[PATCH v2] batman-adv: Close OGM aggregation before transmission

[Chengfeng Ye]

The OGM send worker leaves its forwarding packet on forw_bat_list until
after batadv_iv_ogm_emit() returns. Aggregation holds forw_bat_list_lock,
but emission reads and clones the packet without that lock.

CPU 0 can therefore start emitting a queued packet while CPU 1 takes the
list lock, finds the same packet and appends another OGM. The sender can
observe the new packet length before the corresponding direct-link flag
is set, or clone the skb while its length and payload are being updated.
This can transmit an OGM with incorrect flags or inconsistent data.

KCSAN reported:

  BUG: KCSAN: data-race in batadv_iv_ogm_queue_add / batadv_iv_send_outstanding_bat_ogm_packet
  write to 0xffff888100fedcf8 of 2 bytes by interrupt on cpu 1:
  read to 0xffff888100fedcf8 of 2 bytes by task 70 on cpu 2:
  value changed: 0x00c0 -> 0x00d8

Mark the aggregate as full under forw_bat_list_lock before emission. This
waits for any ongoing append and prevents further aggregation. Emission
uses packet_len to walk the OGMs, so all queued packets are still sent.
Keep the packet on the list so interface purging can find the worker,
wait for it to finish and free the packet.

Fixes: 9b4aec647a92 ("batman-adv: fix rare race conditions on interface removal")
Assisted-by: GPT-6-Astra
Signed-off-by: Chengfeng Ye <nicoyip.dev@xxxxxxxxx>
---
Changes in v2:
- Rebase on batadv/net and route the patch to the batman-adv maintainers
  and mailing list, including the introducing author.
- Add a comment marking the aggregate as full before emission.
- Drop the stable-backport request and defer backport consideration
  pending maintainer assessment of practical impact.
- Add Assisted-by: GPT-6-Astra.

This revision is based on batadv.git, branch batadv/net.

The reproducer archive, batman-ogm-aggregation-race-reproducer.tar.gz,
was sent in a separate reply to this discussion. It contains the PoC,
prebuilt kernels, build scripts, configs and QEMU launcher. Reproduction
uses KCSAN and a conditional kernel-side mdelay(), capped at 50 ms, before
appending the ninth OGM. On Linux master fd179f8a05be, the vulnerable run
reported the target race; the fixed control reported none. Both runs
completed 240 cycles. These are instrumented runs.

The v2 batman-adv subsystem build and git diff --check passed. The code
change relative to v1 is the comment above the existing locking fix.
 net/batman-adv/bat_iv_ogm.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/net/batman-adv/bat_iv_ogm.c b/net/batman-adv/bat_iv_ogm.c
index d8a6a0f64ce2..084970404d3e 100644
--- a/net/batman-adv/bat_iv_ogm.c
+++ b/net/batman-adv/bat_iv_ogm.c
@@ -1909,6 +1909,11 @@ static void batadv_iv_send_outstanding_bat_ogm_packet(struct work_struct *work)
 		goto out;
 	}
 
+	/* Mark aggregate as full before forcing emit. */
+	spin_lock_bh(&bat_priv->forw_bat_list_lock);
+	forw_packet->num_packets = BATADV_MAX_AGGREGATION_PACKETS;
+	spin_unlock_bh(&bat_priv->forw_bat_list_lock);
+
 	batadv_iv_ogm_emit(forw_packet);
 
 	/* we have to have at least one packet in the queue to determine the

base-commit: abfe281aeab4d26b8e262ca9efc979a92c494aed
-- 
2.43.0