[PATCH 09/10] habanalabs: add new opcodes for INFO IOCTL

From: Oded Gabbay
Date: Sun Nov 07 2021 - 15:52:12 EST

Next message: Kurt Garloff: "igb in 5.15: autosuspend hangs netlink"
Previous message: Oded Gabbay: "[PATCH 10/10] habanalabs: make hdev creation code more readable"
In reply to: Oded Gabbay: "[PATCH 10/10] habanalabs: make hdev creation code more readable"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

From: farah kassabri <fkassabri@xxxxxxxxx>

Add implementation for new opcodes in the INFO IOCTL:
1. Retrieve the replaced DRAM rows from f/w.
2. Retrieve the pending DRAM rows from f/w.

Signed-off-by: farah kassabri <fkassabri@xxxxxxxxx>
Reviewed-by: Oded Gabbay <ogabbay@xxxxxxxxxx>
Signed-off-by: Oded Gabbay <ogabbay@xxxxxxxxxx>
---
drivers/misc/habanalabs/common/firmware_if.c | 66 +++++++++++++++++++
drivers/misc/habanalabs/common/habanalabs.h | 3 +
.../misc/habanalabs/common/habanalabs_ioctl.c | 43 ++++++++++++
.../misc/habanalabs/include/common/cpucp_if.h | 33 +++++++++-
include/uapi/misc/habanalabs.h | 4 ++
5 files changed, 148 insertions(+), 1 deletion(-)

diff --git a/drivers/misc/habanalabs/common/firmware_if.c b/drivers/misc/habanalabs/common/firmware_if.c
index 9addcfba6a8b..70e992bdbde7 100644
--- a/drivers/misc/habanalabs/common/firmware_if.c
+++ b/drivers/misc/habanalabs/common/firmware_if.c
@@ -972,6 +972,72 @@ int hl_fw_cpucp_power_get(struct hl_device *hdev, u64 *power)
return rc;
}

+int hl_fw_dram_replaced_row_get(struct hl_device *hdev,
+ struct cpucp_hbm_row_info *info)
+{
+ struct cpucp_hbm_row_info *cpucp_repl_rows_info_cpu_addr;
+ dma_addr_t cpucp_repl_rows_info_dma_addr;
+ struct cpucp_packet pkt = {};
+ u64 result;
+ int rc;
+
+ cpucp_repl_rows_info_cpu_addr =
+ hdev->asic_funcs->cpu_accessible_dma_pool_alloc(hdev,
+ sizeof(struct cpucp_hbm_row_info),
+ &cpucp_repl_rows_info_dma_addr);
+ if (!cpucp_repl_rows_info_cpu_addr) {
+ dev_err(hdev->dev,
+ "Failed to allocate DMA memory for CPU-CP replaced rows info packet\n");
+ return -ENOMEM;
+ }
+
+ memset(cpucp_repl_rows_info_cpu_addr, 0, sizeof(struct cpucp_hbm_row_info));
+
+ pkt.ctl = cpu_to_le32(CPUCP_PACKET_HBM_REPLACED_ROWS_INFO_GET <<
+ CPUCP_PKT_CTL_OPCODE_SHIFT);
+ pkt.addr = cpu_to_le64(cpucp_repl_rows_info_dma_addr);
+ pkt.data_max_size = cpu_to_le32(sizeof(struct cpucp_hbm_row_info));
+
+ rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
+ HL_CPUCP_INFO_TIMEOUT_USEC, &result);
+ if (rc) {
+ dev_err(hdev->dev,
+ "Failed to handle CPU-CP replaced rows info pkt, error %d\n", rc);
+ goto out;
+ }
+
+ memcpy(info, cpucp_repl_rows_info_cpu_addr, sizeof(*info));
+
+out:
+ hdev->asic_funcs->cpu_accessible_dma_pool_free(hdev,
+ sizeof(struct cpucp_hbm_row_info),
+ cpucp_repl_rows_info_cpu_addr);
+
+ return rc;
+}
+
+int hl_fw_dram_pending_row_get(struct hl_device *hdev, u32 *pend_rows_num)
+{
+ struct cpucp_packet pkt;
+ u64 result;
+ int rc;
+
+ memset(&pkt, 0, sizeof(pkt));
+
+ pkt.ctl = cpu_to_le32(CPUCP_PACKET_HBM_PENDING_ROWS_STATUS << CPUCP_PKT_CTL_OPCODE_SHIFT);
+
+ rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt), 0, &result);
+ if (rc) {
+ dev_err(hdev->dev,
+ "Failed to handle CPU-CP pending rows info pkt, error %d\n", rc);
+ goto out;
+ }
+
+ *pend_rows_num = (u32) result;
+out:
+ return rc;
+}
+
void hl_fw_ask_hard_reset_without_linux(struct hl_device *hdev)
{
struct static_fw_load_mgr *static_loader =
diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index fc201537f7a9..a19563c416ac 100644
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -3012,6 +3012,9 @@ int hl_fw_dynamic_send_protocol_cmd(struct hl_device *hdev,
struct fw_load_mgr *fw_loader,
enum comms_cmd cmd, unsigned int size,
bool wait_ok, u32 timeout);
+int hl_fw_dram_replaced_row_get(struct hl_device *hdev,
+ struct cpucp_hbm_row_info *info);
+int hl_fw_dram_pending_row_get(struct hl_device *hdev, u32 *pend_rows_num);
int hl_pci_bars_map(struct hl_device *hdev, const char * const name[3],
bool is_wc[3]);
int hl_pci_elbi_read(struct hl_device *hdev, u64 addr, u32 *data);
diff --git a/drivers/misc/habanalabs/common/habanalabs_ioctl.c b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
index 19726c6b642a..68c655acdec8 100644
--- a/drivers/misc/habanalabs/common/habanalabs_ioctl.c
+++ b/drivers/misc/habanalabs/common/habanalabs_ioctl.c
@@ -503,6 +503,43 @@ static int open_stats_info(struct hl_fpriv *hpriv, struct hl_info_args *args)
min((size_t) max_size, sizeof(open_stats_info))) ? -EFAULT : 0;
}

