[Devel] [PATCH RHEL8 COMMIT] pid_ns: Virtualize pid_max

[Konstantin Khorenko]

The commit is pushed to "branch-rh8-4.18.0-240.1.1.vz8.5.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh8-4.18.0-240.1.1.vz8.5.42
------>
commit 5d7309e62734d363b6c86ad00287d9c0bb49637f
Author: Kirill Tkhai <ktkhai at virtuozzo.com>
Date:   Thu Jun 10 14:52:33 2021 +0300

    pid_ns: Virtualize pid_max
    
    Introduce per pid namespace pid_max.
    
    Extracted from "Initial patch".
    
    Signed-off-by: Kirill Tkhai <ktkhai at virtuozzo.com>
    
    (cherry-picked from vz7 commit 80da36469321 ("pid_ns: Virtualize pid_max"))
    
    https://jira.sw.ru/browse/PSBM-127857
    Signed-off-by: Valeriy Vdovin <valeriy.vdovin at virtuozzo.com>
    
    +++
    pid: Use proper ns in proc_dointvec_pidmax()
    
    Use current task pid ns instead of pid ns for future children
    https://jira.sw.ru/browse/PSBM-121530
    Signed-off-by: Kirill Tkhai <ktkhai at virtuozzo.com>
    
    (cherry-picked from vz7 commit d91a0dc98453 ("pid: Use proper ns in
    proc_dointvec_pidmax()"))
    
    https://jira.sw.ru/browse/PSBM-127857
    Signed-off-by: Valeriy Vdovin <valeriy.vdovin at virtuozzo.com>
---
 include/linux/pid_namespace.h |  1 +
 include/linux/threads.h       |  1 +
 kernel/pid.c                  |  2 +-
 kernel/pid_namespace.c        |  4 ++--
 kernel/sysctl.c               | 15 ++++++++++++---
 5 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/include/linux/pid_namespace.h b/include/linux/pid_namespace.h
index 41a948c37e0f..4115f64f7df6 100644
--- a/include/linux/pid_namespace.h
+++ b/include/linux/pid_namespace.h
@@ -26,6 +26,7 @@ struct pid_namespace {
 	struct idr idr;
 	struct rcu_head rcu;
 	unsigned int pid_allocated;
+	int pid_max;
 	struct task_struct *child_reaper;
 	struct kmem_cache *pid_cachep;
 	unsigned int level;
diff --git a/include/linux/threads.h b/include/linux/threads.h
index 3086dba525e2..9b191fba4a4b 100644
--- a/include/linux/threads.h
+++ b/include/linux/threads.h
@@ -26,6 +26,7 @@
  * This controls the default maximum pid allocated to a process
  */
 #define PID_MAX_DEFAULT (CONFIG_BASE_SMALL ? 0x1000 : 0x8000)
+#define PID_MAX_NS_DEFAULT	(PID_MAX_DEFAULT)
 
 /*
  * A maximum of 4 million PIDs should be enough for a while.
diff --git a/kernel/pid.c b/kernel/pid.c
index 439e0068e4f9..e39cf422c875 100644
--- a/kernel/pid.c
+++ b/kernel/pid.c
@@ -190,7 +190,7 @@ struct pid *alloc_pid(struct pid_namespace *ns)
 		 * a partially initialized PID (see below).
 		 */
 		nr = idr_alloc_cyclic(&tmp->idr, NULL, pid_min,
-				      pid_max, GFP_ATOMIC);
+				      ns->pid_max, GFP_ATOMIC);
 		spin_unlock_irq(&pidmap_lock);
 		idr_preload_end();
 
diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c
index a25046a55dbd..c1a84b83a6f6 100644
--- a/kernel/pid_namespace.c
+++ b/kernel/pid_namespace.c
@@ -118,6 +118,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
 	ns->ucounts = ucounts;
 	ns->pid_allocated = PIDNS_ADDING;
 	INIT_WORK(&ns->proc_work, proc_cleanup_work);
+	ns->pid_max = PID_MAX_NS_DEFAULT;
 
 	return ns;
 
@@ -289,6 +290,7 @@ static int pid_ns_ctl_handler(struct ctl_table *table, int write,
 	next = idr_get_cursor(&pid_ns->idr) - 1;
 
 	tmp.data = &next;
+	tmp.extra2 = &pid_ns->pid_max;
 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
 	if (!ret && write)
 		idr_set_cursor(&pid_ns->idr, next + 1);
@@ -296,7 +298,6 @@ static int pid_ns_ctl_handler(struct ctl_table *table, int write,
 	return ret;
 }
 
-extern int pid_max;
 static struct ctl_table pid_ns_ctl_table[] = {
 	{
 		.procname = "ns_last_pid",
@@ -304,7 +305,6 @@ static struct ctl_table pid_ns_ctl_table[] = {
 		.mode = 0666 | S_ISVTX, /* permissions are checked in the handler */
 		.proc_handler = pid_ns_ctl_handler,
 		.extra1 = SYSCTL_ZERO,
-		.extra2 = &pid_max,
 	},
 	{ }
 };
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 7ec1f1e99431..7051209d7ec9 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -338,6 +338,16 @@ static int min_extfrag_threshold;
 static int max_extfrag_threshold = 1000;
 #endif
 
+static int proc_dointvec_pidmax(struct ctl_table *table, int write,
+		  void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	struct ctl_table tmp;
+
+	tmp = *table;
+	tmp.data = &task_active_pid_ns(current)->pid_max;
+	return proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
+}
+
 #ifdef CONFIG_COREDUMP
 sysctl_virtual(proc_dostring_coredump);
 #endif
@@ -853,10 +863,9 @@ static struct ctl_table kern_table[] = {
 #endif
 	{
 		.procname	= "pid_max",
-		.data		= &pid_max,
 		.maxlen		= sizeof (int),
-		.mode		= 0644,
-		.proc_handler	= proc_dointvec_minmax,
+		.mode		= 0644 | S_ISVTX,
+		.proc_handler	= proc_dointvec_pidmax,
 		.extra1		= &pid_max_min,
 		.extra2		= &pid_max_max,
 	},