Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-12-02 09:01:34 +00:00
Code Issues Packages Projects Releases Wiki Activity
e08293a4cc
linux/net/rxrpc/call_object.c
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

605 lines
16 KiB
C
Raw Blame History

/* RxRPC individual remote procedure call handling
*
* Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/circ_buf.h>
#include <linux/spinlock_types.h>
#include <net/sock.h>
#include <net/af_rxrpc.h>
#include "ar-internal.h"
const char *const rxrpc_call_states[NR__RXRPC_CALL_STATES] = {
[RXRPC_CALL_UNINITIALISED] = "Uninit ",
[RXRPC_CALL_CLIENT_AWAIT_CONN] = "ClWtConn",
[RXRPC_CALL_CLIENT_SEND_REQUEST] = "ClSndReq",
[RXRPC_CALL_CLIENT_AWAIT_REPLY] = "ClAwtRpl",
[RXRPC_CALL_CLIENT_RECV_REPLY] = "ClRcvRpl",
[RXRPC_CALL_SERVER_PREALLOC] = "SvPrealc",
[RXRPC_CALL_SERVER_SECURING] = "SvSecure",
[RXRPC_CALL_SERVER_ACCEPTING] = "SvAccept",
[RXRPC_CALL_SERVER_RECV_REQUEST] = "SvRcvReq",
[RXRPC_CALL_SERVER_ACK_REQUEST] = "SvAckReq",
[RXRPC_CALL_SERVER_SEND_REPLY] = "SvSndRpl",
[RXRPC_CALL_SERVER_AWAIT_ACK] = "SvAwtACK",
[RXRPC_CALL_COMPLETE] = "Complete",
};
const char *const rxrpc_call_completions[NR__RXRPC_CALL_COMPLETIONS] = {
[RXRPC_CALL_SUCCEEDED] = "Complete",
[RXRPC_CALL_REMOTELY_ABORTED] = "RmtAbort",
[RXRPC_CALL_LOCALLY_ABORTED] = "LocAbort",
[RXRPC_CALL_LOCAL_ERROR] = "LocError",
[RXRPC_CALL_NETWORK_ERROR] = "NetError",
};
struct kmem_cache *rxrpc_call_jar;
LIST_HEAD(rxrpc_calls);
DEFINE_RWLOCK(rxrpc_call_lock);
static void rxrpc_call_timer_expired(unsigned long _call)
{
struct rxrpc_call *call = (struct rxrpc_call *)_call;
_enter("%d", call->debug_id);
if (call->state < RXRPC_CALL_COMPLETE)
rxrpc_set_timer(call, rxrpc_timer_expired, ktime_get_real());
}
/*
* find an extant server call
* - called in process context with IRQs enabled
*/
struct rxrpc_call *rxrpc_find_call_by_user_ID(struct rxrpc_sock *rx,
unsigned long user_call_ID)
{
struct rxrpc_call *call;
struct rb_node *p;
_enter("%p,%lx", rx, user_call_ID);
read_lock(&rx->call_lock);
p = rx->calls.rb_node;
while (p) {
call = rb_entry(p, struct rxrpc_call, sock_node);
if (user_call_ID < call->user_call_ID)
p = p->rb_left;
else if (user_call_ID > call->user_call_ID)
p = p->rb_right;
else
goto found_extant_call;
}
read_unlock(&rx->call_lock);
_leave(" = NULL");
return NULL;
found_extant_call:
rxrpc_get_call(call, rxrpc_call_got);
read_unlock(&rx->call_lock);
_leave(" = %p [%d]", call, atomic_read(&call->usage));
return call;
}
/*
* allocate a new call
*/
struct rxrpc_call *rxrpc_alloc_call(gfp_t gfp)
{
struct rxrpc_call *call;
call = kmem_cache_zalloc(rxrpc_call_jar, gfp);
if (!call)
return NULL;
call->rxtx_buffer = kcalloc(RXRPC_RXTX_BUFF_SIZE,
sizeof(struct sk_buff *),
gfp);
if (!call->rxtx_buffer)
goto nomem;
call->rxtx_annotations = kcalloc(RXRPC_RXTX_BUFF_SIZE, sizeof(u8), gfp);
