Refactor maybe_switch_tcbs

Change maybe_switch_tcbs to return NULL when no switching is necessary.
Introduce maybe_switch_current_tcp as a thing wrapper around
maybe_switch_tcbs.

* strace.c (maybe_switch_current_tcp): New function.
(dispatch_event): Use it instead of maybe_switch_tcbs, move comments
and the os_release check before maybe_switch_tcbs invocation ...
(maybe_switch_tcbs): ... here.  Change return value to NULL
if no switching was performed.

diff --git a/strace.c b/strace.c
index f7dfdaa504e919e54e4d05c0b052e0deb6a7b960..d4c1fa878bc89c28af3ad7392badd932c569c249 100644 (file)
--- a/strace.c
+++ b/strace.c
@@ -2029,33 +2029,48 @@ maybe_allocate_tcb(const int pid, int status)
        }
 }
 
+/*
+ * Under Linux, execve changes pid to thread leader's pid, and we see this
+ * changed pid on EVENT_EXEC and later, execve sysexit.  Leader "disappears"
+ * without exit notification.  Let user know that, drop leader's tcb, and fix
+ * up pid in execve thread's tcb.  Effectively, execve thread's tcb replaces
+ * leader's tcb.
+ *
+ * BTW, leader is 'stuck undead' (doesn't report WIFEXITED on exit syscall)
+ * in multi-threaded programs exactly in order to handle this case.
+ */
 static struct tcb *
 maybe_switch_tcbs(struct tcb *tcp, const int pid)
 {
-       FILE *fp;
-       struct tcb *execve_thread;
-       long old_pid = tcb_wait_tab[tcp->wait_data_idx].msg;
+       /*
+        * PTRACE_GETEVENTMSG returns old pid starting from Linux 3.0.
+        * On 2.6 and earlier it can return garbage.
+        */
+       if (os_release < KERNEL_VERSION(3, 0, 0))
+               return NULL;
+
+       const long old_pid = tcb_wait_tab[tcp->wait_data_idx].msg;
 
        /* Avoid truncation in pid2tcb() param passing */
        if (old_pid <= 0 || old_pid == pid)
-               return tcp;
+               return NULL;
        if ((unsigned long) old_pid > UINT_MAX)
-               return tcp;
-       execve_thread = pid2tcb(old_pid);
+               return NULL;
+       struct tcb *execve_thread = pid2tcb(old_pid);
        /* It should be !NULL, but I feel paranoid */
        if (!execve_thread)
-               return tcp;
+               return NULL;
 
        if (execve_thread->curcol != 0) {
                /*
-                * One case we are here is -ff:
-                * try "strace -oLOG -ff test/threaded_execve"
+                * One case we are here is -ff, try
+                * "strace -oLOG -ff test/threaded_execve".
                 */
                fprintf(execve_thread->outf, " <pid changed to %d ...>\n", pid);
                /*execve_thread->curcol = 0; - no need, see code below */
        }
        /* Swap output FILEs (needed for -ff) */
-       fp = execve_thread->outf;
+       FILE *fp = execve_thread->outf;
        execve_thread->outf = tcp->outf;
        tcp->outf = fp;
        /* And their column positions */
@@ -2076,6 +2091,17 @@ maybe_switch_tcbs(struct tcb *tcp, const int pid)
        return tcp;
 }
 
+static struct tcb *
+maybe_switch_current_tcp(void)
+{
+       struct tcb *tcp = maybe_switch_tcbs(current_tcp, current_tcp->pid);
+
+       if (tcp)
+               set_current_tcp(tcp);
+
+       return tcp;
+}
+
 static void
 print_signalled(struct tcb *tcp, const int pid, int status)
 {
@@ -2636,24 +2662,7 @@ dispatch_event(const struct tcb_wait_data *wd)
                        }
                }
 
-               /*
-                * Under Linux, execve changes pid to thread leader's pid,
-                * and we see this changed pid on EVENT_EXEC and later,
-                * execve sysexit. Leader "disappears" without exit
-                * notification. Let user know that, drop leader's tcb,
-                * and fix up pid in execve thread's tcb.
-                * Effectively, execve thread's tcb replaces leader's tcb.
-                *
-                * BTW, leader is 'stuck undead' (doesn't report WIFEXITED
-                * on exit syscall) in multithreaded programs exactly
-                * in order to handle this case.
-                *
-                * PTRACE_GETEVENTMSG returns old pid starting from Linux 3.0.
-                * On 2.6 and earlier, it can return garbage.
-                */
-               if (os_release >= KERNEL_VERSION(3, 0, 0))
-                       set_current_tcp(maybe_switch_tcbs(current_tcp,
-                                                         current_tcp->pid));
+               maybe_switch_current_tcp();
 
                if (detach_on_execve) {
                        if (current_tcp->flags & TCB_SKIP_DETACH_ON_FIRST_EXEC) {