Skip to contentSkip to conversation

RTNL: assertion failed at net/ipv6/addrconf.c (1699)

PATCH Superseded Tommi Rantala 8 participants 34 messages Aug 29, 2014 – Sep 5, 2014

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Hi,

Was fuzzing Linus v3.17-rc2-89-g59753a8 with Trinity as the root user in qemu, when I hit the following assertion failures.

Tommi

[init] Started watchdog process, PID is 4841
[main] Main thread is alive.

[   77.229699] sctp: [Deprecated]: trinity-main (pid 4842) Use of int
in max_burst socket option deprecated.
[   77.229699] Use struct sctp_assoc_value instead
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000
ffff8800382d8220
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20
[   77.308570]  [<ffffffff8117097d>] ? sched_clock_local+0x1d/0x80
[   77.309260]  [<ffffffff810a8a27>] ? kvm_clock_read+0x27/0x40
[   77.309915]  [<ffffffff810736d9>] ? sched_clock+0x9/0x10
[   77.310537]  [<ffffffff815afff8>] ? sock_has_perm+0x168/0x1e0
[   77.311204]  [<ffffffff81170bb8>] ? sched_clock_cpu+0xa8/0xf0
[   77.311866]  [<ffffffff81170d1b>] ? local_clock+0x1b/0x30
[   77.312501]  [<ffffffff811872cd>] ? lock_release_holdtime+0x1d/0x170
[   77.313241]  [<ffffffff815b0010>] ? sock_has_perm+0x180/0x1e0
[   77.313905]  [<ffffffff815afe90>] ?
selinux_msg_queue_alloc_security+0xa0/0xa0
[   77.314746]  [<ffffffff820ce263>] ipv6_setsockopt+0x53/0xb0
[   77.315397]  [<ffffffff820d3135>] udpv6_setsockopt+0x25/0x30
[   77.316058]  [<ffffffff81f9930f>] sock_common_setsockopt+0xf/0x20
[   77.316764]  [<ffffffff81f9305e>] SyS_setsockopt+0x8e/0xd0
[   77.317406]  [<ffffffff823a47e9>] system_call_fastpath+0x16/0x1b

[main] 375 sockets created based on info from socket cachefile. [main] Generating file descriptors
[main] Added 129 filenames from /dev
[main] Added 44048 filenames from /proc
[main] Added 18192 filenames from /sys
[main] Enabled 9 fd providers.
[watchdog] Watchdog is alive. (pid:4841)
[child3:4846] finit_module (313) returned ENOSYS, marking as inactive. [child1:4844] kcmp (312) returned ENOSYS, marking as inactive. [child2:4845] uselib (134) returned ENOSYS, marking as inactive. [child1:4844] nfsservctl (180) returned ENOSYS, marking as inactive. [child2:4845] delete_module (129:[32BIT]) returned ENOSYS, marking as inactive. [child2:4845] init_module (175) returned ENOSYS, marking as inactive.

[   84.126609] trinity-c7: vm86 mode not supported on 64 bit kernel
[child7:4850] vm86 (166:[32BIT]) returned ENOSYS, marking as inactive.
[main] Bailing main loop because ctrl-c.
[   84.345840] RTNL: assertion failed at net/ipv6/addrconf.c (1712)
[   84.346615] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   84.347426] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   84.348102]  ffff88003d76a618 ffff880026133d10 ffffffff8238ba79
ffff8800382d8000
[   84.349018]  ffff880026133d50 ffffffff820bd5db ffffffff81141555
ffff8800382d8220
[   84.349935]  ffff8800382d8000 00000000f76fd1e1 ffff88003d76a618
ffff8800382d8000
[   84.350848] Call Trace:
[   84.351149]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   84.351751]  [<ffffffff820bd5db>] addrconf_leave_solict+0x4b/0xb0
[   84.352574]  [<ffffffff81141555>] ? __local_bh_enable_ip+0xa5/0xf0
[   84.353315]  [<ffffffff820b07b3>] __ipv6_dev_ac_dec+0xc3/0x140
[   84.354019]  [<ffffffff820b08c8>] ipv6_dev_ac_dec+0x98/0xb0
[   84.354687]  [<ffffffff820b0bcd>] ipv6_sock_ac_close+0x10d/0x1a0
[   84.355410]  [<ffffffff820b0aee>] ? ipv6_sock_ac_close+0x2e/0x1a0
[   84.356147]  [<ffffffff820ae9d3>] inet6_release+0x23/0x40
[   84.356789]  [<ffffffff81f91834>] sock_release+0x14/0x80
[   84.357410]  [<ffffffff81f918ad>] sock_close+0xd/0x20
[   84.358042]  [<ffffffff8127fa91>] __fput+0x111/0x1e0
[   84.358622]  [<ffffffff8127fba9>] ____fput+0x9/0x10
[   84.359196]  [<ffffffff8115e3ee>] task_work_run+0x9e/0xd0
[   84.359825]  [<ffffffff8113f4b6>] do_exit+0x456/0xb30
[   84.360419]  [<ffffffff823a541c>] ? retint_swapgs+0x13/0x1b
[   84.361075]  [<ffffffff8113fc54>] do_group_exit+0x84/0xd0
[   84.361705]  [<ffffffff8113fcaf>] SyS_exit_group+0xf/0x10
[   84.362338]  [<ffffffff823a47e9>] system_call_fastpath+0x16/0x1b

[watchdog] [4841] Watchdog exiting because ctrl-c. [init] Ran 775 syscalls. Successes: 179 Failures: 596

#1 lore raw
Quoted 89 lines · Tommi Rantala

Hi,

Was fuzzing Linus v3.17-rc2-89-g59753a8 with Trinity as the root user in qemu, when I hit the following assertion failures.

Tommi

[init] Started watchdog process, PID is 4841
[main] Main thread is alive.

[   77.229699] sctp: [Deprecated]: trinity-main (pid 4842) Use of int
in max_burst socket option deprecated.
[   77.229699] Use struct sctp_assoc_value instead
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000
ffff8800382d8220
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20
[   77.308570]  [<ffffffff8117097d>] ? sched_clock_local+0x1d/0x80
[   77.309260]  [<ffffffff810a8a27>] ? kvm_clock_read+0x27/0x40
[   77.309915]  [<ffffffff810736d9>] ? sched_clock+0x9/0x10
[   77.310537]  [<ffffffff815afff8>] ? sock_has_perm+0x168/0x1e0
[   77.311204]  [<ffffffff81170bb8>] ? sched_clock_cpu+0xa8/0xf0
[   77.311866]  [<ffffffff81170d1b>] ? local_clock+0x1b/0x30
[   77.312501]  [<ffffffff811872cd>] ? lock_release_holdtime+0x1d/0x170
[   77.313241]  [<ffffffff815b0010>] ? sock_has_perm+0x180/0x1e0
[   77.313905]  [<ffffffff815afe90>] ?
selinux_msg_queue_alloc_security+0xa0/0xa0
[   77.314746]  [<ffffffff820ce263>] ipv6_setsockopt+0x53/0xb0
[   77.315397]  [<ffffffff820d3135>] udpv6_setsockopt+0x25/0x30
[   77.316058]  [<ffffffff81f9930f>] sock_common_setsockopt+0xf/0x20
[   77.316764]  [<ffffffff81f9305e>] SyS_setsockopt+0x8e/0xd0
[   77.317406]  [<ffffffff823a47e9>] system_call_fastpath+0x16/0x1b

[main] 375 sockets created based on info from socket cachefile. [main] Generating file descriptors
[main] Added 129 filenames from /dev
[main] Added 44048 filenames from /proc
[main] Added 18192 filenames from /sys
[main] Enabled 9 fd providers.
[watchdog] Watchdog is alive. (pid:4841)
[child3:4846] finit_module (313) returned ENOSYS, marking as inactive. [child1:4844] kcmp (312) returned ENOSYS, marking as inactive. [child2:4845] uselib (134) returned ENOSYS, marking as inactive. [child1:4844] nfsservctl (180) returned ENOSYS, marking as inactive. [child2:4845] delete_module (129:[32BIT]) returned ENOSYS, marking as inactive. [child2:4845] init_module (175) returned ENOSYS, marking as inactive.