if (!call->rxtx_annotations)
goto nomem_2;
mutex_init(&call->user_mutex);
setup_timer(&call->timer, rxrpc_call_timer_expired,
(unsigned long)call);
INIT_WORK(&call->processor, &rxrpc_process_call);
INIT_LIST_HEAD(&call->link);
INIT_LIST_HEAD(&call->chan_wait_link);
INIT_LIST_HEAD(&call->accept_link);
INIT_LIST_HEAD(&call->recvmsg_link);
INIT_LIST_HEAD(&call->sock_link);
init_waitqueue_head(&call->waitq);
spin_lock_init(&call->lock);
rwlock_init(&call->state_lock);
atomic_set(&call->usage, 1);
call->debug_id = atomic_inc_return(&rxrpc_debug_id);
memset(&call->sock_node, 0xed, sizeof(call->sock_node));
/* Leave space in the ring to handle a maxed-out jumbo packet */
call->rx_winsize = rxrpc_rx_window_size;
call->tx_winsize = 16;
call->rx_expect_next = 1;
if (RXRPC_TX_SMSS > 2190)
call->cong_cwnd = 2;
else if (RXRPC_TX_SMSS > 1095)
call->cong_cwnd = 3;
else
call->cong_cwnd = 4;
call->cong_ssthresh = RXRPC_RXTX_BUFF_SIZE - 1;
return call;
nomem_2:
kfree(call->rxtx_buffer);
nomem:
kmem_cache_free(rxrpc_call_jar, call);
return NULL;
}
/*
* Allocate a new client call.
*/
static struct rxrpc_call *rxrpc_alloc_client_call(struct sockaddr_rxrpc *srx,
gfp_t gfp)
{
struct rxrpc_call *call;
ktime_t now;
_enter("");
call = rxrpc_alloc_call(gfp);
if (!call)
return ERR_PTR(-ENOMEM);
call->state = RXRPC_CALL_CLIENT_AWAIT_CONN;
call->service_id = srx->srx_service;
call->tx_phase = true;
now = ktime_get_real();
call->acks_latest_ts = now;
call->cong_tstamp = now;
_leave(" = %p", call);
return call;
}
/*
* Initiate the call ack/resend/expiry timer.
*/
static void rxrpc_start_call_timer(struct rxrpc_call *call)
{
ktime_t now = ktime_get_real(), expire_at;
expire_at = ktime_add_ms(now, rxrpc_max_call_lifetime);
call->expire_at = expire_at;
call->ack_at = expire_at;
call->ping_at = expire_at;
call->resend_at = expire_at;
call->timer.expires = jiffies + LONG_MAX / 2;
rxrpc_set_timer(call, rxrpc_timer_begin, now);
}
/*
* Set up a call for the given parameters.
* - Called with the socket lock held, which it must release.
* - If it returns a call, the call's lock will need releasing by the caller.
*/
struct rxrpc_call *rxrpc_new_client_call(struct rxrpc_sock *rx,
struct rxrpc_conn_parameters *cp,
struct sockaddr_rxrpc *srx,
unsigned long user_call_ID,
gfp_t gfp)
__releases(&rx->sk.sk_lock.slock)
{
struct rxrpc_call *call, *xcall;
struct rb_node *parent, **pp;
const void *here = __builtin_return_address(0);
int ret;
_enter("%p,%lx", rx, user_call_ID);
call = rxrpc_alloc_client_call(srx, gfp);
if (IS_ERR(call)) {
release_sock(&rx->sk);
_leave(" = %ld", PTR_ERR(call));
return call;
}
trace_rxrpc_call(call, rxrpc_call_new_client, atomic_read(&call->usage),
here, (const void *)user_call_ID);
/* We need to protect a partially set up call against the user as we
* will be acting outside the socket lock.
