[bpf PATCH v4 0/4] sockmap/ktls fixes

[bpf PATCH v4 0/4] sockmap/ktls fixes

[John Fastabend]

Series of fixes for sockmap and ktls, see patches for descriptions.

v2: fix build issue for CONFIG_TLS_DEVICE and fixup couple comments
    from Jakub

v3: fix issue where release could call unhash resulting in a use after
    free. Now we detach the ulp pointer before calling into destroy
    or unhash. This way if we get a callback into unhash from destroy
    path there is no ulp to access. The fallout is we must pass the
    ctx into the functions rather than use the sk lookup in each
    routine. This is probably better anyways.

v4: move unhash routine to TLS_SW only, hardware offloads need to
    keep ctx around long enough to free in-flight context. We will
    need a follow up fix for this.

---

John Fastabend (4):
      bpf: tls, implement unhash to avoid transition out of ESTABLISHED
      bpf: sockmap, only stop/flush strp if it was enabled at some point
      bpf: sockmap remove duplicate queue free
      bpf: sockmap fix msg->sg.size account on ingress skb


 include/net/tls.h    |   28 +++++++---
 net/core/skmsg.c     |    7 ++-
 net/ipv4/tcp_bpf.c   |    2 -
 net/tls/tls_device.c |   10 ++--
 net/tls/tls_main.c   |   82 ++++++++++++++++++++---------
 net/tls/tls_sw.c     |  140 +++++++++++++++++++++++++++++---------------------
 6 files changed, 166 insertions(+), 103 deletions(-)

--
Signature

[bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[John Fastabend]

It is possible (via shutdown()) for TCP socks to go through TCP_CLOSE
state via tcp_disconnect() without calling into close callback. This
would allow a kTLS enabled socket to exist outside of ESTABLISHED
state which is not supported.

Solve this the same way we solved the sock{map|hash} case by adding
an unhash hook to remove tear down the TLS state.

In the process we also make the close hook more robust. We add a put
call into the close path, also in the unhash path, to remove the
reference to ulp data after free. Its no longer valid and may confuse
things later if the socket (re)enters kTLS code paths. Second we add
an 'if(ctx)' check to ensure the ctx is still valid and not released
from a previous unhash/close path.

Fixes: d91c3e17f75f2 ("net/tls: Only attach to sockets in ESTABLISHED state")
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
 include/net/tls.h    |   28 +++++++---
 net/tls/tls_device.c |   10 ++--
 net/tls/tls_main.c   |   82 ++++++++++++++++++++---------
 net/tls/tls_sw.c     |  140 +++++++++++++++++++++++++++++---------------------
 4 files changed, 161 insertions(+), 99 deletions(-)

diff --git a/include/net/tls.h b/include/net/tls.h
index 5934246b2c6f..05d8cd5a3297 100644
--- a/include/net/tls.h
+++ b/include/net/tls.h
@@ -250,11 +250,14 @@ struct tls_context {
 	bool in_tcp_sendpages;
 	bool pending_open_record_frags;
 
-	int (*push_pending_record)(struct sock *sk, int flags);
+	int (*push_pending_record)(struct sock *sk,
+				   struct tls_context *ctx, int flags);
 
 	void (*sk_write_space)(struct sock *sk);
 	void (*sk_destruct)(struct sock *sk);
 	void (*sk_proto_close)(struct sock *sk, long timeout);
+	void (*sk_proto_unhash)(struct sock *sk);
+	struct proto *sk_proto;
 
 	int  (*setsockopt)(struct sock *sk, int level,
 			   int optname, char __user *optval,
@@ -292,9 +295,10 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size);
 int tls_sw_sendpage(struct sock *sk, struct page *page,
 		    int offset, size_t size, int flags);
 void tls_sw_close(struct sock *sk, long timeout);
-void tls_sw_free_resources_tx(struct sock *sk);
-void tls_sw_free_resources_rx(struct sock *sk);
-void tls_sw_release_resources_rx(struct sock *sk);
+void tls_sw_free_resources_tx(struct sock *sk,
+			      struct tls_context *ctx, bool locked);
+void tls_sw_free_resources_rx(struct sock *sk, struct tls_context *ctx);
+void tls_sw_release_resources_rx(struct sock *sk, struct tls_context *ctx);
 int tls_sw_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 		   int nonblock, int flags, int *addr_len);
 bool tls_sw_stream_read(const struct sock *sk);
@@ -310,7 +314,7 @@ void tls_device_sk_destruct(struct sock *sk);
 void tls_device_free_resources_tx(struct sock *sk);
 void tls_device_init(void);
 void tls_device_cleanup(void);
-int tls_tx_records(struct sock *sk, int flags);
+int tls_tx_records(struct sock *sk, struct tls_context *ctx, int flags);
 
 struct tls_record_info *tls_get_record(struct tls_offload_context_tx *context,
 				       u32 seq, u64 *p_record_sn);
@@ -416,12 +420,10 @@ static inline struct tls_context *tls_get_ctx(const struct sock *sk)
 }
 
 static inline void tls_advance_record_sn(struct sock *sk,
+					 struct tls_prot_info *prot,
 					 struct cipher_context *ctx,
 					 int version)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
-	struct tls_prot_info *prot = &tls_ctx->prot_info;
-
 	if (tls_bigint_increment(ctx->rec_seq, prot->rec_seq_size))
 		tls_err_abort(sk, EBADMSG);
 
@@ -493,6 +495,16 @@ static inline void xor_iv_with_seq(int version, char *iv, char *seq)
 	}
 }
 
