[PATCH 00/11] wifi: ath12k: Fix endianness handling in QMI

[PATCH 00/11] wifi: ath12k: Fix endianness handling in QMI

[Alexander Wilhelm]

Fix endianness handling in QMI firmware transfer on big-endian
platforms. Without this fix, the firmware download fails due to
misinterpreted data structures exchanged between the host and the
wireless module.

The issue occurs during early bring-up on big endian systems, where QMI
messages are not correctly parsed by the driver, leading to failed
initialization sequences. Ensure all relevant fields are properly
converted between CPU and little-endian format in request and response
messages, as expected by the firmware. Attached logs showing the failure
before the fix:

    ath12k_pci 0001:01:00.0: BAR 0: assigned [mem 0xc00000000-0xc001fffff 64bit]
    ath12k_pci 0001:01:00.0: boot pci_mem 0xcd4148e9
    ath12k_pci 0001:01:00.0: pci probe 17cb:1109 17cb:1109
    ath12k_pci 0001:01:00.0: pci tcsr_soc_hw_version major 2 minor 0
    ath12k_pci 0001:01:00.0: request MSI one vector
    ath12k_pci 0001:01:00.0: MSI vectors: 1
    ath12k_pci 0001:01:00.0: msi base data is 0
    ath12k_pci 0001:01:00.0: Hardware name: qcn9274 hw2.0
    ath12k_pci 0001:01:00.0: boot firmware request ath12k/QCN9274/hw2.0/firmware-2.bin size 15134776
    ath12k_pci 0001:01:00.0: found fw timestamp 1722934582
    ath12k_pci 0001:01:00.0: found m3 image ie (421880 B)
    ath12k_pci 0001:01:00.0: found fw image ie (7229440 B)
    ath12k_pci 0001:01:00.0: found dualmac fw image ie (7483392 B)
    ath12k_pci 0001:01:00.0: found firmware features ie (1 B)
    ath12k_pci 0001:01:00.0: features
    ath12k_pci 0001:01:00.0: using fw api 2
    ath12k_pci 0001:01:00.0: dualmac fw selected for board id: 1005
    ath12k_pci 0001:01:00.0: Assign MSI to user: MHI, num_vectors: 3, user_base_data: 0, base_vector: 0
    ath12k_pci 0001:01:00.0: Number of assigned MSI for MHI is 3, base vector is 0
    ath12k_pci 0001:01:00.0: Assign MSI to user: CE, num_vectors: 1, user_base_data: 0, base_vector: 0
    ath12k_pci 0001:01:00.0: Assign MSI to user: DP, num_vectors: 1, user_base_data: 0, base_vector: 0
    ath12k_pci 0001:01:00.0: irq:18 group:0
    ath12k_pci 0001:01:00.0: irq:18 group:1
    ath12k_pci 0001:01:00.0: irq:18 group:2
    ath12k_pci 0001:01:00.0: irq:18 group:3
    ath12k_pci 0001:01:00.0: irq:18 group:4
    ath12k_pci 0001:01:00.0: irq:18 group:5
    ath12k_pci 0001:01:00.0: irq:18 group:6
    ath12k_pci 0001:01:00.0: irq:18 group:7
    ath12k_pci 0001:01:00.0: pci after request_irq msi_ep_base_data 0
    ath12k_pci 0001:01:00.0: cookie:0x0
    ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
    ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
    ath12k_pci 0001:01:00.0: soc reset cause:0
    ath12k_pci 0001:01:00.0: MHISTATUS 0xff04
    ath12k_pci 0001:01:00.0: pci link_ctl 0x0000 L0s 0 L1 0
    ath12k_pci 0001:01:00.0: pci reg 0x3164 instance 0x11 read val 0x11
    ath12k_pci 0001:01:00.0: setting mhi state: INIT(0)
    ath12k_pci 0001:01:00.0: setting mhi state: POWER_ON(2)
    ath12k_pci 0001:01:00.0: mhi notify status reason UNKNOWN
    ath12k_pci 0001:01:00.0: mhi notify status reason MHI_CB_EE_MISSION_MODE
    ath12k_pci 0001:01:00.0: qmi wifi fw qmi service connected
    ath12k_pci 0001:01:00.0: phy capability resp valid 1 num_phy 2 valid 1 board_id 84934656 valid 1 single_chip_mlo_support 0
    ath12k_pci 0001:01:00.0: intra device MLO is disabled hence skip QMI MLO cap

Alexander Wilhelm (11):
  wifi: ath12k: fix endianness handling in QMI host capability request
  wifi: ath12k: fix endianness handling in QMI phy capability response
  wifi: ath12k: fix endianness handling in QMI firmware indication
  wifi: ath12k: fix endianness handling in QMI firmware memory indication
  wifi: ath12k: fix endianness handling in QMI respond firmware memory
  wifi: ath12k: fix endianness handling in QMI firmware capabilities
  wifi: ath12k: fix endianness handling in QMI bdf download
  wifi: ath12k: fix endianness handling in QMI firmware m3 info
  wifi: ath12k: fix endianness handling in QMI firmware wlan mode
  wifi: ath12k: fix endianness handling in QMI wlan configuration
  wifi: ath12k: fix endianness handling in QMI response

 drivers/net/wireless/ath/ath12k/qmi.c | 149 ++++++++++++++------------
 drivers/net/wireless/ath/ath12k/qmi.h | 106 +++++++++---------
 include/linux/soc/qcom/qmi.h          |   4 +-
 3 files changed, 136 insertions(+), 123 deletions(-)

-- 
2.34.1

[PATCH 01/11] wifi: ath12k: fix endianness handling in QMI host capability request

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI host capability request message and its associated
structures. Add missing byte swaps wherever these structures are
accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c |  9 +++++----
 drivers/net/wireless/ath/ath12k/qmi.h | 24 ++++++++++++------------
 2 files changed, 17 insertions(+), 16 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 7c611a1fd6d0..2287ed87dae8 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2073,14 +2073,14 @@ static int ath12k_host_cap_parse_mlo(struct ath12k_base *ab,
 	req->mlo_capable_valid = 1;
 	req->mlo_capable = 1;
 	req->mlo_chip_id_valid = 1;
-	req->mlo_chip_id = ab->device_id;
+	req->mlo_chip_id = cpu_to_le16(ab->device_id);
 	req->mlo_group_id_valid = 1;
 	req->mlo_group_id = ag->id;
 	req->max_mlo_peer_valid = 1;
 	/* Max peer number generally won't change for the same device
 	 * but needs to be synced with host driver.
 	 */
-	req->max_mlo_peer = ab->hw_params->max_mlo_peer;
+	req->max_mlo_peer = cpu_to_le16(ab->hw_params->max_mlo_peer);
 	req->mlo_num_chips_valid = 1;
 	req->mlo_num_chips = ag->num_devices;
 
@@ -2164,7 +2164,7 @@ int ath12k_qmi_host_cap_send(struct ath12k_base *ab)
 	int ret = 0;
 
 	req.num_clients_valid = 1;
-	req.num_clients = 1;
+	req.num_clients = cpu_to_le32(1);
 	req.mem_cfg_mode = ab->qmi.target_mem_mode;
 	req.mem_cfg_mode_valid = 1;
 	req.bdf_support_valid = 1;
@@ -2182,7 +2182,8 @@ int ath12k_qmi_host_cap_send(struct ath12k_base *ab)
 
 	if (ab->hw_params->qmi_cnss_feature_bitmap) {
 		req.feature_list_valid = 1;
-		req.feature_list = ab->hw_params->qmi_cnss_feature_bitmap;
+		req.feature_list =
+			cpu_to_le64(ab->hw_params->qmi_cnss_feature_bitmap);
 	}
 
 	/* BRINGUP: here we are piggybacking a lot of stuff using
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index abdaade3b542..ed7808fe2b71 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -164,8 +164,8 @@ struct ath12k_qmi {
 #define QMI_WLANFW_MAX_HOST_DDR_RANGE_SIZE_V01		3
 
 struct qmi_wlanfw_host_ddr_range {
-	u64 start;
-	u64 size;
+	__le64 start;
+	__le64 size;
 };
 
 enum ath12k_qmi_target_mem {
@@ -205,12 +205,12 @@ enum ath12k_qmi_cnss_feature {
 
 struct qmi_wlanfw_host_cap_req_msg_v01 {
 	u8 num_clients_valid;
-	u32 num_clients;
+	__le32 num_clients;
 	u8 wake_msi_valid;
-	u32 wake_msi;
+	__le32 wake_msi;
 	u8 gpios_valid;
-	u32 gpios_len;
-	u32 gpios[QMI_WLFW_MAX_NUM_GPIO_V01];
+	__le32 gpios_len;
+	__le32 gpios[QMI_WLFW_MAX_NUM_GPIO_V01];
 	u8 nm_modem_valid;
 	u8 nm_modem;
 	u8 bdf_support_valid;
@@ -228,31 +228,31 @@ struct qmi_wlanfw_host_cap_req_msg_v01 {
 	u8 cal_done_valid;
 	u8 cal_done;
 	u8 mem_bucket_valid;
-	u32 mem_bucket;
+	__le32 mem_bucket;
 	u8 mem_cfg_mode_valid;
 	u8 mem_cfg_mode;
 	u8 cal_duration_valid;
-	u16 cal_duraiton;
+	__le16 cal_duraiton;
 	u8 platform_name_valid;
 	char platform_name[QMI_WLANFW_MAX_PLATFORM_NAME_LEN_V01 + 1];
 	u8 ddr_range_valid;
 	struct qmi_wlanfw_host_ddr_range ddr_range[QMI_WLANFW_MAX_HOST_DDR_RANGE_SIZE_V01];
 	u8 host_build_type_valid;
-	enum qmi_wlanfw_host_build_type host_build_type;
+	__le32 host_build_type;
 	u8 mlo_capable_valid;
 	u8 mlo_capable;
 	u8 mlo_chip_id_valid;
-	u16 mlo_chip_id;
+	__le16 mlo_chip_id;
 	u8 mlo_group_id_valid;
 	u8 mlo_group_id;
 	u8 max_mlo_peer_valid;
-	u16 max_mlo_peer;
+	__le16 max_mlo_peer;
 	u8 mlo_num_chips_valid;
 	u8 mlo_num_chips;
 	u8 mlo_chip_info_valid;
 	struct wlfw_host_mlo_chip_info_s_v01 mlo_chip_info[QMI_WLFW_MAX_NUM_MLO_CHIPS_V01];
 	u8 feature_list_valid;
-	u64 feature_list;
+	__le64 feature_list;
 
 };
 
-- 
2.34.1

[PATCH 02/11] wifi: ath12k: fix endianness handling in QMI phy capability response

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI phy capability response message. Add missing byte
swaps wherever this structure is accessed or modified.

