Merge pull request #1369 from shellysela219/mstreg_syn_gb

show syndrome in case of mstreg failure

Author: ahmadawwad

include/mtcr_ul/mtcr_com_defs.h

@@ -446,6 +446,8 @@ typedef struct icmd_params_t {
     u_int32_t     dma_size;
     int           dma_icmd;
     mtcr_status_e icmd_ready;
+    int           syndrome;
+    int           syndrome_addr;
 } icmd_params;
 
 typedef struct ctx_params_t {

mlxreg/mlxreg_lib/mlxreg_lib.cpp

@@ -240,8 +240,18 @@ MlxRegLibStatus MlxRegLib::sendRegister(string regName, int method, std::vector<
     u_int16_t regId = (u_int16_t)_regAccessMap.find(regName)->second;
     int rc;
     rc = sendMaccessReg(regId, method, data);
-    if (rc) {
-        throw MlxRegException("Failed to send access register: %s", m_err2str((MError)rc));
+    if (rc)
+    {
+        char error_msg[200];
+        snprintf(error_msg, sizeof(error_msg), "Failed to send access register: %s", m_err2str((MError)rc));
+        if (_mf->icmd.syndrome)
+        {
+            snprintf(error_msg + strlen(error_msg),
+                     sizeof(error_msg) - strlen(error_msg),
+                     " and the syndrome number is: 0x%X",
+                     (_mf->icmd.syndrome));
+        }
+        throw MlxRegException(error_msg);
     }
     return MRLS_SUCCESS;
 }
@@ -255,7 +265,16 @@ MlxRegLibStatus MlxRegLib::sendRegister(u_int16_t regId, int method, std::vector
     rc = sendMaccessReg(regId, method, data);
     if (rc)
     {
-        throw MlxRegException("Failed send access register: %s", m_err2str((MError)rc));
+        char error_msg[200];
+        snprintf(error_msg, sizeof(error_msg), "Failed to send access register: %s", m_err2str((MError)rc));
+        if (_mf->icmd.syndrome)
+        {
+            snprintf(error_msg + strlen(error_msg),
+                     sizeof(error_msg) - strlen(error_msg),
+                     " and the syndrome number is: 0x%X",
+                     (_mf->icmd.syndrome));
+        }
+        throw MlxRegException(error_msg);
     }
     return MRLS_SUCCESS;
 }

mtcr_ul/fwctrl.c

@@ -187,13 +187,22 @@ int fwctl_control_access_register(int    fd,
     memcpy(data_in, data, size_in);
     memcpy(&cmd_status, status, sizeof(int));
     if (cmd_status) {
+        u_int32_t syndrome = MLX5_GET(access_register_out, out, syndrome);
+        mf->icmd.syndrome = syndrome;
+        FWCTL_DEBUG_PRINT(mf, "FWCTL_IOCTL_CMD_RPC returned error from FW: reg_id=0x%x, method=0x%x, cmd_status=0x%x, syndrome=0x%x\n",
+                          reg_id, method, cmd_status, syndrome);
+
         if (reg_id == mnvda_reg_id) {
             *reg_status = translate_cmd_status_to_reg_status(cmd_status);
         } else {
             *reg_status = return_by_reg_status(cmd_status);
         }
+        FWCTL_DEBUG_PRINT(mf, "Mapped FW cmd_status=0x%x to reg_status=0x%x (%s)\n", cmd_status, *reg_status, m_err2str(*reg_status));
+    } else {
+        *reg_status = 0;
+        mf->icmd.syndrome = 0;
+        FWCTL_DEBUG_PRINT(mf, "FWCTL_IOCTL_CMD_RPC succeeded: reg_id=0x%x, method=0x%x\n", reg_id, method);
     }
-
     FWCTL_DEBUG_PRINT(mf, "register id = 0x%x, command status = 0x%x, reg status code: 0x%x, reg status: %s\n",
                       reg_id, cmd_status, *reg_status, m_err2str(*reg_status));
 out:

mtcr_ul/mtcr_ul_icmd_cif.c

@@ -121,6 +121,11 @@
 #define VCR_SEMAPHORE62                0x0 /* semaphore Domain */
 #define VCR_CMD_ADDR                   0x100000 /* mailbox addr */
 #define VCR_CMD_SIZE_ADDR              0x1000 /* mailbox size */
+#define VCR_SYNDROME_OFFSET            0x1008
+#define SYNDROME_OFFSET                0x3f8
+#define SYNDROME_BITOFF                0
+#define SYNDROME_BITLEN                24
+
 
