From 7a400eb025dd53883c3560d0fdb069542f7ad3db Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timo=20Ter=C3=A4s?= Date: Wed, 31 Mar 2010 00:17:05 +0000 Subject: [PATCH 12/18] xfrm: remove policy lock when accessing policy->walk.dead All of the code considers ->dead as a hint that the cached policy needs to get refreshed. The read side can just drop the read lock without any side effects. The write side needs to make sure that it's written only exactly once. Only possible race is at xfrm_policy_kill(). This is fixed by checking result of __xfrm_policy_unlink() when needed. It will always succeed if the policy object is looked up from the hash list (so some checks are removed), but it needs to be checked if we are trying to unlink policy via a reference (appropriate checks added). Since policy->walk.dead is written exactly once, it no longer needs to be protected with a write lock. Signed-off-by: Timo Teras Acked-by: Herbert Xu Signed-off-by: David S. Miller (backported from commit ea2dea9dacc256fe927857feb423872051642ae7) --- net/xfrm/xfrm_policy.c | 20 +++++--------------- net/xfrm/xfrm_user.c | 6 +----- 2 files changed, 6 insertions(+), 20 deletions(-) diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c index d75047c..110184f 100644 --- a/net/xfrm/xfrm_policy.c +++ b/net/xfrm/xfrm_policy.c @@ -156,7 +156,7 @@ static void xfrm_policy_timer(unsigned long data) read_lock(&xp->lock); - if (xp->walk.dead) + if (unlikely(xp->walk.dead)) goto out; dir = xfrm_policy_id2dir(xp->index); @@ -297,17 +297,7 @@ static DECLARE_WORK(xfrm_policy_gc_work, xfrm_policy_gc_task); static void xfrm_policy_kill(struct xfrm_policy *policy) { - int dead; - - write_lock_bh(&policy->lock); - dead = policy->walk.dead; policy->walk.dead = 1; - write_unlock_bh(&policy->lock); - - if (unlikely(dead)) { - WARN_ON(1); - return; - } spin_lock_bh(&xfrm_policy_gc_lock); hlist_add_head(&policy->bydst, &xfrm_policy_gc_list); @@ -1115,6 +1105,9 @@ int xfrm_sk_policy_insert(struct sock *sk, int dir, struct xfrm_policy *pol) __xfrm_policy_link(pol, XFRM_POLICY_MAX+dir); } if (old_pol) + /* Unlinking succeeds always. This is the only function + * allowed to delete or replace socket policy. + */ __xfrm_policy_unlink(old_pol, XFRM_POLICY_MAX+dir); write_unlock_bh(&xfrm_policy_lock); @@ -1705,11 +1698,8 @@ restart: goto error; } - for (pi = 0; pi < npols; pi++) { - read_lock_bh(&pols[pi]->lock); + for (pi = 0; pi < npols; pi++) pol_dead |= pols[pi]->walk.dead; - read_unlock_bh(&pols[pi]->lock); - } write_lock_bh(&policy->lock); if (unlikely(pol_dead || stale_bundle(dst))) { diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c index d1e9ee3..f9c56e9 100644 --- a/net/xfrm/xfrm_user.c +++ b/net/xfrm/xfrm_user.c @@ -1617,13 +1617,9 @@ static int xfrm_add_pol_expire(struct sk_buff *skb, struct nlmsghdr *nlh, if (xp == NULL) return -ENOENT; - read_lock(&xp->lock); - if (xp->walk.dead) { - read_unlock(&xp->lock); + if (unlikely(xp->walk.dead)) goto out; - } - read_unlock(&xp->lock); err = 0; if (up->hard) { uid_t loginuid = NETLINK_CB(skb).loginuid; -- 1.7.0.2