The patch below does not apply to the 4.9-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@xxxxxxxxxxxxxxx>.
Possible dependencies:
4a7ba45b1a43 ("memcg: fix possible use-after-free in memcg_write_event_control()")
0a268dbd7932 ("cgroup: move cgroup v1 specific code to kernel/cgroup/cgroup-v1.c")
201af4c0fab0 ("cgroup: move cgroup files under kernel/cgroup/")
5f617ebbdf10 ("cgroup: reorder css_set fields")
2fae98634334 ("cgroup: remove cgroup_pid_fry() and friends")
b4b90a8e86f2 ("cgroup: reimplement reading "cgroup.procs" on cgroup v2")
3007098494be ("cgroup: add support for eBPF programs")
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
>From 4a7ba45b1a435e7097ca0f79a847d0949d0eb088 Mon Sep 17 00:00:00 2001
From: Tejun Heo <tj@xxxxxxxxxx>
Date: Wed, 7 Dec 2022 16:53:15 -1000
Subject: [PATCH] memcg: fix possible use-after-free in
memcg_write_event_control()
memcg_write_event_control() accesses the dentry->d_name of the specified
control fd to route the write call. As a cgroup interface file can't be
renamed, it's safe to access d_name as long as the specified file is a
regular cgroup file. Also, as these cgroup interface files can't be
removed before the directory, it's safe to access the parent too.
Prior to 347c4a874710 ("memcg: remove cgroup_event->cft"), there was a
call to __file_cft() which verified that the specified file is a regular
cgroupfs file before further accesses. The cftype pointer returned from
__file_cft() was no longer necessary and the commit inadvertently dropped
the file type check with it allowing any file to slip through. With the
invarients broken, the d_name and parent accesses can now race against
renames and removals of arbitrary files and cause use-after-free's.
Fix the bug by resurrecting the file type check in __file_cft(). Now that
cgroupfs is implemented through kernfs, checking the file operations needs
to go through a layer of indirection. Instead, let's check the superblock
and dentry type.
Link: https://lkml.kernel.org/r/Y5FRm/cfcKPGzWwl@xxxxxxxxxxxxxxx
Fixes: 347c4a874710 ("memcg: remove cgroup_event->cft")
Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
Reported-by: Jann Horn <jannh@xxxxxxxxxx>
Acked-by: Roman Gushchin <roman.gushchin@xxxxxxxxx>
Acked-by: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxxxxx>
Cc: Muchun Song <songmuchun@xxxxxxxxxxxxx>
Cc: Shakeel Butt <shakeelb@xxxxxxxxxx>
Cc: <stable@xxxxxxxxxxxxxxx> [3.14+]
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 528bd44b59e2..2b7d077de7ef 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -68,6 +68,7 @@ struct css_task_iter {
struct list_head iters_node; /* css_set->task_iters */
};
+extern struct file_system_type cgroup_fs_type;
extern struct cgroup_root cgrp_dfl_root;
extern struct css_set init_css_set;
diff --git a/kernel/cgroup/cgroup-internal.h b/kernel/cgroup/cgroup-internal.h
index fd4020835ec6..367b0a42ada9 100644
--- a/kernel/cgroup/cgroup-internal.h
+++ b/kernel/cgroup/cgroup-internal.h
@@ -167,7 +167,6 @@ struct cgroup_mgctx {
extern spinlock_t css_set_lock;
extern struct cgroup_subsys *cgroup_subsys[];
extern struct list_head cgroup_roots;
-extern struct file_system_type cgroup_fs_type;
/* iterate across the hierarchies */
#define for_each_root(root) \
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index a1a35c12635e..266a1ab05434 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4832,6 +4832,7 @@ static ssize_t memcg_write_event_control(struct kernfs_open_file *of,
unsigned int efd, cfd;
struct fd efile;
struct fd cfile;
+ struct dentry *cdentry;
const char *name;
char *endp;
int ret;
@@ -4885,6 +4886,16 @@ static ssize_t memcg_write_event_control(struct kernfs_open_file *of,
if (ret < 0)
goto out_put_cfile;
+ /*
+ * The control file must be a regular cgroup1 file. As a regular cgroup
+ * file can't be renamed, it's safe to access its name afterwards.
+ */
+ cdentry = cfile.file->f_path.dentry;
+ if (cdentry->d_sb->s_type != &cgroup_fs_type || !d_is_reg(cdentry)) {
+ ret = -EINVAL;
+ goto out_put_cfile;
+ }
+
/*
* Determine the event callbacks and set them in @event. This used
* to be done via struct cftype but cgroup core no longer knows
@@ -4893,7 +4904,7 @@ static ssize_t memcg_write_event_control(struct kernfs_open_file *of,
*
* DO NOT ADD NEW FILES.
*/
- name = cfile.file->f_path.dentry->d_name.name;
+ name = cdentry->d_name.name;
if (!strcmp(name, "memory.usage_in_bytes")) {
event->register_event = mem_cgroup_usage_register_event;
@@ -4917,7 +4928,7 @@ static ssize_t memcg_write_event_control(struct kernfs_open_file *of,
* automatically removed on cgroup destruction but the removal is
* asynchronous, so take an extra ref on @css.
*/
- cfile_css = css_tryget_online_from_dir(cfile.file->f_path.dentry->d_parent,
+ cfile_css = css_tryget_online_from_dir(cdentry->d_parent,
&memory_cgrp_subsys);
ret = -EINVAL;
if (IS_ERR(cfile_css))
