* [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails
@ 2026-06-24 21:36 Sid Kumar
2026-07-21 9:07 ` Jiayuan Chen
0 siblings, 1 reply; 4+ messages in thread
From: Sid Kumar @ 2026-06-24 21:36 UTC (permalink / raw)
To: stable
Cc: sashal, Pablo Neira Ayuso, Vegard Nossum, Greg Kroah-Hartman,
Jakub Acs, Sid Kumar
From: Pablo Neira Ayuso <pablo@netfilter.org>
[ Upstream commit e79b47a8615d42c68aaeb68971593333667382ed ]
>From abort path, nft_mapelem_activate() needs to restore refcounters to
the original state. Currently, it uses the set->ops->walk() to iterate
over these set elements. The existing set iterator skips inactive
elements in the next generation, this does not work from the abort path
to restore the original state since it has to skip active elements
instead (not inactive ones).
This patch moves the check for inactive elements to the set iterator
callback, then it reverses the logic for the .activate case which
needs to skip active elements.
Toggle next generation bit for elements when delete set command is
invoked and call nft_clear() from .activate (abort) path to restore the
next generation bit.
The splat below shows an object in mappings memleak:
[43929.457523] ------------[ cut here ]------------
[43929.457532] WARNING: CPU: 0 PID: 1139 at include/net/netfilter/nf_tables.h:1237 nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
[...]
[43929.458014] RIP: 0010:nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
[43929.458076] Code: 83 f8 01 77 ab 49 8d 7c 24 08 e8 37 5e d0 de 49 8b 6c 24 08 48 8d 7d 50 e8 e9 5c d0 de 8b 45 50 8d 50 ff 89 55 50 85 c0 75 86 <0f> 0b eb 82 0f 0b eb b3 0f 1f 40 00 90 90 90 90 90 90 90 90 90 90
[43929.458081] RSP: 0018:ffff888140f9f4b0 EFLAGS: 00010246
[43929.458086] RAX: 0000000000000000 RBX: ffff8881434f5288 RCX: dffffc0000000000
[43929.458090] RDX: 00000000ffffffff RSI: ffffffffa26d28a7 RDI: ffff88810ecc9550
[43929.458093] RBP: ffff88810ecc9500 R08: 0000000000000001 R09: ffffed10281f3e8f
[43929.458096] R10: 0000000000000003 R11: ffff0000ffff0000 R12: ffff8881434f52a0
[43929.458100] R13: ffff888140f9f5f4 R14: ffff888151c7a800 R15: 0000000000000002
[43929.458103] FS: 00007f0c687c4740(0000) GS:ffff888390800000(0000) knlGS:0000000000000000
[43929.458107] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[43929.458111] CR2: 00007f58dbe5b008 CR3: 0000000123602005 CR4: 00000000001706f0
[43929.458114] Call Trace:
[43929.458118] <TASK>
[43929.458121] ? __warn+0x9f/0x1a0
[43929.458127] ? nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
[43929.458188] ? report_bug+0x1b1/0x1e0
[43929.458196] ? handle_bug+0x3c/0x70
[43929.458200] ? exc_invalid_op+0x17/0x40
[43929.458211] ? nft_setelem_data_deactivate+0xd7/0xf0 [nf_tables]
[43929.458271] ? nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
[43929.458332] nft_mapelem_deactivate+0x24/0x30 [nf_tables]
[43929.458392] nft_rhash_walk+0xdd/0x180 [nf_tables]
[43929.458453] ? __pfx_nft_rhash_walk+0x10/0x10 [nf_tables]
[43929.458512] ? rb_insert_color+0x2e/0x280
[43929.458520] nft_map_deactivate+0xdc/0x1e0 [nf_tables]
[43929.458582] ? __pfx_nft_map_deactivate+0x10/0x10 [nf_tables]
[43929.458642] ? __pfx_nft_mapelem_deactivate+0x10/0x10 [nf_tables]
[43929.458701] ? __rcu_read_unlock+0x46/0x70
[43929.458709] nft_delset+0xff/0x110 [nf_tables]
[43929.458769] nft_flush_table+0x16f/0x460 [nf_tables]
[43929.458830] nf_tables_deltable+0x501/0x580 [nf_tables]
(cherry picked from commit e79b47a8615d42c68aaeb68971593333667382ed)
[Vegard: CVE-2024-27012; fixed conflicts due to missing commits
0e1ea651c9717ddcd8e0648d8468477a31867b0a ("netfilter: nf_tables: shrink
memory consumption of set elements") and
9dad402b89e81a0516bad5e0ac009b7a0a80898f ("netfilter: nf_tables: expose
opaque set element as struct nft_elem_priv") so we pass the correct types
and values to nft_setelem_data_deactivate(), nft_setelem_validate(),
nft_set_elem_ext(), etc.]