+static inline void tls_put_ctx(struct sock *sk)
+{
+	struct inet_connection_sock *icsk = inet_csk(sk);
+	struct tls_context *ctx = icsk->icsk_ulp_data;
+
+	if (!ctx)
+		return;
+	sk->sk_prot = ctx->sk_proto;
+	icsk->icsk_ulp_data = NULL;
+}
 
 static inline struct tls_sw_context_rx *tls_sw_ctx_rx(
 		const struct tls_context *tls_ctx)
diff --git a/net/tls/tls_device.c b/net/tls/tls_device.c
index 14dedb24fa7b..d3cd887f0488 100644
--- a/net/tls/tls_device.c
+++ b/net/tls/tls_device.c
@@ -281,7 +281,8 @@ static int tls_push_record(struct sock *sk,
 	list_add_tail(&record->list, &offload_ctx->records_list);
 	spin_unlock_irq(&offload_ctx->lock);
 	offload_ctx->open_record = NULL;
-	tls_advance_record_sn(sk, &ctx->tx, ctx->crypto_send.info.version);
+	tls_advance_record_sn(sk, prot, &ctx->tx,
+			      ctx->crypto_send.info.version);
 
 	for (i = 0; i < record->num_frags; i++) {
 		frag = &record->frags[i];
@@ -548,7 +549,8 @@ struct tls_record_info *tls_get_record(struct tls_offload_context_tx *context,
 }
 EXPORT_SYMBOL(tls_get_record);
 
-static int tls_device_push_pending_record(struct sock *sk, int flags)
+static int tls_device_push_pending_record(struct sock *sk,
+					  struct tls_context *ctx, int flags)
 {
 	struct iov_iter	msg_iter;
 
@@ -922,7 +924,7 @@ int tls_set_device_offload_rx(struct sock *sk, struct tls_context *ctx)
 
 free_sw_resources:
 	up_read(&device_offload_lock);
-	tls_sw_free_resources_rx(sk);
+	tls_sw_free_resources_rx(sk, ctx);
 	down_read(&device_offload_lock);
 release_ctx:
 	ctx->priv_ctx_rx = NULL;
@@ -958,7 +960,7 @@ void tls_device_offload_cleanup_rx(struct sock *sk)
 	}
 out:
 	up_read(&device_offload_lock);
-	tls_sw_release_resources_rx(sk);
+	tls_sw_release_resources_rx(sk, tls_ctx);
 }
 
 static int tls_device_down(struct net_device *netdev)
diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c
index 478603f43964..7f7982361128 100644
--- a/net/tls/tls_main.c
+++ b/net/tls/tls_main.c
@@ -150,12 +150,11 @@ int tls_push_sg(struct sock *sk,
 	return 0;
 }
 
-static int tls_handle_open_record(struct sock *sk, int flags)
+static int tls_handle_open_record(struct sock *sk,
+				  struct tls_context *ctx, int flags)
 {
-	struct tls_context *ctx = tls_get_ctx(sk);
-
 	if (tls_is_pending_open_record(ctx))
-		return ctx->push_pending_record(sk, flags);
+		return ctx->push_pending_record(sk, ctx, flags);
 
 	return 0;
 }
@@ -163,6 +162,7 @@ static int tls_handle_open_record(struct sock *sk, int flags)
 int tls_proccess_cmsg(struct sock *sk, struct msghdr *msg,
 		      unsigned char *record_type)
 {
+	struct tls_context *ctx;
 	struct cmsghdr *cmsg;
 	int rc = -EINVAL;
 
@@ -180,7 +180,11 @@ int tls_proccess_cmsg(struct sock *sk, struct msghdr *msg,
 			if (msg->msg_flags & MSG_MORE)
 				return -EINVAL;
 
-			rc = tls_handle_open_record(sk, msg->msg_flags);
+			ctx = tls_get_ctx(sk);
+			if (unlikely(!ctx))
+				return -EBUSY;
+
+			rc = tls_handle_open_record(sk, ctx, msg->msg_flags);
 			if (rc)
 				return rc;
 
@@ -261,32 +265,28 @@ static void tls_ctx_free(struct tls_context *ctx)
 	kfree(ctx);
 }
 
-static void tls_sk_proto_close(struct sock *sk, long timeout)
+static bool tls_sk_proto_destroy(struct sock *sk,
+				 struct tls_context *ctx, bool locked)
 {
-	struct tls_context *ctx = tls_get_ctx(sk);
 	long timeo = sock_sndtimeo(sk, 0);
-	void (*sk_proto_close)(struct sock *sk, long timeout);
-	bool free_ctx = false;
-
-	lock_sock(sk);
-	sk_proto_close = ctx->sk_proto_close;
 
 	if (ctx->tx_conf == TLS_HW_RECORD && ctx->rx_conf == TLS_HW_RECORD)
-		goto skip_tx_cleanup;
+		return false;
 
-	if (ctx->tx_conf == TLS_BASE && ctx->rx_conf == TLS_BASE) {
-		free_ctx = true;
-		goto skip_tx_cleanup;
-	}
+	if (ctx->tx_conf == TLS_SW && ctx->rx_conf == TLS_SW)
+		tls_put_ctx(sk);
+
+	if (ctx->tx_conf == TLS_BASE && ctx->rx_conf == TLS_BASE)
+		return true;
 
 	if (!tls_complete_pending_work(sk, ctx, 0, &timeo))
-		tls_handle_open_record(sk, 0);
+		tls_handle_open_record(sk, ctx, 0);
 
 	/* We need these for tls_sw_fallback handling of other packets */
 	if (ctx->tx_conf == TLS_SW) {
 		kfree(ctx->tx.rec_seq);
 		kfree(ctx->tx.iv);
-		tls_sw_free_resources_tx(sk);
+		tls_sw_free_resources_tx(sk, ctx, locked);
 #ifdef CONFIG_TLS_DEVICE
 	} else if (ctx->tx_conf == TLS_HW) {
 		tls_device_free_resources_tx(sk);
@@ -294,21 +294,46 @@ static void tls_sk_proto_close(struct sock *sk, long timeout)
 	}
 
 	if (ctx->rx_conf == TLS_SW)
-		tls_sw_free_resources_rx(sk);
+		tls_sw_free_resources_rx(sk, ctx);
 