*/
mutex_lock(&call->user_mutex);
/* Publish the call, even though it is incompletely set up as yet */
write_lock(&rx->call_lock);
pp = &rx->calls.rb_node;
parent = NULL;
while (*pp) {
parent = *pp;
xcall = rb_entry(parent, struct rxrpc_call, sock_node);
if (user_call_ID < xcall->user_call_ID)
pp = &(*pp)->rb_left;
else if (user_call_ID > xcall->user_call_ID)
pp = &(*pp)->rb_right;
else
goto error_dup_user_ID;
}
rcu_assign_pointer(call->socket, rx);
call->user_call_ID = user_call_ID;
__set_bit(RXRPC_CALL_HAS_USERID, &call->flags);
rxrpc_get_call(call, rxrpc_call_got_userid);
rb_link_node(&call->sock_node, parent, pp);
rb_insert_color(&call->sock_node, &rx->calls);
list_add(&call->sock_link, &rx->sock_calls);
write_unlock(&rx->call_lock);
write_lock(&rxrpc_call_lock);
list_add_tail(&call->link, &rxrpc_calls);
write_unlock(&rxrpc_call_lock);
/* From this point on, the call is protected by its own lock. */
release_sock(&rx->sk);
/* Set up or get a connection record and set the protocol parameters,
* including channel number and call ID.
*/
ret = rxrpc_connect_call(call, cp, srx, gfp);
if (ret < 0)
goto error;
trace_rxrpc_call(call, rxrpc_call_connected, atomic_read(&call->usage),
here, NULL);
spin_lock_bh(&call->conn->params.peer->lock);
hlist_add_head(&call->error_link,
&call->conn->params.peer->error_targets);
spin_unlock_bh(&call->conn->params.peer->lock);
rxrpc_start_call_timer(call);
_net("CALL new %d on CONN %d", call->debug_id, call->conn->debug_id);
_leave(" = %p [new]", call);
return call;
/* We unexpectedly found the user ID in the list after taking
* the call_lock. This shouldn't happen unless the user races
* with itself and tries to add the same user ID twice at the
* same time in different threads.
*/
error_dup_user_ID:
write_unlock(&rx->call_lock);
release_sock(&rx->sk);
ret = -EEXIST;
error:
__rxrpc_set_call_completion(call, RXRPC_CALL_LOCAL_ERROR,
RX_CALL_DEAD, ret);
trace_rxrpc_call(call, rxrpc_call_error, atomic_read(&call->usage),
here, ERR_PTR(ret));
rxrpc_release_call(rx, call);
mutex_unlock(&call->user_mutex);
rxrpc_put_call(call, rxrpc_call_put);
_leave(" = %d", ret);
return ERR_PTR(ret);
}
/*
* Set up an incoming call. call->conn points to the connection.
* This is called in BH context and isn't allowed to fail.
*/
void rxrpc_incoming_call(struct rxrpc_sock *rx,
struct rxrpc_call *call,
struct sk_buff *skb)
{
struct rxrpc_connection *conn = call->conn;
struct rxrpc_skb_priv *sp = rxrpc_skb(skb);
u32 chan;
_enter(",%d", call->conn->debug_id);
rcu_assign_pointer(call->socket, rx);
call->call_id = sp->hdr.callNumber;
call->service_id = sp->hdr.serviceId;
call->cid = sp->hdr.cid;
call->state = RXRPC_CALL_SERVER_ACCEPTING;
if (sp->hdr.securityIndex > 0)
call->state = RXRPC_CALL_SERVER_SECURING;
call->cong_tstamp = skb->tstamp;
/* Set the channel for this call. We don't get channel_lock as we're
* only defending against the data_ready handler (which we're called
* from) and the RESPONSE packet parser (which is only really
* interested in call_counter and can cope with a disagreement with the
* call pointer).
*/
chan = sp->hdr.cid & RXRPC_CHANNELMASK;
conn->channels[chan].call_counter = call->call_id;
conn->channels[chan].call_id = call->call_id;
rcu_assign_pointer(conn->channels[chan].call, call);
spin_lock(&conn->params.peer->lock);
hlist_add_head(&call->error_link, &conn->params.peer->error_targets);
spin_unlock(&conn->params.peer->lock);
_net("CALL incoming %d on CONN %d", call->debug_id, call->conn->debug_id);
rxrpc_start_call_timer(call);
_leave("");
}
/*
* Queue a call's work processor, getting a ref to pass to the work queue.
