]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
ip6mr: Annotate access to mrt->mroute_do_{pim,assert,wrvifwhole}.
authorKuniyuki Iwashima <kuniyu@google.com>
Thu, 4 Jun 2026 22:46:20 +0000 (22:46 +0000)
committerJakub Kicinski <kuba@kernel.org>
Tue, 9 Jun 2026 00:06:22 +0000 (17:06 -0700)
These fields in struct mr_table are updated in ip6_mroute_setsockopt()
under RTNL:

  * mroute_do_pim
  * mroute_do_assert (MRT6_PIM is under RTNL while MRT6_ASSERT is lockless)
  * mroute_do_wrvifwhole

However, ip6_mroute_getsockopt() does not hold RTNL and read the first
two fields locklessly, and ip6_mr_forward() reads all the three under
RCU.

Let's use WRITE_ONCE() and READ_ONCE() for them.

Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260604224712.3209821-3-kuniyu@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/ipv6/ip6mr.c

index 9d748f4c44e97046fa5acae5042f8bd31e13b66d..e3dd5851b87912d9a5697899158d5cf297a404af 100644 (file)
@@ -1778,7 +1778,7 @@ int ip6_mroute_setsockopt(struct sock *sk, int optname, sockptr_t optval,
                        return -EINVAL;
                if (copy_from_sockptr(&v, optval, sizeof(v)))
                        return -EFAULT;
-               mrt->mroute_do_assert = v;
+               WRITE_ONCE(mrt->mroute_do_assert, v);
                return 0;
        }
 
@@ -1798,9 +1798,9 @@ int ip6_mroute_setsockopt(struct sock *sk, int optname, sockptr_t optval,
                rtnl_lock();
                ret = 0;
                if (v != mrt->mroute_do_pim) {
-                       mrt->mroute_do_pim = v;
-                       mrt->mroute_do_assert = v;
-                       mrt->mroute_do_wrvifwhole = do_wrmifwhole;
+                       WRITE_ONCE(mrt->mroute_do_pim, v);
+                       WRITE_ONCE(mrt->mroute_do_assert, v);
+                       WRITE_ONCE(mrt->mroute_do_wrvifwhole, do_wrmifwhole);
                }
                rtnl_unlock();
                return ret;
@@ -1868,11 +1868,11 @@ int ip6_mroute_getsockopt(struct sock *sk, int optname, sockptr_t optval,
                break;
 #ifdef CONFIG_IPV6_PIMSM_V2
        case MRT6_PIM:
-               val = mrt->mroute_do_pim;
+               val = READ_ONCE(mrt->mroute_do_pim);
                break;
 #endif
        case MRT6_ASSERT:
-               val = mrt->mroute_do_assert;
+               val = READ_ONCE(mrt->mroute_do_assert);
                break;
        default:
                return -ENOPROTOOPT;
@@ -2175,20 +2175,20 @@ static void ip6_mr_forward(struct net *net, struct mr_table *mrt,
        if (rcu_access_pointer(mrt->vif_table[vif].dev) != dev) {
                atomic_long_inc(&c->_c.mfc_un.res.wrong_if);
 
-               if (true_vifi >= 0 && mrt->mroute_do_assert &&
+               if (true_vifi >= 0 && READ_ONCE(mrt->mroute_do_assert) &&
                    /* pimsm uses asserts, when switching from RPT to SPT,
                       so that we cannot check that packet arrived on an oif.
                       It is bad, but otherwise we would need to move pretty
                       large chunk of pimd to kernel. Ough... --ANK
                     */
-                   (mrt->mroute_do_pim ||
+                   (READ_ONCE(mrt->mroute_do_pim) ||
                     c->_c.mfc_un.res.ttls[true_vifi] < 255) &&
                    time_after(jiffies,
                               c->_c.mfc_un.res.last_assert +
                               MFC_ASSERT_THRESH)) {
                        c->_c.mfc_un.res.last_assert = jiffies;
                        ip6mr_cache_report(mrt, skb, true_vifi, MRT6MSG_WRONGMIF);
-                       if (mrt->mroute_do_wrvifwhole)
+                       if (READ_ONCE(mrt->mroute_do_wrvifwhole))
                                ip6mr_cache_report(mrt, skb, true_vifi,
                                                   MRT6MSG_WRMIFWHOLE);
                }