[PATCH AUTOSEL 6.14 086/108] wifi: ath12k: make assoc link associate first

[Sasha Levin]

From: Baochen Qiang <quic_bqiang at quicinc.com>

[ Upstream commit ead6d41116b81098061c878d9bfc0b1a6c629090 ]

In MLO scenario WCN7850 firmware requests the assoc link to associate
before any other links. However currently in
ath12k_mac_op_vif_cfg_changed() we are doing association in an ascending
order of link id. If the assoc link does not get assigned the smallest
id, a non-assoc link gets associated first and firmware crashes.

Change to do association for the assoc link first.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0.c5-00481-QCAHMTSWPL_V1.0_V2.0_SILICONZ-3
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00284-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.3.1-00209-QCAHKSWPL_SILICONZ-1

Signed-off-by: Baochen Qiang <quic_bqiang at quicinc.com>
Link: https://patch.msgid.link/20250409-ath12k-wcn7850-mlo-support-v2-5-3801132ca2c3@quicinc.com
Signed-off-by: Jeff Johnson <jeff.johnson at oss.qualcomm.com>
Signed-off-by: Sasha Levin <sashal at kernel.org>
---

**YES** This commit should be backported to stable kernel trees. Here's
my detailed analysis: ## Critical Bug Fix Analysis ### The Problem The
commit fixes a **critical firmware crash** in MLO (Multi-Link Operation)
scenarios for WCN7850 hardware. The commit message clearly states: - "If
the assoc link does not get assigned the smallest id, a non-assoc link
gets associated first and **firmware crashes**" - This indicates a
severe user-affecting bug that causes complete driver/firmware failure
### Code Analysis - What's Being Fixed **Before the fix:** The original
code processed MLO links in ascending link ID order: ```c
for_each_set_bit(link_id, &links, IEEE80211_MLD_MAX_NUM_LINKS) { //
Process any link first, potentially non-assoc links ath12k_bss_assoc(ar,
arvif, info); } ``` **After the fix:** The new code ensures the
association link is processed first: ```c if (vif->cfg.assoc) { //
FIRST: Handle the assoc link specifically arvif =
wiphy_dereference(hw->wiphy, ahvif->link[ahsta->assoc_link_id]);
ath12k_bss_assoc(ar, arvif, info); // THEN: Handle remaining links,
excluding assoc link links &= ~BIT(ahsta->assoc_link_id);
for_each_set_bit(link_id, &links, IEEE80211_MLD_MAX_NUM_LINKS) {
ath12k_bss_assoc(ar, arvif, info); } } ``` ### Why This Qualifies for
Stable Backporting 1. **Fixes User-Affecting Bug**: Firmware crashes
directly impact users by causing connection failures and system
instability. 2. **Critical Hardware Support**: The fix affects WCN7850
and QCN9274 hardware which are actively used Wi-Fi 7 chips in current
devices. 3. **Small, Contained Fix**: The change is minimal - just
reordering the association sequence. It doesn't introduce new features
or architectural changes. 4. **MLO is Production Ready**: MLO (Multi-
Link Operation) is part of the Wi-Fi 7 standard and is being deployed in
production devices. 5. **Firmware Dependency**: The commit addresses a
strict firmware requirement that "requests the assoc link to associate
before any other links" - this is a hardware/firmware constraint, not a
driver preference. 6. **Low Regression Risk**: - The fix only changes
the order of operations - It maintains the same logic but ensures
correct sequencing - The exclusion of the assoc link (`links &=
~BIT(ahsta->assoc_link_id)`) prevents double-processing ### Context from
Similar Commits All the provided similar commits were marked "NO" for
backporting because they were either: - Adding new MLO
features/functionality - Implementing new capabilities - Making
architectural changes This commit is fundamentally different - it's
fixing a **firmware crash bug** in existing MLO functionality. ###
Hardware Impact The fix specifically mentions testing on: - WCN7850
hw2.0 PCI (two different firmware versions) - QCN9274 hw2.0 PCI These
are current-generation Wi-Fi 7 chipsets actively deployed in consumer
and enterprise devices. ### Conclusion This commit fixes a critical
firmware crash that affects users of Wi-Fi 7 devices with MLO
capability. The fix is small, targeted, and addresses a fundamental
firmware requirement. It meets all criteria for stable backporting:
important bug fix, minimal risk, contained change, and affects hardware
in active use.

 drivers/net/wireless/ath/ath12k/mac.c | 31 +++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

diff --git a/drivers/net/wireless/ath/ath12k/mac.c b/drivers/net/wireless/ath/ath12k/mac.c
index 5f6d9896ef613..06b2baccd1335 100644
--- a/drivers/net/wireless/ath/ath12k/mac.c
+++ b/drivers/net/wireless/ath/ath12k/mac.c
@@ -3499,6 +3499,8 @@ static void ath12k_mac_op_vif_cfg_changed(struct ieee80211_hw *hw,
 	unsigned long links = ahvif->links_map;
 	struct ieee80211_bss_conf *info;
 	struct ath12k_link_vif *arvif;
+	struct ieee80211_sta *sta;
+	struct ath12k_sta *ahsta;
 	struct ath12k *ar;
 	u8 link_id;
 
@@ -3511,6 +3513,35 @@ static void ath12k_mac_op_vif_cfg_changed(struct ieee80211_hw *hw,
 	}
 
 	if (changed & BSS_CHANGED_ASSOC) {
+		if (vif->cfg.assoc) {
+			/* only in station mode we can get here, so it's safe
+			 * to use ap_addr
+			 */
+			rcu_read_lock();
+			sta = ieee80211_find_sta(vif, vif->cfg.ap_addr);
+			if (!sta) {
+				rcu_read_unlock();
+				WARN_ONCE(1, "failed to find sta with addr %pM\n",
+					  vif->cfg.ap_addr);
+				return;
+			}
+
+			ahsta = ath12k_sta_to_ahsta(sta);
+			arvif = wiphy_dereference(hw->wiphy,
+						  ahvif->link[ahsta->assoc_link_id]);
+			rcu_read_unlock();
+
+			ar = arvif->ar;
+			/* there is no reason for which an assoc link's
+			 * bss info does not exist
+			 */
+			info = ath12k_mac_get_link_bss_conf(arvif);
+			ath12k_bss_assoc(ar, arvif, info);
+
+			/* exclude assoc link as it is done above */
+			links &= ~BIT(ahsta->assoc_link_id);
+		}
+
 		for_each_set_bit(link_id, &links, IEEE80211_MLD_MAX_NUM_LINKS) {
 			arvif = wiphy_dereference(hw->wiphy, ahvif->link[link_id]);
 			if (!arvif || !arvif->ar)
-- 
2.39.5