 #ifdef CONFIG_TLS_DEVICE
 	if (ctx->rx_conf == TLS_HW)
 		tls_device_offload_cleanup_rx(sk);
-
-	if (ctx->tx_conf != TLS_HW && ctx->rx_conf != TLS_HW) {
-#else
-	{
 #endif
+
+	if (ctx->tx_conf != TLS_HW && ctx->rx_conf != TLS_HW)
+		return true;
+	return false;
+}
+
+static void tls_sk_proto_unhash(struct sock *sk)
+{
+	struct tls_context *ctx = tls_get_ctx(sk);
+	void (*sk_proto_unhash)(struct sock *sk);
+	bool free_ctx;
+
+	if (!ctx)
+		return sk->sk_prot->unhash(sk);
+	sk_proto_unhash = ctx->sk_proto_unhash;
+	free_ctx = tls_sk_proto_destroy(sk, ctx, false);
+	if (sk_proto_unhash)
+		sk_proto_unhash(sk);
+	if (free_ctx)
 		tls_ctx_free(ctx);
-		ctx = NULL;
-	}
+}
 
-skip_tx_cleanup:
+static void tls_sk_proto_close(struct sock *sk, long timeout)
+{
+	void (*sk_proto_close)(struct sock *sk, long timeout);
+	struct tls_context *ctx = tls_get_ctx(sk);
+	bool free_ctx;
+
+	if (!ctx)
+		return sk->sk_prot->destroy(sk);
+
+	lock_sock(sk);
+	sk_proto_close = ctx->sk_proto_close;
+	free_ctx = tls_sk_proto_destroy(sk, ctx, true);
 	release_sock(sk);
 	sk_proto_close(sk, timeout);
 	/* free ctx for TLS_HW_RECORD, used by tcp_set_state
@@ -601,6 +626,8 @@ static struct tls_context *create_ctx(struct sock *sk)
 	ctx->setsockopt = sk->sk_prot->setsockopt;
 	ctx->getsockopt = sk->sk_prot->getsockopt;
 	ctx->sk_proto_close = sk->sk_prot->close;
+	ctx->sk_proto_unhash = sk->sk_prot->unhash;
+	ctx->sk_proto = sk->sk_prot;
 	return ctx;
 }
 
@@ -738,6 +765,7 @@ static void build_protos(struct proto prot[TLS_NUM_CONFIG][TLS_NUM_CONFIG],
 	prot[TLS_SW][TLS_SW].recvmsg		= tls_sw_recvmsg;
 	prot[TLS_SW][TLS_SW].stream_memory_read	= tls_sw_stream_read;
 	prot[TLS_SW][TLS_SW].close		= tls_sk_proto_close;
+	prot[TLS_SW][TLS_SW].unhash		= tls_sk_proto_unhash;
 
 #ifdef CONFIG_TLS_DEVICE
 	prot[TLS_HW][TLS_BASE] = prot[TLS_BASE][TLS_BASE];
diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c
index 29d6af43dd24..2de433232b99 100644
--- a/net/tls/tls_sw.c
+++ b/net/tls/tls_sw.c
@@ -266,9 +266,9 @@ static void tls_trim_both_msgs(struct sock *sk, int target_size)
 	sk_msg_trim(sk, &rec->msg_encrypted, target_size);
 }
 
-static int tls_alloc_encrypted_msg(struct sock *sk, int len)
+static int tls_alloc_encrypted_msg(struct sock *sk,
+				   struct tls_context *tls_ctx,  int len)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct tls_rec *rec = ctx->open_rec;
 	struct sk_msg *msg_en = &rec->msg_encrypted;
@@ -300,9 +300,8 @@ static int tls_clone_plaintext_msg(struct sock *sk, int required)
 	return sk_msg_clone(sk, msg_pl, msg_en, skip, len);
 }
 
-static struct tls_rec *tls_get_rec(struct sock *sk)
+static struct tls_rec *tls_get_rec(struct sock *sk, struct tls_context *tls_ctx)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_prot_info *prot = &tls_ctx->prot_info;
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct sk_msg *msg_pl, *msg_en;
@@ -339,9 +338,8 @@ static void tls_free_rec(struct sock *sk, struct tls_rec *rec)
 	kfree(rec);
 }
 
-static void tls_free_open_rec(struct sock *sk)
+static void tls_free_open_rec(struct sock *sk, struct tls_context *tls_ctx)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct tls_rec *rec = ctx->open_rec;
 
@@ -351,9 +349,8 @@ static void tls_free_open_rec(struct sock *sk)
 	}
 }
 
-int tls_tx_records(struct sock *sk, int flags)
+int tls_tx_records(struct sock *sk, struct tls_context *tls_ctx, int flags)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct tls_rec *rec, *tmp;
 	struct sk_msg *msg_en;
@@ -519,12 +516,13 @@ static int tls_do_encryption(struct sock *sk,
 
 	/* Unhook the record from context if encryption is not failure */
 	ctx->open_rec = NULL;
-	tls_advance_record_sn(sk, &tls_ctx->tx, prot->version);
+	tls_advance_record_sn(sk, prot, &tls_ctx->tx, prot->version);
 	return rc;
 }
 
-static int tls_split_open_record(struct sock *sk, struct tls_rec *from,
-				 struct tls_rec **to, struct sk_msg *msg_opl,
+static int tls_split_open_record(struct sock *sk, struct tls_context *tls_ctx,
+				 struct tls_rec *from, struct tls_rec **to,
+				 struct sk_msg *msg_opl,
 				 struct sk_msg *msg_oen, u32 split_point,
 				 u32 tx_overhead_size, u32 *orig_end)
 {
@@ -536,7 +534,7 @@ static int tls_split_open_record(struct sock *sk, struct tls_rec *from,
 	struct tls_rec *new;
 	int ret;
 