Fixes: 53a65445c144 ("wifi: ath12k: add QMI PHY capability learn support")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 2 +-
 drivers/net/wireless/ath/ath12k/qmi.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 2287ed87dae8..2973edd2735b 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2283,7 +2283,7 @@ static void ath12k_qmi_phy_cap_send(struct ath12k_base *ab)
 		   "phy capability resp valid %d single_chip_mlo_support %d valid %d num_phy %d valid %d board_id %d\n",
 		   resp.single_chip_mlo_support_valid, resp.single_chip_mlo_support,
 		   resp.num_phy_valid, resp.num_phy,
-		   resp.board_id_valid, resp.board_id);
+		   resp.board_id_valid, le32_to_cpu(resp.board_id));
 
 	return;
 
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index ed7808fe2b71..abe2fba25d13 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -273,7 +273,7 @@ struct qmi_wlanfw_phy_cap_resp_msg_v01 {
 	u8 num_phy_valid;
 	u8 num_phy;
 	u8 board_id_valid;
-	u32 board_id;
+	__le32 board_id;
 	u8 single_chip_mlo_support_valid;
 	u8 single_chip_mlo_support;
 };
-- 
2.34.1

[PATCH 03/11] wifi: ath12k: fix endianness handling in QMI firmware indication

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware indication request and response messages
and its associated structures as well. Add missing byte swaps wherever
these structures are accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 2 +-
 drivers/net/wireless/ath/ath12k/qmi.h | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 2973edd2735b..cf05c76cc112 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2315,7 +2315,7 @@ static int ath12k_qmi_fw_ind_register_send(struct ath12k_base *ab)
 	}
 
 	req->client_id_valid = 1;
-	req->client_id = QMI_WLANFW_CLIENT_ID;
+	req->client_id = cpu_to_le32(QMI_WLANFW_CLIENT_ID);
 	req->fw_ready_enable_valid = 1;
 	req->fw_ready_enable = 1;
 	req->request_mem_enable_valid = 1;
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index abe2fba25d13..026dc33a39d8 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -296,7 +296,7 @@ struct qmi_wlanfw_ind_register_req_msg_v01 {
 	u8 pin_connect_result_enable_valid;
 	u8 pin_connect_result_enable;
 	u8 client_id_valid;
-	u32 client_id;
+	__le32 client_id;
 	u8 request_mem_enable_valid;
 	u8 request_mem_enable;
 	u8 fw_mem_ready_enable_valid;
@@ -304,7 +304,7 @@ struct qmi_wlanfw_ind_register_req_msg_v01 {
 	u8 fw_init_done_enable_valid;
 	u8 fw_init_done_enable;
 	u8 rejuvenate_enable_valid;
-	u32 rejuvenate_enable;
+	__le32 rejuvenate_enable;
 	u8 xo_cal_enable_valid;
 	u8 xo_cal_enable;
 	u8 cal_done_enable_valid;
@@ -314,7 +314,7 @@ struct qmi_wlanfw_ind_register_req_msg_v01 {
 struct qmi_wlanfw_ind_register_resp_msg_v01 {
 	struct qmi_response_type_v01 resp;
 	u8 fw_status_valid;
-	u64 fw_status;
+	__le64 fw_status;
 };
 
 #define QMI_WLANFW_REQUEST_MEM_IND_MSG_V01_MAX_LEN	1824
-- 
2.34.1

[PATCH 04/11] wifi: ath12k: fix endianness handling in QMI firmware memory indication

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware memory indication request message and its
associated structures as well. Add missing byte swaps wherever these
structures are accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 15 ++++++++-------
 drivers/net/wireless/ath/ath12k/qmi.h | 12 ++++++------
 2 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index cf05c76cc112..146e9e1c4834 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -3637,18 +3637,19 @@ static void ath12k_qmi_msg_mem_request_cb(struct qmi_handle *qmi_hdl,
 
 	ath12k_dbg(ab, ATH12K_DBG_QMI, "qmi firmware request memory request\n");
 
-	if (msg->mem_seg_len == 0 ||
-	    msg->mem_seg_len > ATH12K_QMI_WLANFW_MAX_NUM_MEM_SEG_V01)
+	if (le32_to_cpu(msg->mem_seg_len) == 0 ||
+	    le32_to_cpu(msg->mem_seg_len) > ATH12K_QMI_WLANFW_MAX_NUM_MEM_SEG_V01)
 		ath12k_warn(ab, "Invalid memory segment length: %u\n",
-			    msg->mem_seg_len);
+			    le32_to_cpu(msg->mem_seg_len));
 
-	ab->qmi.mem_seg_count = msg->mem_seg_len;
+	ab->qmi.mem_seg_count = le32_to_cpu(msg->mem_seg_len);
 
 	for (i = 0; i < qmi->mem_seg_count ; i++) {
-		ab->qmi.target_mem[i].type = msg->mem_seg[i].type;
-		ab->qmi.target_mem[i].size = msg->mem_seg[i].size;
+		ab->qmi.target_mem[i].type = le32_to_cpu(msg->mem_seg[i].type);
+		ab->qmi.target_mem[i].size = le32_to_cpu(msg->mem_seg[i].size);
 		ath12k_dbg(ab, ATH12K_DBG_QMI, "qmi mem seg type %d size %d\n",
-			   msg->mem_seg[i].type, msg->mem_seg[i].size);
+			   le32_to_cpu(msg->mem_seg[i].type),
+			   le32_to_cpu(msg->mem_seg[i].size));
 	}
 
 	if (test_bit(ATH12K_FLAG_FIXED_MEM_REGION, &ab->dev_flags)) {
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index 026dc33a39d8..92993c85cec9 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -327,8 +327,8 @@ struct qmi_wlanfw_ind_register_resp_msg_v01 {
 #define QMI_WLANFW_MAX_STR_LEN_V01                      16
 
 struct qmi_wlanfw_mem_cfg_s_v01 {
-	u64 offset;
-	u32 size;
+	__le64 offset;
+	__le32 size;
 	u8 secure_flag;
 };
 
@@ -344,14 +344,14 @@ enum qmi_wlanfw_mem_type_enum_v01 {
 };
 
 struct qmi_wlanfw_mem_seg_s_v01 {
-	u32 size;
-	enum qmi_wlanfw_mem_type_enum_v01 type;
-	u32 mem_cfg_len;
+	__le32 size;
+	__le32 type;
+	__le32 mem_cfg_len;
 	struct qmi_wlanfw_mem_cfg_s_v01 mem_cfg[QMI_WLANFW_MAX_NUM_MEM_CFG_V01];
 };
 
 struct qmi_wlanfw_request_mem_ind_msg_v01 {
-	u32 mem_seg_len;
+	__le32 mem_seg_len;
 	struct qmi_wlanfw_mem_seg_s_v01 mem_seg[ATH12K_QMI_WLANFW_MAX_NUM_MEM_SEG_V01];
 };
 
-- 
2.34.1

[PATCH 05/11] wifi: ath12k: fix endianness handling in QMI respond firmware memory

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI respond firmware memory request message and its
associated structures as well. Add missing byte swaps wherever these
structures are accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 10 +++++-----
 drivers/net/wireless/ath/ath12k/qmi.h |  8 ++++----
 2 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 146e9e1c4834..2d00ea54e478 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2392,11 +2392,11 @@ int ath12k_qmi_respond_fw_mem_request(struct ath12k_base *ab)
 			   ab->qmi.mem_seg_count);
 	} else {
 		delayed = false;
-		req->mem_seg_len = ab->qmi.mem_seg_count;
-		for (i = 0; i < req->mem_seg_len ; i++) {
-			req->mem_seg[i].addr = ab->qmi.target_mem[i].paddr;
-			req->mem_seg[i].size = ab->qmi.target_mem[i].size;
-			req->mem_seg[i].type = ab->qmi.target_mem[i].type;
+		req->mem_seg_len = cpu_to_le32(ab->qmi.mem_seg_count);
+		for (i = 0; i < ab->qmi.mem_seg_count; i++) {
+			req->mem_seg[i].addr = cpu_to_le64(ab->qmi.target_mem[i].paddr);
+			req->mem_seg[i].size = cpu_to_le32(ab->qmi.target_mem[i].size);
+			req->mem_seg[i].type = cpu_to_le32(ab->qmi.target_mem[i].type);
 			ath12k_dbg(ab, ATH12K_DBG_QMI,
 				   "qmi req mem_seg[%d] %pad %u %u\n", i,
 				   &ab->qmi.target_mem[i].paddr,
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index 92993c85cec9..bd715bca8f23 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -356,14 +356,14 @@ struct qmi_wlanfw_request_mem_ind_msg_v01 {
 };
 
 struct qmi_wlanfw_mem_seg_resp_s_v01 {
-	u64 addr;
-	u32 size;
-	enum qmi_wlanfw_mem_type_enum_v01 type;
+	__le64 addr;
+	__le32 size;
+	__le32 type;
 	u8 restore;
 };
 
 struct qmi_wlanfw_respond_mem_req_msg_v01 {
-	u32 mem_seg_len;
+	__le32 mem_seg_len;
 	struct qmi_wlanfw_mem_seg_resp_s_v01 mem_seg[ATH12K_QMI_WLANFW_MAX_NUM_MEM_SEG_V01];
 };
 