[   84.126609] trinity-c7: vm86 mode not supported on 64 bit kernel
[child7:4850] vm86 (166:[32BIT]) returned ENOSYS, marking as inactive.
[main] Bailing main loop because ctrl-c.
[   84.345840] RTNL: assertion failed at net/ipv6/addrconf.c (1712)
[   84.346615] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   84.347426] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   84.348102]  ffff88003d76a618 ffff880026133d10 ffffffff8238ba79
ffff8800382d8000
[   84.349018]  ffff880026133d50 ffffffff820bd5db ffffffff81141555
ffff8800382d8220
[   84.349935]  ffff8800382d8000 00000000f76fd1e1 ffff88003d76a618
ffff8800382d8000
[   84.350848] Call Trace:
[   84.351149]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   84.351751]  [<ffffffff820bd5db>] addrconf_leave_solict+0x4b/0xb0
[   84.352574]  [<ffffffff81141555>] ? __local_bh_enable_ip+0xa5/0xf0
[   84.353315]  [<ffffffff820b07b3>] __ipv6_dev_ac_dec+0xc3/0x140
[   84.354019]  [<ffffffff820b08c8>] ipv6_dev_ac_dec+0x98/0xb0
[   84.354687]  [<ffffffff820b0bcd>] ipv6_sock_ac_close+0x10d/0x1a0
[   84.355410]  [<ffffffff820b0aee>] ? ipv6_sock_ac_close+0x2e/0x1a0
[   84.356147]  [<ffffffff820ae9d3>] inet6_release+0x23/0x40
[   84.356789]  [<ffffffff81f91834>] sock_release+0x14/0x80
[   84.357410]  [<ffffffff81f918ad>] sock_close+0xd/0x20
[   84.358042]  [<ffffffff8127fa91>] __fput+0x111/0x1e0
[   84.358622]  [<ffffffff8127fba9>] ____fput+0x9/0x10
[   84.359196]  [<ffffffff8115e3ee>] task_work_run+0x9e/0xd0
[   84.359825]  [<ffffffff8113f4b6>] do_exit+0x456/0xb30
[   84.360419]  [<ffffffff823a541c>] ? retint_swapgs+0x13/0x1b
[   84.361075]  [<ffffffff8113fc54>] do_group_exit+0x84/0xd0
[   84.361705]  [<ffffffff8113fcaf>] SyS_exit_group+0xf/0x10
[   84.362338]  [<ffffffff823a47e9>] system_call_fastpath+0x16/0x1b
Quoted 7 lines

[watchdog] [4841] Watchdog exiting because ctrl-c.
[init] Ran 775 syscalls. Successes: 179 Failures: 596
--
To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Yep, looks like ipv6_dev_ac_inc() and __ipv6_dev_ac_dec() are called without RNTL in the socket option path and with RTNL in the address configuration path. So it look like this this can actually trigger list corruptions.

-vlad

#2 lore raw
In reply to Tommi Rantala
Quoted 9 lines · Tommi Rantala
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000

ffff8800382d8220

Quoted 8 lines
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20

I think we should just use rtnl_lock() instead of rcu_read_lock() there, it is not a hot path worth optimization.

Please try the attached patch.

commit 31d83db0b417f705cbb31b2159603b8b53b81ab6

Author: Cong Wang <[email protected]>
Date:   Fri Aug 29 11:02:15 2014 -0700

ipv6: fix rtnl lock assertion in ipv6_sock_ac_join()

include/linux/netdevice.h +2 -2
index 4298013..1ae0e74 100644
@@ -2077,8 +2077,8 @@ void __dev_remove_pack(struct packet_type *pt);
 void dev_add_offload(struct packet_offload *po);
 void dev_remove_offload(struct packet_offload *po);
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short flags,
-					unsigned short mask);
+struct net_device *dev_get_by_flags(struct net *net, unsigned short flags,
+				    unsigned short mask);
 struct net_device *dev_get_by_name(struct net *net, const char *name);
 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
 struct net_device *__dev_get_by_name(struct net *net, const char *name);
net/core/dev.c +5 -4
index 26d296c..73cdb03 100644
@@ -897,23 +897,24 @@ struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
 EXPORT_SYMBOL(dev_getfirstbyhwtype);
 
 /**
- *	dev_get_by_flags_rcu - find any device with given flags
+ *	dev_get_by_flags - find any device with given flags
  *	@net: the applicable net namespace
  *	@if_flags: IFF_* values
  *	@mask: bitmask of bits in if_flags to check
  *
  *	Search for any interface with the given flags. Returns NULL if a device
  *	is not found or a pointer to the device. Must be called inside
- *	rcu_read_lock(), and result refcount is unchanged.
+ *	rtnl_lock(), and result refcount is unchanged.
  */
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
+struct net_device *dev_get_by_flags(struct net *net, unsigned short if_flags,
 				    unsigned short mask)
 {
 	struct net_device *dev, *ret;
 
+	ASSERT_RTNL();
 	ret = NULL;
-	for_each_netdev_rcu(net, dev) {
+	for_each_netdev(net, dev) {
 		if (((dev->flags ^ if_flags) & mask) == 0) {
 			ret = dev;
 			break;
net/ipv6/anycast.c +9 -7
index 2101832..c523c1a 100644
@@ -77,7 +77,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	pac->acl_next = NULL;
 	pac->acl_addr = *addr;
 
-	rcu_read_lock();
+	rtnl_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
 
@@ -90,11 +90,11 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			goto error;
 		} else {
 			/* router, no matching interface: just pick one */
-			dev = dev_get_by_flags_rcu(net, IFF_UP,
+			dev = dev_get_by_flags(net, IFF_UP,
 						   IFF_UP | IFF_LOOPBACK);
 		}
 	} else
-		dev = dev_get_by_index_rcu(net, ifindex);
+		dev = __dev_get_by_index(net, ifindex);
 
 	if (dev == NULL) {
 		err = -ENODEV;
@@ -136,7 +136,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	}
 
 error:
-	rcu_read_unlock();
+	rtnl_unlock();
 	if (pac)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 	return err;
@@ -171,13 +171,15 @@ int ipv6_sock_ac_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
-	rcu_read_lock();
-	dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+	rtnl_lock();
+	dev = __dev_get_by_index(net, pac->acl_ifindex);
 	if (dev)
 		ipv6_dev_ac_dec(dev, &pac->acl_addr);
-	rcu_read_unlock();
+	rtnl_unlock();
 
 	sock_kfree_s(sk, pac, sizeof(*pac));
+	if (!dev)
+		return -ENODEV;
 	return 0;
 }
 
#3 lore raw
Quoted 20 lines · Cong Wang
Quoted 9 lines · Tommi Rantala
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000

ffff8800382d8220

Quoted 8 lines
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20
Quoted 4 lines

I think we should just use rtnl_lock() instead of rcu_read_lock() there, it is not a hot path worth optimization.

Please try the attached patch.

note: it doesn't build as it is now, it needs:

-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);

I just tried your patch with a basic test program (open socket/join/leave/close and open socket/join/close).

I think you need to modify ipv6_sock_ac_close as well, or you can still trigger the assertion when closing the socket without leaving first.

Modified patch attached.

Signature · 1 line
Sabrina
include/linux/netdevice.h +2 -2
index 429801370d0c..1ae0e745b1b1 100644
@@ -2077,8 +2077,8 @@ void __dev_remove_pack(struct packet_type *pt);
 void dev_add_offload(struct packet_offload *po);
 void dev_remove_offload(struct packet_offload *po);
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short flags,
-					unsigned short mask);
+struct net_device *dev_get_by_flags(struct net *net, unsigned short flags,
+				    unsigned short mask);
 struct net_device *dev_get_by_name(struct net *net, const char *name);
 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
 struct net_device *__dev_get_by_name(struct net *net, const char *name);
net/core/dev.c +6 -5
index 443b814db05b..8fede6ef4a39 100644
@@ -897,23 +897,24 @@ struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
 EXPORT_SYMBOL(dev_getfirstbyhwtype);
 
 /**
- *	dev_get_by_flags_rcu - find any device with given flags
+ *	dev_get_by_flags - find any device with given flags
  *	@net: the applicable net namespace
  *	@if_flags: IFF_* values
  *	@mask: bitmask of bits in if_flags to check
  *
  *	Search for any interface with the given flags. Returns NULL if a device
  *	is not found or a pointer to the device. Must be called inside
- *	rcu_read_lock(), and result refcount is unchanged.
+ *	rtnl_lock(), and result refcount is unchanged.
  */
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
+struct net_device *dev_get_by_flags(struct net *net, unsigned short if_flags,
 				    unsigned short mask)
 {
 	struct net_device *dev, *ret;
 
+	ASSERT_RTNL();
 	ret = NULL;
-	for_each_netdev_rcu(net, dev) {
+	for_each_netdev(net, dev) {
 		if (((dev->flags ^ if_flags) & mask) == 0) {
 			ret = dev;
 			break;
@@ -921,7 +922,7 @@ struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags
 	}
 	return ret;
 }
-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);
 
 /**
  *	dev_valid_name - check if name is okay for network device
net/ipv6/anycast.c +12 -10
index 210183244689..6de5caa26ea4 100644
@@ -77,7 +77,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	pac->acl_next = NULL;
 	pac->acl_addr = *addr;
 
-	rcu_read_lock();
+	rtnl_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
 
@@ -90,11 +90,11 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			goto error;
 		} else {
 			/* router, no matching interface: just pick one */
-			dev = dev_get_by_flags_rcu(net, IFF_UP,
+			dev = dev_get_by_flags(net, IFF_UP,
 						   IFF_UP | IFF_LOOPBACK);
 		}
 	} else
-		dev = dev_get_by_index_rcu(net, ifindex);
+		dev = __dev_get_by_index(net, ifindex);
 
 	if (dev == NULL) {
 		err = -ENODEV;
@@ -136,7 +136,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	}
 
 error:
-	rcu_read_unlock();
+	rtnl_unlock();
 	if (pac)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 	return err;
@@ -171,13 +171,15 @@ int ipv6_sock_ac_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
-	rcu_read_lock();
-	dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+	rtnl_lock();
+	dev = __dev_get_by_index(net, pac->acl_ifindex);
 	if (dev)
 		ipv6_dev_ac_dec(dev, &pac->acl_addr);
-	rcu_read_unlock();
+	rtnl_unlock();
 
 	sock_kfree_s(sk, pac, sizeof(*pac));
+	if (!dev)
+		return -ENODEV;
 	return 0;
 }
 
@@ -198,12 +200,12 @@ void ipv6_sock_ac_close(struct sock *sk)
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
 	prev_index = 0;
-	rcu_read_lock();
+	rtnl_lock();
 	while (pac) {
 		struct ipv6_ac_socklist *next = pac->acl_next;
 
 		if (pac->acl_ifindex != prev_index) {
-			dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+			dev = __dev_get_by_index(net, pac->acl_ifindex);
 			prev_index = pac->acl_ifindex;
 		}
 		if (dev)
@@ -211,7 +213,7 @@ void ipv6_sock_ac_close(struct sock *sk)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 		pac = next;
 	}
-	rcu_read_unlock();
+	rtnl_unlock();
 }
 
 static void aca_put(struct ifacaddr6 *ac)
#4 lore raw
Quoted 32 lines · Sabrina Dubroca
Quoted 20 lines · Cong Wang
Quoted 9 lines · Tommi Rantala
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000

ffff8800382d8220

Quoted 8 lines
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20
Quoted 4 lines

I think we should just use rtnl_lock() instead of rcu_read_lock() there, it is not a hot path worth optimization.

Please try the attached patch.

note: it doesn't build as it is now, it needs:

net/core/dev.c +1 -1
-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);
Quoted 5 lines

I just tried your patch with a basic test program (open socket/join/leave/close and open socket/join/close).

I think you need to modify ipv6_sock_ac_close as well, or you can still trigger the assertion when closing the socket without leaving first.

You are absolutely right here.

Can I have your Signed-off-by and Tested-by before sending the patch formally?

Thanks!

#6 lore raw
In reply to Sabrina Dubroca

Hi Sabrina,

Quoted 32 lines · Sabrina Dubroca
Quoted 20 lines · Cong Wang
Quoted 9 lines · Tommi Rantala
[   77.297196] RTNL: assertion failed at net/ipv6/addrconf.c (1699)
[   77.298080] CPU: 0 PID: 4842 Comm: trinity-main Not tainted 3.17.0-rc2+ #30
[   77.299039] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   77.299789]  ffff88003d76a618 ffff880026133c50 ffffffff8238ba79
ffff880037c84520
[   77.300829]  ffff880026133c90 ffffffff820bd52b 0000000000000000
ffffffff82d86c40
[   77.301869]  0000000000000000 00000000f76fd1e1 ffff8800382d8000

ffff8800382d8220

Quoted 8 lines
[   77.302906] Call Trace:
[   77.303246]  [<ffffffff8238ba79>] dump_stack+0x4d/0x66
[   77.303928]  [<ffffffff820bd52b>] addrconf_join_solict+0x4b/0xb0
[   77.304731]  [<ffffffff820b031b>] ipv6_dev_ac_inc+0x2bb/0x330
[   77.305498]  [<ffffffff820b0060>] ? ac6_seq_start+0x260/0x260
[   77.306257]  [<ffffffff820b05fe>] ipv6_sock_ac_join+0x26e/0x360
[   77.307046]  [<ffffffff820b0429>] ? ipv6_sock_ac_join+0x99/0x360
[   77.307798]  [<ffffffff820cdd60>] do_ipv6_setsockopt.isra.5+0xa70/0xf20
Quoted 4 lines

I think we should just use rtnl_lock() instead of rcu_read_lock() there, it is not a hot path worth optimization.

Please try the attached patch.

note: it doesn't build as it is now, it needs:

net/core/dev.c +1 -1
-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);
Quoted 7 lines

I just tried your patch with a basic test program (open socket/join/leave/close and open socket/join/close).

I think you need to modify ipv6_sock_ac_close as well, or you can still trigger the assertion when closing the socket without leaving first.

Modified patch attached.

Sorry, just had time to look at this.

The reason is not to have list corruption but that the calls down to ndo_set_rx_mode expect rtnl to be locked by the drivers. Filter lists are locked by addr_list_lock and that's why I think we never saw any problems with that, but drivers expect rtnl locked for those calls.

But this problem also affects multicast join, so patch seems incomplete to me (and for that matter ssm multicast join, too).

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

So just adding rtnl_lock add appropriate places seems to be ok to me, but still need to review parts of the ssm code.

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Thanks,
Hannes

#7 lore raw

Hello,

Quoted 19 lines · Hannes Frederic Sowa

Hi Sabrina,

[...]

Sorry, just had time to look at this.

The reason is not to have list corruption but that the calls down to ndo_set_rx_mode expect rtnl to be locked by the drivers. Filter lists are locked by addr_list_lock and that's why I think we never saw any problems with that, but drivers expect rtnl locked for those calls.

But this problem also affects multicast join, so patch seems incomplete to me (and for that matter ssm multicast join, too).

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

Quoted 8 lines

So just adding rtnl_lock add appropriate places seems to be ok to me, but still need to review parts of the ssm code.

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Thanks,
Hannes

Thanks for explaining.

I had a look at what you suggested.

So, for anycast, on top of the previous patch, we'd have:

---

net/ipv6/anycast.c +0 -0
index 210183244689..61dd3046b804 100644

static void aca_put(struct ifacaddr6 *ac)
@@ -233,6 +235,8 @@ int ipv6_dev_ac_inc(struct net_device *dev, const struct in6_addr *addr)
struct rt6_info *rt;
int err;

+ ASSERT_RTNL();
+
idev = in6_dev_get(dev);

if (idev == NULL)
@@ -302,6 +306,8 @@ int __ipv6_dev_ac_dec(struct inet6_dev *idev, const struct in6_addr *addr)
{
struct ifacaddr6 *aca, *prev_aca;

+ ASSERT_RTNL();
+
write_lock_bh(&idev->lock);
prev_aca = NULL;
for (aca = idev->ac_list; aca; aca = aca->aca_next) {
@@ -336,6 +342,8 @@ static int ipv6_dev_ac_dec(struct net_device *dev, const struct in6_addr *addr)
{
struct inet6_dev *idev = __in6_dev_get(dev);

+ ASSERT_RTNL();
+
if (idev == NULL)
return -ENODEV;
return __ipv6_dev_ac_dec(idev, addr);

---

And for multicast:
- locking order in the patch below: rtnl -> rcu -> ipv6_sk_mc_lock
- ipv6_sock_mc_join: maybe move all the _unlock()'s together at the end of the function
- do we need to modify rcu_dereference_protected in ipv6_sock_mc_drop/ipv6_sock_mc_close
- I had a look at the other codepaths that call ipv6_dev_mc_inc/dec
- ipv6_mc_destroy_dev, dev_forward_change, ipv6_add_dev, addrconf_join_solict -- all take rtnl or already have an
ASSERT_RTNL()
- pndisc_destructor, called from pneigh_ifdown/pneigh_delete
- pndisc_constructor, called from pneigh_lookup -- pneigh_lookup
has ASSERT_RTNL(), but pneigh_lookup is called from ip6_forward and ndisc_recv_na
- (hope I didn't miss any callers)

As far as I could see, apart maybe from pndisc_constructor, it seems okay, but I'd like to hear your comments.

Current modifications:

---

net/ipv6/mcast.c +16 -0
index 70881795da96..d73ac1ef65f2 100644
@@ -172,6 +172,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	mc_lst->next = NULL;
 	mc_lst->addr = *addr;
 
+	rtnl_lock();
 	rcu_read_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
@@ -185,6 +186,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (dev == NULL) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return -ENODEV;
 	}
@@ -202,6 +204,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (err) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return err;
 	}
@@ -212,6 +215,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	spin_unlock(&ipv6_sk_mc_lock);
 
 	rcu_read_unlock();
+	rtnl_unlock();
 
 	return 0;
 }
@@ -229,6 +233,7 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	if (!ipv6_addr_is_multicast(addr))
 		return -EINVAL;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	for (lnk = &np->ipv6_mc_list;
 	     (mc_lst = rcu_dereference_protected(*lnk,
@@ -252,12 +257,15 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			} else
 				(void) ip6_mc_leave_src(sk, mc_lst, NULL);
 			rcu_read_unlock();
+			rtnl_unlock();
+
 			atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
 			kfree_rcu(mc_lst, rcu);
 			return 0;
 		}
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 
 	return -EADDRNOTAVAIL;
 }
@@ -302,6 +310,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 	if (!rcu_access_pointer(np->ipv6_mc_list))
 		return;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	while ((mc_lst = rcu_dereference_protected(np->ipv6_mc_list,
 				lockdep_is_held(&ipv6_sk_mc_lock))) != NULL) {
@@ -328,6 +337,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 		spin_lock(&ipv6_sk_mc_lock);
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 }
 
 int ip6_mc_source(int add, int omode, struct sock *sk,
@@ -845,6 +855,8 @@ int ipv6_dev_mc_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct ifmcaddr6 *mc;
 	struct inet6_dev *idev;
 
+	ASSERT_RTNL();
+
 	/* we need to take a reference on idev */
 	idev = in6_dev_get(dev);
 
@@ -916,6 +928,8 @@ int __ipv6_dev_mc_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifmcaddr6 *ma, **map;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	for (map = &idev->mc_list; (ma = *map) != NULL; map = &ma->next) {
 		if (ipv6_addr_equal(&ma->mca_addr, addr)) {
@@ -942,6 +956,8 @@ int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr)
 	struct inet6_dev *idev;
 	int err;
 
+	ASSERT_RTNL();
+
 	rcu_read_lock();
 
 	idev = __in6_dev_get(dev);

---

Thanks,

Signature · 1 line
Sabrina
#8 lore raw
Quoted 3 lines · Sabrina Dubroca

- pndisc_constructor, called from pneigh_lookup -- pneigh_lookup has ASSERT_RTNL(), but pneigh_lookup is called from ip6_forward and ndisc_recv_na

Ah, these have creat = 0, so it's fine. I missed that earlier.

