ipvs: changes related to service usecnt
Change the usage of svc usecnt during command execution:
- we check if svc is registered but we do not need to hold usecnt
reference while under __ip_vs_mutex, only the packet handling needs
it during scheduling
- change __ip_vs_service_get to __ip_vs_service_find and
__ip_vs_svc_fwm_get to __ip_vs_svc_fwm_find because now caller
will increase svc->usecnt
- put common code that calls update_service in __ip_vs_update_dest
- put common code in ip_vs_unlink_service() and use it to unregister
the service
- add comment that svc should not be accessed after ip_vs_del_service
anymore
- all IP_VS_WAIT_WHILE calls are now unified: usecnt > 0
- Properly log the app ports
As result, some problems are fixed:
- possible use-after-free of svc in ip_vs_genl_set_cmd after
ip_vs_del_service because our usecnt reference does not guarantee that
svc is not freed on refcnt==0, eg. when no dests are moved to trash
- possible usecnt leak in do_ip_vs_set_ctl after ip_vs_del_service
when the service is not freed now, for example, when some
destionations are moved into trash and svc->refcnt remains above 0.
It is harmless because svc is not in hash anymore.
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Patrick McHardy <kaber@trash.net>
diff --git a/net/netfilter/ipvs/ip_vs_app.c b/net/netfilter/ipvs/ip_vs_app.c
index e76f87f..a475ede 100644
--- a/net/netfilter/ipvs/ip_vs_app.c
+++ b/net/netfilter/ipvs/ip_vs_app.c
@@ -103,8 +103,8 @@
goto out;
list_add(&inc->a_list, &app->incs_list);
- IP_VS_DBG(9, "%s application %s:%u registered\n",
- pp->name, inc->name, inc->port);
+ IP_VS_DBG(9, "%s App %s:%u registered\n",
+ pp->name, inc->name, ntohs(inc->port));
return 0;
@@ -130,7 +130,7 @@
pp->unregister_app(inc);
IP_VS_DBG(9, "%s App %s:%u unregistered\n",
- pp->name, inc->name, inc->port);
+ pp->name, inc->name, ntohs(inc->port));
list_del(&inc->a_list);
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index e637cd0..e4ec8f3 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -405,7 +405,7 @@
* Get service by {proto,addr,port} in the service table.
*/
static inline struct ip_vs_service *
-__ip_vs_service_get(int af, __u16 protocol, const union nf_inet_addr *vaddr,
+__ip_vs_service_find(int af, __u16 protocol, const union nf_inet_addr *vaddr,
__be16 vport)
{
unsigned hash;
@@ -420,7 +420,6 @@
&& (svc->port == vport)
&& (svc->protocol == protocol)) {
/* HIT */
- atomic_inc(&svc->usecnt);
return svc;
}
}
@@ -433,7 +432,7 @@
* Get service by {fwmark} in the service table.
*/
static inline struct ip_vs_service *
-__ip_vs_svc_fwm_get(int af, __u32 fwmark)
+__ip_vs_svc_fwm_find(int af, __u32 fwmark)
{
unsigned hash;
struct ip_vs_service *svc;
@@ -444,7 +443,6 @@
list_for_each_entry(svc, &ip_vs_svc_fwm_table[hash], f_list) {
if (svc->fwmark == fwmark && svc->af == af) {
/* HIT */
- atomic_inc(&svc->usecnt);
return svc;
}
}
@@ -463,14 +461,14 @@
/*
* Check the table hashed by fwmark first
*/
- if (fwmark && (svc = __ip_vs_svc_fwm_get(af, fwmark)))
+ if (fwmark && (svc = __ip_vs_svc_fwm_find(af, fwmark)))
goto out;
/*
* Check the table hashed by <protocol,addr,port>
* for "full" addressed entries
*/
- svc = __ip_vs_service_get(af, protocol, vaddr, vport);
+ svc = __ip_vs_service_find(af, protocol, vaddr, vport);
if (svc == NULL
&& protocol == IPPROTO_TCP
@@ -480,7 +478,7 @@
* Check if ftp service entry exists, the packet
* might belong to FTP data connections.