[acsjakub: clean cherry-pick of the commit 164936b2fc88
("netfilter: nf_tables: restore set elements when delete set fails")
from 6.6.y. Plus, add "[ Upstream commit .." header to the message]
[sidk: clean cherry-pick of the commit faa0deee2721
("netfilter: nf_tables: restore set elements when delete set fails")
from 6.1.y.
Fixes: 628bd3e49cba ("netfilter: nf_tables: drop map element references from preparation phase")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jakub Acs <acsjakub@amazon.de>
Signed-off-by: Sid Kumar <sidkumar1@gmail.com>
---
net/netfilter/nf_tables_api.c | 41 ++++++++++++++++++++++++++++++----
net/netfilter/nft_set_bitmap.c | 4 +---
net/netfilter/nft_set_hash.c | 8 ++-----
net/netfilter/nft_set_pipapo.c | 5 +----
net/netfilter/nft_set_rbtree.c | 4 +---
5 files changed, 42 insertions(+), 20 deletions(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 623b776bf792..021a2a34b8a1 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -594,6 +594,12 @@ static int nft_mapelem_deactivate(const struct nft_ctx *ctx,
const struct nft_set_iter *iter,
struct nft_set_elem *elem)
{
+ struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
+
+ if (!nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
+ nft_set_elem_change_active(ctx->net, set, ext);
nft_setelem_data_deactivate(ctx->net, set, elem);
return 0;
@@ -619,6 +625,7 @@ static void nft_map_catchall_deactivate(const struct nft_ctx *ctx,
continue;
elem.priv = catchall->elem;
+ nft_set_elem_change_active(ctx->net, set, ext);
nft_setelem_data_deactivate(ctx->net, set, &elem);
break;
}
@@ -3522,6 +3529,9 @@ int nft_setelem_validate(const struct nft_ctx *ctx, struct nft_set *set,
const struct nft_data *data;
int err;
+ if (!nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
if (nft_set_ext_exists(ext, NFT_SET_EXT_FLAGS) &&
*nft_set_ext_flags(ext) & NFT_SET_ELEM_INTERVAL_END)
return 0;
@@ -3545,19 +3555,22 @@ int nft_setelem_validate(const struct nft_ctx *ctx, struct nft_set *set,
int nft_set_catchall_validate(const struct nft_ctx *ctx, struct nft_set *set)
{
- u8 genmask = nft_genmask_next(ctx->net);
+ struct nft_set_iter dummy_iter = {
+ .genmask = nft_genmask_next(ctx->net),
+ };
struct nft_set_elem_catchall *catchall;
struct nft_set_elem elem;
+
struct nft_set_ext *ext;
int ret = 0;
list_for_each_entry_rcu(catchall, &set->catchall_list, list) {
ext = nft_set_elem_ext(set, catchall->elem);
- if (!nft_set_elem_active(ext, genmask))
+ if (!nft_set_elem_active(ext, dummy_iter.genmask))
continue;
elem.priv = catchall->elem;
- ret = nft_setelem_validate(ctx, set, NULL, &elem);
+ ret = nft_setelem_validate(ctx, set, &dummy_iter, &elem);
if (ret < 0)
return ret;
}
@@ -5032,6 +5045,11 @@ static int nf_tables_bind_check_setelem(const struct nft_ctx *ctx,
const struct nft_set_iter *iter,
struct nft_set_elem *elem)
{
+ const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
+
+ if (!nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
return nft_setelem_data_validate(ctx, set, elem);
}
@@ -5126,6 +5144,13 @@ static int nft_mapelem_activate(const struct nft_ctx *ctx,
const struct nft_set_iter *iter,
struct nft_set_elem *elem)
{
+ struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
+
+ /* called from abort path, reverse check to undo changes. */
+ if (nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
+ nft_clear(ctx->net, ext);
nft_setelem_data_activate(ctx->net, set, elem);
return 0;
@@ -5144,6 +5169,7 @@ static void nft_map_catchall_activate(const struct nft_ctx *ctx,
if (nft_set_elem_active(ext, genmask))
continue;
+ nft_clear(ctx->net, ext);
elem.priv = catchall->elem;
nft_setelem_data_activate(ctx->net, set, &elem);
