netfilter-devel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Phil Sutter <phil@nwl.cc>
To: Pablo Neira Ayuso <pablo@netfilter.org>
Cc: netfilter-devel@vger.kernel.org
Subject: [iptables PATCH v2 15/24] nft: Optimize flushing all chains of a table
Date: Wed, 25 Sep 2019 23:25:56 +0200	[thread overview]
Message-ID: <20190925212605.1005-16-phil@nwl.cc> (raw)
In-Reply-To: <20190925212605.1005-1-phil@nwl.cc>

Leverage nftables' support for flushing all chains of a table by
omitting NFTNL_RULE_CHAIN attribute in NFT_MSG_DELRULE payload.

The only caveat is with verbose output, as that still requires to have a
list of (existing) chains to iterate over. Apart from that, implementing
this shortcut is pretty straightforward: Don't retrieve a chain list and
just call __nft_rule_flush() directly which doesn't set above attribute
if chain name pointer is NULL.

A bigger deal is keeping rule cache consistent: Instead of just clearing
rule list for each flushed chain, flush_rule_cache() is updated to
iterate over all cached chains of the given table, clearing their rule
lists if not called for a specific chain.

While being at it, sort local variable declarations in nft_rule_flush()
from longest to shortest and drop the loop-local 'chain_name' variable
(but instead use 'chain' function parameter which is not used at that
point).

Signed-off-by: Phil Sutter <phil@nwl.cc>
---
 iptables/nft.c | 54 ++++++++++++++++++++++++++++++++++----------------
 1 file changed, 37 insertions(+), 17 deletions(-)

diff --git a/iptables/nft.c b/iptables/nft.c
index b2b15f08c1637..b50ec12ae9d42 100644
--- a/iptables/nft.c
+++ b/iptables/nft.c
@@ -837,7 +837,7 @@ int nft_init(struct nft_handle *h, const struct builtin_table *t)
 	return 0;
 }
 
-static int __flush_rule_cache(struct nftnl_rule *r, void *data)
+static int ____flush_rule_cache(struct nftnl_rule *r, void *data)
 {
 	nftnl_rule_list_del(r);
 	nftnl_rule_free(r);
@@ -845,9 +845,25 @@ static int __flush_rule_cache(struct nftnl_rule *r, void *data)
 	return 0;
 }
 
-static void flush_rule_cache(struct nftnl_chain *c)
+static int __flush_rule_cache(struct nftnl_chain *c, void *data)
 {
-	nftnl_rule_foreach(c, __flush_rule_cache, NULL);
+	return nftnl_rule_foreach(c, ____flush_rule_cache, NULL);
+}
+
+static int flush_rule_cache(struct nft_handle *h, const char *table,
+			    struct nftnl_chain *c)
+{
+	const struct builtin_table *t;
+
+	if (c)
+		return __flush_rule_cache(c, NULL);
+
+	t = nft_table_builtin_find(h, table);
+	if (!t|| !h->cache->table[t->type].chains)
+		return 0;
+
+	return nftnl_chain_list_foreach(h->cache->table[t->type].chains,
+					__flush_rule_cache, NULL);
 }
 
 static int __flush_chain_cache(struct nftnl_chain *c, void *data)
@@ -1842,7 +1858,7 @@ __nft_rule_flush(struct nft_handle *h, const char *table,
 	struct obj_update *obj;
 	struct nftnl_rule *r;
 
-	if (verbose)
+	if (verbose && chain)
 		fprintf(stdout, "Flushing chain `%s'\n", chain);
 
 	r = nftnl_rule_alloc();
@@ -1850,7 +1866,8 @@ __nft_rule_flush(struct nft_handle *h, const char *table,
 		return;
 
 	nftnl_rule_set_str(r, NFTNL_RULE_TABLE, table);
-	nftnl_rule_set_str(r, NFTNL_RULE_CHAIN, chain);
+	if (chain)
+		nftnl_rule_set_str(r, NFTNL_RULE_CHAIN, chain);
 
 	obj = batch_rule_add(h, NFT_COMPAT_RULE_FLUSH, r);
 	if (!obj) {
@@ -1864,19 +1881,21 @@ __nft_rule_flush(struct nft_handle *h, const char *table,
 int nft_rule_flush(struct nft_handle *h, const char *chain, const char *table,
 		   bool verbose)
 {
-	int ret = 0;
-	struct nftnl_chain_list *list;
 	struct nftnl_chain_list_iter *iter;
-	struct nftnl_chain *c;
+	struct nftnl_chain_list *list;
+	struct nftnl_chain *c = NULL;
+	int ret = 0;
 
 	nft_xt_builtin_init(h, table);
 
 	nft_fn = nft_rule_flush;
 
