a race in pid generation that causes pids to be reused immediatelly

Quoted 8 lines

Review

Quoted 4 lines

(quoted diff) +4 -0

+			if (lc > 0) {
+				printf("last_pid = %d\npid = %d\n", last_pid, pid);
+				distance = PidDistance(last_pid, pid);
+				if (distance == 0 || distance > 30000) {

Hmm are you sure that distance > 30000 is good thing to assert? Who says that two pids keeps at least 2768 distance and even then, who says that 32768 is maximal pid (see below)?

It was using a expert's advise here,
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=5fdee8c4a5e1800489ce61963208f8cc55e42ea1

Quoted 2 lines

Review

Quoted 1 line

(quoted diff) +1 -0

+	return 0;

Rather use tst_exit() here.

check() function is used in main() here,

TEST(check())
cleanup();

As you can see that the test has not finished yet after returned. It might also break existing LTP standard options like,

-c n Run n copies concurrently
-I x Execute test for x seconds

as TEST() seems expecting a non-void function.

Quoted 9 lines

Review

Quoted 7 lines

(quoted diff) +7 -0

+	/* capture signals */
+	tst_sig(FORK, DEF_HANDLER, cleanup);
+
+	/* Pause if that option was specified */
+	TEST_PAUSE;
+
+	tst_tmpdir();

Do you really need temporary directory to test fork() ?

Not really, but as mentioned earlier that those were copied from a standard LTP test template that I don't if some LTP internal things were using them. For example, LTP internal seems make use of tst_sig() and TEST_PAUSE in setup() for those standard options,

-p Pause for SIGUSR1 before starting
-P x Pause for x seconds between iterations

Therefore, I am concerning about remove those temporary directory creating might break other standard options and alike.

Quoted 8 lines

Review

Quoted 5 lines

(quoted diff) +5 -0

+/* The distance mod 32768 between two pids, where the first pid is
+   expected to be smaller than the second. */
+int PidDistance(pid_t first, pid_t second) {
+	return (second + 32768 - first) % 32768;
+}

The comment doesn't hold as when second is smaller than first the negative value would be smaller than 32768.

Again this is what the expert said. What would be your advise here?

Quoted 2 lines

Moreover the maximal pid value may be changed by writing to "/proc/sys/kernel/pid_max", then this code is broken.

One thing I can do here is to backup/restore /proc/sys/kernel/pid_max and set a value explicitly to avoid ambiguity.

Quoted 2 lines

Also I'm not sure if the result of such computation couldn't overflow/underflow as int may not be the same size as pid_t.

Is this still an issue after set pid_max a value explicitly like 32768? Thanks.