+static int dram_pending_rows_info(struct hl_fpriv *hpriv, struct hl_info_args *args)
+{
+ struct hl_device *hdev = hpriv->hdev;
+ u32 max_size = args->return_size;
+ u32 pend_rows_num = 0;
+ void __user *out = (void __user *) (uintptr_t) args->return_pointer;
+ int rc;
+
+ if ((!max_size) || (!out))
+ return -EINVAL;
+
+ rc = hl_fw_dram_pending_row_get(hdev, &pend_rows_num);
+ if (rc)
+ return rc;
+
+ return copy_to_user(out, &pend_rows_num,
+ min_t(size_t, max_size, sizeof(pend_rows_num))) ? -EFAULT : 0;
+}
+
+static int dram_replaced_rows_info(struct hl_fpriv *hpriv, struct hl_info_args *args)
+{
+ struct hl_device *hdev = hpriv->hdev;
+ u32 max_size = args->return_size;
+ struct cpucp_hbm_row_info info = {0};
+ void __user *out = (void __user *) (uintptr_t) args->return_pointer;
+ int rc;
+
+ if ((!max_size) || (!out))
+ return -EINVAL;
+
+ rc = hl_fw_dram_replaced_row_get(hdev, &info);
+ if (rc)
+ return rc;
+
+ return copy_to_user(out, &info, min_t(size_t, max_size, sizeof(info))) ? -EFAULT : 0;
+}
+
static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
struct device *dev)
{
@@ -589,6 +626,12 @@ static int _hl_info_ioctl(struct hl_fpriv *hpriv, void *data,
case HL_INFO_OPEN_STATS:
return open_stats_info(hpriv, args);

