reason, signum);
+
+	/*
+	 * We can't use Genode::printf() as the exception may have occurred in the
+	 * Genode console library itself, which uses a mutex. Therefore, we use
+	 * Genode::snprintf() and call the console hook directly to minimize
+	 * overlaps with other code paths.
+	 */
+	static char msg[128];
+	snprintf(msg, sizeof(msg),
+	         ESC_ERR "%s (signum=%d), see Linux kernel log for details" ESC_END "\n",
+	         reason, signum);
+	stdout_write(msg);
 
 	/*
 	 * We reset the signal handler to SIG_DFL and trigger exception again,