It is necessary to have a different pid_max in different containers.
For example, multiple containers are running on a host, one of which
is Android, and its 32 bit bionic libc only accepts pid <= 65535. So
it requires the global pid_max <= 65535. This will cause configuration
conflicts with other containers and also limit the maximum number of
tasks for the entire system.

Signed-off-by: Yun Zhou <yun.zhou@windriver.com>
---
 include/linux/pid_namespace.h     |  1 +
 kernel/pid.c                      | 12 +++++------
 kernel/pid_namespace.c            | 34 ++++++++++++++++++++++++++-----
 kernel/sysctl.c                   |  9 --------
 kernel/trace/pid_list.c           |  2 +-
 kernel/trace/trace.h              |  2 --
 kernel/trace/trace_sched_switch.c |  2 +-
 7 files changed, 38 insertions(+), 24 deletions(-)

diff --git a/include/linux/pid_namespace.h b/include/linux/pid_namespace.h
index f9f9931e02d6a..064cfe2542fc5 100644
--- a/include/linux/pid_namespace.h
+++ b/include/linux/pid_namespace.h
@@ -38,6 +38,7 @@ struct pid_namespace {
 	struct ucounts *ucounts;
 	int reboot;	/* group exit code if this pidns was rebooted */
 	struct ns_common ns;
+	int pid_max;
 #if defined(CONFIG_SYSCTL) && defined(CONFIG_MEMFD_CREATE)
 	int memfd_noexec_scope;
 #endif
diff --git a/kernel/pid.c b/kernel/pid.c
index 2715afb77eab8..f8026a61436bd 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -60,8 +60,6 @@ struct pid init_struct_pid = {
 	}, }
 };
 
-int pid_max = PID_MAX_DEFAULT;
-
 int pid_max_min = RESERVED_PIDS + 1;
 int pid_max_max = PID_MAX_LIMIT;
 /*
@@ -78,6 +76,7 @@ static u64 pidfs_ino = RESERVED_PIDS;
  */
 struct pid_namespace init_pid_ns = {
 	.ns.count = REFCOUNT_INIT(2),
+	.pid_max = PID_MAX_DEFAULT,
 	.idr = IDR_INIT(init_pid_ns.idr),
 	.pid_allocated = PIDNS_ADDING,
 	.level = 0,
@@ -198,7 +197,7 @@ struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid,
 			tid = set_tid[ns->level - i];
 
 			retval = -EINVAL;
-			if (tid < 1 || tid >= pid_max)
+			if (tid < 1 || tid >= tmp->pid_max)
 				goto out_free;
 			/*
 			 * Also fail if a PID != 1 is requested and
@@ -238,7 +237,7 @@ struct pid *alloc_pid(struct pid_namespace *ns, pid_t *set_tid,
 			 * a partially initialized PID (see below).
 			 */
 			nr = idr_alloc_cyclic(&tmp->idr, NULL, pid_min,
-					      pid_max, GFP_ATOMIC);
+					      tmp->pid_max, GFP_ATOMIC);
 		}
 		spin_unlock_irq(&pidmap_lock);
 		idr_preload_end();
@@ -653,11 +652,12 @@ void __init pid_idr_init(void)
 	BUILD_BUG_ON(PID_MAX_LIMIT >= PIDNS_ADDING);
 
 	/* bump default and minimum pid_max based on number of cpus */
-	pid_max = min(pid_max_max, max_t(int, pid_max,
+	init_pid_ns.pid_max = min(pid_max_max, max_t(int, init_pid_ns.pid_max,
 				PIDS_PER_CPU_DEFAULT * num_possible_cpus()));
 	pid_max_min = max_t(int, pid_max_min,
 				PIDS_PER_CPU_MIN * num_possible_cpus());
-	pr_info("pid_max: default: %u minimum: %u\n", pid_max, pid_max_min);
+	pr_info("pid_max: default: %u minimum: %u\n", init_pid_ns.pid_max,
+			pid_max_min);
 
 	idr_init(&init_pid_ns.idr);
 
diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c
index d70ab49d5b4a6..a5a8254825d55 100644
--- a/kernel/pid_namespace.c
+++ b/kernel/pid_namespace.c
@@ -111,6 +111,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
 	ns->user_ns = get_user_ns(user_ns);
 	ns->ucounts = ucounts;
 	ns->pid_allocated = PIDNS_ADDING;
+	ns->pid_max = parent_pid_ns->pid_max;
 #if defined(CONFIG_SYSCTL) && defined(CONFIG_MEMFD_CREATE)
 	ns->memfd_noexec_scope = pidns_memfd_noexec_scope(parent_pid_ns);
 #endif
@@ -280,19 +281,44 @@ static int pid_ns_ctl_handler(const struct ctl_table *table, int write,
 
 	return ret;
 }
+#endif	/* CONFIG_CHECKPOINT_RESTORE */
+
+static int pid_max_ns_ctl_handler(const struct ctl_table *table, int write,
+		void *buffer, size_t *lenp, loff_t *ppos)
+{
+	struct pid_namespace *pid_ns = task_active_pid_ns(current);
+	struct ctl_table tmp = *table;
+
+	if (write && !checkpoint_restore_ns_capable(pid_ns->user_ns))
+		return -EPERM;
+
+	tmp.data = &pid_ns->pid_max;
+	if (pid_ns->parent)
+		tmp.extra2 = &pid_ns->parent->pid_max;
+
+	return proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
+}
 
