Skip to contentSkip to conversation

fix rule flush operation

PATCH In review Eric Leblond 2 participants 6 messages Oct 2, 2013 – Oct 3, 2013 netfilter

lore
Share & tools
lore
#0 lore raw

cover fix rule flush operation

Hello,

Here's a small patchset which fixes rule flushing. A command like:
nft flush table ip6 nat
was not flushing the rules at all. Patch 4 fixes the problem and patch 3 limit the scope of the flush which was not limited.

Patchset statistics:

 src/mnl.c     |  2 +-
 src/netlink.c | 13 ++++++++-----
 2 files changed, 9 insertions(+), 6 deletions(-)

BR,
--
Eric Leblond

#1 lore raw

PATCH 1/4 mnl: fix typo in comment

---
 src/mnl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
src/mnl.c +1 -1
index bec4218..27b181a 100644
@@ -249,7 +249,7 @@ int mnl_batch_talk(struct mnl_socket *nl, struct list_head *err_list)
 			goto err;
 
 		ret = mnl_cb_run(rcv_buf, ret, 0, portid, NULL, NULL);
-		/* Continue on error, make sure we get all acknoledgments */
+		/* Continue on error, make sure we get all acknowledgments */
 		if (ret == -1)
 			mnl_err_list_node_add(err_list, errno, nlh->nlmsg_seq);
 
Signature · 1 line
1.8.4.rc3
#2 lore raw

PATCH 2/4 netlink: suppress useless variable

In reply to Eric Leblond
---
 src/netlink.c | 3 ---
 1 file changed, 3 deletions(-)
src/netlink.c +0 -3
index b1e44ad..6f3002b 100644
@@ -441,7 +441,6 @@ static int netlink_flush_rules(struct netlink_ctx *ctx, const struct handle *h,
 			       const struct location *loc)
 {
 	struct nft_rule_list *rule_cache;
-	struct nft_rule *nlr;
 
 	rule_cache = mnl_nft_rule_dump(nf_sock, h->family);
 	if (rule_cache == NULL)
@@ -450,9 +449,7 @@ static int netlink_flush_rules(struct netlink_ctx *ctx, const struct handle *h,
 					strerror(errno));
 
 	mnl_batch_begin();
-	nlr = alloc_nft_rule(h);
 	nft_rule_list_foreach(rule_cache, flush_rule_cb, ctx);
-	nft_rule_free(nlr);
 	nft_rule_list_free(rule_cache);
 	mnl_batch_end();
 	return 0;
Signature · 1 line
1.8.4.rc3
#3 lore raw

PATCH 3/4 netlink: only flush asked table/chain

In reply to Eric Leblond

The flush operation was not limiting the flush to the table or chain specified on command line. The result was that all the rules for a given family are flush independantly of the flush command.

---
 src/netlink.c | 8 ++++++++
 1 file changed, 8 insertions(+)
src/netlink.c +8 -0
index 6f3002b..f75cef7 100644
@@ -425,8 +425,15 @@ static int netlink_list_rules(struct netlink_ctx *ctx, const struct handle *h,
 static int flush_rule_cb(struct nft_rule *nlr, void *arg)
 {
 	struct netlink_ctx *ctx = arg;
+	const struct handle *h = ctx->data;
 	int err;
 
+	if ((h->table &&
+	    strcmp(nft_rule_attr_get_str(nlr, NFT_RULE_ATTR_TABLE), h->table) != 0) ||
+	    (h->chain &&
+	     strcmp(nft_rule_attr_get_str(nlr, NFT_RULE_ATTR_CHAIN), h->chain) != 0))
+		return 0;
+
 	netlink_dump_rule(nlr);
 	err = mnl_nft_rule_batch_del(nlr, 0, ctx->seqnum);
 	if (err < 0) {
@@ -448,6 +455,7 @@ static int netlink_flush_rules(struct netlink_ctx *ctx, const struct handle *h,
 					"Could not receive rules from kernel: %s",
 					strerror(errno));
 
+	ctx->data = h;
 	mnl_batch_begin();
 	nft_rule_list_foreach(rule_cache, flush_rule_cb, ctx);
 	nft_rule_list_free(rule_cache);
Signature · 1 line
1.8.4.rc3
#4 lore raw

PATCH 4/4 netlink: fix nft flush operation

In reply to Eric Leblond

nft_netlink function is already calling mnl_batch_end and mnl_batch_begin so it is not necessary to do it in the netlink_flush_rules function. Doing this result in a invalid netlink message which is discarded by the kernel.

---
 src/netlink.c | 2 --
 1 file changed, 2 deletions(-)
src/netlink.c +0 -2
index f75cef7..a62c357 100644
@@ -456,10 +456,8 @@ static int netlink_flush_rules(struct netlink_ctx *ctx, const struct handle *h,
 					strerror(errno));
 
 	ctx->data = h;
-	mnl_batch_begin();
 	nft_rule_list_foreach(rule_cache, flush_rule_cb, ctx);
 	nft_rule_list_free(rule_cache);
-	mnl_batch_end();
 	return 0;
 }
 
Signature · 1 line
1.8.4.rc3
#5 lore raw
In reply to Eric Leblond
Quoted 8 lines · Eric Leblond

Hello,

Here's a small patchset which fixes rule flushing. A command like:
nft flush table ip6 nat
was not flushing the rules at all. Patch 4 fixes the problem and patch 3 limit the scope of the flush which was not limited.

All applied, thanks Eric.