*/
- svc = __ip_vs_service_get(af, protocol, vaddr, FTPPORT);
+ svc = __ip_vs_service_find(af, protocol, vaddr, FTPPORT);
}
if (svc == NULL
@@ -488,10 +486,12 @@
/*
* Check if the catch-all port (port zero) exists
*/
- svc = __ip_vs_service_get(af, protocol, vaddr, 0);
+ svc = __ip_vs_service_find(af, protocol, vaddr, 0);
}
out:
+ if (svc)
+ atomic_inc(&svc->usecnt);
read_unlock(&__ip_vs_svc_lock);
IP_VS_DBG_BUF(9, "lookup service: fwm %u %s %s:%u %s\n",
@@ -510,14 +510,19 @@
dest->svc = svc;
}
-static inline void
+static void
__ip_vs_unbind_svc(struct ip_vs_dest *dest)
{
struct ip_vs_service *svc = dest->svc;
dest->svc = NULL;
- if (atomic_dec_and_test(&svc->refcnt))
+ if (atomic_dec_and_test(&svc->refcnt)) {
+ IP_VS_DBG_BUF(3, "Removing service %u/%s:%u usecnt=%d\n",
+ svc->fwmark,
+ IP_VS_DBG_ADDR(svc->af, &svc->addr),
+ ntohs(svc->port), atomic_read(&svc->usecnt));
kfree(svc);
+ }
}
@@ -762,8 +767,8 @@
* Update a destination in the given service
*/
static void
-__ip_vs_update_dest(struct ip_vs_service *svc,
- struct ip_vs_dest *dest, struct ip_vs_dest_user_kern *udest)
+__ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
+ struct ip_vs_dest_user_kern *udest, int add)
{
int conn_flags;
@@ -818,6 +823,25 @@
dest->flags &= ~IP_VS_DEST_F_OVERLOAD;
dest->u_threshold = udest->u_threshold;
dest->l_threshold = udest->l_threshold;
+
+ if (add)
+ ip_vs_new_estimator(&dest->stats);
+
+ write_lock_bh(&__ip_vs_svc_lock);
+
+ /* Wait until all other svc users go away */
+ IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
+
+ if (add) {
+ list_add(&dest->n_list, &svc->destinations);
+ svc->num_dests++;
+ }
+
+ /* call the update_service, because server weight may be changed */
+ if (svc->scheduler->update_service)
+ svc->scheduler->update_service(svc);
+
+ write_unlock_bh(&__ip_vs_svc_lock);
}
@@ -865,13 +889,12 @@
atomic_set(&dest->activeconns, 0);
atomic_set(&dest->inactconns, 0);
atomic_set(&dest->persistconns, 0);
- atomic_set(&dest->refcnt, 0);
+ atomic_set(&dest->refcnt, 1);
INIT_LIST_HEAD(&dest->d_list);
spin_lock_init(&dest->dst_lock);
spin_lock_init(&dest->stats.lock);
- __ip_vs_update_dest(svc, dest, udest);
- ip_vs_new_estimator(&dest->stats);
+ __ip_vs_update_dest(svc, dest, udest, 1);
*dest_p = dest;
@@ -931,65 +954,22 @@
IP_VS_DBG_ADDR(svc->af, &dest->vaddr),
ntohs(dest->vport));
- __ip_vs_update_dest(svc, dest, udest);
-
/*
* Get the destination from the trash
*/
list_del(&dest->n_list);
- ip_vs_new_estimator(&dest->stats);
-
- write_lock_bh(&__ip_vs_svc_lock);
-
+ __ip_vs_update_dest(svc, dest, udest, 1);
+ ret = 0;
+ } else {
/*
- * Wait until all other svc users go away.
