openkylin
/
platform_kernel-5.15
mirror of https://gitee.com/openkylin/platform_kernel-5.15.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity

bpf, netns: Keep attached programs in bpf_prog_array 

Prepare for having multi-prog attachments for new netns attach types by
storing programs to run in a bpf_prog_array, which is well suited for
iterating over programs and running them in sequence.

After this change bpf(PROG_QUERY) may block to allocate memory in
bpf_prog_array_copy_to_user() for collected program IDs. This forces a
change in how we protect access to the attached program in the query
callback. Because bpf_prog_array_copy_to_user() can sleep, we switch from
an RCU read lock to holding a mutex that serializes updaters.

Because we allow only one BPF flow_dissector program to be attached to
netns at all times, the bpf_prog_array pointed by net->bpf.run_array is
always either detached (null) or one element long.

No functional changes intended.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200625141357.910330-3-jakub@cloudflare.com
This commit is contained in:
Jakub Sitnicki 2020-06-25 16:13:55 +02:00 committed by Alexei Starovoitov
parent 3b7016996c
commit 695c12147a
3 changed files with 96 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
include/net/netns/bpf.h
View File

@ -9,9 +9,12 @@
#include <linux/bpf-netns.h> #include <linux/bpf-netns.h>
struct bpf_prog; struct bpf_prog;
struct bpf_prog_array;
struct netns_bpf { struct netns_bpf {
struct bpf_prog __rcu *progs[MAX_NETNS_BPF_ATTACH_TYPE]; /* Array of programs to run compiled from progs or links */
struct bpf_prog_array __rcu *run_array[MAX_NETNS_BPF_ATTACH_TYPE];
struct bpf_prog *progs[MAX_NETNS_BPF_ATTACH_TYPE];
struct bpf_link *links[MAX_NETNS_BPF_ATTACH_TYPE]; struct bpf_link *links[MAX_NETNS_BPF_ATTACH_TYPE];
}; };

