[RFC PATCH 1/3] misc_cgroup: add support for nofile limit

[RFC PATCH 1/3] misc_cgroup: add support for nofile limit

[brookxu]

From: Chunguang Xu <brookxu@tencent.com>

Since the global open files are limited, in order to avoid the
abnormal behavior of some containers from generating too many
files, causing other containers to be unavailable, we need to
limit the open files of some containers.

Signed-off-by: Chunguang Xu <brookxu@tencent.com>
---
 fs/file_table.c             | 25 +++++++++++++++++++++++--
 include/linux/fs.h          |  4 +++-
 include/linux/misc_cgroup.h |  1 +
 kernel/cgroup/misc.c        |  1 +
 4 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/fs/file_table.c b/fs/file_table.c
index 45437f8e1003..a7848a4cac19 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -29,6 +29,7 @@
 #include <linux/swap.h>
 
 #include <linux/atomic.h>
+#include <linux/misc_cgroup.h>
 
 #include "internal.h"
 
@@ -53,8 +54,14 @@ static void file_free_rcu(struct rcu_head *head)
 static inline void file_free(struct file *f)
 {
 	security_file_free(f);
-	if (!(f->f_mode & FMODE_NOACCOUNT))
+	if (!(f->f_mode & FMODE_NOACCOUNT)) {
+		struct misc_cg *misc_cg = css_misc(f->f_css);
+
+		misc_cg_uncharge(MISC_CG_RES_NOFILE, misc_cg, 1);
+		put_misc_cg(misc_cg);
+
 		percpu_counter_dec(&nr_files);
+	}
 	call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
 }
 
@@ -148,8 +155,20 @@ struct file *alloc_empty_file(int flags, const struct cred *cred)
 	}
 
 	f = __alloc_file(flags, cred);
-	if (!IS_ERR(f))
+	if (!IS_ERR(f)) {
+		struct misc_cg *misc_cg = get_current_misc_cg();
+		int ret;
+
+		ret = misc_cg_try_charge(MISC_CG_RES_NOFILE, misc_cg, 1);
+		if (ret < 0) {
+			put_misc_cg(misc_cg);
+			file_free(f);
+			goto out;
+		}
+
 		percpu_counter_inc(&nr_files);
+		f->f_css = &misc_cg->css;
+	}
 
 	return f;
 
@@ -159,6 +178,7 @@ struct file *alloc_empty_file(int flags, const struct cred *cred)
 		pr_info("VFS: file-max limit %lu reached\n", get_max_files());
 		old_max = get_nr_files();
 	}
+ out:
 	return ERR_PTR(-ENFILE);
 }
 
@@ -397,4 +417,5 @@ void __init files_maxfiles_init(void)
 	n = ((nr_pages - memreserve) * (PAGE_SIZE / 1024)) / 10;
 
 	files_stat.max_files = max_t(unsigned long, n, NR_FILE);
+	misc_cg_set_capacity(MISC_CG_RES_NOFILE, files_stat.max_files);
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index fad6663cd1b0..9ef3dd579ed6 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -947,7 +947,9 @@ struct file {
 #endif
 	/* needed for tty driver, and maybe others */
 	void			*private_data;
-
+#ifdef CONFIG_CGROUP_MISC
+	struct cgroup_subsys_state *f_css;
+#endif
 #ifdef CONFIG_EPOLL
 	/* Used by fs/eventpoll.c to link all the hooks to this file */
 	struct hlist_head	*f_ep;
diff --git a/include/linux/misc_cgroup.h b/include/linux/misc_cgroup.h
index da2367e2ac1e..8450a5e66de0 100644
--- a/include/linux/misc_cgroup.h
+++ b/include/linux/misc_cgroup.h
@@ -18,6 +18,7 @@ enum misc_res_type {
 	/* AMD SEV-ES ASIDs resource */
 	MISC_CG_RES_SEV_ES,
 #endif
+	MISC_CG_RES_NOFILE,
 	MISC_CG_RES_TYPES
 };
 
diff --git a/kernel/cgroup/misc.c b/kernel/cgroup/misc.c
index ec02d963cad1..5d51b8eeece6 100644
--- a/kernel/cgroup/misc.c
+++ b/kernel/cgroup/misc.c
@@ -24,6 +24,7 @@ static const char *const misc_res_name[] = {
 	/* AMD SEV-ES ASIDs resource */
 	"sev_es",
 #endif
+	"nofile"
 };
 
 /* Root misc cgroup */
-- 
2.30.0

[RFC PATCH 2/3] misc_cgroup: add failcnt counter

[brookxu]

From: Chunguang Xu <brookxu@tencent.com>

Instead of printing logs, we should probably track failures through
a failcnt counter, similar to mem_cgroup.