Sorry for the noise.

Signature · 1 line
Sabrina
#9 lore raw
In reply to Sabrina Dubroca

Hi,

Quoted 61 lines · Sabrina Dubroca
Quoted 19 lines · Hannes Frederic Sowa

Hi Sabrina,

[...]

Sorry, just had time to look at this.

The reason is not to have list corruption but that the calls down to ndo_set_rx_mode expect rtnl to be locked by the drivers. Filter lists are locked by addr_list_lock and that's why I think we never saw any problems with that, but drivers expect rtnl locked for those calls.

But this problem also affects multicast join, so patch seems incomplete to me (and for that matter ssm multicast join, too).

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

Quoted 8 lines

So just adding rtnl_lock add appropriate places seems to be ok to me, but still need to review parts of the ssm code.

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Thanks,
Hannes

Thanks for explaining.

I had a look at what you suggested.

So, for anycast, on top of the previous patch, we'd have:

---

net/ipv6/anycast.c +4 -0
index 210183244689..61dd3046b804 100644
 static void aca_put(struct ifacaddr6 *ac)
@@ -233,6 +235,8 @@ int ipv6_dev_ac_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct rt6_info *rt;
 	int err;
 
+	ASSERT_RTNL();
+
 	idev = in6_dev_get(dev);
 
 	if (idev == NULL)
@@ -302,6 +306,8 @@ int __ipv6_dev_ac_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifacaddr6 *aca, *prev_aca;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	prev_aca = NULL;
 	for (aca = idev->ac_list; aca; aca = aca->aca_next) {

net/ipv6/anycast.c static int ipv6_dev_ac_dec(struct net_device *dev, const struct in6_addr *addr) · line 349

Quoted 8 lines
net/ipv6/anycast.c +2 -0
@@ -336,6 +342,8 @@ static int ipv6_dev_ac_dec(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct inet6_dev *idev = __in6_dev_get(dev);
 
+	ASSERT_RTNL();
+
 	if (idev == NULL)
 		return -ENODEV;
 	return __ipv6_dev_ac_dec(idev, addr);

ASSERT_RTNL() still performs a runtime check. While those are not really fast paths, I still think it is better to keep them to a minimum and place them only at places where we know all code which needs to be guarded passes by:

I would suggest to move ASSERT_RTNL to ipv6_dev_mc_inc and __ipv6_dev_mc_dec and even remove the checks from addrconf_join_solict and addrconf_leave_solict. Does that cover all code paths and makes sense?

Quoted 17 lines

---

And for multicast:
- locking order in the patch below: rtnl -> rcu -> ipv6_sk_mc_lock
- ipv6_sock_mc_join: maybe move all the _unlock()'s together at the end of the function
- do we need to modify rcu_dereference_protected in ipv6_sock_mc_drop/ipv6_sock_mc_close
- I had a look at the other codepaths that call ipv6_dev_mc_inc/dec
- ipv6_mc_destroy_dev, dev_forward_change, ipv6_add_dev, addrconf_join_solict -- all take rtnl or already have an
ASSERT_RTNL()
- pndisc_destructor, called from pneigh_ifdown/pneigh_delete
- pndisc_constructor, called from pneigh_lookup -- pneigh_lookup
has ASSERT_RTNL(), but pneigh_lookup is called from ip6_forward and ndisc_recv_na
- (hope I didn't miss any callers)

Quoted 2 lines

As far as I could see, apart maybe from pndisc_constructor, it seems okay, but I'd like to hear your comments.

The rest of the patch looks good.

Can you or Cong post a final patch with the adapted ac_join/drop changes?

Thanks,
Hannes

#10 lore raw

PATCH 1/1 ipv6: fix rtnl locking in setsockopt for anycast and multicast

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

---
I included Cong's Signed-off-by for the first part of the patch, I hope that's OK.

This patch is based on -next, but since the assertion can also be triggered on a current kernel (tested on a 3.16), I think it should also go in stable.

 include/linux/netdevice.h |  4 ++--
 net/core/dev.c            | 11 ++++++-----
 net/ipv6/addrconf.c       | 15 +++++----------
 net/ipv6/anycast.c        | 30 +++++++++++++++++++-----------
 net/ipv6/mcast.c          | 16 ++++++++++++++++
 5 files changed, 48 insertions(+), 28 deletions(-)
include/linux/netdevice.h +2 -2
index 429801370d0c..1ae0e745b1b1 100644
@@ -2077,8 +2077,8 @@ void __dev_remove_pack(struct packet_type *pt);
 void dev_add_offload(struct packet_offload *po);
 void dev_remove_offload(struct packet_offload *po);
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short flags,
-					unsigned short mask);
+struct net_device *dev_get_by_flags(struct net *net, unsigned short flags,
+				    unsigned short mask);
 struct net_device *dev_get_by_name(struct net *net, const char *name);
 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
 struct net_device *__dev_get_by_name(struct net *net, const char *name);
net/core/dev.c +6 -5
index 443b814db05b..8fede6ef4a39 100644
@@ -897,23 +897,24 @@ struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
 EXPORT_SYMBOL(dev_getfirstbyhwtype);
 
 /**
- *	dev_get_by_flags_rcu - find any device with given flags
+ *	dev_get_by_flags - find any device with given flags
  *	@net: the applicable net namespace
  *	@if_flags: IFF_* values
  *	@mask: bitmask of bits in if_flags to check
  *
  *	Search for any interface with the given flags. Returns NULL if a device
  *	is not found or a pointer to the device. Must be called inside
- *	rcu_read_lock(), and result refcount is unchanged.
+ *	rtnl_lock(), and result refcount is unchanged.
  */
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
+struct net_device *dev_get_by_flags(struct net *net, unsigned short if_flags,
 				    unsigned short mask)
 {
 	struct net_device *dev, *ret;
 
+	ASSERT_RTNL();
 	ret = NULL;
-	for_each_netdev_rcu(net, dev) {
+	for_each_netdev(net, dev) {
 		if (((dev->flags ^ if_flags) & mask) == 0) {
 			ret = dev;
 			break;
@@ -921,7 +922,7 @@ struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags
 	}
 	return ret;
 }
-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);
 
 /**
  *	dev_valid_name - check if name is okay for network device
net/ipv6/addrconf.c +5 -10
index 267ce3caee24..7ada65937d23 100644
@@ -1690,14 +1690,12 @@ void addrconf_dad_failure(struct inet6_ifaddr *ifp)
 	addrconf_mod_dad_work(ifp, 0);
 }
 
-/* Join to solicited addr multicast group. */
-
+/* Join to solicited addr multicast group.
+ * caller must hold RTNL */
 void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1705,12 +1703,11 @@ void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 	ipv6_dev_mc_inc(dev, &maddr);
 }
 
+/* caller must hold RTNL */
 void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1718,12 +1715,11 @@ void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 	__ipv6_dev_mc_dec(idev, &maddr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
@@ -1732,12 +1728,11 @@ static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 	ipv6_dev_ac_inc(ifp->idev->dev, &addr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
net/ipv6/anycast.c +19 -11
index 210183244689..572c2faede55 100644
@@ -77,7 +77,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	pac->acl_next = NULL;
 	pac->acl_addr = *addr;
 
-	rcu_read_lock();
+	rtnl_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
 
@@ -90,11 +90,11 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			goto error;
 		} else {
 			/* router, no matching interface: just pick one */
-			dev = dev_get_by_flags_rcu(net, IFF_UP,
-						   IFF_UP | IFF_LOOPBACK);
+			dev = dev_get_by_flags(net, IFF_UP,
+					       IFF_UP | IFF_LOOPBACK);
 		}
 	} else
-		dev = dev_get_by_index_rcu(net, ifindex);
+		dev = __dev_get_by_index(net, ifindex);
 
 	if (dev == NULL) {
 		err = -ENODEV;
@@ -136,7 +136,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	}
 
 error:
-	rcu_read_unlock();
+	rtnl_unlock();
 	if (pac)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 	return err;
@@ -171,13 +171,15 @@ int ipv6_sock_ac_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
-	rcu_read_lock();
-	dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+	rtnl_lock();
+	dev = __dev_get_by_index(net, pac->acl_ifindex);
 	if (dev)
 		ipv6_dev_ac_dec(dev, &pac->acl_addr);
-	rcu_read_unlock();
+	rtnl_unlock();
 
 	sock_kfree_s(sk, pac, sizeof(*pac));
+	if (!dev)
+		return -ENODEV;
 	return 0;
 }
 
@@ -198,12 +200,12 @@ void ipv6_sock_ac_close(struct sock *sk)
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
 	prev_index = 0;
-	rcu_read_lock();
+	rtnl_lock();
 	while (pac) {
 		struct ipv6_ac_socklist *next = pac->acl_next;
 
 		if (pac->acl_ifindex != prev_index) {
-			dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+			dev = __dev_get_by_index(net, pac->acl_ifindex);
 			prev_index = pac->acl_ifindex;
 		}
 		if (dev)
@@ -211,7 +213,7 @@ void ipv6_sock_ac_close(struct sock *sk)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 		pac = next;
 	}
-	rcu_read_unlock();
+	rtnl_unlock();
 }
 
 static void aca_put(struct ifacaddr6 *ac)
@@ -233,6 +235,8 @@ int ipv6_dev_ac_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct rt6_info *rt;
 	int err;
 