*/
bool rxrpc_queue_call(struct rxrpc_call *call)
{
const void *here = __builtin_return_address(0);
int n = __atomic_add_unless(&call->usage, 1, 0);
if (n == 0)
return false;
if (rxrpc_queue_work(&call->processor))
trace_rxrpc_call(call, rxrpc_call_queued, n + 1, here, NULL);
else
rxrpc_put_call(call, rxrpc_call_put_noqueue);
return true;
}
/*
* Queue a call's work processor, passing the callers ref to the work queue.
*/
bool __rxrpc_queue_call(struct rxrpc_call *call)
{
const void *here = __builtin_return_address(0);
int n = atomic_read(&call->usage);
ASSERTCMP(n, >=, 1);
if (rxrpc_queue_work(&call->processor))
trace_rxrpc_call(call, rxrpc_call_queued_ref, n, here, NULL);
else
rxrpc_put_call(call, rxrpc_call_put_noqueue);
return true;
}
/*
* Note the re-emergence of a call.
*/
void rxrpc_see_call(struct rxrpc_call *call)
{
const void *here = __builtin_return_address(0);
if (call) {
int n = atomic_read(&call->usage);
trace_rxrpc_call(call, rxrpc_call_seen, n, here, NULL);
}
}
/*
* Note the addition of a ref on a call.
*/
void rxrpc_get_call(struct rxrpc_call *call, enum rxrpc_call_trace op)
{
const void *here = __builtin_return_address(0);
int n = atomic_inc_return(&call->usage);
trace_rxrpc_call(call, op, n, here, NULL);
}
/*
* Detach a call from its owning socket.
*/
void rxrpc_release_call(struct rxrpc_sock *rx, struct rxrpc_call *call)
{
const void *here = __builtin_return_address(0);
struct rxrpc_connection *conn = call->conn;
bool put = false;
int i;
_enter("{%d,%d}", call->debug_id, atomic_read(&call->usage));
trace_rxrpc_call(call, rxrpc_call_release, atomic_read(&call->usage),
here, (const void *)call->flags);
ASSERTCMP(call->state, ==, RXRPC_CALL_COMPLETE);
spin_lock_bh(&call->lock);
if (test_and_set_bit(RXRPC_CALL_RELEASED, &call->flags))
BUG();
spin_unlock_bh(&call->lock);
del_timer_sync(&call->timer);
/* Make sure we don't get any more notifications */
write_lock_bh(&rx->recvmsg_lock);
if (!list_empty(&call->recvmsg_link)) {
_debug("unlinking once-pending call %p { e=%lx f=%lx }",
call, call->events, call->flags);
list_del(&call->recvmsg_link);
put = true;
}
/* list_empty() must return false in rxrpc_notify_socket() */
call->recvmsg_link.next = NULL;
call->recvmsg_link.prev = NULL;
write_unlock_bh(&rx->recvmsg_lock);
if (put)
rxrpc_put_call(call, rxrpc_call_put);
write_lock(&rx->call_lock);
if (test_and_clear_bit(RXRPC_CALL_HAS_USERID, &call->flags)) {
rb_erase(&call->sock_node, &rx->calls);
memset(&call->sock_node, 0xdd, sizeof(call->sock_node));
rxrpc_put_call(call, rxrpc_call_put_userid);
}
list_del(&call->sock_link);
write_unlock(&rx->call_lock);
_debug("RELEASE CALL %p (%d CONN %p)", call, call->debug_id, conn);
if (conn)
rxrpc_disconnect_call(call);
for (i = 0; i < RXRPC_RXTX_BUFF_SIZE; i++) {
rxrpc_free_skb(call->rxtx_buffer[i],
(call->tx_phase ? rxrpc_skb_tx_cleaned :
rxrpc_skb_rx_cleaned));
