From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.linuxfoundation.org ([140.211.169.12]:46449 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933217AbcHNUof (ORCPT ); Sun, 14 Aug 2016 16:44:35 -0400 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Xin Long , Herbert Xu , Vegard Nossum , Marcelo Ricardo Leitner , "David S. Miller" Subject: [PATCH 4.7 10/41] net/sctp: terminate rhashtable walk correctly Date: Sun, 14 Aug 2016 22:38:36 +0200 Message-Id: <20160814202532.613192555@linuxfoundation.org> In-Reply-To: <20160814202531.818402015@linuxfoundation.org> References: <20160814202531.818402015@linuxfoundation.org> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: stable-owner@vger.kernel.org List-ID: 4.7-stable review patch. If anyone has any objections, please let me know. ------------------ From: Vegard Nossum [ Upstream commit 5fc382d87517707ad77ea4c9c12e2a3fde2c838a ] I was seeing a lot of these: BUG: sleeping function called from invalid context at mm/slab.h:388 in_atomic(): 0, irqs_disabled(): 0, pid: 14971, name: trinity-c2 Preemption disabled at:[] rhashtable_walk_start+0x46/0x150 [] preempt_count_add+0x1fb/0x280 [] _raw_spin_lock+0x12/0x40 [] console_unlock+0x2f7/0x930 [] vprintk_emit+0x2fb/0x520 [] vprintk_default+0x1a/0x20 [] printk+0x94/0xb0 [] print_stack_trace+0xe0/0x170 [] ___might_sleep+0x3be/0x460 [] __might_sleep+0x90/0x1a0 [] kmem_cache_alloc+0x153/0x1e0 [] rhashtable_walk_init+0xfe/0x2d0 [] sctp_transport_walk_start+0x1e/0x60 [] sctp_transport_seq_start+0x4d/0x150 [] seq_read+0x27b/0x1180 [] proc_reg_read+0xbc/0x180 [] __vfs_read+0xdb/0x610 [] vfs_read+0xea/0x2d0 [] SyS_pread64+0x11b/0x150 [] do_syscall_64+0x19c/0x410 [] return_from_SYSCALL_64+0x0/0x6a [] 0xffffffffffffffff Apparently we always need to call rhashtable_walk_stop(), even when rhashtable_walk_start() fails: * rhashtable_walk_start - Start a hash table walk * @iter: Hash table iterator * * Start a hash table walk. Note that we take the RCU lock in all * cases including when we return an error. So you must always call * rhashtable_walk_stop to clean up. otherwise we never call rcu_read_unlock() and we get the splat above. Fixes: 53fa1036 ("sctp: fix some rhashtable functions using in sctp proc/diag") See-also: 53fa1036 ("sctp: fix some rhashtable functions using in sctp proc/diag") See-also: f2dba9c6 ("rhashtable: Introduce rhashtable_walk_*") Cc: Xin Long Cc: Herbert Xu Cc: stable@vger.kernel.org Signed-off-by: Vegard Nossum Acked-by: Marcelo Ricardo Leitner Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman --- net/sctp/socket.c | 1 + 1 file changed, 1 insertion(+) --- a/net/sctp/socket.c +++ b/net/sctp/socket.c @@ -4301,6 +4301,7 @@ int sctp_transport_walk_start(struct rha err = rhashtable_walk_start(iter); if (err && err != -EAGAIN) { + rhashtable_walk_stop(iter); rhashtable_walk_exit(iter); return err; }