break;
@@ -5417,6 +5443,9 @@ static int nf_tables_dump_setelem(const struct nft_ctx *ctx,
const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
struct nft_set_dump_args *args;
+ if (!nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
if (nft_set_elem_expired(ext) || nft_set_elem_is_dead(ext))
return 0;
@@ -6103,7 +6132,7 @@ static void nft_setelem_activate(struct net *net, struct nft_set *set,
struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
if (nft_setelem_is_catchall(set, elem)) {
- nft_set_elem_change_active(net, set, ext);
+ nft_clear(net, ext);
} else {
set->ops->activate(net, set, elem);
}
@@ -6777,9 +6806,13 @@ static int nft_setelem_flush(const struct nft_ctx *ctx,
const struct nft_set_iter *iter,
struct nft_set_elem *elem)
{
+ const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
struct nft_trans *trans;
int err;
+ if (!nft_set_elem_active(ext, iter->genmask))
+ return 0;
+
trans = nft_trans_alloc_gfp(ctx, NFT_MSG_DELSETELEM,
sizeof(struct nft_trans_elem), GFP_ATOMIC);
if (!trans)
diff --git a/net/netfilter/nft_set_bitmap.c b/net/netfilter/nft_set_bitmap.c
index 60122539fee6..028955a9cb09 100644
--- a/net/netfilter/nft_set_bitmap.c
+++ b/net/netfilter/nft_set_bitmap.c
@@ -171,7 +171,7 @@ static void nft_bitmap_activate(const struct net *net,
nft_bitmap_location(set, nft_set_ext_key(&be->ext), &idx, &off);
/* Enter 11 state. */
priv->bitmap[idx] |= (genmask << off);
- nft_set_elem_change_active(net, set, &be->ext);
+ nft_clear(net, &be->ext);
}
static bool nft_bitmap_flush(const struct net *net,
@@ -223,8 +223,6 @@ static void nft_bitmap_walk(const struct nft_ctx *ctx,
list_for_each_entry_rcu(be, &priv->list, head) {
if (iter->count < iter->skip)
goto cont;
- if (!nft_set_elem_active(&be->ext, iter->genmask))
- goto cont;
elem.priv = be;
diff --git a/net/netfilter/nft_set_hash.c b/net/netfilter/nft_set_hash.c
index 9ea4a0990318..5a74ee4b7dfb 100644
--- a/net/netfilter/nft_set_hash.c
+++ b/net/netfilter/nft_set_hash.c
@@ -196,7 +196,7 @@ static void nft_rhash_activate(const struct net *net, const struct nft_set *set,
{
struct nft_rhash_elem *he = elem->priv;
- nft_set_elem_change_active(net, set, &he->ext);
+ nft_clear(net, &he->ext);
}
static bool nft_rhash_flush(const struct net *net,
@@ -285,8 +285,6 @@ static void nft_rhash_walk(const struct nft_ctx *ctx, struct nft_set *set,
if (iter->count < iter->skip)
goto cont;
- if (!nft_set_elem_active(&he->ext, iter->genmask))
- goto cont;
elem.priv = he;
@@ -615,7 +613,7 @@ static void nft_hash_activate(const struct net *net, const struct nft_set *set,
{
struct nft_hash_elem *he = elem->priv;
- nft_set_elem_change_active(net, set, &he->ext);
+ nft_clear(net, &he->ext);
}
static bool nft_hash_flush(const struct net *net,
@@ -669,8 +667,6 @@ static void nft_hash_walk(const struct nft_ctx *ctx, struct nft_set *set,
hlist_for_each_entry_rcu(he, &priv->table[i], node) {
if (iter->count < iter->skip)
goto cont;
- if (!nft_set_elem_active(&he->ext, iter->genmask))
- goto cont;
elem.priv = he;
diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c
index 4bede370c8b8..a965444ce5c5 100644
--- a/net/netfilter/nft_set_pipapo.c
+++ b/net/netfilter/nft_set_pipapo.c
@@ -1819,7 +1819,7 @@ static void nft_pipapo_activate(const struct net *net,
{
struct nft_pipapo_elem *e = elem->priv;
- nft_set_elem_change_active(net, set, &e->ext);
+ nft_clear(net, &e->ext);
}
/**
@@ -2121,9 +2121,6 @@ static void nft_pipapo_walk(const struct nft_ctx *ctx, struct nft_set *set,
e = f->mt[r].e;
- if (!nft_set_elem_active(&e->ext, iter->genmask))
- goto cont;
-
elem.priv = e;
iter->err = iter->fn(ctx, set, iter, &elem);