-- 
2.34.1

[PATCH 06/11] wifi: ath12k: fix endianness handling in QMI firmware capabilities

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware capabilities response message and its
associated structures as well. Add missing byte swaps wherever these
structures are accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 16 ++++++++--------
 drivers/net/wireless/ath/ath12k/qmi.h | 26 +++++++++++++-------------
 2 files changed, 21 insertions(+), 21 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 2d00ea54e478..0bcfb9281d01 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2850,20 +2850,20 @@ int ath12k_qmi_request_target_cap(struct ath12k_base *ab)
 	}
 
 	if (resp.chip_info_valid) {
-		ab->qmi.target.chip_id = resp.chip_info.chip_id;
-		ab->qmi.target.chip_family = resp.chip_info.chip_family;
+		ab->qmi.target.chip_id = le32_to_cpu(resp.chip_info.chip_id);
+		ab->qmi.target.chip_family = le32_to_cpu(resp.chip_info.chip_family);
 	}
 
 	if (resp.board_info_valid)
-		ab->qmi.target.board_id = resp.board_info.board_id;
+		ab->qmi.target.board_id = le32_to_cpu(resp.board_info.board_id);
 	else
 		ab->qmi.target.board_id = board_id;
 
 	if (resp.soc_info_valid)
-		ab->qmi.target.soc_id = resp.soc_info.soc_id;
+		ab->qmi.target.soc_id = le32_to_cpu(resp.soc_info.soc_id);
 
 	if (resp.fw_version_info_valid) {
-		ab->qmi.target.fw_version = resp.fw_version_info.fw_version;
+		ab->qmi.target.fw_version = le32_to_cpu(resp.fw_version_info.fw_version);
 		strscpy(ab->qmi.target.fw_build_timestamp,
 			resp.fw_version_info.fw_build_timestamp,
 			sizeof(ab->qmi.target.fw_build_timestamp));
@@ -2876,9 +2876,9 @@ int ath12k_qmi_request_target_cap(struct ath12k_base *ab)
 	if (resp.dev_mem_info_valid) {
 		for (i = 0; i < ATH12K_QMI_WLFW_MAX_DEV_MEM_NUM_V01; i++) {
 			ab->qmi.dev_mem[i].start =
-				resp.dev_mem[i].start;
+				le64_to_cpu(resp.dev_mem[i].start);
 			ab->qmi.dev_mem[i].size =
-				resp.dev_mem[i].size;
+				le64_to_cpu(resp.dev_mem[i].size);
 			ath12k_dbg(ab, ATH12K_DBG_QMI,
 				   "devmem [%d] start 0x%llx size %llu\n", i,
 				   ab->qmi.dev_mem[i].start,
@@ -2887,7 +2887,7 @@ int ath12k_qmi_request_target_cap(struct ath12k_base *ab)
 	}
 
 	if (resp.eeprom_caldata_read_timeout_valid) {
-		ab->qmi.target.eeprom_caldata = resp.eeprom_caldata_read_timeout;
+		ab->qmi.target.eeprom_caldata = le32_to_cpu(resp.eeprom_caldata_read_timeout);
 		ath12k_dbg(ab, ATH12K_DBG_QMI, "qmi cal data supported from eeprom\n");
 	}
 
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index bd715bca8f23..114324157e84 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -421,26 +421,26 @@ struct qmi_wlanfw_memory_region_info_s_v01 {
 };
 
 struct qmi_wlanfw_rf_chip_info_s_v01 {
-	u32 chip_id;
-	u32 chip_family;
+	__le32 chip_id;
+	__le32 chip_family;
 };
 
 struct qmi_wlanfw_rf_board_info_s_v01 {
-	u32 board_id;
+	__le32 board_id;
 };
 
 struct qmi_wlanfw_soc_info_s_v01 {
-	u32 soc_id;
+	__le32 soc_id;
 };
 
 struct qmi_wlanfw_fw_version_info_s_v01 {
-	u32 fw_version;
+	__le32 fw_version;
 	char fw_build_timestamp[ATH12K_QMI_WLANFW_MAX_TIMESTAMP_LEN_V01 + 1];
 };
 
 struct qmi_wlanfw_dev_mem_info_s_v01 {
-	u64 start;
-	u64 size;
+	__le64 start;
+	__le64 size;
 };
 
 enum qmi_wlanfw_cal_temp_id_enum_v01 {
@@ -475,17 +475,17 @@ struct qmi_wlanfw_cap_resp_msg_v01 {
 	u8 num_macs_valid;
 	u8 num_macs;
 	u8 voltage_mv_valid;
-	u32 voltage_mv;
+	__le32 voltage_mv;
 	u8 time_freq_hz_valid;
-	u32 time_freq_hz;
+	__le32 time_freq_hz;
 	u8 otp_version_valid;
-	u32 otp_version;
+	__le32 otp_version;
 	u8 eeprom_caldata_read_timeout_valid;
-	u32 eeprom_caldata_read_timeout;
+	__le32 eeprom_caldata_read_timeout;
 	u8 fw_caps_valid;
-	u64 fw_caps;
+	__le64 fw_caps;
 	u8 rd_card_chain_cap_valid;
-	enum qmi_wlanfw_rd_card_chain_cap_v01 rd_card_chain_cap;
+	__le32 rd_card_chain_cap;
 	u8 dev_mem_info_valid;
 	struct qmi_wlanfw_dev_mem_info_s_v01 dev_mem[ATH12K_QMI_WLFW_MAX_DEV_MEM_NUM_V01];
 };
-- 
2.34.1

[PATCH 07/11] wifi: ath12k: fix endianness handling in QMI bdf download

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI board data file download request message. Add
missing byte swaps wherever this structure is accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 18 +++++++++---------
 drivers/net/wireless/ath/ath12k/qmi.h |  8 ++++----
 2 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 0bcfb9281d01..d18ffb86f7bc 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2934,9 +2934,9 @@ static int ath12k_qmi_load_file_target_mem(struct ath12k_base *ab,
 	while (remaining) {
 		req->valid = 1;
 		req->file_id_valid = 1;
-		req->file_id = ab->qmi.target.board_id;
+		req->file_id = cpu_to_le32(ab->qmi.target.board_id);
 		req->total_size_valid = 1;
-		req->total_size = remaining;
+		req->total_size = cpu_to_le32(remaining);
 		req->seg_id_valid = 1;
 		req->data_valid = 1;
 		req->bdf_type = type;
@@ -2945,18 +2945,18 @@ static int ath12k_qmi_load_file_target_mem(struct ath12k_base *ab,
 		req->end = 0;
 
 		if (remaining > QMI_WLANFW_MAX_DATA_SIZE_V01) {
-			req->data_len = QMI_WLANFW_MAX_DATA_SIZE_V01;
+			req->data_len = cpu_to_le32(QMI_WLANFW_MAX_DATA_SIZE_V01);
 		} else {
-			req->data_len = remaining;
+			req->data_len = cpu_to_le32(remaining);
 			req->end = 1;
 		}
 
 		if (type == ATH12K_QMI_FILE_TYPE_EEPROM) {
 			req->data_valid = 0;
 			req->end = 1;
-			req->data_len = ATH12K_QMI_MAX_BDF_FILE_NAME_SIZE;
+			req->data_len = cpu_to_le32(ATH12K_QMI_MAX_BDF_FILE_NAME_SIZE);
 		} else {
-			memcpy(req->data, temp, req->data_len);
+			memcpy(req->data, temp, le32_to_cpu(req->data_len));
 		}
 
 		ret = qmi_txn_init(&ab->qmi.handle, &txn,
@@ -2991,9 +2991,9 @@ static int ath12k_qmi_load_file_target_mem(struct ath12k_base *ab,
 		if (type == ATH12K_QMI_FILE_TYPE_EEPROM) {
 			remaining = 0;
 		} else {
-			remaining -= req->data_len;
-			temp += req->data_len;
-			req->seg_id++;
+			remaining -= le32_to_cpu(req->data_len);
+			temp += le32_to_cpu(req->data_len);
+			req->seg_id += cpu_to_le32(1);
 			ath12k_dbg(ab, ATH12K_DBG_QMI,
 				   "qmi bdf download request remaining %i\n",
 				   remaining);
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index 114324157e84..a48850ec0296 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -504,13 +504,13 @@ struct qmi_wlanfw_cap_req_msg_v01 {
 struct qmi_wlanfw_bdf_download_req_msg_v01 {
 	u8 valid;
 	u8 file_id_valid;
-	enum qmi_wlanfw_cal_temp_id_enum_v01 file_id;
+	__le32 file_id;
 	u8 total_size_valid;
-	u32 total_size;
+	__le32 total_size;
 	u8 seg_id_valid;
-	u32 seg_id;
+	__le32 seg_id;
 	u8 data_valid;
-	u32 data_len;
+	__le32 data_len;
 	u8 data[QMI_WLANFW_MAX_DATA_SIZE_V01];
 	u8 end_valid;
 	u8 end;
-- 
2.34.1