-	new = tls_get_rec(sk);
+	new = tls_get_rec(sk, tls_ctx);
 	if (!new)
 		return -ENOMEM;
 	ret = sk_msg_alloc(sk, &new->msg_encrypted, msg_opl->sg.size +
@@ -641,10 +639,9 @@ static void tls_merge_open_record(struct sock *sk, struct tls_rec *to,
 	kfree(from);
 }
 
-static int tls_push_record(struct sock *sk, int flags,
-			   unsigned char record_type)
+static int tls_push_record(struct sock *sk, struct tls_context *tls_ctx,
+			   int flags, unsigned char record_type)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_prot_info *prot = &tls_ctx->prot_info;
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct tls_rec *rec = ctx->open_rec, *tmp = NULL;
@@ -663,7 +660,8 @@ static int tls_push_record(struct sock *sk, int flags,
 	split_point = msg_pl->apply_bytes;
 	split = split_point && split_point < msg_pl->sg.size;
 	if (split) {
-		rc = tls_split_open_record(sk, rec, &tmp, msg_pl, msg_en,
+		rc = tls_split_open_record(sk, tls_ctx, rec, &tmp,
+					   msg_pl, msg_en,
 					   split_point, prot->overhead_size,
 					   &orig_end);
 		if (rc < 0)
@@ -732,14 +730,14 @@ static int tls_push_record(struct sock *sk, int flags,
 		ctx->open_rec = tmp;
 	}
 
-	return tls_tx_records(sk, flags);
+	return tls_tx_records(sk, tls_ctx, flags);
 }
 
 static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
+			       struct tls_context *tls_ctx,
 			       bool full_record, u8 record_type,
 			       size_t *copied, int flags)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct sk_msg msg_redir = { };
 	struct sk_psock *psock;
@@ -752,7 +750,7 @@ static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
 	policy = !(flags & MSG_SENDPAGE_NOPOLICY);
 	psock = sk_psock_get(sk);
 	if (!psock || !policy)
-		return tls_push_record(sk, flags, record_type);
+		return tls_push_record(sk, tls_ctx, flags, record_type);
 more_data:
 	enospc = sk_msg_full(msg);
 	if (psock->eval == __SK_NONE) {
@@ -775,10 +773,10 @@ static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
 
 	switch (psock->eval) {
 	case __SK_PASS:
-		err = tls_push_record(sk, flags, record_type);
+		err = tls_push_record(sk, tls_ctx, flags, record_type);
 		if (err < 0) {
 			*copied -= sk_msg_free(sk, msg);
-			tls_free_open_rec(sk);
+			tls_free_open_rec(sk, tls_ctx);
 			goto out_err;
 		}
 		break;
@@ -799,7 +797,7 @@ static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
 			msg->sg.size = 0;
 		}
 		if (msg->sg.size == 0)
-			tls_free_open_rec(sk);
+			tls_free_open_rec(sk, tls_ctx);
 		break;
 	case __SK_DROP:
 	default:
@@ -809,7 +807,7 @@ static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
 		else
 			msg->apply_bytes -= send;
 		if (msg->sg.size == 0)
-			tls_free_open_rec(sk);
+			tls_free_open_rec(sk, tls_ctx);
 		*copied -= (send + delta);
 		err = -EACCES;
 	}
@@ -838,14 +836,15 @@ static int bpf_exec_tx_verdict(struct sk_msg *msg, struct sock *sk,
 	return err;
 }
 
-static int tls_sw_push_pending_record(struct sock *sk, int flags)
+static int tls_sw_push_pending_record(struct sock *sk,
+				      struct tls_context *tls_ctx, int flags)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
-	struct tls_rec *rec = ctx->open_rec;
 	struct sk_msg *msg_pl;
+	struct tls_rec *rec;
 	size_t copied;
 
+	rec = ctx->open_rec;
 	if (!rec)
 		return 0;
 
@@ -854,31 +853,39 @@ static int tls_sw_push_pending_record(struct sock *sk, int flags)
 	if (!copied)
 		return 0;
 
-	return bpf_exec_tx_verdict(msg_pl, sk, true, TLS_RECORD_TYPE_DATA,
+	return bpf_exec_tx_verdict(msg_pl, sk, tls_ctx, true,
+				   TLS_RECORD_TYPE_DATA,
 				   &copied, flags);
 }
 
 int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 {
 	long timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
-	struct tls_prot_info *prot = &tls_ctx->prot_info;
-	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
-	bool async_capable = ctx->async_capable;
 	unsigned char record_type = TLS_RECORD_TYPE_DATA;
 	bool is_kvec = iov_iter_is_kvec(&msg->msg_iter);
 	bool eor = !(msg->msg_flags & MSG_MORE);
+	bool full_record, async_capable;
 	size_t try_to_copy, copied = 0;
 	struct sk_msg *msg_pl, *msg_en;
+	struct tls_sw_context_tx *ctx;
+	struct tls_context *tls_ctx;
+	struct tls_prot_info *prot;
 	struct tls_rec *rec;
 	int required_size;
 	int num_async = 0;
-	bool full_record;
 	int record_room;
 	int num_zc = 0;
 	int orig_size;
 	int ret = 0;
 
+	tls_ctx = tls_get_ctx(sk);
+	if (unlikely(!tls_ctx))
+		return -EBUSY;
+
+	prot = &tls_ctx->prot_info;
+	ctx = tls_sw_ctx_tx(tls_ctx);
+	async_capable = ctx->async_capable;
+
 	if (msg->msg_flags & ~(MSG_MORE | MSG_DONTWAIT | MSG_NOSIGNAL))
 		return -ENOTSUPP;
 
