FAILED: patch "[PATCH] wifi: mac80211: capture fast-RX rate before mesh reuses" failed to apply to 6.18-stable tree

The patch below does not apply to the 6.18-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.18.y
git checkout FETCH_HEAD
git cherry-pick -x d71c841be5d9e586ee7f36c0dc8ed4db0d9a1349
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2026052854-skimmer-nutlike-eb6e@gregkh' --subject-prefix 'PATCH 6.18.y' HEAD^..

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From d71c841be5d9e586ee7f36c0dc8ed4db0d9a1349 Mon Sep 17 00:00:00 2001
From: Zhao Li <[email protected]>
Date: Sat, 9 May 2026 12:34:28 +0800
Subject: [PATCH] wifi: mac80211: capture fast-RX rate before mesh reuses
skb->cb

ieee80211_invoke_fast_rx() reads RX status through IEEE80211_SKB_RXCB(skb), which aliases the same skb->cb storage that ieee80211_rx_mesh_data() reuses as IEEE80211_TX_INFO. In the unicast forward path, mesh_data does:

	info = IEEE80211_SKB_CB(fwd_skb);
	memset(info, 0, sizeof(*info));

on the same skb the caller still names via rx->skb, then either queues the skb for TX (success) or kfree_skb()'s it (no-route) before returning RX_QUEUED. The caller's RX_QUEUED arm then calls sta_stats_encode_rate(status) on memory that is either zeroed (success path) or freed (no-route path). The latter is KASAN slab-use-after-free in ieee80211_prepare_and_rx_handle.

Fix by encoding the rate from status before invoking ieee80211_rx_mesh_data(), so the RX_QUEUED arm consumes a value captured while status was still backed by valid memory.

index d18e962126ce..3fb40449c6c5 100644
@@ -4984,6 +4984,7 @@ static bool ieee80211_invoke_fast_rx(struct ieee80211_rx_data *rx,
 		u8 sa[ETH_ALEN];
 	} addrs __aligned(2);
 	struct ieee80211_sta_rx_stats *stats;
+	u32 encoded_rate;
 
 	/* for parallel-rx, we need to have DUP_VALIDATED, otherwise we write
 	 * to a common data structure; drivers can implement that per queue

@@ -5091,11 +5092,14 @@ static bool ieee80211_invoke_fast_rx(struct ieee80211_rx_data *rx,
 	/* push the addresses in front */
 	memcpy(skb_push(skb, sizeof(addrs)), &addrs, sizeof(addrs));
 
+	/* capture before mesh forward may memset or free skb->cb */
+	encoded_rate = sta_stats_encode_rate(status);
+
 	res = ieee80211_rx_mesh_data(rx->sdata, rx->sta, rx->skb);
 	switch (res) {
 	case RX_QUEUED:
 		stats->last_rx = jiffies;
-		stats->last_rate = sta_stats_encode_rate(status);
+		stats->last_rate = encoded_rate;
 		return true;
 	case RX_CONTINUE:
 		break;

Hi Greg,

Please drop this for 6.18.y; no backport is needed there.

The failing cherry-pick is expected because the code fixed by d71c841be5d9 ("wifi: mac80211: capture fast-RX rate before mesh reuses skb->cb") is not present in 6.18.y.

In 6.18.y, the fast-RX mesh RX_QUEUED arm is:

	res = ieee80211_rx_mesh_data(rx->sdata, rx->sta, rx->skb);
	switch (res) {
	case RX_QUEUED:
		return true;

The later vulnerable post-mesh stats update:

	stats->last_rx = jiffies;
	stats->last_rate = sta_stats_encode_rate(status);

was introduced by:

cc18fffa3a517 ("wifi: mac80211: fix missing RX bitrate update for mesh forwarding path")

That code is not in 6.18.y, so d71c841be5d9 is not applicable to 6.18.y.

Thanks,
Zhao