+ * Allocate and initialize the dest structure
*/
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
-
- list_add(&dest->n_list, &svc->destinations);
- svc->num_dests++;
-
- /* call the update_service function of its scheduler */
- if (svc->scheduler->update_service)
- svc->scheduler->update_service(svc);
-
- write_unlock_bh(&__ip_vs_svc_lock);
- return 0;
+ ret = ip_vs_new_dest(svc, udest, &dest);
}
-
- /*
- * Allocate and initialize the dest structure
- */
- ret = ip_vs_new_dest(svc, udest, &dest);
- if (ret) {
- return ret;
- }
-
- /*
- * Add the dest entry into the list
- */
- atomic_inc(&dest->refcnt);
-
- write_lock_bh(&__ip_vs_svc_lock);
-
- /*
- * Wait until all other svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
-
- list_add(&dest->n_list, &svc->destinations);
- svc->num_dests++;
-
- /* call the update_service function of its scheduler */
- if (svc->scheduler->update_service)
- svc->scheduler->update_service(svc);
-
- write_unlock_bh(&__ip_vs_svc_lock);
-
LeaveFunction(2);
- return 0;
+ return ret;
}
@@ -1028,19 +1008,7 @@
return -ENOENT;
}
- __ip_vs_update_dest(svc, dest, udest);
-
- write_lock_bh(&__ip_vs_svc_lock);
-
- /* Wait until all other svc users go away */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
-
- /* call the update_service, because server weight may be changed */
- if (svc->scheduler->update_service)
- svc->scheduler->update_service(svc);
-
- write_unlock_bh(&__ip_vs_svc_lock);
-
+ __ip_vs_update_dest(svc, dest, udest, 0);
LeaveFunction(2);
return 0;
@@ -1067,6 +1035,10 @@
* the destination into the trash.
*/
if (atomic_dec_and_test(&dest->refcnt)) {
+ IP_VS_DBG_BUF(3, "Removing destination %u/%s:%u\n",
+ dest->vfwmark,
+ IP_VS_DBG_ADDR(dest->af, &dest->addr),
+ ntohs(dest->port));
ip_vs_dst_reset(dest);
/* simply decrease svc->refcnt here, let the caller check
and release the service if nobody refers to it.
@@ -1133,7 +1105,7 @@
/*
* Wait until all other svc users go away.
*/
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
+ IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
/*
* Unlink dest from the service
@@ -1190,7 +1162,7 @@
}
/* I'm the first user of the service */
- atomic_set(&svc->usecnt, 1);
+ atomic_set(&svc->usecnt, 0);
atomic_set(&svc->refcnt, 0);
svc->af = u->af;
@@ -1284,7 +1256,7 @@
/*
* Wait until all other svc users go away.
*/
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
+ IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
/*
* Set the flags and timeout value
@@ -1383,21 +1355,23 @@
/*
* Free the service if nobody refers to it
*/
- if (atomic_read(&svc->refcnt) == 0)
+ if (atomic_read(&svc->refcnt) == 0) {
+ IP_VS_DBG_BUF(3, "Removing service %u/%s:%u usecnt=%d\n",
+ svc->fwmark,
+ IP_VS_DBG_ADDR(svc->af, &svc->addr),
+ ntohs(svc->port), atomic_read(&svc->usecnt));
kfree(svc);
+ }
/* decrease the module use count */
ip_vs_use_count_dec();
}
/*
- * Delete a service from the service list
+ * Unlink a service from list and try to delete it if its refcnt reached 0
*/
-static int ip_vs_del_service(struct ip_vs_service *svc)
+static void ip_vs_unlink_service(struct ip_vs_service *svc)
{
- if (svc == NULL)
- return -EEXIST;
-
/*
* Unhash it from the service table
*/
@@ -1408,11 +1382,21 @@
/*
* Wait until all the svc users go away.