[PATCH 08/11] wifi: ath12k: fix endianness handling in QMI firmware m3 info

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware M3 info request message. Add missing byte
swaps wherever this structure is accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 4 ++--
 drivers/net/wireless/ath/ath12k/qmi.h | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index d18ffb86f7bc..5043d0c39b56 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -3198,8 +3198,8 @@ int ath12k_qmi_wlanfw_m3_info_send(struct ath12k_base *ab)
 			ath12k_err(ab, "failed to load m3 firmware: %d", ret);
 			return ret;
 		}
-		req.addr = m3_mem->paddr;
-		req.size = m3_mem->size;
+		req.addr = cpu_to_le64(m3_mem->paddr);
+		req.size = cpu_to_le32(m3_mem->size);
 	}
 
 	ret = qmi_txn_init(&ab->qmi.handle, &txn,
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index a48850ec0296..101d91131ef0 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -529,8 +529,8 @@ struct qmi_wlanfw_bdf_download_resp_msg_v01 {
 #define QMI_WLANFW_M3_INFO_REQ_V01		0x003C
 
 struct qmi_wlanfw_m3_info_req_msg_v01 {
-	u64 addr;
-	u32 size;
+	__le64 addr;
+	__le32 size;
 };
 
 struct qmi_wlanfw_m3_info_resp_msg_v01 {
-- 
2.34.1

[PATCH 09/11] wifi: ath12k: fix endianness handling in QMI firmware wlan mode

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware wlan mode request message. Add missing
byte swaps wherever this structure is accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 2 +-
 drivers/net/wireless/ath/ath12k/qmi.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 5043d0c39b56..2fdd5eb128d2 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -3242,7 +3242,7 @@ static int ath12k_qmi_wlanfw_mode_send(struct ath12k_base *ab,
 	struct qmi_txn txn;
 	int ret = 0;
 
-	req.mode = mode;
+	req.mode = cpu_to_le32(mode);
 	req.hw_debug_valid = 1;
 	req.hw_debug = 0;
 
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index 101d91131ef0..d05cd9b47911 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -552,7 +552,7 @@ struct qmi_wlanfw_m3_info_resp_msg_v01 {
 #define QMI_WLANFW_MAX_NUM_SHADOW_REG_V3_V01		60
 
 struct qmi_wlanfw_wlan_mode_req_msg_v01 {
-	u32 mode;
+	__le32 mode;
 	u8 hw_debug_valid;
 	u8 hw_debug;
 };
-- 
2.34.1

[PATCH 10/11] wifi: ath12k: fix endianness handling in QMI wlan configuration

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI firmware wlan configuration request message and
its associated structures as well. Add missing byte swaps wherever these
structures are accessed or modified.

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 22 ++++++++++++----------
 drivers/net/wireless/ath/ath12k/qmi.h | 14 +++++++-------
 2 files changed, 19 insertions(+), 17 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 2fdd5eb128d2..71cf6408c6fb 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -3291,7 +3291,7 @@ static int ath12k_qmi_wlanfw_wlan_cfg_send(struct ath12k_base *ab)
 	struct ce_pipe_config *ce_cfg;
 	struct service_to_pipe *svc_cfg;
 	struct qmi_txn txn;
-	int ret = 0, pipe_num;
+	int ret = 0, pipe_num, sr, len;
 
 	ce_cfg	= (struct ce_pipe_config *)ab->qmi.ce_cfg.tgt_ce;
 	svc_cfg	= (struct service_to_pipe *)ab->qmi.ce_cfg.svc_to_ce_map;
@@ -3306,8 +3306,8 @@ static int ath12k_qmi_wlanfw_wlan_cfg_send(struct ath12k_base *ab)
 
 	req->tgt_cfg_valid = 1;
 	/* This is number of CE configs */
-	req->tgt_cfg_len = ab->qmi.ce_cfg.tgt_ce_len;
-	for (pipe_num = 0; pipe_num < req->tgt_cfg_len ; pipe_num++) {
+	req->tgt_cfg_len = cpu_to_le32(ab->qmi.ce_cfg.tgt_ce_len);
+	for (pipe_num = 0; pipe_num < ab->qmi.ce_cfg.tgt_ce_len; pipe_num++) {
 		req->tgt_cfg[pipe_num].pipe_num = ce_cfg[pipe_num].pipenum;
 		req->tgt_cfg[pipe_num].pipe_dir = ce_cfg[pipe_num].pipedir;
 		req->tgt_cfg[pipe_num].nentries = ce_cfg[pipe_num].nentries;
@@ -3317,8 +3317,8 @@ static int ath12k_qmi_wlanfw_wlan_cfg_send(struct ath12k_base *ab)
 
 	req->svc_cfg_valid = 1;
 	/* This is number of Service/CE configs */
-	req->svc_cfg_len = ab->qmi.ce_cfg.svc_to_ce_map_len;
-	for (pipe_num = 0; pipe_num < req->svc_cfg_len; pipe_num++) {
+	req->svc_cfg_len = cpu_to_le32(ab->qmi.ce_cfg.svc_to_ce_map_len);
+	for (pipe_num = 0; pipe_num < ab->qmi.ce_cfg.svc_to_ce_map_len; pipe_num++) {
 		req->svc_cfg[pipe_num].service_id = svc_cfg[pipe_num].service_id;
 		req->svc_cfg[pipe_num].pipe_dir = svc_cfg[pipe_num].pipedir;
 		req->svc_cfg[pipe_num].pipe_num = svc_cfg[pipe_num].pipenum;
@@ -3327,11 +3327,13 @@ static int ath12k_qmi_wlanfw_wlan_cfg_send(struct ath12k_base *ab)
 	/* set shadow v3 configuration */
 	if (ab->hw_params->supports_shadow_regs) {
 		req->shadow_reg_v3_valid = 1;
-		req->shadow_reg_v3_len = min_t(u32,
-					       ab->qmi.ce_cfg.shadow_reg_v3_len,
-					       QMI_WLANFW_MAX_NUM_SHADOW_REG_V3_V01);
-		memcpy(&req->shadow_reg_v3, ab->qmi.ce_cfg.shadow_reg_v3,
-		       sizeof(u32) * req->shadow_reg_v3_len);
+		len = min_t(u32,
+			    ab->qmi.ce_cfg.shadow_reg_v3_len,
+			    QMI_WLANFW_MAX_NUM_SHADOW_REG_V3_V01);
+		req->shadow_reg_v3_len = cpu_to_le32(len);
+
+		for (sr = 0; sr < len; sr++)
+			req->shadow_reg_v3[sr].addr = cpu_to_le32(ab->qmi.ce_cfg.shadow_reg_v3[sr]);
 	} else {
 		req->shadow_reg_v3_valid = 0;
 	}
diff --git a/drivers/net/wireless/ath/ath12k/qmi.h b/drivers/net/wireless/ath/ath12k/qmi.h
index d05cd9b47911..37840d015a5c 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.h
+++ b/drivers/net/wireless/ath/ath12k/qmi.h
@@ -406,12 +406,12 @@ struct qmi_wlanfw_ce_svc_pipe_cfg_s_v01 {
 };
 
 struct qmi_wlanfw_shadow_reg_cfg_s_v01 {
-	u16 id;
-	u16 offset;
+	__le16 id;
+	__le16 offset;
 };
 
 struct qmi_wlanfw_shadow_reg_v3_cfg_s_v01 {
-	u32 addr;
+	__le32 addr;
 };
 
 struct qmi_wlanfw_memory_region_info_s_v01 {
@@ -565,19 +565,19 @@ struct qmi_wlanfw_wlan_cfg_req_msg_v01 {
 	u8 host_version_valid;
 	char host_version[QMI_WLANFW_MAX_STR_LEN_V01 + 1];
 	u8  tgt_cfg_valid;
-	u32  tgt_cfg_len;
+	__le32 tgt_cfg_len;
 	struct qmi_wlanfw_ce_tgt_pipe_cfg_s_v01
 			tgt_cfg[QMI_WLANFW_MAX_NUM_CE_V01];
 	u8  svc_cfg_valid;
-	u32 svc_cfg_len;
+	__le32 svc_cfg_len;
 	struct qmi_wlanfw_ce_svc_pipe_cfg_s_v01
 			svc_cfg[QMI_WLANFW_MAX_NUM_SVC_V01];
 	u8 shadow_reg_valid;
-	u32 shadow_reg_len;
+	__le32 shadow_reg_len;
 	struct qmi_wlanfw_shadow_reg_cfg_s_v01
 		shadow_reg[QMI_WLANFW_MAX_NUM_SHADOW_REG_V01];
 	u8 shadow_reg_v3_valid;
-	u32 shadow_reg_v3_len;
+	__le32 shadow_reg_v3_len;
 	struct qmi_wlanfw_shadow_reg_v3_cfg_s_v01
 		shadow_reg_v3[QMI_WLANFW_MAX_NUM_SHADOW_REG_V3_V01];
 };
-- 
2.34.1

[PATCH 11/11] wifi: ath12k: fix endianness handling in QMI response

[Alexander Wilhelm]

Ensure proper endianness support for big-endian platforms by correcting
data types in the QMI response. Add missing byte swaps wherever this
structure is accessed or modified.

