[PATCH net-next] tcx: Fix splat during dev unregister

[Daniel Borkmann <daniel@xxxxxxxxxxxxx>]

From: Martin KaFai Lau <martin.lau@xxxxxxxxxx>

During unregister_netdevice_many_notify(), the ordering of our concerned
function calls is like this:

  unregister_netdevice_many_notify
    dev_shutdown
	qdisc_put
            clsact_destroy
    tcx_uninstall

The syzbot reproducer triggered a case that the qdisc refcnt is not
zero during dev_shutdown().

tcx_uninstall() will then WARN_ON_ONCE(tcx_entry(entry)->miniq_active)
because the miniq is still active and the entry should not be freed.
The latter assumed that qdisc destruction happens before tcx teardown.

This fix is to avoid tcx_uninstall() doing tcx_entry_free() when the
miniq is still alive and let the clsact_destroy() do the free later, so
that we do not assume any specific ordering for either of them.

If still active, tcx_uninstall() does clear the entry when flushing out
the prog/link. clsact_destroy() will then notice the "!tcx_entry_is_active()"
and then does the tcx_entry_free() eventually.

Fixes: e420bed02507 ("bpf: Add fd-based tcx multi-prog infra with link support")
Reported-by: syzbot+376a289e86a0fd02b9ba@xxxxxxxxxxxxxxxxxxxxxxxxx
Reported-by: Leon Romanovsky <leonro@xxxxxxxxxx>
Signed-off-by: Martin KaFai Lau <martin.lau@xxxxxxxxxx>
Co-developed-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Tested-by: syzbot+376a289e86a0fd02b9ba@xxxxxxxxxxxxxxxxxxxxxxxxx
---
 [ Sending directly to net-next given the issue was reported there by Leon. ]

 include/linux/bpf_mprog.h | 16 ++++++++++++++++
 kernel/bpf/tcx.c          | 12 ++++++++----
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/include/linux/bpf_mprog.h b/include/linux/bpf_mprog.h
index 2b429488f840..929225f7b095 100644
--- a/include/linux/bpf_mprog.h
+++ b/include/linux/bpf_mprog.h
@@ -256,6 +256,22 @@ static inline void bpf_mprog_entry_copy(struct bpf_mprog_entry *dst,
 	memcpy(dst->fp_items, src->fp_items, sizeof(src->fp_items));
 }
 
+static inline void bpf_mprog_entry_clear(struct bpf_mprog_entry *dst)
+{
+	memset(dst->fp_items, 0, sizeof(dst->fp_items));
+}
+
+static inline void bpf_mprog_clear_all(struct bpf_mprog_entry *entry,
+				       struct bpf_mprog_entry **entry_new)
+{
+	struct bpf_mprog_entry *peer;
+
+	peer = bpf_mprog_peer(entry);
+	bpf_mprog_entry_clear(peer);
+	peer->parent->count = 0;
+	*entry_new = peer;
+}
+
 static inline void bpf_mprog_entry_grow(struct bpf_mprog_entry *entry, int idx)
 {
 	int total = bpf_mprog_total(entry);
diff --git a/kernel/bpf/tcx.c b/kernel/bpf/tcx.c
index 69a272712b29..13f0b5dc8262 100644
--- a/kernel/bpf/tcx.c
+++ b/kernel/bpf/tcx.c
@@ -94,15 +94,19 @@ int tcx_prog_detach(const union bpf_attr *attr, struct bpf_prog *prog)
 
 void tcx_uninstall(struct net_device *dev, bool ingress)
 {
+	struct bpf_mprog_entry *entry, *entry_new = NULL;
 	struct bpf_tuple tuple = {};
-	struct bpf_mprog_entry *entry;
 	struct bpf_mprog_fp *fp;
 	struct bpf_mprog_cp *cp;
+	bool active;
 
 	entry = tcx_entry_fetch(dev, ingress);
 	if (!entry)
 		return;
-	tcx_entry_update(dev, NULL, ingress);
+	active = tcx_entry(entry)->miniq_active;
+	if (active)
+		bpf_mprog_clear_all(entry, &entry_new);
+	tcx_entry_update(dev, entry_new, ingress);
 	tcx_entry_sync();
 	bpf_mprog_foreach_tuple(entry, fp, cp, tuple) {
 		if (tuple.link)
@@ -111,8 +115,8 @@ void tcx_uninstall(struct net_device *dev, bool ingress)
 			bpf_prog_put(tuple.prog);
 		tcx_skeys_dec(ingress);
 	}
-	WARN_ON_ONCE(tcx_entry(entry)->miniq_active);
-	tcx_entry_free(entry);
+	if (!active)
+		tcx_entry_free(entry);
 }
 
 int tcx_prog_query(const union bpf_attr *attr, union bpf_attr __user *uattr)
-- 
2.21.0