*/
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 1);
+ IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
__ip_vs_del_service(svc);
write_unlock_bh(&__ip_vs_svc_lock);
+}
+
+/*
+ * Delete a service from the service list
+ */
+static int ip_vs_del_service(struct ip_vs_service *svc)
+{
+ if (svc == NULL)
+ return -EEXIST;
+ ip_vs_unlink_service(svc);
return 0;
}
@@ -1431,14 +1415,7 @@
*/
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
list_for_each_entry_safe(svc, nxt, &ip_vs_svc_table[idx], s_list) {
- write_lock_bh(&__ip_vs_svc_lock);
- ip_vs_svc_unhash(svc);
- /*
- * Wait until all the svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
- __ip_vs_del_service(svc);
- write_unlock_bh(&__ip_vs_svc_lock);
+ ip_vs_unlink_service(svc);
}
}
@@ -1448,14 +1425,7 @@
for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
list_for_each_entry_safe(svc, nxt,
&ip_vs_svc_fwm_table[idx], f_list) {
- write_lock_bh(&__ip_vs_svc_lock);
- ip_vs_svc_unhash(svc);
- /*
- * Wait until all the svc users go away.
- */
- IP_VS_WAIT_WHILE(atomic_read(&svc->usecnt) > 0);
- __ip_vs_del_service(svc);
- write_unlock_bh(&__ip_vs_svc_lock);
+ ip_vs_unlink_service(svc);
}
}
@@ -2168,15 +2138,15 @@
/* Lookup the exact service by <protocol, addr, port> or fwmark */
if (usvc.fwmark == 0)
- svc = __ip_vs_service_get(usvc.af, usvc.protocol,
- &usvc.addr, usvc.port);
+ svc = __ip_vs_service_find(usvc.af, usvc.protocol,
+ &usvc.addr, usvc.port);
else
- svc = __ip_vs_svc_fwm_get(usvc.af, usvc.fwmark);
+ svc = __ip_vs_svc_fwm_find(usvc.af, usvc.fwmark);
if (cmd != IP_VS_SO_SET_ADD
&& (svc == NULL || svc->protocol != usvc.protocol)) {
ret = -ESRCH;
- goto out_drop_service;
+ goto out_unlock;
}
switch (cmd) {
@@ -2210,10 +2180,6 @@
ret = -EINVAL;
}
-out_drop_service:
- if (svc)
- ip_vs_service_put(svc);
-
out_unlock:
mutex_unlock(&__ip_vs_mutex);
out_dec:
@@ -2306,10 +2272,10 @@
int ret = 0;
if (get->fwmark)
- svc = __ip_vs_svc_fwm_get(AF_INET, get->fwmark);
+ svc = __ip_vs_svc_fwm_find(AF_INET, get->fwmark);
else
- svc = __ip_vs_service_get(AF_INET, get->protocol, &addr,
- get->port);
+ svc = __ip_vs_service_find(AF_INET, get->protocol, &addr,
+ get->port);
if (svc) {
int count = 0;
@@ -2337,7 +2303,6 @@
}
count++;
}
- ip_vs_service_put(svc);
} else
ret = -ESRCH;
return ret;
@@ -2458,15 +2423,14 @@
entry = (struct ip_vs_service_entry *)arg;
addr.ip = entry->addr;
if (entry->fwmark)
- svc = __ip_vs_svc_fwm_get(AF_INET, entry->fwmark);
+ svc = __ip_vs_svc_fwm_find(AF_INET, entry->fwmark);
else
- svc = __ip_vs_service_get(AF_INET, entry->protocol,
- &addr, entry->port);
+ svc = __ip_vs_service_find(AF_INET, entry->protocol,
+ &addr, entry->port);
if (svc) {
ip_vs_copy_service(entry, svc);
if (copy_to_user(user, entry, sizeof(*entry)) != 0)
ret = -EFAULT;
- ip_vs_service_put(svc);
} else
ret = -ESRCH;
}
@@ -2733,10 +2697,12 @@
}
static int ip_vs_genl_parse_service(struct ip_vs_service_user_kern *usvc,
- struct nlattr *nla, int full_entry)
