linux/net
David Howells 540b1c48c3 rxrpc: Fix deadlock between call creation and sendmsg/recvmsg
All the routines by which rxrpc is accessed from the outside are serialised
by means of the socket lock (sendmsg, recvmsg, bind,
rxrpc_kernel_begin_call(), ...) and this presents a problem:

 (1) If a number of calls on the same socket are in the process of
     connection to the same peer, a maximum of four concurrent live calls
     are permitted before further calls need to wait for a slot.

 (2) If a call is waiting for a slot, it is deep inside sendmsg() or
     rxrpc_kernel_begin_call() and the entry function is holding the socket
     lock.

 (3) sendmsg() and recvmsg() or the in-kernel equivalents are prevented
     from servicing the other calls as they need to take the socket lock to
     do so.

 (4) The socket is stuck until a call is aborted and makes its slot
     available to the waiter.

Fix this by:

 (1) Provide each call with a mutex ('user_mutex') that arbitrates access
     by the users of rxrpc separately for each specific call.

 (2) Make rxrpc_sendmsg() and rxrpc_recvmsg() unlock the socket as soon as
     they've got a call and taken its mutex.

     Note that I'm returning EWOULDBLOCK from recvmsg() if MSG_DONTWAIT is
     set but someone else has the lock.  Should I instead only return
     EWOULDBLOCK if there's nothing currently to be done on a socket, and
     sleep in this particular instance because there is something to be
     done, but we appear to be blocked by the interrupt handler doing its
     ping?

 (3) Make rxrpc_new_client_call() unlock the socket after allocating a new
     call, locking its user mutex and adding it to the socket's call tree.
     The call is returned locked so that sendmsg() can add data to it
     immediately.

     From the moment the call is in the socket tree, it is subject to
     access by sendmsg() and recvmsg() - even if it isn't connected yet.

 (4) Lock new service calls in the UDP data_ready handler (in
     rxrpc_new_incoming_call()) because they may already be in the socket's
     tree and the data_ready handler makes them live immediately if a user
     ID has already been preassigned.

     Note that the new call is locked before any notifications are sent
     that it is live, so doing mutex_trylock() *ought* to always succeed.
     Userspace is prevented from doing sendmsg() on calls that are in a
     too-early state in rxrpc_do_sendmsg().

 (5) Make rxrpc_new_incoming_call() return the call with the user mutex
     held so that a ping can be scheduled immediately under it.

     Note that it might be worth moving the ping call into
     rxrpc_new_incoming_call() and then we can drop the mutex there.

 (6) Make rxrpc_accept_call() take the lock on the call it is accepting and
     release the socket after adding the call to the socket's tree.  This
     is slightly tricky as we've dequeued the call by that point and have
     to requeue it.

     Note that requeuing emits a trace event.

 (7) Make rxrpc_kernel_send_data() and rxrpc_kernel_recv_data() take the
     new mutex immediately and don't bother with the socket mutex at all.

This patch has the nice bonus that calls on the same socket are now to some
extent parallelisable.

Note that we might want to move rxrpc_service_prealloc() calls out from the
socket lock and give it its own lock, so that we don't hang progress in
other calls because we're waiting for the allocator.

We probably also want to avoid calling rxrpc_notify_socket() from within
the socket lock (rxrpc_accept_call()).

Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Marc Dionne <marc.c.dionne@auristor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2017-03-01 09:50:58 -08:00
..
6lowpan 6lowpan: use rb_entry() 2017-01-22 16:46:13 -05:00
9p
802 Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
8021q net: remove ndo_neigh_{construct, destroy} from stacked devices 2017-02-06 11:25:57 -05:00
appletalk lib/vsprintf.c: remove %Z support 2017-02-27 18:43:47 -08:00
atm lib/vsprintf.c: remove %Z support 2017-02-27 18:43:47 -08:00
ax25 ax25: Fix segfault after sock connection timeout 2017-01-16 14:39:58 -05:00
batman-adv Here are two fixes for batman-adv for net-next: 2017-01-29 19:21:26 -05:00
bluetooth scripts/spelling.txt: add "an user" pattern and fix typo instances 2017-02-27 18:43:46 -08:00
bridge lib/vsprintf.c: remove %Z support 2017-02-27 18:43:47 -08:00
caif net: caif: Remove unused stats member from struct chnl_net 2017-01-19 11:45:21 -05:00
can can: bcm: fix hrtimer/tasklet termination in bcm op removal 2017-01-30 11:05:04 +01:00
ceph This time around we have: 2017-02-28 15:36:09 -08:00
core bpf: Fix bpf_xdp_event_output 2017-02-23 13:53:42 -05:00
dcb net: dcb: set error code on failures 2016-12-03 23:54:25 -05:00
dccp net/dccp: fix use after free in tw_timer_handler() 2017-02-22 16:14:07 -05:00
decnet Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
dns_resolver
dsa Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-11 02:31:11 -05:00
ethernet Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next 2017-02-16 21:25:49 -05:00
hsr net/hsr: use eth_hw_addr_random() 2017-02-21 13:25:22 -05:00
ieee802154 lib/vsprintf.c: remove %Z support 2017-02-27 18:43:47 -08:00
ife net: Introduce ife encapsulation module 2017-02-03 15:16:45 -05:00
ipv4 Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
ipv6 Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
ipx ktime: Get rid of the union 2016-12-25 17:21:22 +01:00
irda lib/vsprintf.c: remove %Z support 2017-02-27 18:43:47 -08:00
iucv net/af_iucv: don't use paged skbs for TX on HiperSockets 2017-01-10 21:08:29 -05:00
kcm kcm: fix a null pointer dereference in kcm_sendmsg() 2017-02-14 13:06:37 -05:00
key netns: make struct pernet_operations::id unsigned int 2016-11-18 10:59:15 -05:00
l2tp Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
l3mdev
lapb Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
llc net/llc: avoid BUG_ON() in skb_orphan() 2017-02-12 22:14:49 -05:00
mac80211 Merge branch 'idr-4.11' of git://git.infradead.org/users/willy/linux-dax 2017-02-28 20:29:41 -08:00
mac802154 ktime: Cleanup ktime_set() usage 2016-12-25 17:21:22 +01:00
mpls net: mpls: Add support for netconf 2017-02-20 11:13:37 -05:00
ncsi
netfilter Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
netlabel netlabel: add CALIPSO to the list of built-in protocols 2017-01-06 22:20:45 -05:00
netlink net: adjust skb->truesize in pskb_expand_head() 2017-01-27 12:03:29 -05:00
netrom
nfc
openvswitch openvswitch: Set event bit after initializing labels. 2017-02-20 10:18:43 -05:00
packet Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-19 11:18:46 -05:00
phonet netns: make struct pernet_operations::id unsigned int 2016-11-18 10:59:15 -05:00
psample net: Introduce psample, a new genetlink channel for packet sampling 2017-01-24 13:44:28 -05:00
qrtr net: qrtr: Mark 'buf' as little endian 2017-01-10 20:45:04 -05:00
rds Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
rfkill rfkill: remove rfkill-regulator 2017-01-24 11:07:35 +01:00
rose Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
rxrpc rxrpc: Fix deadlock between call creation and sendmsg/recvmsg 2017-03-01 09:50:58 -08:00
sched net sched actions: do not overwrite status of action creation. 2017-02-26 21:31:32 -05:00
sctp Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2017-02-28 10:00:39 -08:00
smc smc: some potential use after free bugs 2017-01-30 16:37:55 -05:00
strparser
sunrpc The nfsd update this round is mainly a lot of miscellaneous cleanups and 2017-02-28 15:39:09 -08:00
switchdev
tipc tipc: move premature initilalization of stack variables 2017-02-24 11:42:54 -05:00
unix unix: add ioctl to open a unix socket file with O_PATH 2017-02-02 21:58:02 -05:00
vmw_vsock Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net 2016-12-17 20:17:04 -08:00
wimax
wireless Some more updates: 2017-02-10 14:31:51 -05:00
x25 Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
xfrm xfrm: provide correct dst in xfrm_neigh_lookup 2017-02-26 21:35:24 -05:00
compat.c Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next 2017-02-22 10:15:09 -08:00
Kconfig bpf: make jited programs visible in traces 2017-02-17 13:40:05 -05:00
Makefile net: Introduce ife encapsulation module 2017-02-03 15:16:45 -05:00
socket.c net: socket: fix recvmmsg not returning error from sock_error 2017-02-21 13:35:25 -05:00
sysctl_net.c