120
kernel/bpf/net_namespace.c
View File

@ -33,6 +33,17 @@ static void __net_exit bpf_netns_link_auto_detach(struct bpf_link *link)
net_link->net = NULL; net_link->net = NULL;
} }
/* Must be called with netns_bpf_mutex held. */
static void netns_bpf_run_array_detach(struct net *net,
enum netns_bpf_attach_type type)
{
struct bpf_prog_array *run_array;
run_array = rcu_replace_pointer(net->bpf.run_array[type], NULL,
lockdep_is_held(&netns_bpf_mutex));
bpf_prog_array_free(run_array);
}
static void bpf_netns_link_release(struct bpf_link *link) static void bpf_netns_link_release(struct bpf_link *link)
{ {
struct bpf_netns_link *net_link = struct bpf_netns_link *net_link =
@ -54,8 +65,8 @@ static void bpf_netns_link_release(struct bpf_link *link)
if (!net) if (!net)
goto out_unlock; goto out_unlock;
netns_bpf_run_array_detach(net, type);
net->bpf.links[type] = NULL; net->bpf.links[type] = NULL;
RCU_INIT_POINTER(net->bpf.progs[type], NULL);
out_unlock: out_unlock:
mutex_unlock(&netns_bpf_mutex); mutex_unlock(&netns_bpf_mutex);
@ -76,6 +87,7 @@ static int bpf_netns_link_update_prog(struct bpf_link *link,
struct bpf_netns_link *net_link = struct bpf_netns_link *net_link =
container_of(link, struct bpf_netns_link, link); container_of(link, struct bpf_netns_link, link);
enum netns_bpf_attach_type type = net_link->netns_type; enum netns_bpf_attach_type type = net_link->netns_type;
struct bpf_prog_array *run_array;
struct net *net; struct net *net;
int ret = 0; int ret = 0;
@ -93,8 +105,11 @@ static int bpf_netns_link_update_prog(struct bpf_link *link,
goto out_unlock; goto out_unlock;
} }
run_array = rcu_dereference_protected(net->bpf.run_array[type],
lockdep_is_held(&netns_bpf_mutex));
WRITE_ONCE(run_array->items[0].prog, new_prog);
old_prog = xchg(&link->prog, new_prog); old_prog = xchg(&link->prog, new_prog);
rcu_assign_pointer(net->bpf.progs[type], new_prog);
bpf_prog_put(old_prog); bpf_prog_put(old_prog);
out_unlock: out_unlock:
@ -142,14 +157,38 @@ static const struct bpf_link_ops bpf_netns_link_ops = {
.show_fdinfo = bpf_netns_link_show_fdinfo, .show_fdinfo = bpf_netns_link_show_fdinfo,
}; };
/* Must be called with netns_bpf_mutex held. */
static int __netns_bpf_prog_query(const union bpf_attr *attr,
union bpf_attr __user *uattr,
struct net *net,
enum netns_bpf_attach_type type)
{
__u32 __user *prog_ids = u64_to_user_ptr(attr->query.prog_ids);
struct bpf_prog_array *run_array;
u32 prog_cnt = 0, flags = 0;
run_array = rcu_dereference_protected(net->bpf.run_array[type],
lockdep_is_held(&netns_bpf_mutex));
if (run_array)
prog_cnt = bpf_prog_array_length(run_array);
if (copy_to_user(&uattr->query.attach_flags, &flags, sizeof(flags)))
return -EFAULT;
if (copy_to_user(&uattr->query.prog_cnt, &prog_cnt, sizeof(prog_cnt)))
return -EFAULT;
if (!attr->query.prog_cnt || !prog_ids || !prog_cnt)
return 0;
return bpf_prog_array_copy_to_user(run_array, prog_ids,
attr->query.prog_cnt);
}
int netns_bpf_prog_query(const union bpf_attr *attr, int netns_bpf_prog_query(const union bpf_attr *attr,
union bpf_attr __user *uattr) union bpf_attr __user *uattr)
{ {
__u32 __user *prog_ids = u64_to_user_ptr(attr->query.prog_ids);
u32 prog_id, prog_cnt = 0, flags = 0;
enum netns_bpf_attach_type type; enum netns_bpf_attach_type type;
struct bpf_prog *attached;
struct net *net; struct net *net;
int ret;
if (attr->query.query_flags) if (attr->query.query_flags)
return -EINVAL; return -EINVAL;
@ -162,32 +201,17 @@ int netns_bpf_prog_query(const union bpf_attr *attr,
if (IS_ERR(net)) if (IS_ERR(net))
return PTR_ERR(net); return PTR_ERR(net);
rcu_read_lock(); mutex_lock(&netns_bpf_mutex);
attached = rcu_dereference(net->bpf.progs[type]); ret = __netns_bpf_prog_query(attr, uattr, net, type);
if (attached) { mutex_unlock(&netns_bpf_mutex);
prog_cnt = 1;
prog_id = attached->aux->id;
}
rcu_read_unlock();
put_net(net); put_net(net);
return ret;
if (copy_to_user(&uattr->query.attach_flags, &flags, sizeof(flags)))
return -EFAULT;
if (copy_to_user(&uattr->query.prog_cnt, &prog_cnt, sizeof(prog_cnt)))
