5.10-stable review patch. If anyone has any objections, please let me know. ------------------ From: Pablo Neira Ayuso [ Upstream commit def602e498a4f951da95c95b1b8ce8ae68aa733a ] In case that the set is full, a new element gets published then removed without waiting for the RCU grace period, while RCU reader can be walking over it already. To address this issue, add the element transaction even if set is full, but toggle the set_full flag to report -ENFILE so the abort path safely unwinds the set to its previous state. As for element updates, decrement set->nelems to restore it. A simpler fix is to call synchronize_rcu() in the error path. However, with a large batch adding elements to already maxed-out set, this could cause noticeable slowdown of such batches. Fixes: 35d0ac9070ef ("netfilter: nf_tables: fix set->nelems counting with no NLM_F_EXCL") Reported-by: Inseo An Signed-off-by: Pablo Neira Ayuso Signed-off-by: Florian Westphal [ Backport to 5.10.y: applies on top of d4b7f29eb85c ("netfilter: nf_tables: always increment set element count"), as in 6.1.y. 5.10 has neither the nft_set_maxsize() helper (8d738c1869f6) nor the NFT_SET_ELEM_CATCHALL element class, so the ceiling is the same expression the prerequisite inlines, set->size ? set->size + set->ndeact : UINT_MAX, and every element is counted. On the full-set path the element is left in the transaction and unwound by the abort path, which removes it, decrements nelems and frees it only after synchronize_rcu(), instead of the in-place set->ops->remove() this replaces. The update-path hunk from the upstream commit does not apply: 5.10 has no in-place element-update path here. ] Assisted-by: Claude:claude-opus-5-5 Signed-off-by: Steffen Pfendtner Signed-off-by: Sasha Levin --- net/netfilter/nf_tables_api.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 867b5e152d677..cde499274fd37 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -5596,6 +5596,8 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, struct nft_data_desc desc; enum nft_registers dreg; struct nft_trans *trans; + unsigned int max, nelems; + bool set_full = false; u32 flags = 0; u64 timeout; u64 expiration; @@ -5806,10 +5808,15 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, goto err_elem_expr; expr = NULL; + max = set->size ? set->size + set->ndeact : UINT_MAX; + nelems = atomic_inc_return(&set->nelems); + if (nelems > max) + set_full = true; + trans = nft_trans_elem_alloc(ctx, NFT_MSG_NEWSETELEM, set); if (trans == NULL) { err = -ENOMEM; - goto err_elem_expr; + goto err_set_size; } ext->genmask = nft_genmask_cur(ctx->net); @@ -5841,20 +5848,15 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, goto err_element_clash; } - if (!atomic_add_unless(&set->nelems, 1, - set->size ? set->size + set->ndeact : UINT_MAX)) { - err = -ENFILE; - goto err_set_full; - } - nft_trans_elem(trans) = elem; nft_trans_commit_list_add_tail(ctx->net, trans); - return 0; -err_set_full: - set->ops->remove(ctx->net, set, &elem); + return set_full ? -ENFILE : 0; + err_element_clash: kfree(trans); +err_set_size: + atomic_dec(&set->nelems); err_elem_expr: nf_tables_set_elem_destroy(ctx, set, elem.priv); err_parse_data: -- 2.53.0