+	ASSERT_RTNL();
+
 	idev = in6_dev_get(dev);
 
 	if (idev == NULL)
@@ -302,6 +306,8 @@ int __ipv6_dev_ac_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifacaddr6 *aca, *prev_aca;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	prev_aca = NULL;
 	for (aca = idev->ac_list; aca; aca = aca->aca_next) {
@@ -336,6 +342,8 @@ static int ipv6_dev_ac_dec(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct inet6_dev *idev = __in6_dev_get(dev);
 
+	ASSERT_RTNL();
+
 	if (idev == NULL)
 		return -ENODEV;
 	return __ipv6_dev_ac_dec(idev, addr);
net/ipv6/mcast.c +16 -0
index 70881795da96..d73ac1ef65f2 100644
@@ -172,6 +172,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	mc_lst->next = NULL;
 	mc_lst->addr = *addr;
 
+	rtnl_lock();
 	rcu_read_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
@@ -185,6 +186,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (dev == NULL) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return -ENODEV;
 	}
@@ -202,6 +204,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (err) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return err;
 	}
@@ -212,6 +215,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	spin_unlock(&ipv6_sk_mc_lock);
 
 	rcu_read_unlock();
+	rtnl_unlock();
 
 	return 0;
 }
@@ -229,6 +233,7 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	if (!ipv6_addr_is_multicast(addr))
 		return -EINVAL;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	for (lnk = &np->ipv6_mc_list;
 	     (mc_lst = rcu_dereference_protected(*lnk,
@@ -252,12 +257,15 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			} else
 				(void) ip6_mc_leave_src(sk, mc_lst, NULL);
 			rcu_read_unlock();
+			rtnl_unlock();
+
 			atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
 			kfree_rcu(mc_lst, rcu);
 			return 0;
 		}
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 
 	return -EADDRNOTAVAIL;
 }
@@ -302,6 +310,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 	if (!rcu_access_pointer(np->ipv6_mc_list))
 		return;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	while ((mc_lst = rcu_dereference_protected(np->ipv6_mc_list,
 				lockdep_is_held(&ipv6_sk_mc_lock))) != NULL) {
@@ -328,6 +337,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 		spin_lock(&ipv6_sk_mc_lock);
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 }
 
 int ip6_mc_source(int add, int omode, struct sock *sk,
@@ -845,6 +855,8 @@ int ipv6_dev_mc_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct ifmcaddr6 *mc;
 	struct inet6_dev *idev;
 
+	ASSERT_RTNL();
+
 	/* we need to take a reference on idev */
 	idev = in6_dev_get(dev);
 
@@ -916,6 +928,8 @@ int __ipv6_dev_mc_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifmcaddr6 *ma, **map;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	for (map = &idev->mc_list; (ma = *map) != NULL; map = &ma->next) {
 		if (ipv6_addr_equal(&ma->mca_addr, addr)) {
@@ -942,6 +956,8 @@ int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr)
 	struct inet6_dev *idev;
 	int err;
 
+	ASSERT_RTNL();
+
 	rcu_read_lock();
 
 	idev = __in6_dev_get(dev);
Signature · 1 line
2.1.0
#11 lore raw

Re: [PATCH] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Hi,

Quoted 13 lines · Sabrina Dubroca

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

Quoted 7 lines

---
I included Cong's Signed-off-by for the first part of the patch, I hope that's OK.

This patch is based on -next, but since the assertion can also be triggered on a current kernel (tested on a 3.16), I think it should also go in stable.

Ack, thus you should base the patch on the net tree.

Quoted 57 lines
 include/linux/netdevice.h |  4 ++--
 net/core/dev.c            | 11 ++++++-----
 net/ipv6/addrconf.c       | 15 +++++----------
 net/ipv6/anycast.c        | 30 +++++++++++++++++++-----------
 net/ipv6/mcast.c          | 16 ++++++++++++++++
 5 files changed, 48 insertions(+), 28 deletions(-)
include/linux/netdevice.h +2 -2
index 429801370d0c..1ae0e745b1b1 100644
@@ -2077,8 +2077,8 @@ void __dev_remove_pack(struct packet_type *pt);
 void dev_add_offload(struct packet_offload *po);
 void dev_remove_offload(struct packet_offload *po);
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short flags,
-					unsigned short mask);
+struct net_device *dev_get_by_flags(struct net *net, unsigned short flags,
+				    unsigned short mask);
 struct net_device *dev_get_by_name(struct net *net, const char *name);
 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
 struct net_device *__dev_get_by_name(struct net *net, const char *name);
net/core/dev.c +5 -4
index 443b814db05b..8fede6ef4a39 100644
@@ -897,23 +897,24 @@ struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
 EXPORT_SYMBOL(dev_getfirstbyhwtype);
 
 /**
- *	dev_get_by_flags_rcu - find any device with given flags
+ *	dev_get_by_flags - find any device with given flags
  *	@net: the applicable net namespace
  *	@if_flags: IFF_* values
  *	@mask: bitmask of bits in if_flags to check
  *
  *	Search for any interface with the given flags. Returns NULL if a device
  *	is not found or a pointer to the device. Must be called inside
- *	rcu_read_lock(), and result refcount is unchanged.
+ *	rtnl_lock(), and result refcount is unchanged.
  */
 
-struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags,
+struct net_device *dev_get_by_flags(struct net *net, unsigned short if_flags,
 				    unsigned short mask)
 {
 	struct net_device *dev, *ret;
 
+	ASSERT_RTNL();
 	ret = NULL;
-	for_each_netdev_rcu(net, dev) {
+	for_each_netdev(net, dev) {
 		if (((dev->flags ^ if_flags) & mask) == 0) {
 			ret = dev;
 			break;

net/core/dev.c struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags · line 925

Quoted 5 lines
net/core/dev.c +1 -1
@@ -921,7 +922,7 @@ struct net_device *dev_get_by_flags_rcu(struct net *net, unsigned short if_flags
 	}
 	return ret;
 }
-EXPORT_SYMBOL(dev_get_by_flags_rcu);
+EXPORT_SYMBOL(dev_get_by_flags);

I don't have a very strong opinion on that, but think we shouldn't touch this function. In general it looks like a useful one and if you force rtnl lock on it you cannot call it from bh anymore. I think we should keep rcu locking here and in the anycast code. It shouldn't matter much.

Quoted 70 lines
net/core/dev.c +0 -0
 /**
  *	dev_valid_name - check if name is okay for network device
net/ipv6/addrconf.c +5 -10
index 267ce3caee24..7ada65937d23 100644
@@ -1690,14 +1690,12 @@ void addrconf_dad_failure(struct inet6_ifaddr *ifp)
 	addrconf_mod_dad_work(ifp, 0);
 }
 
-/* Join to solicited addr multicast group. */
-
+/* Join to solicited addr multicast group.
+ * caller must hold RTNL */
 void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1705,12 +1703,11 @@ void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 	ipv6_dev_mc_inc(dev, &maddr);
 }
 
+/* caller must hold RTNL */
 void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1718,12 +1715,11 @@ void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 	__ipv6_dev_mc_dec(idev, &maddr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
@@ -1732,12 +1728,11 @@ static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 	ipv6_dev_ac_inc(ifp->idev->dev, &addr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
net/ipv6/anycast.c +0 -0
index 210183244689..572c2faede55 100644

net/ipv6/anycast.c int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr) · line 80

Quoted 5 lines
net/ipv6/anycast.c +1 -1
@@ -77,7 +77,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	pac->acl_next = NULL;
 	pac->acl_addr = *addr;
 
-	rcu_read_lock();
+	rtnl_lock();

We would need to keep rcu_read_lock inside rtnl_lock if you decide to keep dev_get_by_flags_rcu around.

Quoted 19 lines
net/ipv6/anycast.c +3 -3
 	if (ifindex == 0) {
 		struct rt6_info *rt;
 
@@ -90,11 +90,11 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			goto error;
 		} else {
 			/* router, no matching interface: just pick one */
-			dev = dev_get_by_flags_rcu(net, IFF_UP,
-						   IFF_UP | IFF_LOOPBACK);
+			dev = dev_get_by_flags(net, IFF_UP,
+					       IFF_UP | IFF_LOOPBACK);
 		}
 	} else
-		dev = dev_get_by_index_rcu(net, ifindex);
+		dev = __dev_get_by_index(net, ifindex);
 
 	if (dev == NULL) {
 		err = -ENODEV;

net/ipv6/anycast.c int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr) · line 139

Quoted 5 lines
net/ipv6/anycast.c +1 -1
@@ -136,7 +136,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	}
 
 error:
-	rcu_read_unlock();
+	rtnl_unlock();

...here, too.