@@ -910,7 +917,7 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 		if (ctx->open_rec)
 			rec = ctx->open_rec;
 		else
-			rec = ctx->open_rec = tls_get_rec(sk);
+			rec = ctx->open_rec = tls_get_rec(sk, tls_ctx);
 		if (!rec) {
 			ret = -ENOMEM;
 			goto send_end;
@@ -935,7 +942,7 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 			goto wait_for_sndbuf;
 
 alloc_encrypted:
-		ret = tls_alloc_encrypted_msg(sk, required_size);
+		ret = tls_alloc_encrypted_msg(sk, tls_ctx, required_size);
 		if (ret) {
 			if (ret != -ENOSPC)
 				goto wait_for_memory;
@@ -962,7 +969,8 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 			copied += try_to_copy;
 
 			sk_msg_sg_copy_set(msg_pl, first);
-			ret = bpf_exec_tx_verdict(msg_pl, sk, full_record,
+			ret = bpf_exec_tx_verdict(msg_pl, sk, tls_ctx,
+						  full_record,
 						  record_type, &copied,
 						  msg->msg_flags);
 			if (ret) {
@@ -1015,7 +1023,8 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 		tls_ctx->pending_open_record_frags = true;
 		copied += try_to_copy;
 		if (full_record || eor) {
-			ret = bpf_exec_tx_verdict(msg_pl, sk, full_record,
+			ret = bpf_exec_tx_verdict(msg_pl, sk, tls_ctx,
+						  full_record,
 						  record_type, &copied,
 						  msg->msg_flags);
 			if (ret) {
@@ -1069,7 +1078,7 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size)
 	/* Transmit if any encryptions have completed */
 	if (test_and_clear_bit(BIT_TX_SCHEDULED, &ctx->tx_bitmask)) {
 		cancel_delayed_work(&ctx->tx_work.work);
-		tls_tx_records(sk, msg->msg_flags);
+		tls_tx_records(sk, tls_ctx, msg->msg_flags);
 	}
 
 send_end:
@@ -1083,10 +1092,10 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 			      int offset, size_t size, int flags)
 {
 	long timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT);
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
-	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
-	struct tls_prot_info *prot = &tls_ctx->prot_info;
 	unsigned char record_type = TLS_RECORD_TYPE_DATA;
+	struct tls_sw_context_tx *ctx;
+	struct tls_context *tls_ctx;
+	struct tls_prot_info *prot;
 	struct sk_msg *msg_pl;
 	struct tls_rec *rec;
 	int num_async = 0;
@@ -1096,6 +1105,13 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 	int ret = 0;
 	bool eor;
 
+	tls_ctx = tls_get_ctx(sk);
+	if (unlikely(!tls_ctx))
+		return -EBUSY;
+
+	ctx = tls_sw_ctx_tx(tls_ctx);
+	prot = &tls_ctx->prot_info;
+
 	eor = !(flags & (MSG_MORE | MSG_SENDPAGE_NOTLAST));
 	sk_clear_bit(SOCKWQ_ASYNC_NOSPACE, sk);
 
@@ -1118,7 +1134,7 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 		if (ctx->open_rec)
 			rec = ctx->open_rec;
 		else
-			rec = ctx->open_rec = tls_get_rec(sk);
+			rec = ctx->open_rec = tls_get_rec(sk, tls_ctx);
 		if (!rec) {
 			ret = -ENOMEM;
 			goto sendpage_end;
@@ -1140,7 +1156,7 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 		if (!sk_stream_memory_free(sk))
 			goto wait_for_sndbuf;
 alloc_payload:
-		ret = tls_alloc_encrypted_msg(sk, required_size);
+		ret = tls_alloc_encrypted_msg(sk, tls_ctx, required_size);
 		if (ret) {
 			if (ret != -ENOSPC)
 				goto wait_for_memory;
@@ -1163,7 +1179,8 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 		tls_ctx->pending_open_record_frags = true;
 		if (full_record || eor || sk_msg_full(msg_pl)) {
 			rec->inplace_crypto = 0;
-			ret = bpf_exec_tx_verdict(msg_pl, sk, full_record,
+			ret = bpf_exec_tx_verdict(msg_pl, sk, tls_ctx,
+						  full_record,
 						  record_type, &copied, flags);
 			if (ret) {
 				if (ret == -EINPROGRESS)
@@ -1194,7 +1211,7 @@ static int tls_sw_do_sendpage(struct sock *sk, struct page *page,
 		/* Transmit if any encryptions have completed */
 		if (test_and_clear_bit(BIT_TX_SCHEDULED, &ctx->tx_bitmask)) {
 			cancel_delayed_work(&ctx->tx_work.work);
-			tls_tx_records(sk, flags);
+			tls_tx_records(sk, tls_ctx, flags);
 		}
 	}
 sendpage_end:
@@ -1479,7 +1496,8 @@ static int decrypt_skb_update(struct sock *sk, struct sk_buff *skb,
 					       async);
 			if (err < 0) {
 				if (err == -EINPROGRESS)
-					tls_advance_record_sn(sk, &tls_ctx->rx,
+					tls_advance_record_sn(sk, prot,
+							      &tls_ctx->rx,
 							      version);
 
 				return err;
@@ -1491,7 +1509,7 @@ static int decrypt_skb_update(struct sock *sk, struct sk_buff *skb,
 		rxm->full_len -= padding_length(ctx, tls_ctx, skb);
 		rxm->offset += prot->prepend_size;
 		rxm->full_len -= prot->overhead_size;
-		tls_advance_record_sn(sk, &tls_ctx->rx, version);
+		tls_advance_record_sn(sk, prot, &tls_ctx->rx, version);
 		ctx->decrypted = true;
 		ctx->saved_data_ready(sk);
 	} else {
@@ -2031,9 +2049,9 @@ static void tls_data_ready(struct sock *sk)
 	}
 }
 
-void tls_sw_free_resources_tx(struct sock *sk)
+void tls_sw_free_resources_tx(struct sock *sk,
+			      struct tls_context *tls_ctx, bool locked)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_tx *ctx = tls_sw_ctx_tx(tls_ctx);
 	struct tls_rec *rec, *tmp;
 
