aboutsummaryrefslogtreecommitdiffstats
path: root/include
diff options
context:
space:
mode:
authorJakub Kicinski <kuba@kernel.org>2022-10-31 20:14:30 -0700
committerJakub Kicinski <kuba@kernel.org>2022-10-31 20:14:30 -0700
commit6f1a298b2e24c703bfcc643e41bc7c0604fe4830 (patch)
tree042e3e6a4c27c96c782f403bd60b1eec3780b3bb /include
parentnet: systemport: Add support for RDMA overflow statistic counter (diff)
parentnet: dropreason: add SKB_DROP_REASON_FRAG_TOO_FAR (diff)
downloadlinux-6f1a298b2e24c703bfcc643e41bc7c0604fe4830.tar.gz
linux-6f1a298b2e24c703bfcc643e41bc7c0604fe4830.zip
Merge branch 'inet-add-drop-monitor-support'
Eric Dumazet says: ==================== inet: add drop monitor support I recently tried to analyse flakes in ip_defrag selftest. This failed miserably. IPv4 and IPv6 reassembly units are causing false kfree_skb() notifications. It is time to deal with this issue. First two patches are changing core networking to better deal with eventual skb frag_list chains, in respect of kfree_skb/consume_skb status. Last three patches are adding three new drop reasons, and make sure skbs that have been reassembled into a large datagram are no longer viewed as dropped ones. After this, understanding why ip_defrag selftest is flaky is possible using standard drop monitoring tools. ==================== Link: https://lore.kernel.org/r/20221029154520.2747444-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Diffstat (limited to 'include')
-rw-r--r--include/net/dropreason.h14
-rw-r--r--include/net/inet_frag.h6
-rw-r--r--include/net/ipv6_frag.h3
3 files changed, 21 insertions, 2 deletions
diff --git a/include/net/dropreason.h b/include/net/dropreason.h
index c1cbcdbaf149..70539288f995 100644
--- a/include/net/dropreason.h
+++ b/include/net/dropreason.h
@@ -68,6 +68,9 @@
FN(IP_INADDRERRORS) \
FN(IP_INNOROUTES) \
FN(PKT_TOO_BIG) \
+ FN(DUP_FRAG) \
+ FN(FRAG_REASM_TIMEOUT) \
+ FN(FRAG_TOO_FAR) \
FNe(MAX)
/**
@@ -80,6 +83,8 @@ enum skb_drop_reason {
* @SKB_NOT_DROPPED_YET: skb is not dropped yet (used for no-drop case)
*/
SKB_NOT_DROPPED_YET = 0,
+ /** @SKB_CONSUMED: packet has been consumed */
+ SKB_CONSUMED,
/** @SKB_DROP_REASON_NOT_SPECIFIED: drop reason is not specified */
SKB_DROP_REASON_NOT_SPECIFIED,
/** @SKB_DROP_REASON_NO_SOCKET: socket not found */
@@ -298,6 +303,15 @@ enum skb_drop_reason {
* MTU)
*/
SKB_DROP_REASON_PKT_TOO_BIG,
+ /** @SKB_DROP_REASON_DUP_FRAG: duplicate fragment */
+ SKB_DROP_REASON_DUP_FRAG,
+ /** @SKB_DROP_REASON_FRAG_REASM_TIMEOUT: fragment reassembly timeout */
+ SKB_DROP_REASON_FRAG_REASM_TIMEOUT,
+ /**
+ * @SKB_DROP_REASON_FRAG_TOO_FAR: ipv4 fragment too far.
+ * (/proc/sys/net/ipv4/ipfrag_max_dist)
+ */
+ SKB_DROP_REASON_FRAG_TOO_FAR,
/**
* @SKB_DROP_REASON_MAX: the maximum of drop reason, which shouldn't be
* used as a real 'reason'
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 0b0876610553..b23ddec3cd5c 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -7,6 +7,7 @@
#include <linux/in6.h>
#include <linux/rbtree_types.h>
#include <linux/refcount.h>
+#include <net/dropreason.h>
/* Per netns frag queues directory */
struct fqdir {
@@ -34,12 +35,14 @@ struct fqdir {
* @INET_FRAG_LAST_IN: final fragment has arrived
* @INET_FRAG_COMPLETE: frag queue has been processed and is due for destruction
* @INET_FRAG_HASH_DEAD: inet_frag_kill() has not removed fq from rhashtable
+ * @INET_FRAG_DROP: if skbs must be dropped (instead of being consumed)
*/
enum {
INET_FRAG_FIRST_IN = BIT(0),
INET_FRAG_LAST_IN = BIT(1),
INET_FRAG_COMPLETE = BIT(2),
INET_FRAG_HASH_DEAD = BIT(3),
+ INET_FRAG_DROP = BIT(4),
};
struct frag_v4_compare_key {
@@ -139,7 +142,8 @@ void inet_frag_destroy(struct inet_frag_queue *q);
struct inet_frag_queue *inet_frag_find(struct fqdir *fqdir, void *key);
/* Free all skbs in the queue; return the sum of their truesizes. */
-unsigned int inet_frag_rbtree_purge(struct rb_root *root);
+unsigned int inet_frag_rbtree_purge(struct rb_root *root,
+ enum skb_drop_reason reason);
static inline void inet_frag_put(struct inet_frag_queue *q)
{
diff --git a/include/net/ipv6_frag.h b/include/net/ipv6_frag.h
index 5052c66e22d2..7321ffe3a108 100644
--- a/include/net/ipv6_frag.h
+++ b/include/net/ipv6_frag.h
@@ -76,6 +76,7 @@ ip6frag_expire_frag_queue(struct net *net, struct frag_queue *fq)
if (fq->q.flags & INET_FRAG_COMPLETE)
goto out;
+ fq->q.flags |= INET_FRAG_DROP;
inet_frag_kill(&fq->q);
dev = dev_get_by_index_rcu(net, fq->iif);
@@ -101,7 +102,7 @@ ip6frag_expire_frag_queue(struct net *net, struct frag_queue *fq)
spin_unlock(&fq->q.lock);
icmpv6_send(head, ICMPV6_TIME_EXCEED, ICMPV6_EXC_FRAGTIME, 0);
- kfree_skb(head);
+ kfree_skb_reason(head, SKB_DROP_REASON_FRAG_REASM_TIMEOUT);
goto out_rcu_unlock;
out: