bpf/flow_dissector: add mode to enforce global BPF flow dissector
Subsystems (6)
v2 cover bpf/flow_dissector: add mode to enforce global BPF flow dissector
While having a per-net-ns flow dissector programs is convenient for testing, security-wise it's better to have only one vetted global flow dissector implementation.
Let's have a convention that when BPF flow dissector is installed in the root namespace, child namespaces can't override it.
The intended use-case is to attach global BPF flow dissector early from the init scripts/systemd. Attaching global dissector is prohibited if some non-root namespace already has flow dissector attached. Also, attaching to non-root namespace is prohibited when there is flow dissector attached to the root namespace.
v2:
* EPERM -> EEXIST (Song Liu)
* Make sure we don't have dissector attached to non-root namespaces when attaching the global one (Andrii Nakryiko)
Stanislav Fomichev (2):
bpf/flow_dissector: add mode to enforce global BPF flow dissector selftests/bpf: add test for BPF flow dissector in the root namespace
Documentation/bpf/prog_flow_dissector.rst | 3 ++ net/core/flow_dissector.c | 42 ++++++++++++++-- .../selftests/bpf/test_flow_dissector.sh | 48 ++++++++++++++++--- 3 files changed, 83 insertions(+), 10 deletions(-)
Signature · 1 line
2.23.0.581.g78d2f28ef7-goog
PATCH v2 1/2 bpf/flow_dissector: add mode to enforce global BPF flow dissector
Always use init_net flow dissector BPF program if it's attached and fall back to the per-net namespace one. Also, deny installing new programs if there is already one attached to the root namespace.
Users can still detach their BPF programs, but can't attach any new ones (-EEXIST).
--- Documentation/bpf/prog_flow_dissector.rst | 3 ++ net/core/flow_dissector.c | 42 ++++++++++++++++++++--- 2 files changed, 41 insertions(+), 4 deletions(-)
Documentation/bpf/prog_flow_dissector.rst +3 -0
@@ -142,3 +142,6 @@ BPF flow dissector doesn't support exporting all the metadata that in-kernel C-based implementation can export. Notable example is single VLAN (802.1Q) and double VLAN (802.1AD) tags. Please refer to the ``struct bpf_flow_keys`` for a set of information that's currently can be exported from the BPF context. + +When BPF flow dissector is attached to the root network namespace (machine-wide +policy), users can't override it in their child network namespaces.
net/core/flow_dissector.c +38 -4
64 lines not on this page · raw message
Signature · 1 line
2.23.0.581.g78d2f28ef7-goog
PATCH v2 2/2 selftests/bpf: add test for BPF flow dissector in the root namespace
In reply to Stanislav FomichevMake sure non-root namespaces get an error if root flow dissector is attached.
--- .../selftests/bpf/test_flow_dissector.sh | 48 ++++++++++++++++--- 1 file changed, 42 insertions(+), 6 deletions(-)
tools/testing/selftests/bpf/test_flow_dissector.sh +42 -6
@@ -18,19 +18,55 @@ fi # this is the case and run it with in_netns.sh if it is being run in the root # namespace. if [[ -z $(ip netns identify $$) ]]; then + err=0 + if bpftool="$(which bpftool)"; then + echo "Testing global flow dissector..." + + $bpftool prog loadall ./bpf_flow.o /sys/fs/bpf/flow \ + type flow_dissector + + if ! unshare --net $bpftool prog attach pinned \ + /sys/fs/bpf/flow/flow_dissector flow_dissector; then + echo "Unexpected unsuccessful attach in namespace" >&2 + err=1 + fi + + $bpftool prog attach pinned /sys/fs/bpf/flow/flow_dissector \ + flow_dissector + + if unshare --net $bpftool prog attach pinned \ + /sys/fs/bpf/flow/flow_dissector flow_dissector; then + echo "Unexpected successful attach in namespace" >&2 + err=1 + fi + + if ! $bpftool prog detach pinned \ + /sys/fs/bpf/flow/flow_dissector flow_dissector; then + echo "Failed to detach flow dissector" >&2 + err=1 + fi + + rm -rf /sys/fs/bpf/flow + else + echo "Skipping root flow dissector test, bpftool not found" >&2 + fi + + # Run the rest of the tests in a net namespace. ../net/in_netns.sh "$0" "$@" - exit $? -fi + err=$(( $err + $? )) -# Determine selftest success via shell exit code -exit_handler() -{ - if (( $? == 0 )); then + if (( $err == 0 )); then echo "selftests: $TESTNAME [PASS]"; else echo "selftests: $TESTNAME [FAILED]"; fi + exit $err +fi + +# Determine selftest success via shell exit code +exit_handler() +{ set +e # Cleanup
Signature · 1 line
2.23.0.581.g78d2f28ef7-goog
Quoted 9 lines · Stanislav Fomichev
Always use
init_netflow dissector BPF program if it's attached and fall back to the per-net namespace one. Also, deny installing new programs if there is already one attached to the root namespace.