Signed-off-by: Chunguang Xu <brookxu@tencent.com>
---
 include/linux/misc_cgroup.h |  1 +
 kernel/cgroup/misc.c        | 30 ++++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/include/linux/misc_cgroup.h b/include/linux/misc_cgroup.h
index 8450a5e66de0..dd1a786f39b8 100644
--- a/include/linux/misc_cgroup.h
+++ b/include/linux/misc_cgroup.h
@@ -37,6 +37,7 @@ struct misc_cg;
 struct misc_res {
 	unsigned long max;
 	atomic_long_t usage;
+	atomic_long_t failcnt;
 	bool failed;
 };
 
diff --git a/kernel/cgroup/misc.c b/kernel/cgroup/misc.c
index 5d51b8eeece6..7c568b619f82 100644
--- a/kernel/cgroup/misc.c
+++ b/kernel/cgroup/misc.c
@@ -165,6 +165,7 @@ int misc_cg_try_charge(enum misc_res_type type, struct misc_cg *cg,
 				pr_cont("\n");
 				res->failed = true;
 			}
+			atomic_long_inc(&res->failcnt);
 			ret = -EBUSY;
 			goto err_charge;
 		}
@@ -312,6 +313,29 @@ static int misc_cg_current_show(struct seq_file *sf, void *v)
 	return 0;
 }
 
+/**
+ * misc_cg_failcnt_show() - Show the fail count of the misc cgroup.
+ * @sf: Interface file
+ * @v: Arguments passed
+ *
+ * Context: Any context.
+ * Return: 0 to denote successful print.
+ */
+static int misc_cg_failcnt_show(struct seq_file *sf, void *v)
+{
+	int i;
+	unsigned long failcnt;
+	struct misc_cg *cg = css_misc(seq_css(sf));
+
+	for (i = 0; i < MISC_CG_RES_TYPES; i++) {
+		failcnt = atomic_long_read(&cg->res[i].failcnt);
+		if (READ_ONCE(misc_res_capacity[i]) || failcnt)
+			seq_printf(sf, "%s %lu\n", misc_res_name[i], failcnt);
+	}
+
+	return 0;
+}
+
 /**
  * misc_cg_capacity_show() - Show the total capacity of misc res on the host.
  * @sf: Interface file
@@ -349,6 +373,11 @@ static struct cftype misc_cg_files[] = {
 		.seq_show = misc_cg_current_show,
 		.flags = CFTYPE_NOT_ON_ROOT,
 	},
+	{
+		.name = "failcnt",
+		.seq_show = misc_cg_failcnt_show,
+		.flags = CFTYPE_NOT_ON_ROOT,
+	},
 	{
 		.name = "capacity",
 		.seq_show = misc_cg_capacity_show,
@@ -383,6 +412,7 @@ misc_cg_alloc(struct cgroup_subsys_state *parent_css)
 	for (i = 0; i < MISC_CG_RES_TYPES; i++) {
 		WRITE_ONCE(cg->res[i].max, MAX_NUM);
 		atomic_long_set(&cg->res[i].usage, 0);
+		atomic_long_set(&cg->res[i].failcnt, 0);
 	}
 
 	return &cg->css;
-- 
2.30.0

[RFC PATCH 3/3] misc_cgroup: delete failed logs to avoid log flooding

[brookxu]

From: Chunguang Xu <brookxu@tencent.com>

Since the upper-level logic will constantly retry when it fails, in
high-stress scenarios, a large number of failure logs may affect
performance. Therefore, we can replace it with the failcnt counter.

Signed-off-by: Chunguang Xu <brookxu@tencent.com>
---
 kernel/cgroup/misc.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/kernel/cgroup/misc.c b/kernel/cgroup/misc.c
index 7c568b619f82..b7de0fafa48a 100644
--- a/kernel/cgroup/misc.c
+++ b/kernel/cgroup/misc.c
@@ -159,8 +159,6 @@ int misc_cg_try_charge(enum misc_res_type type, struct misc_cg *cg,
 		if (new_usage > READ_ONCE(res->max) ||
 		    new_usage > READ_ONCE(misc_res_capacity[type])) {
 			if (!res->failed) {
-				pr_info("cgroup: charge rejected by the misc controller for %s resource in ",
-					misc_res_name[type]);
 				pr_cont_cgroup_path(i->css.cgroup);
 				pr_cont("\n");
 				res->failed = true;
-- 
2.30.0