-extern int pid_max;
 static struct ctl_table pid_ns_ctl_table[] = {
+#ifdef CONFIG_CHECKPOINT_RESTORE
 	{
 		.procname = "ns_last_pid",
 		.maxlen = sizeof(int),
 		.mode = 0666, /* permissions are checked in the handler */
 		.proc_handler = pid_ns_ctl_handler,
 		.extra1 = SYSCTL_ZERO,
-		.extra2 = &pid_max,
+		.extra2 = &init_pid_ns.pid_max,
 	},
-};
 #endif	/* CONFIG_CHECKPOINT_RESTORE */
+	{
+		.procname = "pid_max",
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = pid_max_ns_ctl_handler,
+		.extra1 = &pid_max_min,
+		.extra2 = &pid_max_max,
+	},
+};
 
 int reboot_pid_ns(struct pid_namespace *pid_ns, int cmd)
 {
@@ -449,9 +475,7 @@ static __init int pid_namespaces_init(void)
 {
 	pid_ns_cachep = KMEM_CACHE(pid_namespace, SLAB_PANIC | SLAB_ACCOUNT);
 
-#ifdef CONFIG_CHECKPOINT_RESTORE
 	register_sysctl_init("kernel", pid_ns_ctl_table);
-#endif
 
 	register_pid_ns_sysctl_table_vm();
 	return 0;
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 79e6cb1d5c48f..676a0d675e7f8 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -1804,15 +1804,6 @@ static struct ctl_table kern_table[] = {
 		.proc_handler	= proc_dointvec,
 	},
 #endif
-	{
-		.procname	= "pid_max",
-		.data		= &pid_max,
-		.maxlen		= sizeof (int),
-		.mode		= 0644,
-		.proc_handler	= proc_dointvec_minmax,
-		.extra1		= &pid_max_min,
-		.extra2		= &pid_max_max,
-	},
 	{
 		.procname	= "panic_on_oops",
 		.data		= &panic_on_oops,
diff --git a/kernel/trace/pid_list.c b/kernel/trace/pid_list.c
index 4966e6bbdf6f3..c62b9b3cfb3d8 100644
--- a/kernel/trace/pid_list.c
+++ b/kernel/trace/pid_list.c
@@ -414,7 +414,7 @@ struct trace_pid_list *trace_pid_list_alloc(void)
 	int i;
 
 	/* According to linux/thread.h, pids can be no bigger that 30 bits */
-	WARN_ON_ONCE(pid_max > (1 << 30));
+	WARN_ON_ONCE(init_pid_ns.pid_max > (1 << 30));
 
 	pid_list = kzalloc(sizeof(*pid_list), GFP_KERNEL);
 	if (!pid_list)
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index c866991b9c78b..e51851d64e4d9 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -715,8 +715,6 @@ extern unsigned long tracing_thresh;
 
 /* PID filtering */
 
-extern int pid_max;
-
 bool trace_find_filtered_pid(struct trace_pid_list *filtered_pids,
 			     pid_t search_pid);
 bool trace_ignore_this_task(struct trace_pid_list *filtered_pids,
diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c
index 8a407adb0e1c4..c20c80abe065b 100644
--- a/kernel/trace/trace_sched_switch.c
+++ b/kernel/trace/trace_sched_switch.c
@@ -442,7 +442,7 @@ int trace_alloc_tgid_map(void)
 	if (tgid_map)
 		return 0;
 
-	tgid_map_max = pid_max;
+	tgid_map_max = init_pid_ns.pid_max;
 	map = kvcalloc(tgid_map_max + 1, sizeof(*tgid_map),
 		       GFP_KERNEL);
 	if (!map)
-- 
2.27.0

On Mon, 4 Nov 2024 08:54:08 +0800
Yun Zhou <yun.zhou@windriver.com> wrote:

> It is necessary to have a different pid_max in different containers.
> For example, multiple containers are running on a host, one of which
> is Android, and its 32 bit bionic libc only accepts pid <= 65535. So
> it requires the global pid_max <= 65535. This will cause configuration
> conflicts with other containers and also limit the maximum number of
> tasks for the entire system.
> 
> Signed-off-by: Yun Zhou <yun.zhou@windriver.com>

I acked the first patch. You don't mention what's different about this one
(which is something you should do below the '---' line and above the diffstat.

Like:

Changes since v1: https://lore.kernel.org/all/20241030052933.1041408-1-yun.zhou@windriver.com/

- Describe what is different.

What changed? Is it big enough for me to re-review this one so that my ack
needs to be reevaluated?

-- Steve