Users can still detach their BPF programs, but can't attach any new ones (-EEXIST).Cc Petar Penkov <[email protected]>Signed-off-by Stanislav Fomichev <[email protected]>
Re: [PATCH bpf-next v2 2/2] selftests/bpf: add test for BPF flow dissector in the root namespace
In reply to Stanislav FomichevQuoted 6 lines · Stanislav Fomichev
Make sure non-root namespaces get an error if root flow dissector is attached.
Cc Petar Penkov <[email protected]>Signed-off-by Stanislav Fomichev <[email protected]>
Quoted 10 lines · Stanislav Fomichev
Always use
init_netflow dissector BPF program if it's attached and fall back to the per-net namespace one. Also, deny installing new programs if there is already one attached to the root namespace.
Users can still detach their BPF programs, but can't attach any new ones (-EEXIST).Cc Petar Penkov <[email protected]>Signed-off-by Stanislav Fomichev <[email protected]>---
Looks good, but see my note below. Regardless:
Quoted 29 lines
Documentation/bpf/prog_flow_dissector.rst | 3 ++ net/core/flow_dissector.c | 42 ++++++++++++++++++++--- 2 files changed, 41 insertions(+), 4 deletions(-)Documentation/bpf/prog_flow_dissector.rst +3
-0@@ -142,3 +142,6 @@ BPF flow dissector doesn't support exporting all the metadata that in-kernel C-based implementation can export. Notable example is single VLAN (802.1Q) and double VLAN (802.1AD) tags. Please refer to the ``struct bpf_flow_keys`` for a set of information that's currently can be exported from the BPF context. + +When BPF flow dissector is attached to the root network namespace (machine-wide +policy), users can't override it in their child network namespaces.net/core/flow_dissector.c +3
-0@@ -114,19 +114,50 @@ int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, {struct bpf_prog *attached;struct net *net; +int ret =0 ; net = current->nsproxy->net_ns;mutex_lock (&flow_dissector_mutex); + +if (net == &init_net) {
net/core/flow_dissector.c int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, · line 132
Quoted 10 lines
net/core/flow_dissector.c +10
-0+/* BPF flow dissector in the root namespace overrides +* any per-net-namespace one. When attaching to root, +* make sure we don't have any BPF program attached +* to the non-root namespaces. +*/ +struct net *ns; + +for_each_net (ns) { +if (net == &init_net) +continue ;
You don't need this condition, if something is attached to init_net, you will return -EEXIST anyway. Or is this a performance optimization?
net/core/flow_dissector.c int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, · line 143
Quoted 11 lines
net/core/flow_dissector.c +11
-0+ +if (rcu_access_pointer (ns->flow_dissector_prog)) { + ret = -EEXIST ; + goto out; + } + } + }else { +/* Make sure root flow dissector is not attached +* when attaching to the non-root namespace. +*/ +
nit: extra empty line
Quoted 38 lines
net/core/flow_dissector.c +14
-5+if (rcu_access_pointer (init_net.flow_dissector_prog)) { + ret = -EEXIST ; + goto out; + } + } + attached =rcu_dereference_protected (net->flow_dissector_prog,lockdep_is_held (&flow_dissector_mutex));if (attached) {/* Only one BPF program can be attached at a time */ -mutex_unlock (&flow_dissector_mutex); -return -EEXIST ; + ret = -EEXIST ; + goto out; }rcu_assign_pointer (net->flow_dissector_prog, prog); +out:mutex_unlock (&flow_dissector_mutex); -return 0 ; +return ret; }int skb_flow_dissector_bpf_prog_detach (const union bpf_attr *attr)@@ -910,7 +941,10 @@ bool __skb_flow_dissect(const struct net *net,WARN_ON_ONCE (!net);if (net) {rcu_read_lock (); - attached =rcu_dereference (net->flow_dissector_prog); + attached =rcu_dereference (init_net.flow_dissector_prog); + +if (!attached) + attached =rcu_dereference (net->flow_dissector_prog);if (attached) {struct bpf_flow_keys flow_keys; --2.23.0.581.g78d2f28ef7-goog
Quoted 50 lines · Andrii Nakryiko
Quoted 10 lines · Stanislav Fomichev
Always use
init_netflow dissector BPF program if it's attached and fall back to the per-net namespace one. Also, deny installing new programs if there is already one attached to the root namespace.
