From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from Chamillionaire.breakpoint.cc (Chamillionaire.breakpoint.cc [91.216.245.30]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id BB0E83D9DAE; Wed, 25 Mar 2026 13:12:05 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=91.216.245.30 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774444327; cv=none; b=pn7Dbfy/S4byNyRgN+bOoK3re8csYgcXd7aXMxhN48CNKq6UjE4/Fr9Ine2Zgl4oVhc8Fd3d525CklC+0C72NxXzjXqc94DKjiuN+WvuuuXtPSTo3FJi91mjb7JK4dvZFjzdKpPNH9JgeZyhtDR6lYQUsBguFf7M05aV/AdYovU= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774444327; c=relaxed/simple; bh=ovcQQBtcsTdRkB76v3VnpXD2u7e715pTSJBCrbUkGlg=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=D0n2KA9ZzIa7iVfoTJkrTDVeASwM2hDzKVCfVQt0v4LESZFpfItY9Rjgq3oTN2MqIKEd7Dw1MUGgnNVm8CoDp2moErbUHCxaw7hiAgnCOUcbnvJXJP1M3COaEs46boTosPIQ9fPuuMHCQ2V9p7qKMN/3E44P5p7sZwEfZ8Tw8kI= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=strlen.de; spf=pass smtp.mailfrom=Chamillionaire.breakpoint.cc; arc=none smtp.client-ip=91.216.245.30 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=strlen.de Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=Chamillionaire.breakpoint.cc Received: by Chamillionaire.breakpoint.cc (Postfix, from userid 1003) id D9AC460A0F; Wed, 25 Mar 2026 14:12:03 +0100 (CET) From: Florian Westphal To: Cc: Paolo Abeni , "David S. Miller" , Eric Dumazet , Jakub Kicinski , , pablo@netfilter.org Subject: [PATCH net 07/14] netfilter: nft_set_rbtree: revisit array resize logic Date: Wed, 25 Mar 2026 14:11:01 +0100 Message-ID: <20260325131108.23045-8-fw@strlen.de> X-Mailer: git-send-email 2.52.0 In-Reply-To: <20260325131108.23045-1-fw@strlen.de> References: <20260325131108.23045-1-fw@strlen.de> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit From: Pablo Neira Ayuso Chris Arges reports high memory consumption with thousands of containers, this patch revisits the array allocation logic. For anonymous sets, start by 16 slots (which takes 256 bytes on x86_64). Expand it by x2 until threshold of 512 slots is reached, over that threshold, expand it by x1.5. For non-anonymous set, start by 1024 slots in the array (which takes 16 Kbytes initially on x86_64). Expand it by x1.5. Use set->ndeact to subtract deactivated elements when calculating the number of the slots in the array, otherwise the array size array gets increased artifically. Add special case shrink logic to deal with flush set too. The shrink logic is skipped by anonymous sets. Use check_add_overflow() to calculate the new array size. Add a WARN_ON_ONCE check to make sure elements fit into the new array size. Reported-by: Chris Arges Fixes: 7e43e0a1141d ("netfilter: nft_set_rbtree: translate rbtree to array for binary search") Signed-off-by: Pablo Neira Ayuso Signed-off-by: Florian Westphal --- net/netfilter/nft_set_rbtree.c | 92 +++++++++++++++++++++++++++------- 1 file changed, 75 insertions(+), 17 deletions(-) diff --git a/net/netfilter/nft_set_rbtree.c b/net/netfilter/nft_set_rbtree.c index fe8bd497d74a..737c339decd0 100644 --- a/net/netfilter/nft_set_rbtree.c +++ b/net/netfilter/nft_set_rbtree.c @@ -572,14 +572,12 @@ static struct nft_array *nft_array_alloc(u32 max_intervals) return array; } -#define NFT_ARRAY_EXTRA_SIZE 10240 - /* Similar to nft_rbtree_{u,k}size to hide details to userspace, but consider * packed representation coming from userspace for anonymous sets too. */ static u32 nft_array_elems(const struct nft_set *set) { - u32 nelems = atomic_read(&set->nelems); + u32 nelems = atomic_read(&set->nelems) - set->ndeact; /* Adjacent intervals are represented