A reference underflow is found in TLS handshake subsystem that causes a
direct use-after-free. Part of the crash log is like below:
[ 2.022114] ------------[ cut here ]------------
[ 2.022193] refcount_t: underflow; use-after-free.
[ 2.022288] WARNING: CPU: 0 PID: 60 at lib/refcount.c:28 refcount_warn_saturate+0xbe/0x110
[ 2.022432] Modules linked in:
[ 2.022848] RIP: 0010:refcount_warn_saturate+0xbe/0x110
[ 2.023231] RSP: 0018:ffffc900001bfe18 EFLAGS: 00000286
[ 2.023325] RAX: 0000000000000000 RBX: 0000000000000007 RCX: 00000000ffffdfff
[ 2.023438] RDX: 0000000000000000 RSI: 00000000ffffffea RDI: 0000000000000001
[ 2.023555] RBP: ffff888004c20098 R08: ffffffff82b392c8 R09: 00000000ffffdfff
[ 2.023693] R10: ffffffff82a592e0 R11: ffffffff82b092e0 R12: ffff888004c200d8
[ 2.023813] R13: 0000000000000000 R14: ffff888004c20000 R15: ffffc90000013ca8
[ 2.023930] FS: 0000000000000000(0000) GS:ffff88807dc00000(0000) knlGS:0000000000000000
[ 2.024062] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2.024161] CR2: ffff888003601000 CR3: 0000000002a2e000 CR4: 00000000000006f0
[ 2.024275] Call Trace:
[ 2.024322] <TASK>
[ 2.024367] ? __warn+0x7f/0x130
[ 2.024430] ? refcount_warn_saturate+0xbe/0x110
[ 2.024513] ? report_bug+0x199/0x1b0
[ 2.024585] ? handle_bug+0x3c/0x70
[ 2.024676] ? exc_invalid_op+0x18/0x70
[ 2.024750] ? asm_exc_invalid_op+0x1a/0x20
[ 2.024830] ? refcount_warn_saturate+0xbe/0x110
[ 2.024916] ? refcount_warn_saturate+0xbe/0x110
[ 2.024998] __tcp_close+0x2f4/0x3d0
[ 2.025065] ? __pfx_kunit_generic_run_threadfn_adapter+0x10/0x10
[ 2.025168] tcp_close+0x1f/0x70
[ 2.025231] inet_release+0x33/0x60
[ 2.025297] sock_release+0x1f/0x80
[ 2.025361] handshake_req_cancel_test2+0x100/0x2d0
[ 2.025457] kunit_try_run_case+0x4c/0xa0
[ 2.025532] kunit_generic_run_threadfn_adapter+0x15/0x20
[ 2.025644] kthread+0xe1/0x110
[ 2.025708] ? __pfx_kthread+0x10/0x10
[ 2.025780] ret_from_fork+0x2c/0x50
One can enable CONFIG_NET_HANDSHAKE_KUNIT_TEST config to reproduce above
crash.
The root cause of this bug is that the commit 1ce77c998f
("net/handshake: Unpin sock->file if a handshake is cancelled") adds one
additional fput() function. That patch claims that the fput() is used to
enable sock->file to be freed even when user space never calls DONE.
However, it seems that the intended DONE routine will never give an
additional fput() of ths sock->file. The existing two of them are just
used to balance the reference added in sockfd_lookup().
This patch revert the mentioned commit to avoid the use-after-free. The
patched kernel could successfully pass the KUNIT test and boot to shell.
[ 0.733613] # Subtest: Handshake API tests
[ 0.734029] 1..11
[ 0.734255] KTAP version 1
[ 0.734542] # Subtest: req_alloc API fuzzing
[ 0.736104] ok 1 handshake_req_alloc NULL proto
[ 0.736114] ok 2 handshake_req_alloc CLASS_NONE
[ 0.736559] ok 3 handshake_req_alloc CLASS_MAX
[ 0.737020] ok 4 handshake_req_alloc no callbacks
[ 0.737488] ok 5 handshake_req_alloc no done callback
[ 0.737988] ok 6 handshake_req_alloc excessive privsize
[ 0.738529] ok 7 handshake_req_alloc all good
[ 0.739036] # req_alloc API fuzzing: pass:7 fail:0 skip:0 total:7
[ 0.739444] ok 1 req_alloc API fuzzing
[ 0.740065] ok 2 req_submit NULL req arg
[ 0.740436] ok 3 req_submit NULL sock arg
[ 0.740834] ok 4 req_submit NULL sock->file
[ 0.741236] ok 5 req_lookup works
[ 0.741621] ok 6 req_submit max pending
[ 0.741974] ok 7 req_submit multiple
[ 0.742382] ok 8 req_cancel before accept
[ 0.742764] ok 9 req_cancel after accept
[ 0.743151] ok 10 req_cancel after done
[ 0.743510] ok 11 req_destroy works
[ 0.743882] # Handshake API tests: pass:11 fail:0 skip:0 total:11
[ 0.744205] # Totals: pass:17 fail:0 skip:0 total:17
Acked-by: Chuck Lever <chuck.lever@oracle.com>
Fixes: 1ce77c998f ("net/handshake: Unpin sock->file if a handshake is cancelled")
Signed-off-by: Lin Ma <linma@zju.edu.cn>
Link: https://lore.kernel.org/r/20230613083204.633896-1-linma@zju.edu.cn
Link: https://lore.kernel.org/r/20230614015249.987448-1-linma@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If user space never calls DONE, sock->file's reference count remains
elevated. Enable sock->file to be freed eventually in this case.
Reported-by: Jakub Kacinski <kuba@kernel.org>
Fixes: 3b3009ea8a ("net/handshake: Create a NETLINK service for handling handshake requests")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
These verify the API contracts and help exercise lifetime rules for
consumer sockets and handshake_req structures.
One way to run these tests:
./tools/testing/kunit/kunit.py run --kunitconfig ./net/handshake/.kunitconfig
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When a kernel consumer needs a transport layer security session, it
first needs a handshake to negotiate and establish a session. This
negotiation can be done in user space via one of the several
existing library implementations, or it can be done in the kernel.
No in-kernel handshake implementations yet exist. In their absence,
we add a netlink service that can:
a. Notify a user space daemon that a handshake is needed.
b. Once notified, the daemon calls the kernel back via this
netlink service to get the handshake parameters, including an
open socket on which to establish the session.
c. Once the handshake is complete, the daemon reports the
session status and other information via a second netlink
operation. This operation marks that it is safe for the
kernel to use the open socket and the security session
established there.
The notification service uses a multicast group. Each handshake
mechanism (eg, tlshd) adopts its own group number so that the
handshake services are completely independent of one another. The
kernel can then tell via netlink_has_listeners() whether a handshake
service is active and prepared to handle a handshake request.
A new netlink operation, ACCEPT, acts like accept(2) in that it
instantiates a file descriptor in the user space daemon's fd table.
If this operation is successful, the reply carries the fd number,
which can be treated as an open and ready file descriptor.
While user space is performing the handshake, the kernel keeps its
muddy paws off the open socket. A second new netlink operation,
DONE, indicates that the user space daemon is finished with the
socket and it is safe for the kernel to use again. The operation
also indicates whether a session was established successfully.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