return -EFAULT;
if (!attr->query.prog_cnt || !prog_ids || !prog_cnt)
return 0;
if (copy_to_user(prog_ids, &prog_id, sizeof(u32)))
return -EFAULT;
return 0;
} }
int netns_bpf_prog_attach(const union bpf_attr *attr, struct bpf_prog *prog) int netns_bpf_prog_attach(const union bpf_attr *attr, struct bpf_prog *prog)
{ {
struct bpf_prog_array *run_array;
enum netns_bpf_attach_type type; enum netns_bpf_attach_type type;
struct bpf_prog *attached; struct bpf_prog *attached;
struct net *net; struct net *net;
@ -217,14 +241,28 @@ int netns_bpf_prog_attach(const union bpf_attr *attr, struct bpf_prog *prog)
if (ret) if (ret)
goto out_unlock; goto out_unlock;
attached = rcu_dereference_protected(net->bpf.progs[type], attached = net->bpf.progs[type];
lockdep_is_held(&netns_bpf_mutex));
if (attached == prog) { if (attached == prog) {
/* The same program cannot be attached twice */ /* The same program cannot be attached twice */
ret = -EINVAL; ret = -EINVAL;
goto out_unlock; goto out_unlock;
} }
rcu_assign_pointer(net->bpf.progs[type], prog);
run_array = rcu_dereference_protected(net->bpf.run_array[type],
lockdep_is_held(&netns_bpf_mutex));
if (run_array) {
WRITE_ONCE(run_array->items[0].prog, prog);
} else {
run_array = bpf_prog_array_alloc(1, GFP_KERNEL);
if (!run_array) {
ret = -ENOMEM;
goto out_unlock;
}
run_array->items[0].prog = prog;
rcu_assign_pointer(net->bpf.run_array[type], run_array);
}
net->bpf.progs[type] = prog;
if (attached) if (attached)
bpf_prog_put(attached); bpf_prog_put(attached);
@ -244,11 +282,11 @@ static int __netns_bpf_prog_detach(struct net *net,
if (net->bpf.links[type]) if (net->bpf.links[type])
return -EINVAL; return -EINVAL;
attached = rcu_dereference_protected(net->bpf.progs[type], attached = net->bpf.progs[type];
lockdep_is_held(&netns_bpf_mutex));
if (!attached) if (!attached)
return -ENOENT; return -ENOENT;
RCU_INIT_POINTER(net->bpf.progs[type], NULL); netns_bpf_run_array_detach(net, type);
net->bpf.progs[type] = NULL;
bpf_prog_put(attached); bpf_prog_put(attached);
return 0; return 0;
} }
@ -272,7 +310,7 @@ int netns_bpf_prog_detach(const union bpf_attr *attr)
static int netns_bpf_link_attach(struct net *net, struct bpf_link *link, static int netns_bpf_link_attach(struct net *net, struct bpf_link *link,
enum netns_bpf_attach_type type) enum netns_bpf_attach_type type)
{ {
struct bpf_prog *prog; struct bpf_prog_array *run_array;
int err; int err;
mutex_lock(&netns_bpf_mutex); mutex_lock(&netns_bpf_mutex);
@ -283,9 +321,7 @@ static int netns_bpf_link_attach(struct net *net, struct bpf_link *link,
goto out_unlock; goto out_unlock;
} }
/* Links are not compatible with attaching prog directly */ /* Links are not compatible with attaching prog directly */
prog = rcu_dereference_protected(net->bpf.progs[type], if (net->bpf.progs[type]) {
lockdep_is_held(&netns_bpf_mutex));
if (prog) {
err = -EEXIST; err = -EEXIST;
goto out_unlock; goto out_unlock;
} }
@ -301,7 +337,14 @@ static int netns_bpf_link_attach(struct net *net, struct bpf_link *link,
if (err) if (err)
goto out_unlock; goto out_unlock;
rcu_assign_pointer(net->bpf.progs[type], link->prog); run_array = bpf_prog_array_alloc(1, GFP_KERNEL);
if (!run_array) {
err = -ENOMEM;
goto out_unlock;
}
run_array->items[0].prog = link->prog;
rcu_assign_pointer(net->bpf.run_array[type], run_array);
net->bpf.links[type] = link; net->bpf.links[type] = link;
out_unlock: out_unlock:
@ -368,11 +411,12 @@ static void __net_exit netns_bpf_pernet_pre_exit(struct net *net)
mutex_lock(&netns_bpf_mutex); mutex_lock(&netns_bpf_mutex);
for (type = 0; type < MAX_NETNS_BPF_ATTACH_TYPE; type++) { for (type = 0; type < MAX_NETNS_BPF_ATTACH_TYPE; type++) {
netns_bpf_run_array_detach(net, type);
link = net->bpf.links[type]; link = net->bpf.links[type];
if (link) if (link)
bpf_netns_link_auto_detach(link); bpf_netns_link_auto_detach(link);
else else if (net->bpf.progs[type])
__netns_bpf_prog_detach(net, type); bpf_prog_put(net->bpf.progs[type]);
} }
mutex_unlock(&netns_bpf_mutex); mutex_unlock(&netns_bpf_mutex);
} }