with a single start element in * anonymous sets, use the current element counter as is. @@ -595,27 +593,87 @@ static u32 nft_array_elems(const struct nft_set *set) return (nelems / 2) + 2; } -static int nft_array_may_resize(const struct nft_set *set) +#define NFT_ARRAY_INITIAL_SIZE 1024 +#define NFT_ARRAY_INITIAL_ANON_SIZE 16 +#define NFT_ARRAY_INITIAL_ANON_THRESH (8192U / sizeof(struct nft_array_interval)) + +static int nft_array_may_resize(const struct nft_set *set, bool flush) { - u32 nelems = nft_array_elems(set), new_max_intervals; + u32 initial_intervals, max_intervals, new_max_intervals, delta; + u32 shrinked_max_intervals, nelems = nft_array_elems(set); struct nft_rbtree *priv = nft_set_priv(set); struct nft_array *array; - if (!priv->array_next) { - array = nft_array_alloc(nelems + NFT_ARRAY_EXTRA_SIZE); - if (!array) - return -ENOMEM; + if (nft_set_is_anonymous(set)) + initial_intervals = NFT_ARRAY_INITIAL_ANON_SIZE; + else + initial_intervals = NFT_ARRAY_INITIAL_SIZE; + + if (priv->array_next) { + max_intervals = priv->array_next->max_intervals; + new_max_intervals = priv->array_next->max_intervals; + } else { + if (priv->array) { + max_intervals = priv->array->max_intervals; + new_max_intervals = priv->array->max_intervals; + } else { + max_intervals = 0; + new_max_intervals = initial_intervals; + } + } - priv->array_next = array; + if (nft_set_is_anonymous(set)) + goto maybe_grow; + + if (flush) { + /* Set flush just started, nelems still report elements.*/ + nelems = 0; + new_max_intervals = NFT_ARRAY_INITIAL_SIZE; + goto realloc_array; } - if (nelems < priv->array_next->max_intervals) - return 0; + if (check_add_overflow(new_max_intervals, new_max_intervals, + &shrinked_max_intervals)) + return -EOVERFLOW; + + shrinked_max_intervals = DIV_ROUND_UP(shrinked_max_intervals, 3); - new_max_intervals = priv->array_next->max_intervals + NFT_ARRAY_EXTRA_SIZE; - if (nft_array_intervals_alloc(priv->array_next, new_max_intervals) < 0) + if (shrinked_max_intervals > NFT_ARRAY_INITIAL_SIZE && + nelems < shrinked_max_intervals) { + new_max_intervals = shrinked_max_intervals; + goto realloc_array; + } +maybe_grow: + if (nelems > new_max_intervals) { + if (nft_set_is_anonymous(set) && + new_max_intervals < NFT_ARRAY_INITIAL_ANON_THRESH) { + new_max_intervals <<= 1; + } else { + delta = new_max_intervals >> 1; + if (check_add_overflow(new_max_intervals, delta, + &new_max_intervals)) + return -EOVERFLOW; + } + } + +realloc_array: + if (WARN_ON_ONCE(nelems > new_max_intervals)) return -ENOMEM; + if (priv->array_next) { + if (max_intervals == new_max_intervals) + return 0; + + if (nft_array_intervals_alloc(priv->array_next, new_max_intervals) < 0) + return -ENOMEM; + } else { + array = nft_array_alloc(new_max_intervals); + if (!array) + return -ENOMEM; + + priv->array_next = array; + } + return 0; } @@ -630,7 +688,7 @@ static int nft_rbtree_insert(const struct net *net, const struct nft_set *set, nft_rbtree_maybe_reset_start_cookie(priv, tstamp); - if (nft_array_may_resize(set) < 0) + if (nft_array_may_resize(set, false) < 0) return -ENOMEM; do { @@ -741,7 +799,7 @@ nft_rbtree_deactivate(const struct net *net, const struct nft_set *set, nft_rbtree_interval_null(set, this)) priv->start_rbe_cookie = 0; - if (nft_array_may_resize(set) < 0) + if (nft_array_may_resize(set, false) < 0) return NULL; while (parent != NULL) { @@ -811,7 +869,7 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx, switch (iter->type) { case NFT_ITER_UPDATE_CLONE: - if (nft_array_may_resize(set) < 0) { + if (nft_array_may_resize(set, true) < 0) { iter->err = -ENOMEM; break; } -- 2.52.0