Fixes: 9b8a11e82615 ("soc: qcom: Introduce QMI encoder/decoder")
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
---
 drivers/net/wireless/ath/ath12k/qmi.c | 49 ++++++++++++++++-----------
 include/linux/soc/qcom/qmi.h          |  4 +--
 2 files changed, 31 insertions(+), 22 deletions(-)

diff --git a/drivers/net/wireless/ath/ath12k/qmi.c b/drivers/net/wireless/ath/ath12k/qmi.c
index 71cf6408c6fb..d421e7b3fc87 100644
--- a/drivers/net/wireless/ath/ath12k/qmi.c
+++ b/drivers/net/wireless/ath/ath12k/qmi.c
@@ -2227,9 +2227,10 @@ int ath12k_qmi_host_cap_send(struct ath12k_base *ab)
 	if (ret < 0)
 		goto out;
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "Host capability request failed, result: %d, err: %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2264,7 +2265,7 @@ static void ath12k_qmi_phy_cap_send(struct ath12k_base *ab)
 	if (ret < 0)
 		goto out;
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ret = -EOPNOTSUPP;
 		goto out;
 	}
@@ -2352,9 +2353,10 @@ static int ath12k_qmi_fw_ind_register_send(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp->resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp->resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "FW Ind register request failed, result: %d, err: %d\n",
-			    resp->resp.result, resp->resp.error);
+			    le16_to_cpu(resp->resp.result),
+			    le16_to_cpu(resp->resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2427,15 +2429,16 @@ int ath12k_qmi_respond_fw_mem_request(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		/* the error response is expected when
 		 * target_mem_delayed is true.
 		 */
-		if (delayed && resp.resp.error == 0)
+		if (delayed && le16_to_cpu(resp.resp.error) == 0)
 			goto out;
 
 		ath12k_warn(ab, "Respond mem req failed, result: %d, err: %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2842,9 +2845,10 @@ int ath12k_qmi_request_target_cap(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "qmi targetcap req failed, result: %d, err: %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2981,9 +2985,10 @@ static int ath12k_qmi_load_file_target_mem(struct ath12k_base *ab,
 		if (ret < 0)
 			goto out;
 
-		if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+		if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 			ath12k_warn(ab, "qmi BDF download failed, result: %d, err: %d\n",
-				    resp.resp.result, resp.resp.error);
+				    le16_to_cpu(resp.resp.result),
+				    le16_to_cpu(resp.resp.error));
 			ret = -EINVAL;
 			goto out;
 		}
@@ -3224,9 +3229,10 @@ int ath12k_qmi_wlanfw_m3_info_send(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "qmi M3 info request failed, result: %d, err: %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -3273,9 +3279,10 @@ static int ath12k_qmi_wlanfw_mode_send(struct ath12k_base *ab,
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "Mode request failed, mode: %d, result: %d err: %d\n",
-			    mode, resp.resp.result, resp.resp.error);
+			    mode, le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -3360,9 +3367,10 @@ static int ath12k_qmi_wlanfw_wlan_cfg_send(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "qmi wlan config request failed, result: %d, err: %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
@@ -3404,9 +3412,10 @@ static int ath12k_qmi_wlanfw_wlan_ini_send(struct ath12k_base *ab)
 		goto out;
 	}
 
-	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
+	if (le16_to_cpu(resp.resp.result) != QMI_RESULT_SUCCESS_V01) {
 		ath12k_warn(ab, "QMI wlan ini response failure: %d %d\n",
-			    resp.resp.result, resp.resp.error);
+			    le16_to_cpu(resp.resp.result),
+			    le16_to_cpu(resp.resp.error));
 		ret = -EINVAL;
 		goto out;
 	}
diff --git a/include/linux/soc/qcom/qmi.h b/include/linux/soc/qcom/qmi.h
index 469e02d2aa0d..a487b724abd2 100644
--- a/include/linux/soc/qcom/qmi.h
+++ b/include/linux/soc/qcom/qmi.h
@@ -98,8 +98,8 @@ struct qmi_elem_info {
  * @error:	error value, when @result is QMI_RESULT_FAILURE_V01
  */
 struct qmi_response_type_v01 {
-	u16 result;
-	u16 error;
+	__le16 result;
+	__le16 error;
 };
 
 extern const struct qmi_elem_info qmi_response_type_v01_ei[];
-- 
2.34.1

Re: [PATCH 00/11] wifi: ath12k: Fix endianness handling in QMI

[Jeff Johnson]

On 7/16/2025 12:50 AM, Alexander Wilhelm wrote:
> Fix endianness handling in QMI firmware transfer on big-endian
> platforms. Without this fix, the firmware download fails due to
> misinterpreted data structures exchanged between the host and the
> wireless module.
> 
> The issue occurs during early bring-up on big endian systems, where QMI
> messages are not correctly parsed by the driver, leading to failed
> initialization sequences. Ensure all relevant fields are properly
> converted between CPU and little-endian format in request and response
> messages, as expected by the firmware. Attached logs showing the failure
> before the fix:
> 
>     ath12k_pci 0001:01:00.0: BAR 0: assigned [mem 0xc00000000-0xc001fffff 64bit]
>     ath12k_pci 0001:01:00.0: boot pci_mem 0xcd4148e9
>     ath12k_pci 0001:01:00.0: pci probe 17cb:1109 17cb:1109
>     ath12k_pci 0001:01:00.0: pci tcsr_soc_hw_version major 2 minor 0
>     ath12k_pci 0001:01:00.0: request MSI one vector
>     ath12k_pci 0001:01:00.0: MSI vectors: 1
>     ath12k_pci 0001:01:00.0: msi base data is 0
>     ath12k_pci 0001:01:00.0: Hardware name: qcn9274 hw2.0
>     ath12k_pci 0001:01:00.0: boot firmware request ath12k/QCN9274/hw2.0/firmware-2.bin size 15134776
>     ath12k_pci 0001:01:00.0: found fw timestamp 1722934582
>     ath12k_pci 0001:01:00.0: found m3 image ie (421880 B)
>     ath12k_pci 0001:01:00.0: found fw image ie (7229440 B)
>     ath12k_pci 0001:01:00.0: found dualmac fw image ie (7483392 B)
>     ath12k_pci 0001:01:00.0: found firmware features ie (1 B)
>     ath12k_pci 0001:01:00.0: features
>     ath12k_pci 0001:01:00.0: using fw api 2
>     ath12k_pci 0001:01:00.0: dualmac fw selected for board id: 1005
>     ath12k_pci 0001:01:00.0: Assign MSI to user: MHI, num_vectors: 3, user_base_data: 0, base_vector: 0
>     ath12k_pci 0001:01:00.0: Number of assigned MSI for MHI is 3, base vector is 0
>     ath12k_pci 0001:01:00.0: Assign MSI to user: CE, num_vectors: 1, user_base_data: 0, base_vector: 0
>     ath12k_pci 0001:01:00.0: Assign MSI to user: DP, num_vectors: 1, user_base_data: 0, base_vector: 0
>     ath12k_pci 0001:01:00.0: irq:18 group:0
>     ath12k_pci 0001:01:00.0: irq:18 group:1
>     ath12k_pci 0001:01:00.0: irq:18 group:2
>     ath12k_pci 0001:01:00.0: irq:18 group:3
>     ath12k_pci 0001:01:00.0: irq:18 group:4
>     ath12k_pci 0001:01:00.0: irq:18 group:5
>     ath12k_pci 0001:01:00.0: irq:18 group:6
>     ath12k_pci 0001:01:00.0: irq:18 group:7
>     ath12k_pci 0001:01:00.0: pci after request_irq msi_ep_base_data 0
>     ath12k_pci 0001:01:00.0: cookie:0x0
>     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
>     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
>     ath12k_pci 0001:01:00.0: soc reset cause:0
>     ath12k_pci 0001:01:00.0: MHISTATUS 0xff04
>     ath12k_pci 0001:01:00.0: pci link_ctl 0x0000 L0s 0 L1 0
>     ath12k_pci 0001:01:00.0: pci reg 0x3164 instance 0x11 read val 0x11
>     ath12k_pci 0001:01:00.0: setting mhi state: INIT(0)
>     ath12k_pci 0001:01:00.0: setting mhi state: POWER_ON(2)
>     ath12k_pci 0001:01:00.0: mhi notify status reason UNKNOWN
>     ath12k_pci 0001:01:00.0: mhi notify status reason MHI_CB_EE_MISSION_MODE
>     ath12k_pci 0001:01:00.0: qmi wifi fw qmi service connected
>     ath12k_pci 0001:01:00.0: phy capability resp valid 1 num_phy 2 valid 1 board_id 84934656 valid 1 single_chip_mlo_support 0
>     ath12k_pci 0001:01:00.0: intra device MLO is disabled hence skip QMI MLO cap
> 
> Alexander Wilhelm (11):
>   wifi: ath12k: fix endianness handling in QMI host capability request
>   wifi: ath12k: fix endianness handling in QMI phy capability response
>   wifi: ath12k: fix endianness handling in QMI firmware indication
>   wifi: ath12k: fix endianness handling in QMI firmware memory indication
>   wifi: ath12k: fix endianness handling in QMI respond firmware memory
>   wifi: ath12k: fix endianness handling in QMI firmware capabilities
>   wifi: ath12k: fix endianness handling in QMI bdf download
>   wifi: ath12k: fix endianness handling in QMI firmware m3 info
>   wifi: ath12k: fix endianness handling in QMI firmware wlan mode
>   wifi: ath12k: fix endianness handling in QMI wlan configuration
>   wifi: ath12k: fix endianness handling in QMI response
> 
>  drivers/net/wireless/ath/ath12k/qmi.c | 149 ++++++++++++++------------
>  drivers/net/wireless/ath/ath12k/qmi.h | 106 +++++++++---------
>  include/linux/soc/qcom/qmi.h          |   4 +-
>  3 files changed, 136 insertions(+), 123 deletions(-)
> 

Frankly I'm shocked that the low-level QMI encode/decode is not doing the
endian conversion. Since the Qualcomm internal tool that generates the data
structures has always generated structs with cpu endianess (i.e. u8, u16, u32,
etc) I just assumed that endian conversion was handled at a low level.