@@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
 	if (atomic_read(&ctx->encrypt_pending))
 		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
 
-	release_sock(sk);
+	if (locked)
+		release_sock(sk);
 	cancel_delayed_work_sync(&ctx->tx_work.work);
-	lock_sock(sk);
+	if (locked)
+		lock_sock(sk);
 
 	/* Tx whatever records we can transmit and abandon the rest */
-	tls_tx_records(sk, -1);
+	tls_tx_records(sk, tls_ctx, -1);
 
 	/* Free up un-sent records in tx_list. First, free
 	 * the partially sent record if any at head of tx_list.
@@ -2067,15 +2087,17 @@ void tls_sw_free_resources_tx(struct sock *sk)
 		kfree(rec);
 	}
 
-	crypto_free_aead(ctx->aead_send);
-	tls_free_open_rec(sk);
+	if (ctx->aead_send) {
+		crypto_free_aead(ctx->aead_send);
+		ctx->aead_send = NULL;
+	}
+	tls_free_open_rec(sk, tls_ctx);
 
 	kfree(ctx);
 }
 
-void tls_sw_release_resources_rx(struct sock *sk)
+void tls_sw_release_resources_rx(struct sock *sk, struct tls_context *tls_ctx)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_rx *ctx = tls_sw_ctx_rx(tls_ctx);
 
 	kfree(tls_ctx->rx.rec_seq);
@@ -2096,13 +2118,11 @@ void tls_sw_release_resources_rx(struct sock *sk)
 	}
 }
 
-void tls_sw_free_resources_rx(struct sock *sk)
+void tls_sw_free_resources_rx(struct sock *sk, struct tls_context *tls_ctx)
 {
-	struct tls_context *tls_ctx = tls_get_ctx(sk);
 	struct tls_sw_context_rx *ctx = tls_sw_ctx_rx(tls_ctx);
 
-	tls_sw_release_resources_rx(sk);
-
+	tls_sw_release_resources_rx(sk, tls_ctx);
 	kfree(ctx);
 }
 
@@ -2120,7 +2140,7 @@ static void tx_work_handler(struct work_struct *work)
 		return;
 
 	lock_sock(sk);
-	tls_tx_records(sk, -1);
+	tls_tx_records(sk, tls_ctx, -1);
 	release_sock(sk);
 }
 

[bpf PATCH v4 2/4] bpf: sockmap, only stop/flush strp if it was enabled at some point

[John Fastabend]

If we try to call strp_done on a parser that has never been
initialized, because the sockmap user is only using TX side for
example we get the following error.


  [  883.422081] WARNING: CPU: 1 PID: 208 at kernel/workqueue.c:3030 __flush_work+0x1ca/0x1e0
  ...
  [  883.422095] Workqueue: events sk_psock_destroy_deferred
  [  883.422097] RIP: 0010:__flush_work+0x1ca/0x1e0


This had been wrapped in a 'if (psock->parser.enabled)' logic which
was broken because the strp_done() was never actually being called
because we do a strp_stop() earlier in the tear down logic will
set parser.enabled to false. This could result in a use after free
if work was still in the queue and was resolved by the patch here,
1d79895aef18f ("sk_msg: Always cancel strp work before freeing the
psock"). However, calling strp_stop(), done by the patch marked in
the fixes tag, only is useful if we never initialized a strp parser
program and never initialized the strp to start with. Because if
we had initialized a stream parser strp_stop() would have been called
by sk_psock_drop() earlier in the tear down process.  By forcing the
strp to stop we get past the WARNING in strp_done that checks
the stopped flag but calling cancel_work_sync on work that has never
been initialized is also wrong and generates the warning above.

To fix check if the parser program exists. If the program exists
then the strp work has been initialized and must be sync'd and
cancelled before free'ing any structures. If no program exists we
never initialized the stream parser in the first place so skip the
sync/cancel logic implemented by strp_done.

Finally, remove the strp_done its not needed and in the case where we are using the
stream parser has already been called.

Fixes: e8e3437762ad9 ("bpf: Stop the psock parser before canceling its work")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
 net/core/skmsg.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/core/skmsg.c b/net/core/skmsg.c
index cc94d921476c..49d1efa329d7 100644
--- a/net/core/skmsg.c
+++ b/net/core/skmsg.c
@@ -554,8 +554,10 @@ static void sk_psock_destroy_deferred(struct work_struct *gc)
 	struct sk_psock *psock = container_of(gc, struct sk_psock, gc);
 
 	/* No sk_callback_lock since already detached. */
-	strp_stop(&psock->parser.strp);
-	strp_done(&psock->parser.strp);
+
+	/* Parser has been stopped */
+	if (psock->progs.skb_parser)
+		strp_done(&psock->parser.strp);
 
 	cancel_work_sync(&psock->work);
 

[bpf PATCH v4 3/4] bpf: sockmap remove duplicate queue free

[John Fastabend]

In tcp bpf remove we free the cork list and purge the ingress msg
list. However we do this before the ref count reaches zero so it
could be possible some other access is in progress. In this case
(tcp close and/or tcp_unhash) we happen to also hold the sock
lock so no path exists but lets fix it otherwise it is extremely
fragile and breaks the reference counting rules. Also we already
check the cork list and ingress msg queue and free them once the
ref count reaches zero so its wasteful to check twice.

