tls: fix NULL deref on tls_sw_splice_eof() with empty record
authorJann Horn <jannh@google.com>
Wed, 22 Nov 2023 21:44:47 +0000 (22:44 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 3 Dec 2023 06:33:07 +0000 (07:33 +0100)
commit2214e2bb5489145aba944874d0ee1652a0a63dc8
treeb97ad7537f91a4439e14519bb17aa7a6bad9f187
parentfe5b55bcfe242fc4a02770a3f79ac76e28334271
tls: fix NULL deref on tls_sw_splice_eof() with empty record

commit 53f2cb491b500897a619ff6abd72f565933760f0 upstream.

syzkaller discovered that if tls_sw_splice_eof() is executed as part of
sendfile() when the plaintext/ciphertext sk_msg are empty, the send path
gets confused because the empty ciphertext buffer does not have enough
space for the encryption overhead. This causes tls_push_record() to go on
the `split = true` path (which is only supposed to be used when interacting
with an attached BPF program), and then get further confused and hit the
tls_merge_open_record() path, which then assumes that there must be at
least one populated buffer element, leading to a NULL deref.

It is possible to have empty plaintext/ciphertext buffers if we previously
bailed from tls_sw_sendmsg_locked() via the tls_trim_both_msgs() path.
tls_sw_push_pending_record() already handles this case correctly; let's do
the same check in tls_sw_splice_eof().

Fixes: df720d288dbb ("tls/sw: Use splice_eof() to flush")
Cc: stable@vger.kernel.org
Reported-by: syzbot+40d43509a099ea756317@syzkaller.appspotmail.com
Signed-off-by: Jann Horn <jannh@google.com>
Link: https://lore.kernel.org/r/20231122214447.675768-1-jannh@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/tls/tls_sw.c