[PATCH 2/3] selftests/namespaces: Fix waitpid race in listns_efault_test cleanup

[Ricardo B. Marlière]

The efault tests spawn two categories of child processes: namespace
children (each in its own mount namespace, for concurrent destruction) and
an iterator child that calls listns() in a tight loop.  The cleanup loop
used waitpid(-1), which reaps any child in any order. If the iterator child
exits early (e.g. because listns() returned ENOSYS) before all namespace
children have been reaped, waitpid(-1) may consume it instead.  The
subsequent targeted waitpid(iter_pid) would then block indefinitely.

Track the PIDs of the namespace children explicitly and use targeted
waitpid() calls in the cleanup loop so the iterator child cannot be
inadvertently reaped during namespace cleanup.

Signed-off-by: Ricardo B. Marlière <rbm@xxxxxxxx>
---
 tools/testing/selftests/namespaces/listns_efault_test.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/tools/testing/selftests/namespaces/listns_efault_test.c b/tools/testing/selftests/namespaces/listns_efault_test.c
index c7ed4023d7a8..ac3b208264f5 100644
--- a/tools/testing/selftests/namespaces/listns_efault_test.c
+++ b/tools/testing/selftests/namespaces/listns_efault_test.c
@@ -39,7 +39,7 @@ TEST(listns_partial_fault_with_ns_cleanup)
 	__u64 *ns_ids;
 	ssize_t ret;
 	long page_size;
-	pid_t pid, iter_pid;
+	pid_t pid, iter_pid, ns_pids[5];
 	int pidfds[5];
 	int sv[5][2];
 	int iter_pidfd;
@@ -115,6 +115,7 @@ TEST(listns_partial_fault_with_ns_cleanup)
 
 		pid = create_child(&pidfds[i], CLONE_NEWNS);
 		ASSERT_NE(pid, -1);
+		ns_pids[i] = pid;
 
 		if (pid == 0) {
 			close(sv[i][0]); /* Close parent end */
@@ -165,7 +166,7 @@ TEST(listns_partial_fault_with_ns_cleanup)
 
 	/* Wait for all mount namespace children to exit and cleanup */
 	for (i = 0; i < 5; i++) {
-		waitpid(-1, NULL, 0);
+		waitpid(ns_pids[i], NULL, 0);
 		close(sv[i][0]);
 		close(pidfds[i]);
 	}
@@ -251,7 +252,7 @@ TEST(listns_late_fault_with_ns_cleanup)
 	__u64 *ns_ids;
 	ssize_t ret;
 	long page_size;
-	pid_t pid, iter_pid;
+	pid_t pid, iter_pid, ns_pids[10];
 	int pidfds[10];
 	int sv[10][2];
 	int iter_pidfd;
@@ -321,6 +322,7 @@ TEST(listns_late_fault_with_ns_cleanup)
 
 		pid = create_child(&pidfds[i], CLONE_NEWNS);
 		ASSERT_NE(pid, -1);
+		ns_pids[i] = pid;
 
 		if (pid == 0) {
 			close(sv[i][0]); /* Close parent end */
@@ -374,7 +376,7 @@ TEST(listns_late_fault_with_ns_cleanup)
 
 	/* Wait for all children and cleanup */
 	for (i = 0; i < 10; i++) {
-		waitpid(-1, NULL, 0);
+		waitpid(ns_pids[i], NULL, 0);
 		close(sv[i][0]);
 		close(pidfds[i]);
 	}
@@ -403,7 +405,7 @@ TEST(listns_mnt_ns_cleanup_on_fault)
 	__u64 *ns_ids;
 	ssize_t ret;
 	long page_size;
-	pid_t pid, iter_pid;
+	pid_t pid, iter_pid, ns_pids[8];
 	int pidfds[8];
 	int sv[8][2];
 	int iter_pidfd;
@@ -463,6 +465,7 @@ TEST(listns_mnt_ns_cleanup_on_fault)
 
 		pid = create_child(&pidfds[i], CLONE_NEWNS);
 		ASSERT_NE(pid, -1);
+		ns_pids[i] = pid;
 
 		if (pid == 0) {
 			close(sv[i][0]); /* Close parent end */
@@ -509,7 +512,7 @@ TEST(listns_mnt_ns_cleanup_on_fault)
 
 	/* Wait for children and cleanup */
 	for (i = 0; i < 8; i++) {
-		waitpid(-1, NULL, 0);
+		waitpid(ns_pids[i], NULL, 0);
 		close(sv[i][0]);
 		close(pidfds[i]);
 	}

-- 
2.53.0