Re: [PATCH net-next 09/15] batman-adv: annotate originator last_seen with READ/WRITE_ONCE

[Sven Eckelmann <sven@narfation.org>]

[-- Attachment #1: Type: text/plain, Size: 958 bytes --]

On Thursday, 12 November 2026 11:34:43 CET sashiko-bot@kernel.org wrote:
[...]
> [Severity: Medium]
> This problem wasn't introduced by this patch, but can
> batadv_v_neigh_dump_neigh() and batadv_iv_neigh_dump_neigh()

batadv_iv_neigh_dump_neigh? There is no function with this name in the source 
code. There is only batadv_iv_ogm_neigh_dump_neigh()

> suffer from
> the same timestamp underflow [1]?
> 
> net/batman-adv/bat_v.c:batadv_v_neigh_dump_neigh() {
>     ...
>         last_seen_msecs = jiffies_to_msecs(jiffies - hardif_neigh->last_seen);
>     ...
> }
> 
> Because hardif_neigh->last_seen is updated locklessly on packet reception
> and read without READ_ONCE() or jiffies_delta_to_msecs(), a concurrent
> update can leave hardif_neigh->last_seen ahead of the sampled jiffies
> value, wrapping the subtraction and exporting a huge last_seen_msecs
> value over netlink.

This is adjusted in the next two patches in this patchset.

Regards,
	Sven

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 228 bytes --]