Quoted 168 lines
net/ipv6/anycast.c +14 -6
 	if (pac)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 	return err;
@@ -171,13 +171,15 @@ int ipv6_sock_ac_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
-	rcu_read_lock();
-	dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+	rtnl_lock();
+	dev = __dev_get_by_index(net, pac->acl_ifindex);
 	if (dev)
 		ipv6_dev_ac_dec(dev, &pac->acl_addr);
-	rcu_read_unlock();
+	rtnl_unlock();
 
 	sock_kfree_s(sk, pac, sizeof(*pac));
+	if (!dev)
+		return -ENODEV;
 	return 0;
 }
 
@@ -198,12 +200,12 @@ void ipv6_sock_ac_close(struct sock *sk)
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
 	prev_index = 0;
-	rcu_read_lock();
+	rtnl_lock();
 	while (pac) {
 		struct ipv6_ac_socklist *next = pac->acl_next;
 
 		if (pac->acl_ifindex != prev_index) {
-			dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
+			dev = __dev_get_by_index(net, pac->acl_ifindex);
 			prev_index = pac->acl_ifindex;
 		}
 		if (dev)
@@ -211,7 +213,7 @@ void ipv6_sock_ac_close(struct sock *sk)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 		pac = next;
 	}
-	rcu_read_unlock();
+	rtnl_unlock();
 }
 
 static void aca_put(struct ifacaddr6 *ac)
@@ -233,6 +235,8 @@ int ipv6_dev_ac_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct rt6_info *rt;
 	int err;
 
+	ASSERT_RTNL();
+
 	idev = in6_dev_get(dev);
 
 	if (idev == NULL)
@@ -302,6 +306,8 @@ int __ipv6_dev_ac_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifacaddr6 *aca, *prev_aca;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	prev_aca = NULL;
 	for (aca = idev->ac_list; aca; aca = aca->aca_next) {
@@ -336,6 +342,8 @@ static int ipv6_dev_ac_dec(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct inet6_dev *idev = __in6_dev_get(dev);
 
+	ASSERT_RTNL();
+
 	if (idev == NULL)
 		return -ENODEV;
 	return __ipv6_dev_ac_dec(idev, addr);
net/ipv6/mcast.c +14 -0
index 70881795da96..d73ac1ef65f2 100644
@@ -172,6 +172,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	mc_lst->next = NULL;
 	mc_lst->addr = *addr;
 
+	rtnl_lock();
 	rcu_read_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
@@ -185,6 +186,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (dev == NULL) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return -ENODEV;
 	}
@@ -202,6 +204,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (err) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return err;
 	}
@@ -212,6 +215,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	spin_unlock(&ipv6_sk_mc_lock);
 
 	rcu_read_unlock();
+	rtnl_unlock();
 
 	return 0;
 }
@@ -229,6 +233,7 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	if (!ipv6_addr_is_multicast(addr))
 		return -EINVAL;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	for (lnk = &np->ipv6_mc_list;
 	     (mc_lst = rcu_dereference_protected(*lnk,
@@ -252,12 +257,15 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			} else
 				(void) ip6_mc_leave_src(sk, mc_lst, NULL);
 			rcu_read_unlock();
+			rtnl_unlock();
+
 			atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
 			kfree_rcu(mc_lst, rcu);
 			return 0;
 		}
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 
 	return -EADDRNOTAVAIL;
 }
@@ -302,6 +310,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 	if (!rcu_access_pointer(np->ipv6_mc_list))
 		return;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	while ((mc_lst = rcu_dereference_protected(np->ipv6_mc_list,
 				lockdep_is_held(&ipv6_sk_mc_lock))) != NULL) {
@@ -328,6 +337,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 		spin_lock(&ipv6_sk_mc_lock);
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 }
 
 int ip6_mc_source(int add, int omode, struct sock *sk,
@@ -845,6 +855,8 @@ int ipv6_dev_mc_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct ifmcaddr6 *mc;
 	struct inet6_dev *idev;
 
+	ASSERT_RTNL();
+
 	/* we need to take a reference on idev */
 	idev = in6_dev_get(dev);
 
@@ -916,6 +928,8 @@ int __ipv6_dev_mc_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifmcaddr6 *ma, **map;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	for (map = &idev->mc_list; (ma = *map) != NULL; map = &ma->next) {
 		if (ipv6_addr_equal(&ma->mca_addr, addr)) {

net/ipv6/mcast.c int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr) · line 960

Quoted 5 lines
net/ipv6/mcast.c +2 -0
@@ -942,6 +956,8 @@ int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr)
 	struct inet6_dev *idev;
 	int err;
 
+	ASSERT_RTNL();
+

Minor nit:
This one is not necessary and will be guarded by the __ipv6_dev_mc_dec one.

net/ipv6/mcast.c int ipv6_dev_mc_dec(struct net_device *dev, const struct in6_addr *addr) · line 963

Quoted 3 lines
net/ipv6/mcast.c +0 -0
 	rcu_read_lock();
 
 	idev = __in6_dev_get(dev);

Rest looks good to me, thanks,
Hannes

#12 lore raw

PATCH v2 1/1 ipv6: fix rtnl locking in setsockopt for anycast and multicast

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

---
As was said earlier, this should go in stable.

v2:
- based on net
- keep dev_get_by_flags_rcu and RCU in ipv6_sock_ac_*
- remove two ASSERT_RTNL() that are not necessary

Thank you for your help, Hannes!

 net/ipv6/addrconf.c | 15 +++++----------
 net/ipv6/anycast.c  | 12 ++++++++++++
 net/ipv6/mcast.c    | 14 ++++++++++++++
 3 files changed, 31 insertions(+), 10 deletions(-)
net/ipv6/addrconf.c +5 -10
index 0b239fc1816e..aa0e135b808c 100644
@@ -1690,14 +1690,12 @@ void addrconf_dad_failure(struct inet6_ifaddr *ifp)
 	addrconf_mod_dad_work(ifp, 0);
 }
 
-/* Join to solicited addr multicast group. */
-
+/* Join to solicited addr multicast group.
+ * caller must hold RTNL */
 void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1705,12 +1703,11 @@ void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr)
 	ipv6_dev_mc_inc(dev, &maddr);
 }
 
+/* caller must hold RTNL */
 void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct in6_addr maddr;
 
-	ASSERT_RTNL();
-
 	if (idev->dev->flags&(IFF_LOOPBACK|IFF_NOARP))
 		return;
 
@@ -1718,12 +1715,11 @@ void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr)
 	__ipv6_dev_mc_dec(idev, &maddr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
@@ -1732,12 +1728,11 @@ static void addrconf_join_anycast(struct inet6_ifaddr *ifp)
 	ipv6_dev_ac_inc(ifp->idev->dev, &addr);
 }
 
+/* caller must hold RTNL */
 static void addrconf_leave_anycast(struct inet6_ifaddr *ifp)
 {
 	struct in6_addr addr;
 
-	ASSERT_RTNL();
-
 	if (ifp->prefix_len >= 127) /* RFC 6164 */
 		return;
 	ipv6_addr_prefix(&addr, &ifp->addr, ifp->prefix_len);
net/ipv6/anycast.c +12 -0
index 210183244689..45b9d81d91e8 100644
@@ -77,6 +77,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	pac->acl_next = NULL;
 	pac->acl_addr = *addr;
 
+	rtnl_lock();
 	rcu_read_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
@@ -137,6 +138,7 @@ int ipv6_sock_ac_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 error:
 	rcu_read_unlock();
+	rtnl_unlock();
 	if (pac)
 		sock_kfree_s(sk, pac, sizeof(*pac));
 	return err;
@@ -171,13 +173,17 @@ int ipv6_sock_ac_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
+	rtnl_lock();
 	rcu_read_lock();
 	dev = dev_get_by_index_rcu(net, pac->acl_ifindex);
 	if (dev)
 		ipv6_dev_ac_dec(dev, &pac->acl_addr);
 	rcu_read_unlock();
+	rtnl_unlock();
 
 	sock_kfree_s(sk, pac, sizeof(*pac));
+	if (!dev)
+		return -ENODEV;
 	return 0;
 }
 
@@ -198,6 +204,7 @@ void ipv6_sock_ac_close(struct sock *sk)
 	spin_unlock_bh(&ipv6_sk_ac_lock);
 
 	prev_index = 0;
+	rtnl_lock();
 	rcu_read_lock();
 	while (pac) {
 		struct ipv6_ac_socklist *next = pac->acl_next;
@@ -212,6 +219,7 @@ void ipv6_sock_ac_close(struct sock *sk)
 		pac = next;
 	}
 	rcu_read_unlock();
+	rtnl_unlock();
 }
 
 static void aca_put(struct ifacaddr6 *ac)
@@ -233,6 +241,8 @@ int ipv6_dev_ac_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct rt6_info *rt;
 	int err;
 
+	ASSERT_RTNL();
+
 	idev = in6_dev_get(dev);
 
 	if (idev == NULL)
@@ -302,6 +312,8 @@ int __ipv6_dev_ac_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifacaddr6 *aca, *prev_aca;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	prev_aca = NULL;
 	for (aca = idev->ac_list; aca; aca = aca->aca_next) {
net/ipv6/mcast.c +14 -0
index 617f0958e164..a23b655a7627 100644
@@ -172,6 +172,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	mc_lst->next = NULL;
 	mc_lst->addr = *addr;
 
+	rtnl_lock();
 	rcu_read_lock();
 	if (ifindex == 0) {
 		struct rt6_info *rt;
@@ -185,6 +186,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (dev == NULL) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return -ENODEV;
 	}
@@ -202,6 +204,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 
 	if (err) {
 		rcu_read_unlock();
+		rtnl_unlock();
 		sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
 		return err;
 	}
@@ -212,6 +215,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	spin_unlock(&ipv6_sk_mc_lock);
 
 	rcu_read_unlock();
+	rtnl_unlock();
 
 	return 0;
 }
@@ -229,6 +233,7 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 	if (!ipv6_addr_is_multicast(addr))
 		return -EINVAL;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	for (lnk = &np->ipv6_mc_list;
 	     (mc_lst = rcu_dereference_protected(*lnk,
@@ -252,12 +257,15 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
 			} else
 				(void) ip6_mc_leave_src(sk, mc_lst, NULL);
 			rcu_read_unlock();
+			rtnl_unlock();
+
 			atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
 			kfree_rcu(mc_lst, rcu);
 			return 0;
 		}
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 
 	return -EADDRNOTAVAIL;
 }
@@ -302,6 +310,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 	if (!rcu_access_pointer(np->ipv6_mc_list))
 		return;
 