Fixes: 604326b41a6fb ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
 net/ipv4/tcp_bpf.c |    2 --
 1 file changed, 2 deletions(-)

diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c
index 1bb7321a256d..4a619c85daed 100644
--- a/net/ipv4/tcp_bpf.c
+++ b/net/ipv4/tcp_bpf.c
@@ -528,8 +528,6 @@ static void tcp_bpf_remove(struct sock *sk, struct sk_psock *psock)
 {
 	struct sk_psock_link *link;
 
-	sk_psock_cork_free(psock);
-	__sk_psock_purge_ingress_msg(psock);
 	while ((link = sk_psock_link_pop(psock))) {
 		sk_psock_unlink(sk, link);
 		sk_psock_free_link(link);

[bpf PATCH v4 4/4] bpf: sockmap fix msg->sg.size account on ingress skb

[John Fastabend]

When converting a skb to msg->sg we forget to set the size after the
latest ktls/tls code conversion. This patch can be reached by doing
a redir into ingress path from BPF skb sock recv hook. Then trying to
read the size fails.

Fix this by setting the size.

Fixes: 604326b41a6fb ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
 net/core/skmsg.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/net/core/skmsg.c b/net/core/skmsg.c
index 49d1efa329d7..93bffaad2135 100644
--- a/net/core/skmsg.c
+++ b/net/core/skmsg.c
@@ -411,6 +411,7 @@ static int sk_psock_skb_ingress(struct sk_psock *psock, struct sk_buff *skb)
 	sk_mem_charge(sk, skb->len);
 	copied = skb->len;
 	msg->sg.start = 0;
+	msg->sg.size = copied;
 	msg->sg.end = num_sge == MAX_MSG_FRAGS ? 0 : num_sge;
 	msg->skb = skb;
 

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[Jakub Kicinski]

On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:
>  #ifdef CONFIG_TLS_DEVICE
>  	if (ctx->rx_conf == TLS_HW)
>  		tls_device_offload_cleanup_rx(sk);
> -
> -	if (ctx->tx_conf != TLS_HW && ctx->rx_conf != TLS_HW) {
> -#else
> -	{
>  #endif
> +
> +	if (ctx->tx_conf != TLS_HW && ctx->rx_conf != TLS_HW)

Did you try to build without CONFIG_TLS_DEVICE?

I think someone spent too much time in Verilog land and decided 
it's a good idea to hide enum values under an ifdef:

$ git grep -C4 TLS_HW,
include/net/tls.h-enum {
include/net/tls.h-      TLS_BASE,
include/net/tls.h-      TLS_SW,
include/net/tls.h-#ifdef CONFIG_TLS_DEVICE
include/net/tls.h:      TLS_HW,
include/net/tls.h-#endif
include/net/tls.h-      TLS_HW_RECORD,
include/net/tls.h-      TLS_NUM_CONFIG,
include/net/tls.h-};

:(

> +		return true;
> +	return false;
> +}

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[Jakub Kicinski]

On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:
> @@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
>  	if (atomic_read(&ctx->encrypt_pending))
>  		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
>  
> -	release_sock(sk);
> +	if (locked)
> +		release_sock(sk);
>  	cancel_delayed_work_sync(&ctx->tx_work.work);

So in the splat I got (on a slightly hacked up kernel) it seemed like
unhash may be called in atomic context:

[  783.232150]  tls_sk_proto_unhash+0x72/0x110 [tls]
[  783.237497]  tcp_set_state+0x484/0x640
[  783.241776]  ? __sk_mem_reduce_allocated+0x72/0x4a0
[  783.247317]  ? tcp_recv_timestamp+0x5c0/0x5c0
[  783.252265]  ? tcp_write_queue_purge+0xa6a/0x1180
[  783.257614]  tcp_done+0xac/0x260
[  783.261309]  tcp_reset+0xbe/0x350
[  783.265101]  tcp_validate_incoming+0xd9d/0x1530

I may have been unclear off-list, I only tested the patch no longer
crashes the offload :(

> -	lock_sock(sk);
> +	if (locked)
> +		lock_sock(sk);
>  
>  	/* Tx whatever records we can transmit and abandon the rest */
> -	tls_tx_records(sk, -1);
> +	tls_tx_records(sk, tls_ctx, -1);
>  
>  	/* Free up un-sent records in tx_list. First, free
>  	 * the partially sent record if any at head of tx_list.

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[John Fastabend]

Jakub Kicinski wrote:
> On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:
> > @@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
> >  	if (atomic_read(&ctx->encrypt_pending))
> >  		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
> >  
> > -	release_sock(sk);
> > +	if (locked)
> > +		release_sock(sk);
> >  	cancel_delayed_work_sync(&ctx->tx_work.work);
> 
> So in the splat I got (on a slightly hacked up kernel) it seemed like
> unhash may be called in atomic context:
> 
> [  783.232150]  tls_sk_proto_unhash+0x72/0x110 [tls]
> [  783.237497]  tcp_set_state+0x484/0x640
> [  783.241776]  ? __sk_mem_reduce_allocated+0x72/0x4a0
> [  783.247317]  ? tcp_recv_timestamp+0x5c0/0x5c0
> [  783.252265]  ? tcp_write_queue_purge+0xa6a/0x1180
> [  783.257614]  tcp_done+0xac/0x260
> [  783.261309]  tcp_reset+0xbe/0x350
> [  783.265101]  tcp_validate_incoming+0xd9d/0x1530
> 
> I may have been unclear off-list, I only tested the patch no longer
> crashes the offload :(
> 

Yep, I misread and thought it was resolved here as well. OK I'll dig into
it. I'm not seeing it from selftests but I guess that means we are missing
a testcase. :( yet another version I guess.