diff --git a/net/netfilter/nft_set_rbtree.c b/net/netfilter/nft_set_rbtree.c
index 426becaad1b9..23e4e656f7f0 100644
--- a/net/netfilter/nft_set_rbtree.c
+++ b/net/netfilter/nft_set_rbtree.c
@@ -548,7 +548,7 @@ static void nft_rbtree_activate(const struct net *net,
{
struct nft_rbtree_elem *rbe = elem->priv;
- nft_set_elem_change_active(net, set, &rbe->ext);
+ nft_clear(net, &rbe->ext);
}
static bool nft_rbtree_flush(const struct net *net,
@@ -618,8 +618,6 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx,
if (iter->count < iter->skip)
goto cont;
- if (!nft_set_elem_active(&rbe->ext, iter->genmask))
- goto cont;
elem.priv = rbe;
--
2.39.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails
2026-06-24 21:36 [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails Sid Kumar
@ 2026-07-21 9:07 ` Jiayuan Chen
2026-07-21 12:08 ` Sasha Levin
0 siblings, 1 reply; 4+ messages in thread
From: Jiayuan Chen @ 2026-07-21 9:07 UTC (permalink / raw)
To: Sid Kumar, stable, Pablo Neira Ayuso
Cc: sashal, Pablo Neira Ayuso, Vegard Nossum, Greg Kroah-Hartman,
Jakub Acs
On 6/25/26 5:36 AM, Sid Kumar wrote:
> From: Pablo Neira Ayuso <pablo@netfilter.org>
>
> [ Upstream commit e79b47a8615d42c68aaeb68971593333667382ed ]
>
> >From abort path, nft_mapelem_activate() needs to restore refcounters to
> the original state. Currently, it uses the set->ops->walk() to iterate
> over these set elements. The existing set iterator skips inactive
> elements in the next generation, this does not work from the abort path
> to restore the original state since it has to skip active elements
> instead (not inactive ones).
>
> This patch moves the check for inactive elements to the set iterator
> callback, then it reverses the logic for the .activate case which
> needs to skip active elements.
>
> Toggle next generation bit for elements when delete set command is
> invoked and call nft_clear() from .activate (abort) path to restore the
> next generation bit.
>
> The splat below shows an object in mappings memleak:
>
> [43929.457523] ------------[ cut here ]------------
> [43929.457532] WARNING: CPU: 0 PID: 1139 at include/net/netfilter/nf_tables.h:1237 nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
> [...]
> [43929.458014] RIP: 0010:nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
> [43929.458076] Code: 83 f8 01 77 ab 49 8d 7c 24 08 e8 37 5e d0 de 49 8b 6c 24 08 48 8d 7d 50 e8 e9 5c d0 de 8b 45 50 8d 50 ff 89 55 50 85 c0 75 86 <0f> 0b eb 82 0f 0b eb b3 0f 1f 40 00 90 90 90 90 90 90 90 90 90 90
> [43929.458081] RSP: 0018:ffff888140f9f4b0 EFLAGS: 00010246
> [43929.458086] RAX: 0000000000000000 RBX: ffff8881434f5288 RCX: dffffc0000000000
> [43929.458090] RDX: 00000000ffffffff RSI: ffffffffa26d28a7 RDI: ffff88810ecc9550
> [43929.458093] RBP: ffff88810ecc9500 R08: 0000000000000001 R09: ffffed10281f3e8f
> [43929.458096] R10: 0000000000000003 R11: ffff0000ffff0000 R12: ffff8881434f52a0
> [43929.458100] R13: ffff888140f9f5f4 R14: ffff888151c7a800 R15: 0000000000000002
> [43929.458103] FS: 00007f0c687c4740(0000) GS:ffff888390800000(0000) knlGS:0000000000000000
> [43929.458107] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [43929.458111] CR2: 00007f58dbe5b008 CR3: 0000000123602005 CR4: 00000000001706f0
> [43929.458114] Call Trace:
> [43929.458118] <TASK>
> [43929.458121] ? __warn+0x9f/0x1a0
> [43929.458127] ? nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
> [43929.458188] ? report_bug+0x1b1/0x1e0
> [43929.458196] ? handle_bug+0x3c/0x70
> [43929.458200] ? exc_invalid_op+0x17/0x40
> [43929.458211] ? nft_setelem_data_deactivate+0xd7/0xf0 [nf_tables]
> [43929.458271] ? nft_setelem_data_deactivate+0xe4/0xf0 [nf_tables]
