bpf: cgroup: properly use bpf_prog_array api

Now that we don't have __rcu markers on the bpf_prog_array helpers,
let's use proper rcu_dereference_protected to obtain array pointer
under mutex.

We also don't need __rcu annotations on cgroup_bpf.inactive since
it's not read/updated concurrently.

v4:
* drop cgroup_rcu_xyz wrappers and use rcu APIs directly; presumably
  should be more clear to understand which mutex/refcount protects
  each particular place

v3:
* amend cgroup_rcu_dereference to include percpu_ref_is_dying;
  cgroup_bpf is now reference counted and we don't hold cgroup_mutex
  anymore in cgroup_bpf_release

v2:
* replace xchg with rcu_swap_protected

Cc: Roman Gushchin <guro@fb.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Acked-by: Roman Gushchin <guro@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
index 9f100fc..b631ee7 100644 (file)
--- a/include/linux/bpf-cgroup.h
+++ b/include/linux/bpf-cgroup.h
@@ -72,7 +72,7 @@ struct cgroup_bpf {
        u32 flags[MAX_BPF_ATTACH_TYPE];
 
        /* temp storage for effective prog array used by prog_attach/detach */
-       struct bpf_prog_array __rcu *inactive;
+       struct bpf_prog_array *inactive;
 
        /* reference counter used to detach bpf programs after cgroup removal */
        struct percpu_ref refcnt;

diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index d995edb..ff594eb 100644 (file)
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -38,6 +38,7 @@ static void cgroup_bpf_release(struct work_struct *work)
        struct cgroup *cgrp = container_of(work, struct cgroup,
                                           bpf.release_work);
        enum bpf_cgroup_storage_type stype;
+       struct bpf_prog_array *old_array;
        unsigned int type;
 
        for (type = 0; type < ARRAY_SIZE(cgrp->bpf.progs); type++) {
@@ -54,7 +55,10 @@ static void cgroup_bpf_release(struct work_struct *work)
                        kfree(pl);
                        static_branch_dec(&cgroup_bpf_enabled_key);
                }
-               bpf_prog_array_free(cgrp->bpf.effective[type]);
+               old_array = rcu_dereference_protected(
+                               cgrp->bpf.effective[type],
+                               percpu_ref_is_dying(&cgrp->bpf.refcnt));
+               bpf_prog_array_free(old_array);
        }
 
        percpu_ref_exit(&cgrp->bpf.refcnt);
@@ -126,7 +130,7 @@ static bool hierarchy_allows_attach(struct cgroup *cgrp,
  */
 static int compute_effective_progs(struct cgroup *cgrp,
                                   enum bpf_attach_type type,
-                                  struct bpf_prog_array __rcu **array)
+                                  struct bpf_prog_array **array)
 {
        enum bpf_cgroup_storage_type stype;
        struct bpf_prog_array *progs;
@@ -164,17 +168,16 @@ static int compute_effective_progs(struct cgroup *cgrp,
                }
        } while ((p = cgroup_parent(p)));
 
-       rcu_assign_pointer(*array, progs);
+       *array = progs;
        return 0;
 }
 
 static void activate_effective_progs(struct cgroup *cgrp,
                                     enum bpf_attach_type type,
-                                    struct bpf_prog_array __rcu *array)
+                                    struct bpf_prog_array *old_array)
 {
-       struct bpf_prog_array __rcu *old_array;
-
-       old_array = xchg(&cgrp->bpf.effective[type], array);
+       rcu_swap_protected(cgrp->bpf.effective[type], old_array,
+                          lockdep_is_held(&cgroup_mutex));
        /* free prog array after grace period, since __cgroup_bpf_run_*()
         * might be still walking the array
         */
@@ -191,7 +194,7 @@ int cgroup_bpf_inherit(struct cgroup *cgrp)
  * that array below is variable length
  */
 #define        NR ARRAY_SIZE(cgrp->bpf.effective)
-       struct bpf_prog_array __rcu *arrays[NR] = {};
+       struct bpf_prog_array *arrays[NR] = {};
        int ret, i;
 
        ret = percpu_ref_init(&cgrp->bpf.refcnt, cgroup_bpf_release_fn, 0,
@@ -477,10 +480,14 @@ int __cgroup_bpf_query(struct cgroup *cgrp, const union bpf_attr *attr,
        enum bpf_attach_type type = attr->query.attach_type;
        struct list_head *progs = &cgrp->bpf.progs[type];
        u32 flags = cgrp->bpf.flags[type];
+       struct bpf_prog_array *effective;
        int cnt, ret = 0, i;
 
+       effective = rcu_dereference_protected(cgrp->bpf.effective[type],
+                                             lockdep_is_held(&cgroup_mutex));
+
        if (attr->query.query_flags & BPF_F_QUERY_EFFECTIVE)
-               cnt = bpf_prog_array_length(cgrp->bpf.effective[type]);
+               cnt = bpf_prog_array_length(effective);
        else
                cnt = prog_list_length(progs);
 
@@ -497,8 +504,7 @@ int __cgroup_bpf_query(struct cgroup *cgrp, const union bpf_attr *attr,
        }
 
        if (attr->query.query_flags & BPF_F_QUERY_EFFECTIVE) {
-               return bpf_prog_array_copy_to_user(cgrp->bpf.effective[type],
-                                                  prog_ids, cnt);
+               return bpf_prog_array_copy_to_user(effective, prog_ids, cnt);
        } else {
                struct bpf_prog_list *pl;
                u32 id;