So should this issue be pushed down to the QMI encode/decode rather than foist
it upon every client's read & write?

/jeff

Re: [PATCH 11/11] wifi: ath12k: fix endianness handling in QMI response

[kernel test robot]

Hi Alexander,

kernel test robot noticed the following build warnings:

[auto build test WARNING on ath/ath-next]
[also build test WARNING on linus/master v6.16-rc6 next-20250716]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Alexander-Wilhelm/wifi-ath12k-fix-endianness-handling-in-QMI-host-capability-request/20250716-162058
base:   https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git ath-next
patch link:    https://lore.kernel.org/r/20250716075100.1447352-12-alexander.wilhelm%40westermo.com
patch subject: [PATCH 11/11] wifi: ath12k: fix endianness handling in QMI response
config: mips-randconfig-r123-20250717 (https://download.01.org/0day-ci/archive/20250717/202507171640.30pUvpPv-lkp@intel.com/config)
compiler: mips-linux-gcc (GCC) 8.5.0
reproduce: (https://download.01.org/0day-ci/archive/20250717/202507171640.30pUvpPv-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202507171640.30pUvpPv-lkp@intel.com/

sparse warnings: (new ones prefixed by >>)
>> drivers/soc/qcom/pdr_interface.c:612:22: sparse: sparse: restricted __le16 degrades to integer
   drivers/soc/qcom/pdr_interface.c:613:22: sparse: sparse: restricted __le16 degrades to integer

vim +612 drivers/soc/qcom/pdr_interface.c

fbe639b44a8275 Sibi Sankar           2020-03-12  549  
fbe639b44a8275 Sibi Sankar           2020-03-12  550  /**
fbe639b44a8275 Sibi Sankar           2020-03-12  551   * pdr_restart_pd() - restart PD
fbe639b44a8275 Sibi Sankar           2020-03-12  552   * @pdr:	PDR client handle
fbe639b44a8275 Sibi Sankar           2020-03-12  553   * @pds:	PD service handle
fbe639b44a8275 Sibi Sankar           2020-03-12  554   *
fbe639b44a8275 Sibi Sankar           2020-03-12  555   * Restarts the PD tracked by the PDR client handle for a given service path.
fbe639b44a8275 Sibi Sankar           2020-03-12  556   *
fbe639b44a8275 Sibi Sankar           2020-03-12  557   * Return: 0 on success, negative errno on failure.
fbe639b44a8275 Sibi Sankar           2020-03-12  558   */
fbe639b44a8275 Sibi Sankar           2020-03-12  559  int pdr_restart_pd(struct pdr_handle *pdr, struct pdr_service *pds)
fbe639b44a8275 Sibi Sankar           2020-03-12  560  {
fbe639b44a8275 Sibi Sankar           2020-03-12  561  	struct servreg_restart_pd_resp resp;
a161ffe4b87772 Tom Rix               2020-08-19  562  	struct servreg_restart_pd_req req = { 0 };
fbe639b44a8275 Sibi Sankar           2020-03-12  563  	struct sockaddr_qrtr addr;
fbe639b44a8275 Sibi Sankar           2020-03-12  564  	struct pdr_service *tmp;
fbe639b44a8275 Sibi Sankar           2020-03-12  565  	struct qmi_txn txn;
fbe639b44a8275 Sibi Sankar           2020-03-12  566  	int ret;
fbe639b44a8275 Sibi Sankar           2020-03-12  567  
fbe639b44a8275 Sibi Sankar           2020-03-12  568  	if (IS_ERR_OR_NULL(pdr) || IS_ERR_OR_NULL(pds))
fbe639b44a8275 Sibi Sankar           2020-03-12  569  		return -EINVAL;
fbe639b44a8275 Sibi Sankar           2020-03-12  570  
fbe639b44a8275 Sibi Sankar           2020-03-12  571  	mutex_lock(&pdr->list_lock);
fbe639b44a8275 Sibi Sankar           2020-03-12  572  	list_for_each_entry(tmp, &pdr->lookups, node) {
fbe639b44a8275 Sibi Sankar           2020-03-12  573  		if (tmp != pds)
fbe639b44a8275 Sibi Sankar           2020-03-12  574  			continue;
fbe639b44a8275 Sibi Sankar           2020-03-12  575  
fbe639b44a8275 Sibi Sankar           2020-03-12  576  		if (!pds->service_connected)
fbe639b44a8275 Sibi Sankar           2020-03-12  577  			break;
fbe639b44a8275 Sibi Sankar           2020-03-12  578  
fbe639b44a8275 Sibi Sankar           2020-03-12  579  		/* Prepare req message */
26bc7a6a0beed8 Len Baker             2021-08-08  580  		strscpy(req.service_path, pds->service_path, sizeof(req.service_path));
fbe639b44a8275 Sibi Sankar           2020-03-12  581  		addr = pds->addr;
fbe639b44a8275 Sibi Sankar           2020-03-12  582  		break;
fbe639b44a8275 Sibi Sankar           2020-03-12  583  	}
fbe639b44a8275 Sibi Sankar           2020-03-12  584  	mutex_unlock(&pdr->list_lock);
fbe639b44a8275 Sibi Sankar           2020-03-12  585  
fbe639b44a8275 Sibi Sankar           2020-03-12  586  	if (!req.service_path[0])
fbe639b44a8275 Sibi Sankar           2020-03-12  587  		return -EINVAL;
fbe639b44a8275 Sibi Sankar           2020-03-12  588  
fbe639b44a8275 Sibi Sankar           2020-03-12  589  	ret = qmi_txn_init(&pdr->notifier_hdl, &txn,
fbe639b44a8275 Sibi Sankar           2020-03-12  590  			   servreg_restart_pd_resp_ei,
fbe639b44a8275 Sibi Sankar           2020-03-12  591  			   &resp);
fbe639b44a8275 Sibi Sankar           2020-03-12  592  	if (ret < 0)
fbe639b44a8275 Sibi Sankar           2020-03-12  593  		return ret;
fbe639b44a8275 Sibi Sankar           2020-03-12  594  
fbe639b44a8275 Sibi Sankar           2020-03-12  595  	ret = qmi_send_request(&pdr->notifier_hdl, &addr,
fbe639b44a8275 Sibi Sankar           2020-03-12  596  			       &txn, SERVREG_RESTART_PD_REQ,
fbe639b44a8275 Sibi Sankar           2020-03-12  597  			       SERVREG_RESTART_PD_REQ_MAX_LEN,
fbe639b44a8275 Sibi Sankar           2020-03-12  598  			       servreg_restart_pd_req_ei, &req);
fbe639b44a8275 Sibi Sankar           2020-03-12  599  	if (ret < 0) {
fbe639b44a8275 Sibi Sankar           2020-03-12  600  		qmi_txn_cancel(&txn);
fbe639b44a8275 Sibi Sankar           2020-03-12  601  		return ret;
fbe639b44a8275 Sibi Sankar           2020-03-12  602  	}
fbe639b44a8275 Sibi Sankar           2020-03-12  603  
fbe639b44a8275 Sibi Sankar           2020-03-12  604  	ret = qmi_txn_wait(&txn, 5 * HZ);
fbe639b44a8275 Sibi Sankar           2020-03-12  605  	if (ret < 0) {
fbe639b44a8275 Sibi Sankar           2020-03-12  606  		pr_err("PDR: %s PD restart txn wait failed: %d\n",
fbe639b44a8275 Sibi Sankar           2020-03-12  607  		       req.service_path, ret);
fbe639b44a8275 Sibi Sankar           2020-03-12  608  		return ret;
fbe639b44a8275 Sibi Sankar           2020-03-12  609  	}
fbe639b44a8275 Sibi Sankar           2020-03-12  610  
fbe639b44a8275 Sibi Sankar           2020-03-12  611  	/* Check response if PDR is disabled */
fbe639b44a8275 Sibi Sankar           2020-03-12 @612  	if (resp.resp.result == QMI_RESULT_FAILURE_V01 &&
fbe639b44a8275 Sibi Sankar           2020-03-12  613  	    resp.resp.error == QMI_ERR_DISABLED_V01) {
fbe639b44a8275 Sibi Sankar           2020-03-12  614  		pr_err("PDR: %s PD restart is disabled: 0x%x\n",
fbe639b44a8275 Sibi Sankar           2020-03-12  615  		       req.service_path, resp.resp.error);
fbe639b44a8275 Sibi Sankar           2020-03-12  616  		return -EOPNOTSUPP;
fbe639b44a8275 Sibi Sankar           2020-03-12  617  	}
fbe639b44a8275 Sibi Sankar           2020-03-12  618  
fbe639b44a8275 Sibi Sankar           2020-03-12  619  	/* Check the response for other error case*/
fbe639b44a8275 Sibi Sankar           2020-03-12  620  	if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
fbe639b44a8275 Sibi Sankar           2020-03-12  621  		pr_err("PDR: %s request for PD restart failed: 0x%x\n",
fbe639b44a8275 Sibi Sankar           2020-03-12  622  		       req.service_path, resp.resp.error);
fbe639b44a8275 Sibi Sankar           2020-03-12  623  		return -EREMOTEIO;
fbe639b44a8275 Sibi Sankar           2020-03-12  624  	}
fbe639b44a8275 Sibi Sankar           2020-03-12  625  
fbe639b44a8275 Sibi Sankar           2020-03-12  626  	return 0;
fbe639b44a8275 Sibi Sankar           2020-03-12  627  }
9b09c0f289c5a8 Unnathi Chalicheemala 2023-09-22  628  EXPORT_SYMBOL_GPL(pdr_restart_pd);
fbe639b44a8275 Sibi Sankar           2020-03-12  629  

