[PATCH net] tcp: reject net_iov in zerocopy receive mapping hints

[PATCH net] tcp: reject net_iov in zerocopy receive mapping hints

[Daehyeon Ko]

After copying a readable prefix, receive_fallback_to_copy() asks
tcp_zerocopy_set_hint_for_skb() where page mapping can resume. If the
next skb is unreadable, find_next_mappable_frag() passes its net_iov
fragment to can_map_frag().

skb_frag_page() returns NULL for a net_iov, but can_map_frag()
dereferences it in PageCompound(). A v7.2 KASAN run on a connected TCP
socket with 64 readable bytes followed by a 4096-byte NET_IOV_DMABUF
fragment reported:

  BUG: KASAN: null-ptr-deref in can_map_frag
  tcp_zerocopy_receive -> can_map_frag
  Kernel panic - not syncing: KASAN: panic_on_warn set

The diagnostic inserted the net_iov directly because the test host has
no devmem-capable NIC. Hardware end-to-end reachability remains untested
and requires CONFIG_NET_DEVMEM plus a supported DMA-buf-bound RX queue.

Reject all net_iov fragments before skb_frag_page(). This covers both
DMABUF and IOURING net_iov types while leaving page-backed checks
unchanged. With the guard, the same queue copied the readable prefix,
returned a 4096-byte skip hint, and completed without a fault.

Fixes: 9f6b619edf2e ("net: support non paged skb frags")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Signed-off-by: Daehyeon Ko <4ncienth@gmail.com>
---
Affected since v6.12-rc1; present in v7.2 and the pinned 2026-09-30
Torvalds, net, net-next and linux-next snapshots.

The existing combined fixed v7.2 build was warning-free. No isolated
kernel build was run. Tested config SHA-256:
f992c9fdb881ca95c467896791c52554719f6204cf06268cd92ac37c14c14f83

The reproducer is available privately on request and is omitted from
this public AI-assisted report.
---
 net/ipv4/tcp.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 562752352afe..87ef6d5cbfeb 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1908,6 +1908,8 @@ static bool can_map_frag(const skb_frag_t *frag)
 
 	if (skb_frag_size(frag) != PAGE_SIZE || skb_frag_off(frag))
 		return false;
+	if (skb_frag_is_net_iov(frag))
+		return false;
 
 	page = skb_frag_page(frag);
 

base-commit: 54518e0e827f4ca9229ae657022c60bf60f5c1bf
-- 
2.55.0

Re: [PATCH net] tcp: reject net_iov in zerocopy receive mapping hints

[Mina Almasry]

On Wed, Sep 30, 2026 at 3:25 AM Daehyeon Ko <4ncienth@gmail.com> wrote:
>
> After copying a readable prefix, receive_fallback_to_copy() asks
> tcp_zerocopy_set_hint_for_skb() where page mapping can resume. If the
> next skb is unreadable, find_next_mappable_frag() passes its net_iov
> fragment to can_map_frag().
>
> skb_frag_page() returns NULL for a net_iov, but can_map_frag()
> dereferences it in PageCompound(). A v7.2 KASAN run on a connected TCP
> socket with 64 readable bytes followed by a 4096-byte NET_IOV_DMABUF
> fragment reported:
>
>   BUG: KASAN: null-ptr-deref in can_map_frag
>   tcp_zerocopy_receive -> can_map_frag
>   Kernel panic - not syncing: KASAN: panic_on_warn set
>
> The diagnostic inserted the net_iov directly because the test host has
> no devmem-capable NIC. Hardware end-to-end reachability remains untested
> and requires CONFIG_NET_DEVMEM plus a supported DMA-buf-bound RX queue.
>
> Reject all net_iov fragments before skb_frag_page(). This covers both
> DMABUF and IOURING net_iov types while leaving page-backed checks
> unchanged. With the guard, the same queue copied the readable prefix,
> returned a 4096-byte skip hint, and completed without a fault.
>
> Fixes: 9f6b619edf2e ("net: support non paged skb frags")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Signed-off-by: Daehyeon Ko <4ncienth@gmail.com>
> ---
> Affected since v6.12-rc1; present in v7.2 and the pinned 2026-09-30
> Torvalds, net, net-next and linux-next snapshots.
>
> The existing combined fixed v7.2 build was warning-free. No isolated
> kernel build was run. Tested config SHA-256:
> f992c9fdb881ca95c467896791c52554719f6204cf06268cd92ac37c14c14f83
>
> The reproducer is available privately on request and is omitted from
> this public AI-assisted report.
> ---
>  net/ipv4/tcp.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
> index 562752352afe..87ef6d5cbfeb 100644
> --- a/net/ipv4/tcp.c
> +++ b/net/ipv4/tcp.c
> @@ -1908,6 +1908,8 @@ static bool can_map_frag(const skb_frag_t *frag)
>
>         if (skb_frag_size(frag) != PAGE_SIZE || skb_frag_off(frag))
>                 return false;
> +       if (skb_frag_is_net_iov(frag))
> +               return false;

I think this should use skb_is_readable(). But you don't need this
helper for frags. I wonder if we should add a readable helper for
frags? I'm still holding out that in the future someone will add a
net_iov type that is actually kmappable and readable.

But this is good enough.

Reviewed-by: Mina Almasry <almasrymina@google.com>

-- 
Thanks,
Mina