[PATCH ath-next] wifi: ath12k: fix error handling in creating hardware group

[Baochen Qiang <baochen.qiang@oss.qualcomm.com>]

In ath12k_core_init() when ath12k_core_hw_group_create() fails,
ath12k_core_hw_group_destroy() is called where for each device below
path would get executed

	ath12k_core_soc_destroy()
		ath12k_qmi_deinit_service()
			qmi_handle_release()

This results in kernel crash in case one of the device fails at
qmi_handle_init() when creating hardware group:

ath12k_pci 0000:10:00.0: failed to initialize qmi handle
ath12k_pci 0000:10:00.0: failed to initialize qmi :-517
ath12k_pci 0000:10:00.0: failed to create soc core: -517
ath12k_pci 0000:10:00.0: unable to create hw group
BUG: unable to handle page fault for address: ffffffffffffffb7
RIP: 0010:qmi_handle_release
Call Trace:
 <TASK>
 ath12k_qmi_deinit_service
 ath12k_core_hw_group_destroy
 ath12k_core_init
 ath12k_pci_probe

The detailed reason is, when qmi_handle_init() fails for a device
ab->qmi.handle is not correctly initialized. Then
ath12k_core_hw_group_create() returns failure, since error handing
is done for all device, eventually qmi_handle_release() is called for the
issue device and finally kernel crashes due to the uninitialized
ab->qmi.handle.

Fix this by moving error handling to ath12k_core_hw_group_create(), this
way the issue device can be skipped.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c5-00284.1-QCAHMTSWPL_V1.0_V2.0_SILICONZ-3

Fixes: 6f245ea0ec6c ("wifi: ath12k: introduce device group abstraction")
Link: https://lore.kernel.org/ath12k/fabc97122016d1a66a53ddedd965d134@posteo.net
Reported-by: a-development <a-development@posteo.de>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220518
Tested-by: a-development <a-development@posteo.de>
Signed-off-by: Baochen Qiang <baochen.qiang@oss.qualcomm.com>
---
 drivers/net/wireless/ath/ath12k/core.c | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/core.c b/drivers/net/wireless/ath/ath12k/core.c
index 5d494c5cdc0da3189640751b8d191fa939ac3ff5..a2137b363c2fea4deef724b682c1e41788777c06 100644
--- a/drivers/net/wireless/ath/ath12k/core.c
+++ b/drivers/net/wireless/ath/ath12k/core.c
@@ -2106,14 +2106,27 @@ static int ath12k_core_hw_group_create(struct ath12k_hw_group *ag)
 		ret = ath12k_core_soc_create(ab);
 		if (ret) {
 			mutex_unlock(&ab->core_lock);
-			ath12k_err(ab, "failed to create soc core: %d\n", ret);
-			return ret;
+			ath12k_err(ab, "failed to create soc %d core: %d\n", i, ret);
+			goto destroy;
 		}
 
 		mutex_unlock(&ab->core_lock);
 	}
 
 	return 0;
+
+destroy:
+	for (i--; i >= 0; i--) {
+		ab = ag->ab[i];
+		if (!ab)
+			continue;
+
+		mutex_lock(&ab->core_lock);
+		ath12k_core_soc_destroy(ab);
+		mutex_unlock(&ab->core_lock);
+	}
+
+	return ret;
 }
 
 void ath12k_core_hw_group_set_mlo_capable(struct ath12k_hw_group *ag)
@@ -2188,7 +2201,7 @@ int ath12k_core_init(struct ath12k_base *ab)
 		if (ret) {
 			mutex_unlock(&ag->mutex);
 			ath12k_warn(ab, "unable to create hw group\n");
-			goto err_destroy_hw_group;
+			goto err_unassign_hw_group;
 		}
 	}
 
@@ -2196,8 +2209,7 @@ int ath12k_core_init(struct ath12k_base *ab)
 
 	return 0;
 
-err_destroy_hw_group:
-	ath12k_core_hw_group_destroy(ab->ag);
+err_unassign_hw_group:
 	ath12k_core_hw_group_unassign(ab);
 err_unregister_notifier:
 	ath12k_core_panic_notifier_unregister(ab);

---
base-commit: 43d31f3f26f766f357e95513ba75c5126ce17d4b
change-id: 20250902-fix-hw-group-create-err-handling-58ff01c84d78

Best regards,
-- 
Baochen Qiang <baochen.qiang@oss.qualcomm.com>