mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] ipconfig: fix use-after-free of device list after ic_close_devs()
@ 2026-10-09  7:45 Henry Martin
  2026-10-09  8:04 ` Eric Dumazet
  0 siblings, 1 reply; 2+ messages in thread
From: Henry Martin @ 2026-10-09  7:45 UTC (permalink / raw)
  To: David Ahern, Ido Schimmel, David S . Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Simon Horman
  Cc: netdev, linux-kernel, Henry Martin, stable

ic_close_devs() kfree()s every ic_device node but leaves the static
ic_first_dev and ic_dev pointers dangling behind.  When a DHCP reply
consists of an OFFER that is never followed by an ACK (a misbehaving
or lossy DHCP server, or a network that drops the ACK), the boot-time
autoconfiguration loop becomes vulnerable:

  1. ic_bootp_recv() accepts the OFFER and records ic_dev = <list
     node> ("We have a winner!" state);
  2. the retransmit round times out in ic_dynamic(), so
     ic_close_devs() frees the whole list — node included;
  3. the dynamic retry path (try_try_again) runs ic_open_devs() and
     ic_dynamic() again; on failure it falls through ic_close_devs()
     once more, where the entry point evaluates

         struct net_device *selected_dev = ic_dev ? ic_dev->dev : NULL;

     — an 8-byte use-after-free read against the freed ic_device.

KASAN confirms (boot-time bootload, no userspace yet):

  BUG: KASAN: slab-use-after-free in ic_close_devs+0x214/0x220
  Read of size 8 by task swapper/0/1
   Call Trace:
    ic_close_devs+0x214/0x220
    ip_auto_config+0x172b/0x34e0
    do_one_initcall+0x9f/0x370
    kernel_init_freeable+0x432/0x760
    kernel_init+0x24/0x1f0
  Allocated by task 1:
    __kmalloc_cache_noprof+0x1b5/0x430 <- ip_auto_config+0x587/0x34e0
    (ic_open_devs' kmalloc_obj of the ic_device list node)
  Freed by task 1:
    kfree+0x166/0x450 <- ic_close_devs+0xcb/0x220
    (ic_close_devs freeing the same node at round end)

NULL both pointers after the free loop so subsequent retry rounds
take the clean empty-list path instead of acting on freed entries.

This vulnerability was discovered by Tencent CodeBuddy Security.

Cc: stable@vger.kernel.org
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Henry Martin <bsdhenrymartin@gmail.com>
---
 net/ipv4/ipconfig.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index 1b8585404a41..ac4851e59c3b 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -346,6 +346,13 @@ static void __init ic_close_devs(void)
 		kfree(d);
 	}
 	rtnl_unlock();
+
+	/* The whole ic_device list was just kfree()'d above; clear the
+	 * static pointers so retry rounds (e.g. dynamic retry after an
+	 * OFFER without ACK) don't dereference them (use-after-free).
+	 */
+	ic_first_dev = NULL;
+	ic_dev = NULL;
 }

 /*
--
2.43.7


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] ipconfig: fix use-after-free of device list after ic_close_devs()
  2026-10-09  7:45 [PATCH] ipconfig: fix use-after-free of device list after ic_close_devs() Henry Martin
@ 2026-10-09  8:04 ` Eric Dumazet
  0 siblings, 0 replies; 2+ messages in thread
From: Eric Dumazet @ 2026-10-09  8:04 UTC (permalink / raw)
  To: Henry Martin
  Cc: David Ahern, Ido Schimmel, David S . Miller, Jakub Kicinski,
	Paolo Abeni, Simon Horman, netdev, linux-kernel, stable

Le ven. 9 oct. 2026 à 09:45, Henry Martin <bsdhenrymartin@gmail.com> a écrit :
>
> ic_close_devs() kfree()s every ic_device node but leaves the static
> ic_first_dev and ic_dev pointers dangling behind.  When a DHCP reply
> consists of an OFFER that is never followed by an ACK (a misbehaving
> or lossy DHCP server, or a network that drops the ACK), the boot-time
> autoconfiguration loop becomes vulnerable:
>
>   1. ic_bootp_recv() accepts the OFFER and records ic_dev = <list
>      node> ("We have a winner!" state);
>   2. the retransmit round times out in ic_dynamic(), so
>      ic_close_devs() frees the whole list — node included;
>   3. the dynamic retry path (try_try_again) runs ic_open_devs() and
>      ic_dynamic() again; on failure it falls through ic_close_devs()
>      once more, where the entry point evaluates
>
>          struct net_device *selected_dev = ic_dev ? ic_dev->dev : NULL;
>
>      — an 8-byte use-after-free read against the freed ic_device.
>
>
> NULL both pointers after the free loop so subsequent retry rounds
> take the clean empty-list path instead of acting on freed entries.
>
> This vulnerability was discovered by Tencent CodeBuddy Security.

This is __init code, on a boot that already failed to get a lease.
I would not call it a vulnerability.

>
> Cc: stable@vger.kernel.org
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")

Wrong tag. In 2.6.12 ic_dev was a 'struct net_device *' that was only
compared, never dereferenced. The dereference came in v5.12 with:

Fixes: 46acf7bdbc72 ("Revert "net: ipv4: handle DSA enabled master
network devices"")

> Signed-off-by: Henry Martin <bsdhenrymartin@gmail.com>
> ---
>  net/ipv4/ipconfig.c | 7 +++++++
>  1 file changed, 7 insertions(+)
>
> diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
> index 1b8585404a41..ac4851e59c3b 100644
> --- a/net/ipv4/ipconfig.c
> +++ b/net/ipv4/ipconfig.c
> @@ -346,6 +346,13 @@ static void __init ic_close_devs(void)
>                 kfree(d);
>         }
>         rtnl_unlock();
> +
> +       /* The whole ic_device list was just kfree()'d above; clear the
> +        * static pointers so retry rounds (e.g. dynamic retry after an
> +        * OFFER without ACK) don't dereference them (use-after-free).
> +        */
> +       ic_first_dev = NULL;
> +       ic_dev = NULL;
>  }

The change looks correct, but the comment is not needed.

Please send a V2 (in more than 24 hours) targeting the net tree
([PATCH net v2]).

Thanks.

pw-bot: cr

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-10-09  8:04 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-10-09  7:45 [PATCH] ipconfig: fix use-after-free of device list after ic_close_devs() Henry Martin
2026-10-09  8:04 ` Eric Dumazet

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®