19
net/core/flow_dissector.c
View File

@ -86,14 +86,14 @@ int flow_dissector_bpf_prog_attach_check(struct net *net,
for_each_net(ns) { for_each_net(ns) {
if (ns == &init_net) if (ns == &init_net)
continue; continue;
if (rcu_access_pointer(ns->bpf.progs[type])) if (rcu_access_pointer(ns->bpf.run_array[type]))
return -EEXIST; return -EEXIST;
} }
} else { } else {
/* Make sure root flow dissector is not attached /* Make sure root flow dissector is not attached
* when attaching to the non-root namespace. * when attaching to the non-root namespace.
*/ */
if (rcu_access_pointer(init_net.bpf.progs[type])) if (rcu_access_pointer(init_net.bpf.run_array[type]))
return -EEXIST; return -EEXIST;
} }
@ -894,7 +894,6 @@ bool __skb_flow_dissect(const struct net *net,
struct flow_dissector_key_addrs *key_addrs; struct flow_dissector_key_addrs *key_addrs;
struct flow_dissector_key_tags *key_tags; struct flow_dissector_key_tags *key_tags;
struct flow_dissector_key_vlan *key_vlan; struct flow_dissector_key_vlan *key_vlan;
struct bpf_prog *attached = NULL;
enum flow_dissect_ret fdret; enum flow_dissect_ret fdret;
enum flow_dissector_key_id dissector_vlan = FLOW_DISSECTOR_KEY_MAX; enum flow_dissector_key_id dissector_vlan = FLOW_DISSECTOR_KEY_MAX;
bool mpls_el = false; bool mpls_el = false;
@ -951,14 +950,14 @@ bool __skb_flow_dissect(const struct net *net,
WARN_ON_ONCE(!net); WARN_ON_ONCE(!net);
if (net) { if (net) {
enum netns_bpf_attach_type type = NETNS_BPF_FLOW_DISSECTOR; enum netns_bpf_attach_type type = NETNS_BPF_FLOW_DISSECTOR;
struct bpf_prog_array *run_array;
rcu_read_lock(); rcu_read_lock();
attached = rcu_dereference(init_net.bpf.progs[type]); run_array = rcu_dereference(init_net.bpf.run_array[type]);
if (!run_array)
run_array = rcu_dereference(net->bpf.run_array[type]);
if (!attached) if (run_array) {
attached = rcu_dereference(net->bpf.progs[type]);
if (attached) {
struct bpf_flow_keys flow_keys; struct bpf_flow_keys flow_keys;
struct bpf_flow_dissector ctx = { struct bpf_flow_dissector ctx = {
.flow_keys = &flow_keys, .flow_keys = &flow_keys,
@ -966,6 +965,7 @@ bool __skb_flow_dissect(const struct net *net,
.data_end = data + hlen, .data_end = data + hlen,
}; };
__be16 n_proto = proto; __be16 n_proto = proto;
struct bpf_prog *prog;
if (skb) { if (skb) {
ctx.skb = skb; ctx.skb = skb;
@ -976,7 +976,8 @@ bool __skb_flow_dissect(const struct net *net,
n_proto = skb->protocol; n_proto = skb->protocol;
} }
ret = bpf_flow_dissect(attached, &ctx, n_proto, nhoff, prog = READ_ONCE(run_array->items[0].prog);
ret = bpf_flow_dissect(prog, &ctx, n_proto, nhoff,
hlen, flags); hlen, flags);
__skb_flow_bpf_to_target(&flow_keys, flow_dissector, __skb_flow_bpf_to_target(&flow_keys, flow_dissector,
target_container); target_container);