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH 07/11] wifi: ath12k: fix endianness handling in QMI bdf download

[kernel test robot]

Hi Alexander,

kernel test robot noticed the following build warnings:

[auto build test WARNING on ath/ath-next]
[also build test WARNING on linus/master v6.16-rc6 next-20250717]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Alexander-Wilhelm/wifi-ath12k-fix-endianness-handling-in-QMI-host-capability-request/20250716-162058
base:   https://git.kernel.org/pub/scm/linux/kernel/git/ath/ath.git ath-next
patch link:    https://lore.kernel.org/r/20250716075100.1447352-8-alexander.wilhelm%40westermo.com
patch subject: [PATCH 07/11] wifi: ath12k: fix endianness handling in QMI bdf download
config: alpha-randconfig-r122-20250717 (https://download.01.org/0day-ci/archive/20250717/202507172333.LZcmFwWs-lkp@intel.com/config)
compiler: alpha-linux-gcc (GCC) 8.5.0
reproduce: (https://download.01.org/0day-ci/archive/20250717/202507172333.LZcmFwWs-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202507172333.LZcmFwWs-lkp@intel.com/

sparse warnings: (new ones prefixed by >>)
>> drivers/net/wireless/ath/ath12k/qmi.c:2996:37: sparse: sparse: bad assignment (+=) to restricted __le32

vim +2996 drivers/net/wireless/ath/ath12k/qmi.c

  2919	
  2920	static int ath12k_qmi_load_file_target_mem(struct ath12k_base *ab,
  2921						   const u8 *data, u32 len, u8 type)
  2922	{
  2923		struct qmi_wlanfw_bdf_download_req_msg_v01 *req;
  2924		struct qmi_wlanfw_bdf_download_resp_msg_v01 resp = {};
  2925		struct qmi_txn txn;
  2926		const u8 *temp = data;
  2927		int ret = 0;
  2928		u32 remaining = len;
  2929	
  2930		req = kzalloc(sizeof(*req), GFP_KERNEL);
  2931		if (!req)
  2932			return -ENOMEM;
  2933	
  2934		while (remaining) {
  2935			req->valid = 1;
  2936			req->file_id_valid = 1;
  2937			req->file_id = cpu_to_le32(ab->qmi.target.board_id);
  2938			req->total_size_valid = 1;
  2939			req->total_size = cpu_to_le32(remaining);
  2940			req->seg_id_valid = 1;
  2941			req->data_valid = 1;
  2942			req->bdf_type = type;
  2943			req->bdf_type_valid = 1;
  2944			req->end_valid = 1;
  2945			req->end = 0;
  2946	
  2947			if (remaining > QMI_WLANFW_MAX_DATA_SIZE_V01) {
  2948				req->data_len = cpu_to_le32(QMI_WLANFW_MAX_DATA_SIZE_V01);
  2949			} else {
  2950				req->data_len = cpu_to_le32(remaining);
  2951				req->end = 1;
  2952			}
  2953	
  2954			if (type == ATH12K_QMI_FILE_TYPE_EEPROM) {
  2955				req->data_valid = 0;
  2956				req->end = 1;
  2957				req->data_len = cpu_to_le32(ATH12K_QMI_MAX_BDF_FILE_NAME_SIZE);
  2958			} else {
  2959				memcpy(req->data, temp, le32_to_cpu(req->data_len));
  2960			}
  2961	
  2962			ret = qmi_txn_init(&ab->qmi.handle, &txn,
  2963					   qmi_wlanfw_bdf_download_resp_msg_v01_ei,
  2964					   &resp);
  2965			if (ret < 0)
  2966				goto out;
  2967	
  2968			ath12k_dbg(ab, ATH12K_DBG_QMI, "qmi bdf download req fixed addr type %d\n",
  2969				   type);
  2970	
  2971			ret = qmi_send_request(&ab->qmi.handle, NULL, &txn,
  2972					       QMI_WLANFW_BDF_DOWNLOAD_REQ_V01,
  2973					       QMI_WLANFW_BDF_DOWNLOAD_REQ_MSG_V01_MAX_LEN,
  2974					       qmi_wlanfw_bdf_download_req_msg_v01_ei, req);
  2975			if (ret < 0) {
  2976				qmi_txn_cancel(&txn);
  2977				goto out;
  2978			}
  2979	
  2980			ret = qmi_txn_wait(&txn, msecs_to_jiffies(ATH12K_QMI_WLANFW_TIMEOUT_MS));
  2981			if (ret < 0)
  2982				goto out;
  2983	
  2984			if (resp.resp.result != QMI_RESULT_SUCCESS_V01) {
  2985				ath12k_warn(ab, "qmi BDF download failed, result: %d, err: %d\n",
  2986					    resp.resp.result, resp.resp.error);
  2987				ret = -EINVAL;
  2988				goto out;
  2989			}
  2990	
  2991			if (type == ATH12K_QMI_FILE_TYPE_EEPROM) {
  2992				remaining = 0;
  2993			} else {
  2994				remaining -= le32_to_cpu(req->data_len);
  2995				temp += le32_to_cpu(req->data_len);
> 2996				req->seg_id += cpu_to_le32(1);
  2997				ath12k_dbg(ab, ATH12K_DBG_QMI,
  2998					   "qmi bdf download request remaining %i\n",
  2999					   remaining);
  3000			}
  3001		}
  3002	
  3003	out:
  3004		kfree(req);
  3005		return ret;
  3006	}
  3007	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH 00/11] wifi: ath12k: Fix endianness handling in QMI

[Bjorn Andersson]

On Wed, Jul 16, 2025 at 08:13:20AM -0700, Jeff Johnson wrote:
> On 7/16/2025 12:50 AM, Alexander Wilhelm wrote:
> > Fix endianness handling in QMI firmware transfer on big-endian
> > platforms. Without this fix, the firmware download fails due to
> > misinterpreted data structures exchanged between the host and the
> > wireless module.
> > 
> > The issue occurs during early bring-up on big endian systems, where QMI
> > messages are not correctly parsed by the driver, leading to failed
> > initialization sequences. Ensure all relevant fields are properly
> > converted between CPU and little-endian format in request and response
> > messages, as expected by the firmware. Attached logs showing the failure
> > before the fix:
> > 
> >     ath12k_pci 0001:01:00.0: BAR 0: assigned [mem 0xc00000000-0xc001fffff 64bit]
> >     ath12k_pci 0001:01:00.0: boot pci_mem 0xcd4148e9
> >     ath12k_pci 0001:01:00.0: pci probe 17cb:1109 17cb:1109
> >     ath12k_pci 0001:01:00.0: pci tcsr_soc_hw_version major 2 minor 0
> >     ath12k_pci 0001:01:00.0: request MSI one vector
> >     ath12k_pci 0001:01:00.0: MSI vectors: 1
> >     ath12k_pci 0001:01:00.0: msi base data is 0
> >     ath12k_pci 0001:01:00.0: Hardware name: qcn9274 hw2.0
> >     ath12k_pci 0001:01:00.0: boot firmware request ath12k/QCN9274/hw2.0/firmware-2.bin size 15134776
> >     ath12k_pci 0001:01:00.0: found fw timestamp 1722934582
> >     ath12k_pci 0001:01:00.0: found m3 image ie (421880 B)
> >     ath12k_pci 0001:01:00.0: found fw image ie (7229440 B)
> >     ath12k_pci 0001:01:00.0: found dualmac fw image ie (7483392 B)
> >     ath12k_pci 0001:01:00.0: found firmware features ie (1 B)
> >     ath12k_pci 0001:01:00.0: features
> >     ath12k_pci 0001:01:00.0: using fw api 2
> >     ath12k_pci 0001:01:00.0: dualmac fw selected for board id: 1005
> >     ath12k_pci 0001:01:00.0: Assign MSI to user: MHI, num_vectors: 3, user_base_data: 0, base_vector: 0
> >     ath12k_pci 0001:01:00.0: Number of assigned MSI for MHI is 3, base vector is 0
> >     ath12k_pci 0001:01:00.0: Assign MSI to user: CE, num_vectors: 1, user_base_data: 0, base_vector: 0
> >     ath12k_pci 0001:01:00.0: Assign MSI to user: DP, num_vectors: 1, user_base_data: 0, base_vector: 0
> >     ath12k_pci 0001:01:00.0: irq:18 group:0
> >     ath12k_pci 0001:01:00.0: irq:18 group:1
> >     ath12k_pci 0001:01:00.0: irq:18 group:2
> >     ath12k_pci 0001:01:00.0: irq:18 group:3
> >     ath12k_pci 0001:01:00.0: irq:18 group:4
> >     ath12k_pci 0001:01:00.0: irq:18 group:5
> >     ath12k_pci 0001:01:00.0: irq:18 group:6
> >     ath12k_pci 0001:01:00.0: irq:18 group:7
> >     ath12k_pci 0001:01:00.0: pci after request_irq msi_ep_base_data 0
> >     ath12k_pci 0001:01:00.0: cookie:0x0
> >     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
> >     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
> >     ath12k_pci 0001:01:00.0: soc reset cause:0
> >     ath12k_pci 0001:01:00.0: MHISTATUS 0xff04
> >     ath12k_pci 0001:01:00.0: pci link_ctl 0x0000 L0s 0 L1 0
> >     ath12k_pci 0001:01:00.0: pci reg 0x3164 instance 0x11 read val 0x11
> >     ath12k_pci 0001:01:00.0: setting mhi state: INIT(0)
> >     ath12k_pci 0001:01:00.0: setting mhi state: POWER_ON(2)
> >     ath12k_pci 0001:01:00.0: mhi notify status reason UNKNOWN
> >     ath12k_pci 0001:01:00.0: mhi notify status reason MHI_CB_EE_MISSION_MODE
> >     ath12k_pci 0001:01:00.0: qmi wifi fw qmi service connected
> >     ath12k_pci 0001:01:00.0: phy capability resp valid 1 num_phy 2 valid 1 board_id 84934656 valid 1 single_chip_mlo_support 0
> >     ath12k_pci 0001:01:00.0: intra device MLO is disabled hence skip QMI MLO cap
> > 
> > Alexander Wilhelm (11):
> >   wifi: ath12k: fix endianness handling in QMI host capability request
> >   wifi: ath12k: fix endianness handling in QMI phy capability response
> >   wifi: ath12k: fix endianness handling in QMI firmware indication
> >   wifi: ath12k: fix endianness handling in QMI firmware memory indication
> >   wifi: ath12k: fix endianness handling in QMI respond firmware memory
> >   wifi: ath12k: fix endianness handling in QMI firmware capabilities
> >   wifi: ath12k: fix endianness handling in QMI bdf download
> >   wifi: ath12k: fix endianness handling in QMI firmware m3 info
> >   wifi: ath12k: fix endianness handling in QMI firmware wlan mode
> >   wifi: ath12k: fix endianness handling in QMI wlan configuration
> >   wifi: ath12k: fix endianness handling in QMI response
> > 
> >  drivers/net/wireless/ath/ath12k/qmi.c | 149 ++++++++++++++------------
> >  drivers/net/wireless/ath/ath12k/qmi.h | 106 +++++++++---------
> >  include/linux/soc/qcom/qmi.h          |   4 +-
> >  3 files changed, 136 insertions(+), 123 deletions(-)
> > 
> 
> Frankly I'm shocked that the low-level QMI encode/decode is not doing the
> endian conversion. Since the Qualcomm internal tool that generates the data
> structures has always generated structs with cpu endianess (i.e. u8, u16, u32,
> etc) I just assumed that endian conversion was handled at a low level.
> 

