[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20240822043609.141992-1-takakura@valinux.co.jp>
Date: Thu, 22 Aug 2024 13:36:09 +0900
From: takakura@...inux.co.jp
To: pablo@...filter.org,
kadlec@...filter.org,
davem@...emloft.net,
dsahern@...nel.org,
edumazet@...gle.com,
kuba@...nel.org,
pabeni@...hat.com,
fw@...len.de
Cc: netfilter-devel@...r.kernel.org,
coreteam@...filter.org,
netdev@...r.kernel.org,
linux-kernel@...r.kernel.org,
Ryo Takakura <takakura@...inux.co.jp>
Subject: [PATCH] netfilter: Don't track counter updates of do_add_counters()
From: Ryo Takakura <takakura@...inux.co.jp>
While adding counters in do_add_counters(), we call
xt_write_recseq_begin/end to indicate that counters are being updated.
Updates are being tracked so that the counters retrieved by get_counters()
will reflect concurrent updates.
However, there is no need to track the updates done by do_add_counters() as
both do_add_counters() and get_counters() acquire per ipv4,ipv6,arp mutex
beforehand which prevents concurrent update and retrieval between the two.
Moreover, as the xt_write_recseq_begin/end is shared among ipv4,ipv6,arp,
do_add_counters() called by one of ipv4,ipv6,arp can falsely delay the
synchronization of concurrent get_counters() or xt_replace_table() called
by any other than the one calling do_add_counters().
So remove xt_write_recseq_begin/end from do_add_counters() for ipv4,ipv6,arp.
Signed-off-by: Ryo Takakura <takakura@...inux.co.jp>
---
net/ipv4/netfilter/arp_tables.c | 4 ----
net/ipv4/netfilter/ip_tables.c | 3 ---
net/ipv6/netfilter/ip6_tables.c | 3 ---
3 files changed, 10 deletions(-)
diff --git a/net/ipv4/netfilter/arp_tables.c b/net/ipv4/netfilter/arp_tables.c
index 14365b20f1c5..20de048d3e0c 100644
--- a/net/ipv4/netfilter/arp_tables.c
+++ b/net/ipv4/netfilter/arp_tables.c
@@ -1009,7 +1009,6 @@ static int do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
const struct xt_table_info *private;
int ret = 0;
struct arpt_entry *iter;
- unsigned int addend;
paddc = xt_copy_counters(arg, len, &tmp);
if (IS_ERR(paddc))
@@ -1029,8 +1028,6 @@ static int do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
}
i = 0;
-
- addend = xt_write_recseq_begin();
xt_entry_foreach(iter, private->entries, private->size) {
struct xt_counters *tmp;
@@ -1038,7 +1035,6 @@ static int do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
++i;
}
- xt_write_recseq_end(addend);
unlock_up_free:
local_bh_enable();
xt_table_unlock(t);
diff --git a/net/ipv4/netfilter/ip_tables.c b/net/ipv4/netfilter/ip_tables.c
index fe89a056eb06..f54dea2a8fcd 100644
--- a/net/ipv4/netfilter/ip_tables.c
+++ b/net/ipv4/netfilter/ip_tables.c
@@ -1162,7 +1162,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
const struct xt_table_info *private;
int ret = 0;
struct ipt_entry *iter;
- unsigned int addend;
paddc = xt_copy_counters(arg, len, &tmp);
if (IS_ERR(paddc))
@@ -1182,7 +1181,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
}
i = 0;
- addend = xt_write_recseq_begin();
xt_entry_foreach(iter, private->entries, private->size) {
struct xt_counters *tmp;
@@ -1190,7 +1188,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
++i;
}
- xt_write_recseq_end(addend);
unlock_up_free:
local_bh_enable();
xt_table_unlock(t);
diff --git a/net/ipv6/netfilter/ip6_tables.c b/net/ipv6/netfilter/ip6_tables.c
index 131f7bb2110d..f1d3bb74eb16 100644
--- a/net/ipv6/netfilter/ip6_tables.c
+++ b/net/ipv6/netfilter/ip6_tables.c
@@ -1179,7 +1179,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
const struct xt_table_info *private;
int ret = 0;
struct ip6t_entry *iter;
- unsigned int addend;
paddc = xt_copy_counters(arg, len, &tmp);
if (IS_ERR(paddc))
@@ -1198,7 +1197,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
}
i = 0;
- addend = xt_write_recseq_begin();
xt_entry_foreach(iter, private->entries, private->size) {
struct xt_counters *tmp;
@@ -1206,7 +1204,6 @@ do_add_counters(struct net *net, sockptr_t arg, unsigned int len)
ADD_COUNTER(*tmp, paddc[i].bcnt, paddc[i].pcnt);
++i;
}
- xt_write_recseq_end(addend);
unlock_up_free:
local_bh_enable();
xt_table_unlock(t);
--
2.34.1
Powered by blists - more mailing lists