+	rtnl_lock();
 	spin_lock(&ipv6_sk_mc_lock);
 	while ((mc_lst = rcu_dereference_protected(np->ipv6_mc_list,
 				lockdep_is_held(&ipv6_sk_mc_lock))) != NULL) {
@@ -328,6 +337,7 @@ void ipv6_sock_mc_close(struct sock *sk)
 		spin_lock(&ipv6_sk_mc_lock);
 	}
 	spin_unlock(&ipv6_sk_mc_lock);
+	rtnl_unlock();
 }
 
 int ip6_mc_source(int add, int omode, struct sock *sk,
@@ -845,6 +855,8 @@ int ipv6_dev_mc_inc(struct net_device *dev, const struct in6_addr *addr)
 	struct ifmcaddr6 *mc;
 	struct inet6_dev *idev;
 
+	ASSERT_RTNL();
+
 	/* we need to take a reference on idev */
 	idev = in6_dev_get(dev);
 
@@ -916,6 +928,8 @@ int __ipv6_dev_mc_dec(struct inet6_dev *idev, const struct in6_addr *addr)
 {
 	struct ifmcaddr6 *ma, **map;
 
+	ASSERT_RTNL();
+
 	write_lock_bh(&idev->lock);
 	for (map = &idev->mc_list; (ma=*map) != NULL; map = &ma->next) {
 		if (ipv6_addr_equal(&ma->mca_addr, addr)) {
Signature · 1 line
2.1.0
Acked-by on v2 1/1
#13 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Quoted 16 lines · Sabrina Dubroca

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

This patch moves ASSERT_RTNL() up a level in the call stack.

---
As was said earlier, this should go in stable.

Quoted 6 lines

v2:
- based on net
- keep dev_get_by_flags_rcu and RCU in ipv6_sock_ac_*
- remove two ASSERT_RTNL() that are not necessary

Thank you for your help, Hannes!

Thanks for fixing! ;)

#14 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

In reply to Sabrina Dubroca
Quoted 5 lines · Sabrina Dubroca

v2:
- based on net
- keep dev_get_by_flags_rcu and RCU in ipv6_sock_ac_*
- remove two ASSERT_RTNL() that are not necessary

There is no point to keep RCU here. Hannes' reply doesn't make any sense.

maintainer
#15 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

In reply to Sabrina Dubroca

From: Sabrina Dubroca <[email protected]>
Date: Tue, 2 Sep 2014 10:29:29 +0200

Quoted 8 lines

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

Quoted 5 lines

This patch moves ASSERT_RTNL() up a level in the call stack.

Applied and queued up for -stable, thanks.

#16 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Quoted 11 lines · David Miller

From: Sabrina Dubroca <[email protected]>
Date: Tue, 2 Sep 2014 10:29:29 +0200

Quoted 8 lines

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

Quoted 7 lines
Quoted 5 lines

This patch moves ASSERT_RTNL() up a level in the call stack.

Applied and queued up for -stable, thanks.

I believe you applied a wrong version, at least the following is not correct:

+ if (!dev)
+ return -ENODEV;

Sabrina took that from my draft patch, but they all don't realize this is wrong.

(I did provide a correct version which is just ignored by you.)

#17 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Quoted 26 lines · Cong Wang
Quoted 11 lines · David Miller

From: Sabrina Dubroca <[email protected]>
Date: Tue, 2 Sep 2014 10:29:29 +0200

Quoted 8 lines

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

Quoted 7 lines
Quoted 5 lines

This patch moves ASSERT_RTNL() up a level in the call stack.

Applied and queued up for -stable, thanks.

I believe you applied a wrong version, at least the following is not correct:

net/ipv6/anycast.c +2 -0
+       if (!dev)
+               return -ENODEV;
Quoted 4 lines

Sabrina took that from my draft patch, but they all don't realize this is wrong.

(I did provide a correct version which is just ignored by you.)

What games are you playing? You know how patches are processed by David and I even let him the choice by pointing out a problem in your patch so that you could an update and send v2.

I really feel miffed about your behavior!

Anyway, I saw the hunk adding the return -ENODEV and didn't see any problems with it. Sure it might be better if it would gone into a separate patch. Can you elaborate what problems you see?

Thanks,
Hannes

#18 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Quoted 4 lines · Hannes Frederic Sowa

What games are you playing? You know how patches are processed by David and I even let him the choice by pointing out a problem in your patch so that you could an update and send v2.

I assume David goes over all the discussion before applying any patch. So in the original discussion, obviously I disagree with your point on RCU, and sent my own version.

I do understand David may miss something due to massive emails in this list, but you make no point here nor I understand your broken English "let him the choice...".

Quoted 3 lines

I really feel miffed about your behavior!

I SEE. Please do redirect my email to your /dev/null. Thanks a lot!

Quoted 4 lines

Anyway, I saw the hunk adding the return -ENODEV and didn't see any problems with it. Sure it might be better if it would gone into a separate patch. Can you elaborate what problems you see?

Yes, it should be not in this patch at the very very least.

maintainer
#19 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

From: Cong Wang <[email protected]>
Date: Fri, 5 Sep 2014 12:23:37 -0700

Quoted 10 lines
Quoted 4 lines · Hannes Frederic Sowa

What games are you playing? You know how patches are processed by David and I even let him the choice by pointing out a problem in your patch so that you could an update and send v2.

I assume David goes over all the discussion before applying any patch. So in the original discussion, obviously I disagree with your point on RCU, and sent my own version.

I thought the retention of RCU locking was reasonable, and that your feedback was something I disagreed with.

This is different from ignoring your feedback.

#20 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

Quoted 5 lines · David Miller

I thought the retention of RCU locking was reasonable, and that your feedback was something I disagreed with.

This is different from ignoring your feedback.

The point is we don't need to backport the patch that far, as I already mentioned, up to commit c15b1ccadb323ea
("ipv6: move DAD and addrconf_verify processing to workqueue"), simply because no one reports any bug.

So we don't have to adjust the patch just for stable in this case.

maintainer
#21 lore raw

Re: [PATCH net v2] ipv6: fix rtnl locking in setsockopt for anycast and multicast

In reply to Cong Wang

From: Cong Wang <[email protected]>
Date: Fri, 5 Sep 2014 11:58:31 -0700

Quoted 26 lines · David Miller
Quoted 11 lines

From: Sabrina Dubroca <[email protected]>
Date: Tue, 2 Sep 2014 10:29:29 +0200

Quoted 8 lines

Calling setsockopt with IPV6_JOIN_ANYCAST or IPV6_LEAVE_ANYCAST triggers the assertion in addrconf_join_solict()/addrconf_leave_solict()

ipv6_sock_ac_join(), ipv6_sock_ac_drop(), ipv6_sock_ac_close() need to take RTNL before calling ipv6_dev_ac_inc/dec. Same thing with ipv6_sock_mc_join(), ipv6_sock_mc_drop(), ipv6_sock_mc_close() before calling ipv6_dev_mc_inc/dec.

Quoted 7 lines
Quoted 5 lines

This patch moves ASSERT_RTNL() up a level in the call stack.

Applied and queued up for -stable, thanks.

I believe you applied a wrong version, at least the following is not correct:

net/ipv6/anycast.c +2 -0
+       if (!dev)
+               return -ENODEV;
Quoted 4 lines

Sabrina took that from my draft patch, but they all don't realize this is wrong.

(I did provide a correct version which is just ignored by you.)

Not ignored, but rather it was hard to interpret the situation due to poor communication in the feedback emails.

The onus is on you guys to communicate things precisely so that I understand what patch is in what state.

#22 lore raw
In reply to Hannes Frederic Sowa
Quoted 5 lines · Hannes Frederic Sowa

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

There is no point to keep RCU read lock if we have rtnl lock, I don't know why you don't want to change dev_get_by_flags(), it is pretty easy to do since it only has one caller.

Even if you really need RCU in future, you are always welcome to bring it back when you do, sorry we should never be blocked by code NOT merged yet.

Quoted 4 lines

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Make it another patch.

#23 lore raw

Hi Cong,

Quoted 11 lines · Cong Wang
Quoted 5 lines · Hannes Frederic Sowa

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

There is no point to keep RCU read lock if we have rtnl lock, I don't know why you don't want to change dev_get_by_flags(), it is pretty easy to do since it only has one caller.

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

Quoted 10 lines

Even if you really need RCU in future, you are always welcome to bring it back when you do, sorry we should never be blocked by code NOT merged yet.

Quoted 4 lines

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Make it another patch.

It is just one logical change, moving ASSERT_RTNLs to places where they better catch invalid callstacks.

Bye,
Hannes

#24 lore raw
Quoted 11 lines · Hannes Frederic Sowa

Hi Cong,

Quoted 8 lines · Cong Wang
Quoted 5 lines · Hannes Frederic Sowa

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

Quoted 7 lines
Quoted 3 lines

There is no point to keep RCU read lock if we have rtnl lock, I don't know why you don't want to change dev_get_by_flags(), it is pretty easy to do since it only has one caller.

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

git grep should show you we only have one caller. Apparently we don't care about any out-of-tree module.

Quoted 6 lines
Quoted 6 lines

Even if you really need RCU in future, you are always welcome to bring it back when you do, sorry we should never be blocked by code NOT merged yet.

Quoted 1 line
Quoted 9 lines
Quoted 5 lines
Quoted 3 lines

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Make it another patch.

It is just one logical change, moving ASSERT_RTNLs to places where they better catch invalid callstacks.

Conflicts with what you claimed above. :)

#25 lore raw
Quoted 8 lines · Cong Wang

On Tue, Sep 2, 2014 at 10:58 AM, Hannes Frederic Sowa

Quoted 3 lines

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

git grep should show you we only have one caller. Apparently we don't care about any out-of-tree module.

Point is : you did not check if some stable versions had more callers.

Its very nice you checked current version, but it is not enough for a stable candidate.

#26 lore raw
Quoted 7 lines · Eric Dumazet
Quoted 6 lines · Cong Wang

On Tue, Sep 2, 2014 at 10:58 AM, Hannes Frederic Sowa

Quoted 3 lines

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

Quoted 7 lines
Quoted 2 lines

git grep should show you we only have one caller. Apparently we don't care about any out-of-tree module.

Point is : you did not check if some stable versions had more callers.

Its very nice you checked current version, but it is not enough for a stable candidate.

That is what we do when backporting patches, I can do that if David asks me to backport it, but you know for netdev that is David's work.

(I am not saying I don't want to help him, I just want to point out the fact. I am very pleased to help David for stable backports as long as he asks)

#27 lore raw
Quoted 5 lines · Cong Wang

That is what we do when backporting patches, I can do that if David asks me to backport it, but you know for netdev that is David's work.

(I am not saying I don't want to help him, I just want to point out the fact. I am very pleased to help David for stable backports as long as he asks)

Problem is : your patch submission do not identify bug origin.

You claim you want to help, but you do not provide the basic thing that _really_ helps.

The proper way to identify bug origin is to add in the headers one line :

Fixes 12-digit-SHA1 ("patch title")
#28 lore raw
Quoted 10 lines · Eric Dumazet
Quoted 5 lines · Cong Wang

That is what we do when backporting patches, I can do that if David asks me to backport it, but you know for netdev that is David's work.

(I am not saying I don't want to help him, I just want to point out the fact. I am very pleased to help David for stable backports as long as he asks)

Problem is : your patch submission do not identify bug origin.

Quoted 8 lines

You claim you want to help, but you do not provide the basic thing that _really_ helps.

The proper way to identify bug origin is to add in the headers one line :

Fixes 12-digit-SHA1 ("patch title")

Since when "Fixes:" tag becomes mandatory for a stable patch? At least netdev-FAQ is not updated. ;-) I 100% agree "Fixes:" is helpful when backporting patches, but it is not mandatory currently.

For this patch, I was too lazy to dig the history, it looks like this is caused by the following commit:

commit c15b1ccadb323ea50023e8f1cca2954129a62b51

Author: Hannes Frederic Sowa <[email protected]>
Date:   Thu Mar 27 18:28:07 2014 +0100

ipv6: move DAD and addrconf_verify processing to workqueue

Thanks.

#29 lore raw
In reply to Cong Wang
Quoted 13 lines · Cong Wang
Quoted 12 lines · Eric Dumazet
Quoted 8 lines · Cong Wang

On Tue, Sep 2, 2014 at 10:58 AM, Hannes Frederic Sowa

Quoted 3 lines

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

git grep should show you we only have one caller. Apparently we don't care about any out-of-tree module.

Point is : you did not check if some stable versions had more callers.

Quoted 8 lines
Quoted 2 lines

Its very nice you checked current version, but it is not enough for a stable candidate.

That is what we do when backporting patches, I can do that if David asks me to backport it, but you know for netdev that is David's work.

(I am not saying I don't want to help him, I just want to point out the fact. I am very pleased to help David for stable backports as long as he asks)

Instead of helping after the fact, why not arrange the patches so that it's not such a big issue. Leave the _rcu variant alone. Add an _rtnl variant of the function and use that in the patch. Have a follow-on patch that removes the _rcu variant all by itself. This way backports become easier, and if anyone wants the _rcu variant back, all they have to do is revert a very simple commit.

-vlad

#30 lore raw
In reply to Cong Wang
Quoted 17 lines · Cong Wang
Quoted 15 lines · Hannes Frederic Sowa

Hi Cong,

Quoted 11 lines · Cong Wang
Quoted 5 lines · Hannes Frederic Sowa

Also rtnl_lock and rcu_read_lock compose in that order, so we don't need to change dev_get_by_flags, but as this is the only user it sure is possible. RCU locked version is just easier composeable, so I wouldn't touch that if needed in future, just also take rcu lock as before.

There is no point to keep RCU read lock if we have rtnl lock, I don't know why you don't want to change dev_get_by_flags(), it is pretty easy to do since it only has one caller.

Quoted 6 lines
Quoted 3 lines

I definitely don't have a problem cleaning this up in net-next. I wanted a minimal patch for stable because I didn't check history where and when additional users of dev_get_by_flags_rcu were removed.

git grep should show you we only have one caller. Apparently we don't care about any out-of-tree module.

Sure, I don't care about out-of-tree modules either. I just wanted to make it easier to backport. Current patch is almost headache free to backport.

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

Even if you really need RCU in future, you are always welcome to bring it back when you do, sorry we should never be blocked by code NOT merged yet.

Quoted 4 lines

Also we should move ASSERT_RTNL checks from addrconf_join_solict to ipv6_dev_mc_inc/dec.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Make it another patch.

It is just one logical change, moving ASSERT_RTNLs to places where they better catch invalid callstacks.

Conflicts with what you claimed above. :)

Those ASSERT_RTNLs were misplaced and only caught the callers mostly from addrconf.c. I don't mind getting reports from stable kernel users and fixing those, too (or help fixing those). ASSERT_RTNL is not dangerous.

We had a long history in not correctly using rtnl lock in ipv6/multicast code and those wrongfully placed ASSERT_RTNLs were my bad when I fixed the duplicate address detection handling.

If enough multicast addresses are subscribed to an interface we might again get those splats because enabling promisc mode on an interface will also check for rtnl lock.

Bye,
Hannes

#31 lore raw
Quoted 5 lines · Hannes Frederic Sowa

Those ASSERT_RTNLs were misplaced and only caught the callers mostly from addrconf.c. I don't mind getting reports from stable kernel users and fixing those, too (or help fixing those). ASSERT_RTNL is not dangerous.

Quoted 8 lines

We had a long history in not correctly using rtnl lock in ipv6/multicast code and those wrongfully placed ASSERT_RTNLs were my bad when I fixed the duplicate address detection handling.

If enough multicast addresses are subscribed to an interface we might again get those splats because enabling promisc mode on an interface will also check for rtnl lock.

Sure, I never doubt adding ASSERT_RTNL() is helpful, I just still think this should be for net-next, or at least a separated patch. I don't want my patch to be blamed in others' "Fixes:". :)

#32 lore raw
Quoted 11 lines · Cong Wang
Quoted 9 lines · Hannes Frederic Sowa

Those ASSERT_RTNLs were misplaced and only caught the callers mostly from addrconf.c. I don't mind getting reports from stable kernel users and fixing those, too (or help fixing those). ASSERT_RTNL is not dangerous.

We had a long history in not correctly using rtnl lock in ipv6/multicast code and those wrongfully placed ASSERT_RTNLs were my bad when I fixed the duplicate address detection handling.

Quoted 8 lines
Quoted 4 lines

If enough multicast addresses are subscribed to an interface we might again get those splats because enabling promisc mode on an interface will also check for rtnl lock.

Sure, I never doubt adding ASSERT_RTNL() is helpful, I just still think this should be for net-next, or at least a separated patch. I don't want my patch to be blamed in others' "Fixes:". :)

Come on, that's why we have community review. Nobody blames anyone because of added regressions. It's more a fault of the community then, and it works out fairly good I think! Even others are keen on fixing your bugs sometimes. ;)

If fixes tag is well researched, it won't point to the addition of ASSERT_RTNL() but your patch would help to discover a bug somewhere else in the stack.

I think for this patch a fixes-tag is hard to find because it is hard to find because it dates back to the beginning of the git history IMHO.

Bye,
Hannes

#33 lore raw
Quoted 8 lines · Hannes Frederic Sowa

If fixes tag is well researched, it won't point to the addition of ASSERT_RTNL() but your patch would help to discover a bug somewhere else in the stack.

I think for this patch a fixes-tag is hard to find because it is hard to find because it dates back to the beginning of the git history IMHO.

As I replied to Eric, this warning is probably caused by the following commit:

commit c15b1ccadb323ea50023e8f1cca2954129a62b51

Author: Hannes Frederic Sowa <[email protected]>
Date:   Thu Mar 27 18:28:07 2014 +0100

ipv6: move DAD and addrconf_verify processing to workqueue

HOWEVER, you can definitely argue that the code without your ASSERT_RTNL() was already broken, it's a little hard to tell without digging more, that might date back to the beginning of git as you said.

For safety, I think we can simply assume it's that commit to be fixed so that we don't fix older kernels until someone really reports a bug.