I'm suspecting that those tools, just like this implementation, is
exclusively tested on little endian machines...

> So should this issue be pushed down to the QMI encode/decode rather than foist
> it upon every client's read & write?
> 

It's been a while since I looked at the implementation, but conceptually
I'm in favor of this.

Regards,
Bjorn

> /jeff

Re: [PATCH 00/11] wifi: ath12k: Fix endianness handling in QMI

[Alexander Wilhelm]

Am Sat, Jul 19, 2025 at 11:24:51PM -0500 schrieb Bjorn Andersson:
> On Wed, Jul 16, 2025 at 08:13:20AM -0700, Jeff Johnson wrote:
> > On 7/16/2025 12:50 AM, Alexander Wilhelm wrote:
> > > Fix endianness handling in QMI firmware transfer on big-endian
> > > platforms. Without this fix, the firmware download fails due to
> > > misinterpreted data structures exchanged between the host and the
> > > wireless module.
> > > 
> > > The issue occurs during early bring-up on big endian systems, where QMI
> > > messages are not correctly parsed by the driver, leading to failed
> > > initialization sequences. Ensure all relevant fields are properly
> > > converted between CPU and little-endian format in request and response
> > > messages, as expected by the firmware. Attached logs showing the failure
> > > before the fix:
> > > 
> > >     ath12k_pci 0001:01:00.0: BAR 0: assigned [mem 0xc00000000-0xc001fffff 64bit]
> > >     ath12k_pci 0001:01:00.0: boot pci_mem 0xcd4148e9
> > >     ath12k_pci 0001:01:00.0: pci probe 17cb:1109 17cb:1109
> > >     ath12k_pci 0001:01:00.0: pci tcsr_soc_hw_version major 2 minor 0
> > >     ath12k_pci 0001:01:00.0: request MSI one vector
> > >     ath12k_pci 0001:01:00.0: MSI vectors: 1
> > >     ath12k_pci 0001:01:00.0: msi base data is 0
> > >     ath12k_pci 0001:01:00.0: Hardware name: qcn9274 hw2.0
> > >     ath12k_pci 0001:01:00.0: boot firmware request ath12k/QCN9274/hw2.0/firmware-2.bin size 15134776
> > >     ath12k_pci 0001:01:00.0: found fw timestamp 1722934582
> > >     ath12k_pci 0001:01:00.0: found m3 image ie (421880 B)
> > >     ath12k_pci 0001:01:00.0: found fw image ie (7229440 B)
> > >     ath12k_pci 0001:01:00.0: found dualmac fw image ie (7483392 B)
> > >     ath12k_pci 0001:01:00.0: found firmware features ie (1 B)
> > >     ath12k_pci 0001:01:00.0: features
> > >     ath12k_pci 0001:01:00.0: using fw api 2
> > >     ath12k_pci 0001:01:00.0: dualmac fw selected for board id: 1005
> > >     ath12k_pci 0001:01:00.0: Assign MSI to user: MHI, num_vectors: 3, user_base_data: 0, base_vector: 0
> > >     ath12k_pci 0001:01:00.0: Number of assigned MSI for MHI is 3, base vector is 0
> > >     ath12k_pci 0001:01:00.0: Assign MSI to user: CE, num_vectors: 1, user_base_data: 0, base_vector: 0
> > >     ath12k_pci 0001:01:00.0: Assign MSI to user: DP, num_vectors: 1, user_base_data: 0, base_vector: 0
> > >     ath12k_pci 0001:01:00.0: irq:18 group:0
> > >     ath12k_pci 0001:01:00.0: irq:18 group:1
> > >     ath12k_pci 0001:01:00.0: irq:18 group:2
> > >     ath12k_pci 0001:01:00.0: irq:18 group:3
> > >     ath12k_pci 0001:01:00.0: irq:18 group:4
> > >     ath12k_pci 0001:01:00.0: irq:18 group:5
> > >     ath12k_pci 0001:01:00.0: irq:18 group:6
> > >     ath12k_pci 0001:01:00.0: irq:18 group:7
> > >     ath12k_pci 0001:01:00.0: pci after request_irq msi_ep_base_data 0
> > >     ath12k_pci 0001:01:00.0: cookie:0x0
> > >     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
> > >     ath12k_pci 0001:01:00.0: WLAON_WARM_SW_ENTRY 0x2
> > >     ath12k_pci 0001:01:00.0: soc reset cause:0
> > >     ath12k_pci 0001:01:00.0: MHISTATUS 0xff04
> > >     ath12k_pci 0001:01:00.0: pci link_ctl 0x0000 L0s 0 L1 0
> > >     ath12k_pci 0001:01:00.0: pci reg 0x3164 instance 0x11 read val 0x11
> > >     ath12k_pci 0001:01:00.0: setting mhi state: INIT(0)
> > >     ath12k_pci 0001:01:00.0: setting mhi state: POWER_ON(2)
> > >     ath12k_pci 0001:01:00.0: mhi notify status reason UNKNOWN
> > >     ath12k_pci 0001:01:00.0: mhi notify status reason MHI_CB_EE_MISSION_MODE
> > >     ath12k_pci 0001:01:00.0: qmi wifi fw qmi service connected
> > >     ath12k_pci 0001:01:00.0: phy capability resp valid 1 num_phy 2 valid 1 board_id 84934656 valid 1 single_chip_mlo_support 0
> > >     ath12k_pci 0001:01:00.0: intra device MLO is disabled hence skip QMI MLO cap
> > > 
> > > Alexander Wilhelm (11):
> > >   wifi: ath12k: fix endianness handling in QMI host capability request
> > >   wifi: ath12k: fix endianness handling in QMI phy capability response
> > >   wifi: ath12k: fix endianness handling in QMI firmware indication
> > >   wifi: ath12k: fix endianness handling in QMI firmware memory indication
> > >   wifi: ath12k: fix endianness handling in QMI respond firmware memory
> > >   wifi: ath12k: fix endianness handling in QMI firmware capabilities
> > >   wifi: ath12k: fix endianness handling in QMI bdf download
> > >   wifi: ath12k: fix endianness handling in QMI firmware m3 info
> > >   wifi: ath12k: fix endianness handling in QMI firmware wlan mode
> > >   wifi: ath12k: fix endianness handling in QMI wlan configuration
> > >   wifi: ath12k: fix endianness handling in QMI response
> > > 
> > >  drivers/net/wireless/ath/ath12k/qmi.c | 149 ++++++++++++++------------
> > >  drivers/net/wireless/ath/ath12k/qmi.h | 106 +++++++++---------
> > >  include/linux/soc/qcom/qmi.h          |   4 +-
> > >  3 files changed, 136 insertions(+), 123 deletions(-)
> > > 
> > 
> > Frankly I'm shocked that the low-level QMI encode/decode is not doing the
> > endian conversion. Since the Qualcomm internal tool that generates the data
> > structures has always generated structs with cpu endianess (i.e. u8, u16, u32,
> > etc) I just assumed that endian conversion was handled at a low level.
> > 
> 
> I'm suspecting that those tools, just like this implementation, is
> exclusively tested on little endian machines...
> 
> > So should this issue be pushed down to the QMI encode/decode rather than foist
> > it upon every client's read & write?
> > 
> 
> It's been a while since I looked at the implementation, but conceptually
> I'm in favor of this.

I could certainly implement the endianness fixes in the QMI code. However, I’m
concerned that other drivers might depend on the current behavior. These changes
could potentially introduce regressions, which I would want to avoid.
Additionally, I’m already seeing some Sparse warnings, and I’m not entirely sure
how to handle them properly at this point.


Best regards
Alexander Wilhelm