> [43929.458332] nft_mapelem_deactivate+0x24/0x30 [nf_tables]
> [43929.458392] nft_rhash_walk+0xdd/0x180 [nf_tables]
> [43929.458453] ? __pfx_nft_rhash_walk+0x10/0x10 [nf_tables]
> [43929.458512] ? rb_insert_color+0x2e/0x280
> [43929.458520] nft_map_deactivate+0xdc/0x1e0 [nf_tables]
> [43929.458582] ? __pfx_nft_map_deactivate+0x10/0x10 [nf_tables]
> [43929.458642] ? __pfx_nft_mapelem_deactivate+0x10/0x10 [nf_tables]
> [43929.458701] ? __rcu_read_unlock+0x46/0x70
> [43929.458709] nft_delset+0xff/0x110 [nf_tables]
> [43929.458769] nft_flush_table+0x16f/0x460 [nf_tables]
> [43929.458830] nf_tables_deltable+0x501/0x580 [nf_tables]
>
> (cherry picked from commit e79b47a8615d42c68aaeb68971593333667382ed)
> [Vegard: CVE-2024-27012; fixed conflicts due to missing commits
> 0e1ea651c9717ddcd8e0648d8468477a31867b0a ("netfilter: nf_tables: shrink
> memory consumption of set elements") and
> 9dad402b89e81a0516bad5e0ac009b7a0a80898f ("netfilter: nf_tables: expose
> opaque set element as struct nft_elem_priv") so we pass the correct types
> and values to nft_setelem_data_deactivate(), nft_setelem_validate(),
> nft_set_elem_ext(), etc.]
> [acsjakub: clean cherry-pick of the commit 164936b2fc88
> ("netfilter: nf_tables: restore set elements when delete set fails")
> from 6.6.y. Plus, add "[ Upstream commit .." header to the message]
> [sidk: clean cherry-pick of the commit faa0deee2721
> ("netfilter: nf_tables: restore set elements when delete set fails")
> from 6.1.y.
>
> Fixes: 628bd3e49cba ("netfilter: nf_tables: drop map element references from preparation phase")
> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
> Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> Signed-off-by: Jakub Acs <acsjakub@amazon.de>
> Signed-off-by: Sid Kumar <sidkumar1@gmail.com>
Is this patch associated with cve-2026-23111
(https://fuzzinglabs.com/repro-cve-2026-23111/) ?
I'm verifying cve-2026-23111 in latest 5.15 stable, but I can still get
garbled text by running 'nft list maps' like this:
table inet mytable {
map triggermap {
type ipv4_addr : verdict
elements = { * : goto �,� �����*� ���� } <== garbled
text, uaf
}
}
Does that means cve-2026-23111 is still exist in latest 5.15 ?
>
> ---
> net/netfilter/nf_tables_api.c | 41 ++++++++++++++++++++++++++++++----
> net/netfilter/nft_set_bitmap.c | 4 +---
> net/netfilter/nft_set_hash.c | 8 ++-----
> net/netfilter/nft_set_pipapo.c | 5 +----
> net/netfilter/nft_set_rbtree.c | 4 +---
> 5 files changed, 42 insertions(+), 20 deletions(-)
>
> diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
> index 623b776bf792..021a2a34b8a1 100644
> --- a/net/netfilter/nf_tables_api.c
> +++ b/net/netfilter/nf_tables_api.c
> @@ -594,6 +594,12 @@ static int nft_mapelem_deactivate(const struct nft_ctx *ctx,
> const struct nft_set_iter *iter,
> struct nft_set_elem *elem)
> {
> + struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
> +
> + if (!nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> + nft_set_elem_change_active(ctx->net, set, ext);
> nft_setelem_data_deactivate(ctx->net, set, elem);
>
> return 0;
> @@ -619,6 +625,7 @@ static void nft_map_catchall_deactivate(const struct nft_ctx *ctx,
> continue;
>
> elem.priv = catchall->elem;
> + nft_set_elem_change_active(ctx->net, set, ext);
> nft_setelem_data_deactivate(ctx->net, set, &elem);
> break;
> }
> @@ -3522,6 +3529,9 @@ int nft_setelem_validate(const struct nft_ctx *ctx, struct nft_set *set,
> const struct nft_data *data;
> int err;
>
> + if (!nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> if (nft_set_ext_exists(ext, NFT_SET_EXT_FLAGS) &&
> *nft_set_ext_flags(ext) & NFT_SET_ELEM_INTERVAL_END)
> return 0;