 #define EXT_MBOX_DMA_OFF 0x8
 
@@ -359,6 +364,15 @@ static MError set_busy_bit(mfile* mf, u_int32_t* reg, int busy_bit_offset)
     return MWRITE4_ICMD(mf, mf->icmd.ctrl_addr, *reg);
 }
 
+static MError get_syndrome(mfile* mf, u_int32_t* reg, int syndrome_bit_offset, int syndrome_bit_len)
+{
+    DBG_PRINTF("Reading syndrome from addr=0x%x\n", mf->icmd.syndrome_addr);
+    int rc = MREAD4_ICMD(mf, mf->icmd.syndrome_addr, reg);
+    CHECK_RC(rc);
+
+    return EXTRACT((*reg), syndrome_bit_offset, syndrome_bit_len);
+}
+
 static int set_sleep()
 {
     char* icmd_sleep_env;
@@ -801,6 +815,12 @@ static int icmd_send_command_com(mfile     * mf,
 
     /* get status */
     ret = translate_status(EXTRACT(reg, STATUS_BITOFF, STATUS_BITLEN));
+
+    if (ret != ME_OK)
+    {
+        mf->icmd.syndrome = get_syndrome(mf, &reg, SYNDROME_BITOFF, SYNDROME_BITLEN);
+    }
+
     CHECK_RC_GO_TO(ret, cleanup);
 
     DBG_PRINTF("-D- Reading command from mailbox");
@@ -976,6 +996,7 @@ static int icmd_init_cr(mfile* mf)
     u_int32_t cmd_ptr_addr;
     u_int32_t reg = 0x0;
     u_int32_t hw_id = 0x0;
+    mf->icmd.syndrome = 0;
 
 #ifndef __FreeBSD__
     u_int32_t dev_type = 0;
@@ -1135,6 +1156,8 @@ static int icmd_init_cr(mfile* mf)
 
         mf->icmd.cmd_addr = EXTRACT(reg, CMD_PTR_BITOFF, mf->icmd.cmd_ptr_bitlen);
         mf->icmd.ctrl_addr = mf->icmd.cmd_addr + CTRL_OFFSET;
+        mf->icmd.syndrome_addr = mf->icmd.cmd_addr + SYNDROME_OFFSET;
+        DBG_PRINTF("-D- iCMD syndrom addr: 0x%x\n", mf->icmd.syndrome_addr);
         break;
 
     case ME_ICMD_STATUS_CR_FAIL:
@@ -1243,6 +1266,8 @@ static int icmd_init_vcr(mfile* mf)
     mf->icmd.cmd_addr = VCR_CMD_ADDR;
     mf->icmd.ctrl_addr = VCR_CTRL_ADDR;
     mf->icmd.semaphore_addr = VCR_SEMAPHORE62;
+    mf->icmd.syndrome_addr = VCR_SYNDROME_OFFSET;
+    mf->icmd.syndrome = 0;
     DBG_PRINTF("-D- Getting VCR_CMD_SIZE_ADDR\n");
 
     rc = icmd_take_semaphore_com(mf, pid);
@@ -1262,6 +1287,7 @@ static int icmd_init_vcr(mfile* mf)
     mf->icmd.icmd_opened = 1;
     DBG_PRINTF("-D- iCMD command addr: 0x%x\n", mf->icmd.cmd_addr);
     DBG_PRINTF("-D- iCMD ctrl addr: 0x%x\n", mf->icmd.ctrl_addr);
+    DBG_PRINTF("-D- iCMD syndrom addr: 0x%x\n", mf->icmd.syndrome_addr);
     DBG_PRINTF("-D- iCMD semaphore addr(semaphore space): 0x%x\n", mf->icmd.semaphore_addr);
     DBG_PRINTF("-D- iCMD max mailbox size: 0x%x  size %d\n", mf->icmd.max_cmd_size, size);
     DBG_PRINTF("-D- iCMD stat_cfg_not_done addr: 0x%x:%d\n", mf->icmd.static_cfg_not_done_addr,