Processes to record topic generations before execution

The sigchld generation expresses the idea that, if we receive a sigchld
signal, the generation will be different than when we last recorded it. A
process cannot exit before it has launched, so check the generation count
before process launch. This is an optimization that reduces failing
waitpid calls.

src/exec.cpp

@@ -899,6 +899,7 @@ static bool exec_process_in_job(parser_t &parser, process_t *p, std::shared_ptr<
     }
 
     // Execute the process.
+    p->check_generations_before_launch();
     switch (p->type) {
         case INTERNAL_FUNCTION:
         case INTERNAL_BLOCK_NODE: {

src/proc.cpp

@@ -324,7 +324,11 @@ static void handle_child_status(pid_t pid, int status) {
     }
 }
 
-process_t::process_t() {}
+process_t::process_t() = default;
+
+void process_t::check_generations_before_launch() {
+    gens_ = topic_monitor_t::principal().current_generations();
+}
 
 job_t::job_t(job_id_t jobid, io_chain_t bio, std::shared_ptr<job_t> parent)
     : block_io(std::move(bio)),

src/proc.h

@@ -113,6 +113,11 @@ class process_t {
 
     void set_io_chain(const io_chain_t &chain) { this->process_io_chain = chain; }
 
+    /// Store the current topic generations. That is, right before the process is launched, record
+    /// the generations of all topics; then we can tell which generation values have changed after
+    /// launch. This helps us avoid spurious waitpid calls.
+    void check_generations_before_launch();
+
     /// Actual command to pass to exec in case of EXTERNAL or INTERNAL_EXEC.
     wcstring actual_cmd;