[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20150807181038.2A32029003CC@tardy>
Date: Fri, 7 Aug 2015 11:10:37 -0700 (PDT)
From: raj@...dy.usa.hp.com (Rick Jones)
To: <netdev@...r.kernel.org>
Cc: <davem@...emloft.net>
Subject: [PATCH net-next] net: add explicit logging and stat for neighbour table overflow
From: Rick Jones <rick.jones2@...com>
Add an explicit neighbour table overflow message (ratelimited) and
statistic to make diagnosing neighbour table overflows tractable in
the wild.
Diagnosing a neighbour table overflow can be quite difficult in the wild
because there is no explicit dmesg logged. Callers to neighbour code
seem to use net_dbg_ratelimit when the neighbour call fails which means
the "base message" is not emitted and the callback suppressed messages
from the ratelimiting can end-up juxtaposed with unrelated messages.
Further, a forced garbage collection will increment a stat on each call
whether it was successful in freeing-up a table entry or not, so that
statistic is only a hint. So, add a net_info_ratelimited message and
explicit statistic to the neighbour code.
Signed-off-by: Rick Jones <rick.jones2@...com>
---
Tested by cutting back the gc_threshN sysctls and attempting to ping a
number of target IPs greater than the maximum size of the ARP table.
diff --git a/include/net/neighbour.h b/include/net/neighbour.h
index bd33e66..8b68384 100644
--- a/include/net/neighbour.h
+++ b/include/net/neighbour.h
@@ -125,6 +125,7 @@ struct neigh_statistics {
unsigned long forced_gc_runs; /* number of forced GC runs */
unsigned long unres_discards; /* number of unresolved drops */
+ unsigned long table_fulls; /* times even gc couldn't help */
};
#define NEIGH_CACHE_STAT_INC(tbl, field) this_cpu_inc((tbl)->stats->field)
diff --git a/include/uapi/linux/neighbour.h b/include/uapi/linux/neighbour.h
index 2e35c61..788655b 100644
--- a/include/uapi/linux/neighbour.h
+++ b/include/uapi/linux/neighbour.h
@@ -106,6 +106,7 @@ struct ndt_stats {
__u64 ndts_rcv_probes_ucast;
__u64 ndts_periodic_gc_runs;
__u64 ndts_forced_gc_runs;
+ __u64 ndts_table_fulls;
};
enum {
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index 84195da..2b515ba 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -274,8 +274,12 @@ static struct neighbour *neigh_alloc(struct neigh_table *tbl, struct net_device
(entries >= tbl->gc_thresh2 &&
time_after(now, tbl->last_flush + 5 * HZ))) {
if (!neigh_forced_gc(tbl) &&
- entries >= tbl->gc_thresh3)
+ entries >= tbl->gc_thresh3) {
+ net_info_ratelimited("%s: neighbor table overflow!\n",
+ tbl->id);
+ NEIGH_CACHE_STAT_INC(tbl, table_fulls);
goto out_entries;
+ }
}
n = kzalloc(tbl->entry_size + dev->neigh_priv_len, GFP_ATOMIC);
@@ -1849,6 +1853,7 @@ static int neightbl_fill_info(struct sk_buff *skb, struct neigh_table *tbl,
ndst.ndts_rcv_probes_ucast += st->rcv_probes_ucast;
ndst.ndts_periodic_gc_runs += st->periodic_gc_runs;
ndst.ndts_forced_gc_runs += st->forced_gc_runs;
+ ndst.ndts_table_fulls += st->table_fulls;
}
if (nla_put(skb, NDTA_STATS, sizeof(ndst), &ndst))
@@ -2717,12 +2722,12 @@ static int neigh_stat_seq_show(struct seq_file *seq, void *v)
struct neigh_statistics *st = v;
if (v == SEQ_START_TOKEN) {
- seq_printf(seq, "entries allocs destroys hash_grows lookups hits res_failed rcv_probes_mcast rcv_probes_ucast periodic_gc_runs forced_gc_runs unresolved_discards\n");
+ seq_printf(seq, "entries allocs destroys hash_grows lookups hits res_failed rcv_probes_mcast rcv_probes_ucast periodic_gc_runs forced_gc_runs unresolved_discards table_fulls\n");
return 0;
}
seq_printf(seq, "%08x %08lx %08lx %08lx %08lx %08lx %08lx "
- "%08lx %08lx %08lx %08lx %08lx\n",
+ "%08lx %08lx %08lx %08lx %08lx %08lx\n",
atomic_read(&tbl->entries),
st->allocs,
@@ -2739,7 +2744,8 @@ static int neigh_stat_seq_show(struct seq_file *seq, void *v)
st->periodic_gc_runs,
st->forced_gc_runs,
- st->unres_discards
+ st->unres_discards,
+ st->table_fulls
);
return 0;
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists