netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: David Ahern <dsahern@kernel.org>
To: netdev@vger.kernel.org
Cc: davem@davemloft.net, kuba@kernel.org,
	nikolay@cumulusnetworks.com, David Ahern <dsahern@gmail.com>
Subject: [PATCH net v2 4/5] ipv4: Refactor nhc evaluation in fib_table_lookup
Date: Tue, 26 May 2020 12:56:17 -0600	[thread overview]
Message-ID: <20200526185618.43748-5-dsahern@kernel.org> (raw)
In-Reply-To: <20200526185618.43748-1-dsahern@kernel.org>

From: David Ahern <dsahern@gmail.com>

FIB lookups can return an entry that references an external nexthop.
While walking the nexthop struct we do not want to make multiple calls
into the nexthop code which can result in 2 different structs getting
accessed - one returning the number of paths the rest of the loop
seeing a different nh_grp struct. If the nexthop group shrunk, the
result is an attempt to access a fib_nh_common that does not exist for
the new nh_grp struct but did for the old one.

To fix that move the device evaluation code to a helper that can be
used for inline fib_nh path as well as external nexthops.

Update the existing check for fi->nh in fib_table_lookup to call a
new helper, nexthop_get_nhc_lookup, which walks the external nexthop
with a single rcu dereference.

Fixes: 430a049190de ("nexthop: Add support for nexthop groups")
Signed-off-by: David Ahern <dsahern@gmail.com>
Acked-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
 include/net/ip_fib.h  |  2 ++
 include/net/nexthop.h | 33 ++++++++++++++++++++++++++++
 net/ipv4/fib_trie.c   | 51 ++++++++++++++++++++++++++++++-------------
 3 files changed, 71 insertions(+), 15 deletions(-)

diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index b219a8fe0950..771ce068bc96 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -479,6 +479,8 @@ void fib_nh_common_release(struct fib_nh_common *nhc);
 void fib_alias_hw_flags_set(struct net *net, const struct fib_rt_info *fri);
 void fib_trie_init(void);
 struct fib_table *fib_trie_table(u32 id, struct fib_table *alias);
+bool fib_lookup_good_nhc(const struct fib_nh_common *nhc, int fib_flags,
+			 const struct flowi4 *flp);
 
 static inline void fib_combine_itag(u32 *itag, const struct fib_result *res)
 {
diff --git a/include/net/nexthop.h b/include/net/nexthop.h
index f09e8d7d9886..9414ae46fc1c 100644
--- a/include/net/nexthop.h
+++ b/include/net/nexthop.h
@@ -233,6 +233,39 @@ struct fib_nh_common *nexthop_fib_nhc(struct nexthop *nh, int nhsel)
 	return &nhi->fib_nhc;
 }
 
+/* called from fib_table_lookup with rcu_lock */
+static inline
+struct fib_nh_common *nexthop_get_nhc_lookup(const struct nexthop *nh,
+					     int fib_flags,
+					     const struct flowi4 *flp,
+					     int *nhsel)
+{
+	struct nh_info *nhi;
+
+	if (nh->is_group) {
+		struct nh_group *nhg = rcu_dereference(nh->nh_grp);
+		int i;
+
+		for (i = 0; i < nhg->num_nh; i++) {
+			struct nexthop *nhe = nhg->nh_entries[i].nh;
+
+			nhi = rcu_dereference(nhe->nh_info);
+			if (fib_lookup_good_nhc(&nhi->fib_nhc, fib_flags, flp)) {
+				*nhsel = i;
+				return &nhi->fib_nhc;
+			}
+		}
+	} else {
+		nhi = rcu_dereference(nh->nh_info);
+		if (fib_lookup_good_nhc(&nhi->fib_nhc, fib_flags, flp)) {
+			*nhsel = 0;
+			return &nhi->fib_nhc;
+		}
+	}
+
+	return NULL;
+}
+
 static inline unsigned int fib_info_num_path(const struct fib_info *fi)
 {
 	if (unlikely(fi->nh))
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 4f334b425538..248f1c1959a6 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1371,6 +1371,26 @@ static inline t_key prefix_mismatch(t_key key, struct key_vector *n)
 	return (key ^ prefix) & (prefix | -prefix);
 }
 