+ struct nlattr *nla, int full_entry,
+ struct ip_vs_service **ret_svc)
{
struct nlattr *attrs[IPVS_SVC_ATTR_MAX + 1];
struct nlattr *nla_af, *nla_port, *nla_fwmark, *nla_protocol, *nla_addr;
+ struct ip_vs_service *svc;
/* Parse mandatory identifying service fields first */
if (nla == NULL ||
@@ -2772,12 +2738,18 @@
usvc->fwmark = 0;
}
+ if (usvc->fwmark)
+ svc = __ip_vs_svc_fwm_find(usvc->af, usvc->fwmark);
+ else
+ svc = __ip_vs_service_find(usvc->af, usvc->protocol,
+ &usvc->addr, usvc->port);
+ *ret_svc = svc;
+
/* If a full entry was requested, check for the additional fields */
if (full_entry) {
struct nlattr *nla_sched, *nla_flags, *nla_timeout,
*nla_netmask;
struct ip_vs_flags flags;
- struct ip_vs_service *svc;
nla_sched = attrs[IPVS_SVC_ATTR_SCHED_NAME];
nla_flags = attrs[IPVS_SVC_ATTR_FLAGS];
@@ -2790,16 +2762,8 @@
nla_memcpy(&flags, nla_flags, sizeof(flags));
/* prefill flags from service if it already exists */
- if (usvc->fwmark)
- svc = __ip_vs_svc_fwm_get(usvc->af, usvc->fwmark);
- else
- svc = __ip_vs_service_get(usvc->af, usvc->protocol,
- &usvc->addr, usvc->port);
- if (svc) {
+ if (svc)
usvc->flags = svc->flags;
- ip_vs_service_put(svc);
- } else
- usvc->flags = 0;
/* set new flags from userland */
usvc->flags = (usvc->flags & ~flags.mask) |
@@ -2815,17 +2779,11 @@
static struct ip_vs_service *ip_vs_genl_find_service(struct nlattr *nla)
{
struct ip_vs_service_user_kern usvc;
+ struct ip_vs_service *svc;
int ret;
- ret = ip_vs_genl_parse_service(&usvc, nla, 0);
- if (ret)
- return ERR_PTR(ret);
-
- if (usvc.fwmark)
- return __ip_vs_svc_fwm_get(usvc.af, usvc.fwmark);
- else
- return __ip_vs_service_get(usvc.af, usvc.protocol,
- &usvc.addr, usvc.port);
+ ret = ip_vs_genl_parse_service(&usvc, nla, 0, &svc);
+ return ret ? ERR_PTR(ret) : svc;
}
static int ip_vs_genl_fill_dest(struct sk_buff *skb, struct ip_vs_dest *dest)
@@ -2916,7 +2874,6 @@
nla_put_failure:
cb->args[0] = idx;
- ip_vs_service_put(svc);
out_err:
mutex_unlock(&__ip_vs_mutex);
@@ -3129,17 +3086,10 @@
ret = ip_vs_genl_parse_service(&usvc,
info->attrs[IPVS_CMD_ATTR_SERVICE],
- need_full_svc);
+ need_full_svc, &svc);
if (ret)
goto out;
- /* Lookup the exact service by <protocol, addr, port> or fwmark */
- if (usvc.fwmark == 0)
- svc = __ip_vs_service_get(usvc.af, usvc.protocol,
- &usvc.addr, usvc.port);
- else
- svc = __ip_vs_svc_fwm_get(usvc.af, usvc.fwmark);
-
/* Unless we're adding a new service, the service must already exist */
if ((cmd != IPVS_CMD_NEW_SERVICE) && (svc == NULL)) {
ret = -ESRCH;
@@ -3173,6 +3123,7 @@
break;
case IPVS_CMD_DEL_SERVICE:
ret = ip_vs_del_service(svc);
+ /* do not use svc, it can be freed */
break;
case IPVS_CMD_NEW_DEST:
ret = ip_vs_add_dest(svc, &udest);
@@ -3191,8 +3142,6 @@
}
out:
- if (svc)
- ip_vs_service_put(svc);
mutex_unlock(&__ip_vs_mutex);
return ret;
@@ -3238,7 +3187,6 @@
goto out_err;
} else if (svc) {
ret = ip_vs_genl_fill_service(msg, svc);
- ip_vs_service_put(svc);
if (ret)
goto nla_put_failure;
} else {