> @@ -3545,19 +3555,22 @@ int nft_setelem_validate(const struct nft_ctx *ctx, struct nft_set *set,
>
> int nft_set_catchall_validate(const struct nft_ctx *ctx, struct nft_set *set)
> {
> - u8 genmask = nft_genmask_next(ctx->net);
> + struct nft_set_iter dummy_iter = {
> + .genmask = nft_genmask_next(ctx->net),
> + };
> struct nft_set_elem_catchall *catchall;
> struct nft_set_elem elem;
> +
> struct nft_set_ext *ext;
> int ret = 0;
>
> list_for_each_entry_rcu(catchall, &set->catchall_list, list) {
> ext = nft_set_elem_ext(set, catchall->elem);
> - if (!nft_set_elem_active(ext, genmask))
> + if (!nft_set_elem_active(ext, dummy_iter.genmask))
> continue;
>
> elem.priv = catchall->elem;
> - ret = nft_setelem_validate(ctx, set, NULL, &elem);
> + ret = nft_setelem_validate(ctx, set, &dummy_iter, &elem);
> if (ret < 0)
> return ret;
> }
> @@ -5032,6 +5045,11 @@ static int nf_tables_bind_check_setelem(const struct nft_ctx *ctx,
> const struct nft_set_iter *iter,
> struct nft_set_elem *elem)
> {
> + const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
> +
> + if (!nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> return nft_setelem_data_validate(ctx, set, elem);
> }
>
> @@ -5126,6 +5144,13 @@ static int nft_mapelem_activate(const struct nft_ctx *ctx,
> const struct nft_set_iter *iter,
> struct nft_set_elem *elem)
> {
> + struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
> +
> + /* called from abort path, reverse check to undo changes. */
> + if (nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> + nft_clear(ctx->net, ext);
> nft_setelem_data_activate(ctx->net, set, elem);
>
> return 0;
> @@ -5144,6 +5169,7 @@ static void nft_map_catchall_activate(const struct nft_ctx *ctx,
> if (nft_set_elem_active(ext, genmask))
> continue;
>
> + nft_clear(ctx->net, ext);
> elem.priv = catchall->elem;
> nft_setelem_data_activate(ctx->net, set, &elem);
> break;
> @@ -5417,6 +5443,9 @@ static int nf_tables_dump_setelem(const struct nft_ctx *ctx,
> const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
> struct nft_set_dump_args *args;
>
> + if (!nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> if (nft_set_elem_expired(ext) || nft_set_elem_is_dead(ext))
> return 0;
>
> @@ -6103,7 +6132,7 @@ static void nft_setelem_activate(struct net *net, struct nft_set *set,
> struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
>
> if (nft_setelem_is_catchall(set, elem)) {
> - nft_set_elem_change_active(net, set, ext);
> + nft_clear(net, ext);
> } else {
> set->ops->activate(net, set, elem);
> }
> @@ -6777,9 +6806,13 @@ static int nft_setelem_flush(const struct nft_ctx *ctx,
> const struct nft_set_iter *iter,
> struct nft_set_elem *elem)
> {
> + const struct nft_set_ext *ext = nft_set_elem_ext(set, elem->priv);
> struct nft_trans *trans;
> int err;
>
> + if (!nft_set_elem_active(ext, iter->genmask))
> + return 0;
> +
> trans = nft_trans_alloc_gfp(ctx, NFT_MSG_DELSETELEM,
> sizeof(struct nft_trans_elem), GFP_ATOMIC);
> if (!trans)
> diff --git a/net/netfilter/nft_set_bitmap.c b/net/netfilter/nft_set_bitmap.c
> index 60122539fee6..028955a9cb09 100644
> --- a/net/netfilter/nft_set_bitmap.c
> +++ b/net/netfilter/nft_set_bitmap.c
> @@ -171,7 +171,7 @@ static void nft_bitmap_activate(const struct net *net,
> nft_bitmap_location(set, nft_set_ext_key(&be->ext), &idx, &off);
> /* Enter 11 state. */
> priv->bitmap[idx] |= (genmask << off);
> - nft_set_elem_change_active(net, set, &be->ext);
> + nft_clear(net, &be->ext);
> }
>
> static bool nft_bitmap_flush(const struct net *net,
> @@ -223,8 +223,6 @@ static void nft_bitmap_walk(const struct nft_ctx *ctx,
> list_for_each_entry_rcu(be, &priv->list, head) {
> if (iter->count < iter->skip)
> goto cont;