+bool fib_lookup_good_nhc(const struct fib_nh_common *nhc, int fib_flags,
+			 const struct flowi4 *flp)
+{
+	if (nhc->nhc_flags & RTNH_F_DEAD)
+		return false;
+
+	if (ip_ignore_linkdown(nhc->nhc_dev) &&
+	    nhc->nhc_flags & RTNH_F_LINKDOWN &&
+	    !(fib_flags & FIB_LOOKUP_IGNORE_LINKSTATE))
+		return false;
+
+	if (!(flp->flowi4_flags & FLOWI_FLAG_SKIP_NH_OIF)) {
+		if (flp->flowi4_oif &&
+		    flp->flowi4_oif != nhc->nhc_oif)
+			return false;
+	}
+
+	return true;
+}
+
 /* should be called with rcu_read_lock */
 int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
 		     struct fib_result *res, int fib_flags)
@@ -1503,6 +1523,7 @@ int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
 	/* Step 3: Process the leaf, if that fails fall back to backtracing */
 	hlist_for_each_entry_rcu(fa, &n->leaf, fa_list) {
 		struct fib_info *fi = fa->fa_info;
+		struct fib_nh_common *nhc;
 		int nhsel, err;
 
 		if ((BITS_PER_LONG > KEYLENGTH) || (fa->fa_slen < KEYLENGTH)) {
@@ -1528,26 +1549,25 @@ int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
 		if (fi->fib_flags & RTNH_F_DEAD)
 			continue;
 
-		if (unlikely(fi->nh && nexthop_is_blackhole(fi->nh))) {
-			err = fib_props[RTN_BLACKHOLE].error;
-			goto out_reject;
+		if (unlikely(fi->nh)) {
+			if (nexthop_is_blackhole(fi->nh)) {
+				err = fib_props[RTN_BLACKHOLE].error;
+				goto out_reject;
+			}
+
+			nhc = nexthop_get_nhc_lookup(fi->nh, fib_flags, flp,
+						     &nhsel);
+			if (nhc)
+				goto set_result;
+			goto miss;
 		}
 
 		for (nhsel = 0; nhsel < fib_info_num_path(fi); nhsel++) {
-			struct fib_nh_common *nhc = fib_info_nhc(fi, nhsel);
+			nhc = fib_info_nhc(fi, nhsel);
 
-			if (nhc->nhc_flags & RTNH_F_DEAD)
+			if (!fib_lookup_good_nhc(nhc, fib_flags, flp))
 				continue;
-			if (ip_ignore_linkdown(nhc->nhc_dev) &&
-			    nhc->nhc_flags & RTNH_F_LINKDOWN &&
-			    !(fib_flags & FIB_LOOKUP_IGNORE_LINKSTATE))
-				continue;
-			if (!(flp->flowi4_flags & FLOWI_FLAG_SKIP_NH_OIF)) {
-				if (flp->flowi4_oif &&
-				    flp->flowi4_oif != nhc->nhc_oif)
-					continue;
-			}
-
+set_result:
 			if (!(fib_flags & FIB_LOOKUP_NOREF))
 				refcount_inc(&fi->fib_clntref);
 
@@ -1568,6 +1588,7 @@ int fib_table_lookup(struct fib_table *tb, const struct flowi4 *flp,
 			return err;
 		}
 	}
+miss:
 #ifdef CONFIG_IP_FIB_TRIE_STATS
 	this_cpu_inc(stats->semantic_match_miss);
 #endif
-- 
2.21.1 (Apple Git-122.3)


  parent reply	other threads:[~2020-05-26 19:33 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-05-26 18:56 [PATCH net v2 0/5] nexthops: Fix 2 fundamental flaws with nexthop groups David Ahern
2020-05-26 18:56 ` [PATCH net v2 1/5] nexthops: Move code from remove_nexthop_from_groups to remove_nh_grp_entry David Ahern
2020-05-26 18:56 ` [PATCH net v2 2/5] nexthops: don't modify published nexthop groups David Ahern
2020-05-26 18:56 ` [PATCH net v2 3/5] nexthop: Expand nexthop_is_multipath in a few places David Ahern
2020-05-26 18:56 ` David Ahern [this message]
2020-05-26 18:56 ` [PATCH net v2 5/5] ipv4: nexthop version of fib_info_nh_uses_dev David Ahern
2020-05-26 23:07 ` [PATCH net v2 0/5] nexthops: Fix 2 fundamental flaws with nexthop groups David Miller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200526185618.43748-5-dsahern@kernel.org \
    --to=dsahern@kernel.org \
    --cc=davem@davemloft.net \
    --cc=dsahern@gmail.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=nikolay@cumulusnetworks.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).