From 4d42cd6bc2ac1b9be50ade13771daec90c9d18b1 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Mon, 27 Feb 2023 10:12:01 -0800 Subject: tls: rx: fix return value for async crypto Gaurav reports that TLS Rx is broken with async crypto accelerators. The commit under fixes missed updating the retval byte counting logic when updating how records are stored. Even tho both before and after the change 'decrypted' was updated inside the main loop, it was completely overwritten when processing the async completions. Now that the rx_list only holds non-zero-copy records we need to add, not overwrite. Reported-and-bisected-by: Gaurav Jain Fixes: cbbdee9918a2 ("tls: rx: async: don't put async zc on the list") Link: https://bugzilla.kernel.org/show_bug.cgi?id=217064 Tested-by: Gaurav Jain Reviewed-by: Simon Horman Link: https://lore.kernel.org/r/20230227181201.1793772-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/tls/tls_sw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/tls/tls_sw.c') diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c index 782d3701b86f..021d760f9133 100644 --- a/net/tls/tls_sw.c +++ b/net/tls/tls_sw.c @@ -2127,7 +2127,7 @@ recv_end: else err = process_rx_list(ctx, msg, &control, 0, async_copy_bytes, is_peek); - decrypted = max(err, 0); + decrypted += max(err, 0); } copied += decrypted; -- cgit v1.2.3 From f3221361dc85d4de22586ce8441ec2c67b454f5d Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 28 Feb 2023 16:28:57 -0800 Subject: net: tls: avoid hanging tasks on the tx_lock syzbot sent a hung task report and Eric explains that adversarial receiver may keep RWIN at 0 for a long time, so we are not guaranteed to make forward progress. Thread which took tx_lock and went to sleep may not release tx_lock for hours. Use interruptible sleep where possible and reschedule the work if it can't take the lock. Testing: existing selftest passes Reported-by: syzbot+9c0268252b8ef967c62e@syzkaller.appspotmail.com Fixes: 79ffe6087e91 ("net/tls: add a TX lock") Link: https://lore.kernel.org/all/000000000000e412e905f5b46201@google.com/ Cc: stable@vger.kernel.org # wait 4 weeks Reviewed-by: Eric Dumazet Link: https://lore.kernel.org/r/20230301002857.2101894-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/tls/tls_sw.c | 26 +++++++++++++++++++------- 1 file changed, 19 insertions(+), 7 deletions(-) (limited to 'net/tls/tls_sw.c') diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c index 021d760f9133..635b8bf6b937 100644 --- a/net/tls/tls_sw.c +++ b/net/tls/tls_sw.c @@ -956,7 +956,9 @@ int tls_sw_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) MSG_CMSG_COMPAT)) return -EOPNOTSUPP; - mutex_lock(&tls_ctx->tx_lock); + ret = mutex_lock_interruptible(&tls_ctx->tx_lock); + if (ret) + return ret; lock_sock(sk); if (unlikely(msg->msg_controllen)) { @@ -1290,7 +1292,9 @@ int tls_sw_sendpage(struct sock *sk, struct page *page, MSG_SENDPAGE_NOTLAST | MSG_SENDPAGE_NOPOLICY)) return -EOPNOTSUPP; - mutex_lock(&tls_ctx->tx_lock); + ret = mutex_lock_interruptible(&tls_ctx->tx_lock); + if (ret) + return ret; lock_sock(sk); ret = tls_sw_do_sendpage(sk, page, offset, size, flags); release_sock(sk); @@ -2435,11 +2439,19 @@ static void tx_work_handler(struct work_struct *work) if (!test_and_clear_bit(BIT_TX_SCHEDULED, &ctx->tx_bitmask)) return; - mutex_lock(&tls_ctx->tx_lock); - lock_sock(sk); - tls_tx_records(sk, -1); - release_sock(sk); - mutex_unlock(&tls_ctx->tx_lock); + + if (mutex_trylock(&tls_ctx->tx_lock)) { + lock_sock(sk); + tls_tx_records(sk, -1); + release_sock(sk); + mutex_unlock(&tls_ctx->tx_lock); + } else if (!test_and_set_bit(BIT_TX_SCHEDULED, &ctx->tx_bitmask)) { + /* Someone is holding the tx_lock, they will likely run Tx + * and cancel the work on their way out of the lock section. + * Schedule a long delay just in case. + */ + schedule_delayed_work(&ctx->tx_work.work, msecs_to_jiffies(10)); + } } static bool tls_is_tx_ready(struct tls_sw_context_tx *ctx) -- cgit v1.2.3