> - if (!nft_set_elem_active(&be->ext, iter->genmask))
> - goto cont;
>
> elem.priv = be;
>
> diff --git a/net/netfilter/nft_set_hash.c b/net/netfilter/nft_set_hash.c
> index 9ea4a0990318..5a74ee4b7dfb 100644
> --- a/net/netfilter/nft_set_hash.c
> +++ b/net/netfilter/nft_set_hash.c
> @@ -196,7 +196,7 @@ static void nft_rhash_activate(const struct net *net, const struct nft_set *set,
> {
> struct nft_rhash_elem *he = elem->priv;
>
> - nft_set_elem_change_active(net, set, &he->ext);
> + nft_clear(net, &he->ext);
> }
>
> static bool nft_rhash_flush(const struct net *net,
> @@ -285,8 +285,6 @@ static void nft_rhash_walk(const struct nft_ctx *ctx, struct nft_set *set,
>
> if (iter->count < iter->skip)
> goto cont;
> - if (!nft_set_elem_active(&he->ext, iter->genmask))
> - goto cont;
>
> elem.priv = he;
>
> @@ -615,7 +613,7 @@ static void nft_hash_activate(const struct net *net, const struct nft_set *set,
> {
> struct nft_hash_elem *he = elem->priv;
>
> - nft_set_elem_change_active(net, set, &he->ext);
> + nft_clear(net, &he->ext);
> }
>
> static bool nft_hash_flush(const struct net *net,
> @@ -669,8 +667,6 @@ static void nft_hash_walk(const struct nft_ctx *ctx, struct nft_set *set,
> hlist_for_each_entry_rcu(he, &priv->table[i], node) {
> if (iter->count < iter->skip)
> goto cont;
> - if (!nft_set_elem_active(&he->ext, iter->genmask))
> - goto cont;
>
> elem.priv = he;
>
> diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c
> index 4bede370c8b8..a965444ce5c5 100644
> --- a/net/netfilter/nft_set_pipapo.c
> +++ b/net/netfilter/nft_set_pipapo.c
> @@ -1819,7 +1819,7 @@ static void nft_pipapo_activate(const struct net *net,
> {
> struct nft_pipapo_elem *e = elem->priv;
>
> - nft_set_elem_change_active(net, set, &e->ext);
> + nft_clear(net, &e->ext);
> }
>
> /**
> @@ -2121,9 +2121,6 @@ static void nft_pipapo_walk(const struct nft_ctx *ctx, struct nft_set *set,
>
> e = f->mt[r].e;
>
> - if (!nft_set_elem_active(&e->ext, iter->genmask))
> - goto cont;
> -
> elem.priv = e;
>
> iter->err = iter->fn(ctx, set, iter, &elem);
> diff --git a/net/netfilter/nft_set_rbtree.c b/net/netfilter/nft_set_rbtree.c
> index 426becaad1b9..23e4e656f7f0 100644
> --- a/net/netfilter/nft_set_rbtree.c
> +++ b/net/netfilter/nft_set_rbtree.c
> @@ -548,7 +548,7 @@ static void nft_rbtree_activate(const struct net *net,
> {
> struct nft_rbtree_elem *rbe = elem->priv;
>
> - nft_set_elem_change_active(net, set, &rbe->ext);
> + nft_clear(net, &rbe->ext);
> }
>
> static bool nft_rbtree_flush(const struct net *net,
> @@ -618,8 +618,6 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx,
>
> if (iter->count < iter->skip)
> goto cont;
> - if (!nft_set_elem_active(&rbe->ext, iter->genmask))
> - goto cont;
>
> elem.priv = rbe;
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails
2026-07-21 9:07 ` Jiayuan Chen
@ 2026-07-21 12:08 ` Sasha Levin
2026-07-21 12:31 ` Jiayuan Chen
0 siblings, 1 reply; 4+ messages in thread
From: Sasha Levin @ 2026-07-21 12:08 UTC (permalink / raw)
To: Sid Kumar, stable, Pablo Neira Ayuso
Cc: Sasha Levin, Vegard Nossum, Greg Kroah-Hartman, Jakub Acs,
Jiayuan Chen
On Tue, Jul 21, 2026 at 05:07:00PM +0800, Jiayuan Chen wrote:
>Is this patch associated with cve-2026-23111
>(https://fuzzinglabs.com/repro-cve-2026-23111/) ?
No - this patch is the 5.15 backport of the CVE-2024-27012 fix
(upstream e79b47a8615d), a different vulnerability. CVE-2026-23111 is
the inverted genmask check in nft_map_catchall_activate(), fixed
upstream by f41c5d151078 ("netfilter: nf_tables: fix inverted genmask
check in nft_map_catchall_activate()"), and that fix is already in
linux-5.15.y as commit 8c760ba4e36c (released back in February).