call->rxtx_buffer[i] = NULL;
}
_leave("");
}
/*
* release all the calls associated with a socket
*/
void rxrpc_release_calls_on_socket(struct rxrpc_sock *rx)
{
struct rxrpc_call *call;
_enter("%p", rx);
while (!list_empty(&rx->to_be_accepted)) {
call = list_entry(rx->to_be_accepted.next,
struct rxrpc_call, accept_link);
list_del(&call->accept_link);
rxrpc_abort_call("SKR", call, 0, RX_CALL_DEAD, ECONNRESET);
rxrpc_put_call(call, rxrpc_call_put);
}
while (!list_empty(&rx->sock_calls)) {
call = list_entry(rx->sock_calls.next,
struct rxrpc_call, sock_link);
rxrpc_get_call(call, rxrpc_call_got);
rxrpc_abort_call("SKT", call, 0, RX_CALL_DEAD, ECONNRESET);
rxrpc_send_abort_packet(call);
rxrpc_release_call(rx, call);
rxrpc_put_call(call, rxrpc_call_put);
}
_leave("");
}
/*
* release a call
*/
void rxrpc_put_call(struct rxrpc_call *call, enum rxrpc_call_trace op)
{
const void *here = __builtin_return_address(0);
int n;
ASSERT(call != NULL);
n = atomic_dec_return(&call->usage);
trace_rxrpc_call(call, op, n, here, NULL);
ASSERTCMP(n, >=, 0);
if (n == 0) {
_debug("call %d dead", call->debug_id);
ASSERTCMP(call->state, ==, RXRPC_CALL_COMPLETE);
write_lock(&rxrpc_call_lock);
list_del_init(&call->link);
write_unlock(&rxrpc_call_lock);
rxrpc_cleanup_call(call);
}
}
/*
* Final call destruction under RCU.
*/
static void rxrpc_rcu_destroy_call(struct rcu_head *rcu)
{
struct rxrpc_call *call = container_of(rcu, struct rxrpc_call, rcu);
rxrpc_put_peer(call->peer);
kfree(call->rxtx_buffer);
kfree(call->rxtx_annotations);
kmem_cache_free(rxrpc_call_jar, call);
}
/*
* clean up a call
*/
void rxrpc_cleanup_call(struct rxrpc_call *call)
{
int i;
_net("DESTROY CALL %d", call->debug_id);
memset(&call->sock_node, 0xcd, sizeof(call->sock_node));
del_timer_sync(&call->timer);
ASSERTCMP(call->state, ==, RXRPC_CALL_COMPLETE);
ASSERT(test_bit(RXRPC_CALL_RELEASED, &call->flags));
ASSERTCMP(call->conn, ==, NULL);
/* Clean up the Rx/Tx buffer */
for (i = 0; i < RXRPC_RXTX_BUFF_SIZE; i++)
rxrpc_free_skb(call->rxtx_buffer[i],
(call->tx_phase ? rxrpc_skb_tx_cleaned :
rxrpc_skb_rx_cleaned));
rxrpc_free_skb(call->tx_pending, rxrpc_skb_tx_cleaned);
call_rcu(&call->rcu, rxrpc_rcu_destroy_call);
}
/*
* Make sure that all calls are gone.
*/
void __exit rxrpc_destroy_all_calls(void)
{
struct rxrpc_call *call;
_enter("");
if (list_empty(&rxrpc_calls))
return;
write_lock(&rxrpc_call_lock);
while (!list_empty(&rxrpc_calls)) {
call = list_entry(rxrpc_calls.next, struct rxrpc_call, link);
_debug("Zapping call %p", call);
rxrpc_see_call(call);
list_del_init(&call->link);
pr_err("Call %p still in use (%d,%s,%lx,%lx)!\n",
call, atomic_read(&call->usage),
rxrpc_call_states[call->state],
call->flags, call->events);
write_unlock(&rxrpc_call_lock);
cond_resched();
write_lock(&rxrpc_call_lock);
}
write_unlock(&rxrpc_call_lock);
}
Reference in New Issue View Git Blame Copy Permalink