Users can still detach their BPF programs, but can't attach any new ones (-EEXIST).Cc Petar Penkov <[email protected]>Signed-off-by Stanislav Fomichev <[email protected]>---
Looks good, but see my note below. Regardless:
Acked-by Andrii Nakryiko <[email protected]>Quoted 34 lines
Documentation/bpf/prog_flow_dissector.rst | 3 ++ net/core/flow_dissector.c | 42 ++++++++++++++++++++--- 2 files changed, 41 insertions(+), 4 deletions(-)Documentation/bpf/prog_flow_dissector.rst +3
-0@@ -142,3 +142,6 @@ BPF flow dissector doesn't support exporting all the metadata that in-kernel C-based implementation can export. Notable example is single VLAN (802.1Q) and double VLAN (802.1AD) tags. Please refer to the ``struct bpf_flow_keys`` for a set of information that's currently can be exported from the BPF context. + +When BPF flow dissector is attached to the root network namespace (machine-wide +policy), users can't override it in their child network namespaces.net/core/flow_dissector.c +8
-0@@ -114,19 +114,50 @@ int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, {struct bpf_prog *attached;struct net *net; +int ret =0 ; net = current->nsproxy->net_ns;mutex_lock (&flow_dissector_mutex); + +if (net == &init_net) { +/* BPF flow dissector in the root namespace overrides +* any per-net-namespace one. When attaching to root, +* make sure we don't have any BPF program attached +* to the non-root namespaces. +*/
Quoted 8 lines
net/core/flow_dissector.c int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, · line 132
Quoted 5 lines
net/core/flow_dissector.c +5
-0+struct net *ns; + +for_each_net (ns) { +if (net == &init_net) +continue ;You don't need this condition, if something is attached to
init_net, you will return-EEXISTanyway. Or is this a performance optimization?
Ah, I agree, will remove an respin.
Quoted 5 lines
Quoted 5 lines
net/core/flow_dissector.c +5
-0+ +if (rcu_access_pointer (ns->flow_dissector_prog)) { + ret = -EEXIST ; + goto out; + }
Quoted 8 lines
net/core/flow_dissector.c int skb_flow_dissector_bpf_prog_attach(const union bpf_attr *attr, · line 143
Quoted 6 lines
net/core/flow_dissector.c +6
-0+ } + }else { +/* Make sure root flow dissector is not attached +* when attaching to the non-root namespace. +*/ +nit: extra empty line
Thx, will fix.
Quoted 38 lines
Quoted 38 lines
net/core/flow_dissector.c +14
-5+if (rcu_access_pointer (init_net.flow_dissector_prog)) { + ret = -EEXIST ; + goto out; + } + } + attached =rcu_dereference_protected (net->flow_dissector_prog,lockdep_is_held (&flow_dissector_mutex));if (attached) {/* Only one BPF program can be attached at a time */ -mutex_unlock (&flow_dissector_mutex); -return -EEXIST ; + ret = -EEXIST ; + goto out; }rcu_assign_pointer (net->flow_dissector_prog, prog); +out:mutex_unlock (&flow_dissector_mutex); -return 0 ; +return ret; }int skb_flow_dissector_bpf_prog_detach (const union bpf_attr *attr)@@ -910,7 +941,10 @@ bool __skb_flow_dissect(const struct net *net,WARN_ON_ONCE (!net);if (net) {rcu_read_lock (); - attached =rcu_dereference (net->flow_dissector_prog); + attached =rcu_dereference (init_net.flow_dissector_prog); + +if (!attached) + attached =rcu_dereference (net->flow_dissector_prog);if (attached) {struct bpf_flow_keys flow_keys; --2.23.0.581.g78d2f28ef7-goog