Thanks,
John


> > -	lock_sock(sk);
> > +	if (locked)
> > +		lock_sock(sk);
> >  
> >  	/* Tx whatever records we can transmit and abandon the rest */
> > -	tls_tx_records(sk, -1);
> > +	tls_tx_records(sk, tls_ctx, -1);
> >  
> >  	/* Free up un-sent records in tx_list. First, free
> >  	 * the partially sent record if any at head of tx_list.
> 

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[John Fastabend]

John Fastabend wrote:
> Jakub Kicinski wrote:
> > On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:
> > > @@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
> > >  	if (atomic_read(&ctx->encrypt_pending))
> > >  		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
> > >  
> > > -	release_sock(sk);
> > > +	if (locked)
> > > +		release_sock(sk);
> > >  	cancel_delayed_work_sync(&ctx->tx_work.work);
> > 
> > So in the splat I got (on a slightly hacked up kernel) it seemed like
> > unhash may be called in atomic context:
> > 
> > [  783.232150]  tls_sk_proto_unhash+0x72/0x110 [tls]
> > [  783.237497]  tcp_set_state+0x484/0x640
> > [  783.241776]  ? __sk_mem_reduce_allocated+0x72/0x4a0
> > [  783.247317]  ? tcp_recv_timestamp+0x5c0/0x5c0
> > [  783.252265]  ? tcp_write_queue_purge+0xa6a/0x1180
> > [  783.257614]  tcp_done+0xac/0x260
> > [  783.261309]  tcp_reset+0xbe/0x350
> > [  783.265101]  tcp_validate_incoming+0xd9d/0x1530
> > 
> > I may have been unclear off-list, I only tested the patch no longer
> > crashes the offload :(
> > 
> 
> Yep, I misread and thought it was resolved here as well. OK I'll dig into
> it. I'm not seeing it from selftests but I guess that means we are missing
> a testcase. :( yet another version I guess.
> 

Seems we need to call release_sock in the unhash case as well. Will
send a new patch shortly.

.John

> Thanks,
> John
> 

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[Jakub Kicinski]

On Tue, 14 May 2019 15:34:55 -0700, John Fastabend wrote:
> John Fastabend wrote:
> > Jakub Kicinski wrote:  
> > > On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:  
> > > > @@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
> > > >  	if (atomic_read(&ctx->encrypt_pending))
> > > >  		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
> > > >  
> > > > -	release_sock(sk);
> > > > +	if (locked)
> > > > +		release_sock(sk);
> > > >  	cancel_delayed_work_sync(&ctx->tx_work.work);  
> > > 
> > > So in the splat I got (on a slightly hacked up kernel) it seemed like
> > > unhash may be called in atomic context:
> > > 
> > > [  783.232150]  tls_sk_proto_unhash+0x72/0x110 [tls]
> > > [  783.237497]  tcp_set_state+0x484/0x640
> > > [  783.241776]  ? __sk_mem_reduce_allocated+0x72/0x4a0
> > > [  783.247317]  ? tcp_recv_timestamp+0x5c0/0x5c0
> > > [  783.252265]  ? tcp_write_queue_purge+0xa6a/0x1180
> > > [  783.257614]  tcp_done+0xac/0x260
> > > [  783.261309]  tcp_reset+0xbe/0x350
> > > [  783.265101]  tcp_validate_incoming+0xd9d/0x1530
> > > 
> > > I may have been unclear off-list, I only tested the patch no longer
> > > crashes the offload :(
> > >   
> > 
> > Yep, I misread and thought it was resolved here as well. OK I'll dig into
> > it. I'm not seeing it from selftests but I guess that means we are missing
> > a testcase. :( yet another version I guess.
> >   
> 
> Seems we need to call release_sock in the unhash case as well. Will
> send a new patch shortly.

My reading of the stack trace was that unhash gets called from
tcp_reset(), IOW from soft IRQ, so we can't cancel_delayed_work_sync()
in tls_sw_free_resources_tx(), no?

Re: [bpf PATCH v4 1/4] bpf: tls, implement unhash to avoid transition out of ESTABLISHED

[John Fastabend]

Jakub Kicinski wrote:
> On Tue, 14 May 2019 15:34:55 -0700, John Fastabend wrote:
> > John Fastabend wrote:
> > > Jakub Kicinski wrote:  
> > > > On Thu, 09 May 2019 21:57:49 -0700, John Fastabend wrote:  
> > > > > @@ -2042,12 +2060,14 @@ void tls_sw_free_resources_tx(struct sock *sk)
> > > > >  	if (atomic_read(&ctx->encrypt_pending))
> > > > >  		crypto_wait_req(-EINPROGRESS, &ctx->async_wait);
> > > > >  
> > > > > -	release_sock(sk);
> > > > > +	if (locked)
> > > > > +		release_sock(sk);
> > > > >  	cancel_delayed_work_sync(&ctx->tx_work.work);  
> > > > 
> > > > So in the splat I got (on a slightly hacked up kernel) it seemed like
> > > > unhash may be called in atomic context:
> > > > 
> > > > [  783.232150]  tls_sk_proto_unhash+0x72/0x110 [tls]
> > > > [  783.237497]  tcp_set_state+0x484/0x640
> > > > [  783.241776]  ? __sk_mem_reduce_allocated+0x72/0x4a0
> > > > [  783.247317]  ? tcp_recv_timestamp+0x5c0/0x5c0
> > > > [  783.252265]  ? tcp_write_queue_purge+0xa6a/0x1180
> > > > [  783.257614]  tcp_done+0xac/0x260
> > > > [  783.261309]  tcp_reset+0xbe/0x350
> > > > [  783.265101]  tcp_validate_incoming+0xd9d/0x1530
> > > > 
> > > > I may have been unclear off-list, I only tested the patch no longer
> > > > crashes the offload :(
> > > >   
> > > 
> > > Yep, I misread and thought it was resolved here as well. OK I'll dig into
> > > it. I'm not seeing it from selftests but I guess that means we are missing
> > > a testcase. :( yet another version I guess.
> > >   
> > 
> > Seems we need to call release_sock in the unhash case as well. Will
> > send a new patch shortly.
> 
> My reading of the stack trace was that unhash gets called from
> tcp_reset(), IOW from soft IRQ, so we can't cancel_delayed_work_sync()
> in tls_sw_free_resources_tx(), no?

Well the tcp_close() path has the lock held and can also call unhash(). Anyways
this dropping the sock lock in the middle of the block seems a bit suspect
to me anyways. I think we can defer the free until after sock is released this
is how it was solved on sockmap side.