-	list = nft_chain_list_get(h, table, chain);
-	if (list == NULL) {
-		ret = 1;
-		goto err;
+	if (chain || verbose) {
+		list = nft_chain_list_get(h, table, chain);
+		if (list == NULL) {
+			ret = 1;
+			goto err;
+		}
 	}
 
 	if (chain) {
@@ -1885,9 +1904,11 @@ int nft_rule_flush(struct nft_handle *h, const char *chain, const char *table,
 			errno = ENOENT;
 			return 0;
 		}
+	}
 
+	if (chain || !verbose) {
 		__nft_rule_flush(h, table, chain, verbose, false);
-		flush_rule_cache(c);
+		flush_rule_cache(h, table, c);
 		return 1;
 	}
 
@@ -1899,11 +1920,10 @@ int nft_rule_flush(struct nft_handle *h, const char *chain, const char *table,
 
 	c = nftnl_chain_list_iter_next(iter);
 	while (c != NULL) {
-		const char *chain_name =
-			nftnl_chain_get_str(c, NFTNL_CHAIN_NAME);
+		chain = nftnl_chain_get_str(c, NFTNL_CHAIN_NAME);
 
-		__nft_rule_flush(h, table, chain_name, verbose, false);
-		flush_rule_cache(c);
+		__nft_rule_flush(h, table, chain, verbose, false);
+		flush_rule_cache(h, table, c);
 		c = nftnl_chain_list_iter_next(iter);
 	}
 	nftnl_chain_list_iter_destroy(iter);
-- 
2.23.0


  parent reply	other threads:[~2019-09-25 21:26 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-25 21:25 [iptables PATCH v2 00/24] Improve iptables-nft performance with large rulesets Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 01/24] xtables_error() does not return Phil Sutter
2019-09-25 21:31   ` Florian Westphal
2019-09-25 21:47     ` Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 02/24] tests/shell: Speed up ipt-restore/0004-restore-race_0 Phil Sutter
2019-09-26  9:07   ` Florian Westphal
2019-09-25 21:25 ` [iptables PATCH v2 03/24] tests: shell: Support running for legacy/nft only Phil Sutter
2019-09-27 14:19   ` Florian Westphal
2019-09-30 16:20     ` Pablo Neira Ayuso
2019-09-25 21:25 ` [iptables PATCH v2 04/24] nft: Fix for add and delete of same rule in single batch Phil Sutter
2019-09-27 14:20   ` Florian Westphal
2019-09-30 16:36     ` Pablo Neira Ayuso
2019-09-25 21:25 ` [iptables PATCH v2 05/24] nft: Make nftnl_table_list_get() fetch only tables Phil Sutter
2019-09-27 14:25   ` Florian Westphal
2019-09-25 21:25 ` [iptables PATCH v2 06/24] xtables-restore: Minimize caching when flushing Phil Sutter
2019-09-27 14:27   ` Florian Westphal
2019-09-25 21:25 ` [iptables PATCH v2 07/24] nft: Support fetch_rule_cache() per chain Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 08/24] nft: Fetch only chains in nft_chain_list_get() Phil Sutter
2019-09-30 16:57   ` Pablo Neira Ayuso
2019-09-30 17:12   ` Pablo Neira Ayuso
2019-09-25 21:25 ` [iptables PATCH v2 09/24] nft: Support fetch_chain_cache() per table Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 10/24] nft: Support fetch_chain_cache() per chain Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 11/24] nft: Support nft_chain_list_get() " Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 12/24] nft: Reduce cache overhead of adding a custom chain Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 13/24] nft: Reduce cache overhead of nft_chain_builtin_init() Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 14/24] nft: Support nft_is_table_compatible() per chain Phil Sutter
2019-09-25 21:25 ` Phil Sutter [this message]
2019-09-25 21:25 ` [iptables PATCH v2 16/24] xtables-restore: Introduce rule counter tokenizer function Phil Sutter
2019-09-25 21:25 ` [iptables PATCH v2 17/24] xtables-restore: Carry in_table in struct nft_xt_restore_parse Phil Sutter
2019-09-30 16:30   ` Pablo Neira Ayuso
2019-09-30 16:31     ` Pablo Neira Ayuso
2019-09-25 21:25 ` [iptables PATCH v2 18/24] xtables-restore: Use xt_params->program_name Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 19/24] xtables-restore: Carry curtable in struct nft_xt_restore_parse Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 20/24] xtables-restore: Introduce line parsing function Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 21/24] tests: shell: Add ipt-restore/0007-flush-noflush_0 Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 22/24] xtables-restore: Remove some pointless linebreaks Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 23/24] xtables-restore: Allow lines without trailing newline character Phil Sutter
2019-09-25 21:26 ` [iptables PATCH v2 24/24] xtables-restore: Improve performance of --noflush operation Phil Sutter

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=20190925212605.1005-16-phil@nwl.cc \
    --to=phil@nwl.cc \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=pablo@netfilter.org \
    /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).