+ case HL_INFO_DRAM_REPLACED_ROWS:
+ return dram_replaced_rows_info(hpriv, args);
+
+ case HL_INFO_DRAM_PENDING_ROWS:
+ return dram_pending_rows_info(hpriv, args);
+
default:
dev_err(dev, "Invalid request %d\n", args->op);
rc = -ENOTTY;
diff --git a/drivers/misc/habanalabs/include/common/cpucp_if.h b/drivers/misc/habanalabs/include/common/cpucp_if.h
index 17927968e19a..5e19c763f3f0 100644
--- a/drivers/misc/habanalabs/include/common/cpucp_if.h
+++ b/drivers/misc/habanalabs/include/common/cpucp_if.h
@@ -1,6 +1,6 @@
/* SPDX-License-Identifier: GPL-2.0
*
- * Copyright 2020 HabanaLabs, Ltd.
+ * Copyright 2021 HabanaLabs, Ltd.
* All Rights Reserved.
*
*/
@@ -377,6 +377,13 @@ enum pq_init_status {
* a different engine or QMAN according to enum cpucp_idle_mask.
* The bit will be 1 if the engine is NOT idle.
*
+ * CPUCP_PACKET_HBM_REPLACED_ROWS_INFO_GET -
+ * Fetch all HBM replaced-rows and prending to be replaced rows data.
+ *
+ * CPUCP_PACKET_HBM_PENDING_ROWS_STATUS -
+ * Fetch status of HBM rows pending replacement and need a reboot to
+ * be replaced.
+ *
* CPUCP_PACKET_POWER_SET -
* Resets power history of device to 0
*/
@@ -424,6 +431,8 @@ enum cpucp_packet_id {
CPUCP_PACKET_NIC_STAT_REGS_CLR, /* internal */
CPUCP_PACKET_NIC_STAT_REGS_ALL_GET, /* internal */
CPUCP_PACKET_IS_IDLE_CHECK, /* internal */
+ CPUCP_PACKET_HBM_REPLACED_ROWS_INFO_GET,/* internal */
+ CPUCP_PACKET_HBM_PENDING_ROWS_STATUS, /* internal */
CPUCP_PACKET_POWER_SET, /* internal */
};

@@ -692,6 +701,7 @@ struct eq_generic_event {
#define CPUCP_MAX_NIC_LANES (CPUCP_MAX_NICS * CPUCP_LANES_PER_NIC)
#define CPUCP_NIC_MASK_ARR_LEN ((CPUCP_MAX_NICS + 63) / 64)
#define CPUCP_NIC_POLARITY_ARR_LEN ((CPUCP_MAX_NIC_LANES + 63) / 64)
+#define CPUCP_HBM_ROW_REPLACE_MAX 32

struct cpucp_sensor {
__le32 type;
@@ -837,4 +847,25 @@ struct cpucp_nic_status {
__le32 high_ber_cnt;
};

+enum cpucp_hbm_row_replace_cause {
+ REPLACE_CAUSE_DOUBLE_ECC_ERR,
+ REPLACE_CAUSE_MULTI_SINGLE_ECC_ERR,
+};
+
+struct cpucp_hbm_row_info {
+ __u8 hbm_idx;
+ __u8 pc;
+ __u8 sid;
+ __u8 bank_idx;
+ __le16 row_addr;
+ __u8 replaced_row_cause; /* enum cpucp_hbm_row_replace_cause */
+ __u8 pad;
+};
+
+struct cpucp_hbm_row_replaced_rows_info {
+ __le16 num_replaced_rows;
+ __u8 pad[6];
+ struct cpucp_hbm_row_info replaced_rows[CPUCP_HBM_ROW_REPLACE_MAX];
+};
+
#endif /* CPUCP_IF_H */
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index 257b9630773e..9b4d72897061 100644
--- a/include/uapi/misc/habanalabs.h
+++ b/include/uapi/misc/habanalabs.h
@@ -334,6 +334,8 @@ enum hl_server_type {
* HL_INFO_TOTAL_ENERGY - Retrieve total energy consumption
* HL_INFO_PLL_FREQUENCY - Retrieve PLL frequency
* HL_INFO_OPEN_STATS - Retrieve info regarding recent device open calls
+ * HL_INFO_DRAM_REPLACED_ROWS - Retrieve DRAM replaced rows info
+ * HL_INFO_DRAM_PENDING_ROWS - Retrieve DRAM pending rows num
*/
#define HL_INFO_HW_IP_INFO 0
#define HL_INFO_HW_EVENTS 1
@@ -353,6 +355,8 @@ enum hl_server_type {
#define HL_INFO_PLL_FREQUENCY 16
#define HL_INFO_POWER 17
#define HL_INFO_OPEN_STATS 18
+#define HL_INFO_DRAM_REPLACED_ROWS 21
+#define HL_INFO_DRAM_PENDING_ROWS 22

#define HL_INFO_VERSION_MAX_LEN 128
#define HL_INFO_CARD_NAME_MAX_LEN 16
--
2.25.1

Next message: Kurt Garloff: "igb in 5.15: autosuspend hangs netlink"
Previous message: Oded Gabbay: "[PATCH 10/10] habanalabs: make hdev creation code more readable"
In reply to: Oded Gabbay: "[PATCH 10/10] habanalabs: make hdev creation code more readable"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]