misc_cgroup: add support for nofile limit

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

@@ -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);
 }

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;

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
 };
 

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 */