>I'm verifying cve-2026-23111 in latest 5.15 stable, but I can still get
>garbled text by running 'nft list maps' like this:
[...]
>Does that means cve-2026-23111 is still exist in latest 5.15 ?
I don't believe so. What you're most likely reproducing is
CVE-2024-27012 itself: released 5.15.y still lacks this patch, so
nft_map_catchall_activate() doesn't restore catchall map elements on
the abort path (there's no nft_clear() there), which leaves a freed
element visible to 'nft list maps'. That's exactly the bug this
backport fixes, and it doesn't reintroduce the inverted check - the
nft_set_elem_active() test it builds on is the corrected one.
So this patch stays queued for 5.15. Could you retest with it applied
and confirm the UAF is gone?
--
Thanks,
Sasha
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails
2026-07-21 12:08 ` Sasha Levin
@ 2026-07-21 12:31 ` Jiayuan Chen
0 siblings, 0 replies; 4+ messages in thread
From: Jiayuan Chen @ 2026-07-21 12:31 UTC (permalink / raw)
To: Sasha Levin, Sid Kumar, stable, Pablo Neira Ayuso
Cc: Vegard Nossum, Greg Kroah-Hartman, Jakub Acs
On 7/21/26 8:08 PM, Sasha Levin wrote:
> On Tue, Jul 21, 2026 at 05:07:00PM +0800, Jiayuan Chen wrote:
>> Is this patch associated with cve-2026-23111
>> (https://fuzzinglabs.com/repro-cve-2026-23111/) ?
> No - this patch is the 5.15 backport of the CVE-2024-27012 fix
> (upstream e79b47a8615d), a different vulnerability. CVE-2026-23111 is
> the inverted genmask check in nft_map_catchall_activate(), fixed
> upstream by f41c5d151078 ("netfilter: nf_tables: fix inverted genmask
> check in nft_map_catchall_activate()"), and that fix is already in
> linux-5.15.y as commit 8c760ba4e36c (released back in February).
>
>> I'm verifying cve-2026-23111 in latest 5.15 stable, but I can still get
>> garbled text by running 'nft list maps' like this:
> [...]
>> Does that means cve-2026-23111 is still exist in latest 5.15 ?
> I don't believe so. What you're most likely reproducing is
> CVE-2024-27012 itself: released 5.15.y still lacks this patch, so
> nft_map_catchall_activate() doesn't restore catchall map elements on
> the abort path (there's no nft_clear() there), which leaves a freed
> element visible to 'nft list maps'. That's exactly the bug this
> backport fixes, and it doesn't reintroduce the inverted check - the
> nft_set_elem_active() test it builds on is the corrected one.
>
> So this patch stays queued for 5.15. Could you retest with it applied
> and confirm the UAF is gone?
Yes, after applying this thread, the UAF disappeared.
Operation 'nft delete chain inet mytable mychain' will rejected just
like 6.6 did (I'm also testing 6.6 stable).
~# FILE_BASH="file_bash"
~# nft flush ruleset
~# rm -f "$FILE_BASH"
~# nft add table inet mytable
~# nft add chain inet mytable mychain
~# nft add map inet mytable mymap { type ipv4_addr : verdict \; }
~# nft add element inet mytable mymap { \* : goto mychain }
~# nft add map inet mytable triggermap { type ipv4_addr : verdict \; }
~# nft add element inet mytable triggermap { \* : goto mychain }
~# cat > "$FILE_BASH" << 'BATCH_CONTENT'
delete map inet mytable triggermap
delete map inet mytable bonjour
BATCH_CONTENT
~# nft -f "$FILE_BASH"
file_bash:2:25-31: Error: Could not process rule: No such file or directory
delete map inet mytable bonjour
^^^^^^^
~# nft delete map inet mytable mymap
~# nft delete chain inet mytable mychain
Error: Could not process rule: Device or resource busy
delete chain inet mytable mychain
^^^^^^^
~# nft list chains
table inet mytable {
chain mychain {
}
}
~# nft list maps
table inet mytable {
map triggermap {
type ipv4_addr : verdict
elements = { * : goto mychain }
}
}
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2026-07-21 12:32 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-06-24 21:36 [PATCH 5.15.y] netfilter: nf_tables: restore set elements when delete set fails Sid Kumar
2026-07-21 9:07 ` Jiayuan Chen
2026-07-21 12:08 ` Sasha Levin
2026-07-21 12:31 ` Jiayuan Chen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox