[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20190308124915.803670841@linuxfoundation.org>
Date: Fri, 8 Mar 2019 13:49:40 +0100
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
stable@...r.kernel.org, Ido Schimmel <idosch@...lanox.com>,
Amit Cohen <amitc@...lanox.com>,
Jiri Pirko <jiri@...lanox.com>,
"David S. Miller" <davem@...emloft.net>
Subject: [PATCH 4.20 28/76] team: Free BPF filter when unregistering netdev
4.20-stable review patch. If anyone has any objections, please let me know.
------------------
From: Ido Schimmel <idosch@...lanox.com>
[ Upstream commit 692c31bd4054212312396b1d303bffab2c5b93a7 ]
When team is used in loadbalance mode a BPF filter can be used to
provide a hash which will determine the Tx port.
When the netdev is later unregistered the filter is not freed which
results in memory leaks [1].
Fix by freeing the program and the corresponding filter when
unregistering the netdev.
[1]
unreferenced object 0xffff8881dbc47cc8 (size 16):
comm "teamd", pid 3068, jiffies 4294997779 (age 438.247s)
hex dump (first 16 bytes):
a3 00 6b 6b 6b 6b 6b 6b 88 a5 82 e1 81 88 ff ff ..kkkkkk........
backtrace:
[<000000008a3b47e3>] team_nl_cmd_options_set+0x88f/0x11b0
[<00000000c4f4f27e>] genl_family_rcv_msg+0x78f/0x1080
[<00000000610ef838>] genl_rcv_msg+0xca/0x170
[<00000000a281df93>] netlink_rcv_skb+0x132/0x380
[<000000004d9448a2>] genl_rcv+0x29/0x40
[<000000000321b2f4>] netlink_unicast+0x4c0/0x690
[<000000008c25dffb>] netlink_sendmsg+0x929/0xe10
[<00000000068298c5>] sock_sendmsg+0xc8/0x110
[<0000000082a61ff0>] ___sys_sendmsg+0x77a/0x8f0
[<00000000663ae29d>] __sys_sendmsg+0xf7/0x250
[<0000000027c5f11a>] do_syscall_64+0x14d/0x610
[<000000006cfbc8d3>] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[<00000000e23197e2>] 0xffffffffffffffff
unreferenced object 0xffff8881e182a588 (size 2048):
comm "teamd", pid 3068, jiffies 4294997780 (age 438.247s)
hex dump (first 32 bytes):
20 00 00 00 02 00 00 00 30 00 00 00 28 f0 ff ff .......0...(...
07 00 00 00 00 00 00 00 28 00 00 00 00 00 00 00 ........(.......
backtrace:
[<000000002daf01fb>] lb_bpf_func_set+0x45c/0x6d0
[<000000008a3b47e3>] team_nl_cmd_options_set+0x88f/0x11b0
[<00000000c4f4f27e>] genl_family_rcv_msg+0x78f/0x1080
[<00000000610ef838>] genl_rcv_msg+0xca/0x170
[<00000000a281df93>] netlink_rcv_skb+0x132/0x380
[<000000004d9448a2>] genl_rcv+0x29/0x40
[<000000000321b2f4>] netlink_unicast+0x4c0/0x690
[<000000008c25dffb>] netlink_sendmsg+0x929/0xe10
[<00000000068298c5>] sock_sendmsg+0xc8/0x110
[<0000000082a61ff0>] ___sys_sendmsg+0x77a/0x8f0
[<00000000663ae29d>] __sys_sendmsg+0xf7/0x250
[<0000000027c5f11a>] do_syscall_64+0x14d/0x610
[<000000006cfbc8d3>] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[<00000000e23197e2>] 0xffffffffffffffff
Fixes: 01d7f30a9f96 ("team: add loadbalance mode")
Signed-off-by: Ido Schimmel <idosch@...lanox.com>
Reported-by: Amit Cohen <amitc@...lanox.com>
Acked-by: Jiri Pirko <jiri@...lanox.com>
Signed-off-by: David S. Miller <davem@...emloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
drivers/net/team/team_mode_loadbalance.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
--- a/drivers/net/team/team_mode_loadbalance.c
+++ b/drivers/net/team/team_mode_loadbalance.c
@@ -325,6 +325,20 @@ static int lb_bpf_func_set(struct team *
return 0;
}
+static void lb_bpf_func_free(struct team *team)
+{
+ struct lb_priv *lb_priv = get_lb_priv(team);
+ struct bpf_prog *fp;
+
+ if (!lb_priv->ex->orig_fprog)
+ return;
+
+ __fprog_destroy(lb_priv->ex->orig_fprog);
+ fp = rcu_dereference_protected(lb_priv->fp,
+ lockdep_is_held(&team->lock));
+ bpf_prog_destroy(fp);
+}
+
static int lb_tx_method_get(struct team *team, struct team_gsetter_ctx *ctx)
{
struct lb_priv *lb_priv = get_lb_priv(team);
@@ -639,6 +653,7 @@ static void lb_exit(struct team *team)
team_options_unregister(team, lb_options,
ARRAY_SIZE(lb_options));
+ lb_bpf_func_free(team);
cancel_delayed_work_sync(&lb_priv->ex->stats.refresh_dw);
free_percpu(lb_priv->pcpu_stats);
kfree(lb_priv->ex);
Powered by blists - more mailing lists