LayerscapeのトピックRe: CWTAP connection failed when using CW to debug LS1043A

CWTAP connection failed when using CW to debug LS1043A

uncoldice — Wed, 12 Apr 2017 03:53:01 GMT

Hi,

I am using CW to debug LS1043ardb, and CWTAP is used.

After building source code, and trying to debug the application on the board, I will met below error when dowloading:

Errors in Debug Server
---------------------------------
Target connection failed. Please re-check the settings.
//
Processor: LS1043A
Probe: //
Additional error details:
[CCS: subcore error during multicore operation]

I am using the default LS1043A_RDB target connection configuration which is provided in CW. version of CW is:

CodeWarrior Development Studio for QorIQ LS series - ARM V8 ISA

(c) Freescale Semiconductor, Inc. 2014. All rights reserved.
Learn more about CodeWarrior at
http://www.freescale.com/codewarrior

Version: 11.2.0
Build Id: 160115

Also I tried to run "source IDcode.tcl" on ccs, below is the output:

(bin) 57 % source IDcode.tcl

Scanning for available TAPs connected via USB.....

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+
+ Available Remote Connections
+
+ 1 - CodeWarriorTAP - 00:04:9f:03:ae:f8
+ 2 - CodeWarriorTAP - <Specify IP Address>
+ 3 - GigabitTAP - <Specify IP Address>
+
+ x - Exit Script without Changes
+
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Specify connection:
1

Configuring TAP Interface....

Configured Connection: cwtap : 00:04:9f:03:ae:f8

TDO -----
|
* Device 0 IDCODE: 5BA00477 Device: Unknown Device
* Device 1 IDCODE: 06B1001D Device: Unmapped FSL Device
|
TDI -----

###################################################
#
# configTAP - Redefine TAP interface
#
# scanboard - Scans the target system
# and returns the JTAG IDCode
#
# ir - Loopback test
#
###################################################

Please help to have a look what's the problem. Thank you very much.

Re: CWTAP connection failed when using CW to debug LS1043A

ufedor — Wed, 12 Apr 2017 04:44:20 GMT

Please run "Diagnose Connection" and provide the log

Re: CWTAP connection failed when using CW to debug LS1043A

uncoldice — Wed, 12 Apr 2017 04:55:52 GMT

This is the diagnose result, please help to have a look.

Also, the connection configuration is:

Re: CWTAP connection failed when using CW to debug LS1043A

ufedor — Wed, 12 Apr 2017 05:38:43 GMT

Please try to modify the LS1043A_RDB Target Initialization File in the Main change the line:

useSafeRCW = False

useSafeRCW = True

and run "Diagnose Connection"

Re: CWTAP connection failed when using CW to debug LS1043A

uncoldice — Wed, 12 Apr 2017 06:10:10 GMT

After that change, I will meet this problem when "connect":

Errors in Debug Server
---------------------------------
Target connection failed. Please re-check the settings.
//
Processor: LS1043A
Probe: CodeWarrior TAP (00:04:9f:03:ae:f8)
//
Additional error details:
[GTA: all targeted cores are in unsupported aarch32 mode]

"diagnose connection" will return same result as I attached in last reply.

BTW, the initialization file content is like below:

python
###################################################################
# Copyright (C) 2015, Freescale Semiconductor, Inc.
# All Rights Reserved
###################################################################
import gdb
import time
import ctypes
CORE_CONTEXT = ":ccs:LS1043A:CortexA53#0"

###################################################################
# Utility functions
###################################################################
#This function should be used instead of gdb.execute in all cases when the value returned is used in the script
def gdb_exec(command, from_tty, to_string):
"""Execute a gdb command, remove the echo if it appears, handle the exception"""
try:
string = gdb.execute(command, from_tty, to_string)
echo_command = "+++" + command + "\n"
if string is not None and string.startswith(echo_command):
return string[len(echo_command):]
else:
return string
except StandardError as msg:
print("\nError when executing gdb command: " + command)
print(msg)
return None
# Swap utility function
def Swap32(x):
return (((x << 24) & 0xFF000000) |
((x << 8) & 0x00FF0000) |
((x >> 8) & 0x0000FF00) |
((x >> 24) & 0x000000FF))
# Memory Modify
def MM(context, address, accessSize, space, value):
gdb.execute("monitor mem write %s %#x %d %s %#x" % (context, address, accessSize, space, value))
return
# Memory Display
def MD(context, address, accessSize, space):
return int(gdb_exec("monitor mem read %s %#x %d %s 1" % (context, address, accessSize, space), False, True), 16)
# CCSR BigEndian Modify
def CCSR_BE_M(address, value):
MM(CORE_CONTEXT, address, 4, "physical_noncoherent", value)
return
# CCSR BigEndian Display
def CCSR_BE_D(address):
return MD(CORE_CONTEXT, address, 4, "physical_noncoherent")
# CCSR LittleEndian Modify
def CCSR_LE_M(address, value):
return CCSR_BE_M(address, Swap32(value))
# CCSR LittleEndian Display
def CCSR_LE_D(address):
return Swap32(CCSR_BE_D(address))
# Register Modify
def RM(context, registerId, size, value):
gdb.execute("monitor reg write %s %#x %d %#x" % (context, registerId, size, value))
return
# Register Display
def RD(context, registerId, size):
return int(gdb_exec("monitor reg read %s %#x %d" % (context, registerId, size), False, True), 16)
# Template Config
def TMP_CFG(context, reg, data):
gdb.execute("monitor template config %s %#x %#x" % (context, reg, data))
return

###################################################################
# Reset
###################################################################
def Reset(useSafeRCW):
if useSafeRCW:
# Set rcw_src to hard-coded RCW option
gdb.execute("monitor rcw source set 0x9F")
# add here if you need to override other particular RCW values
# gdb.execute("monitor rcw data set RCW1 0x40282830 RCW9-10 0x00C12980 0x00002580")
gdb.execute("monitor rcw apply")
else:
try:
# This init requires target reset, irrespective of the user preference
#user_reset = gdb.parse_and_eval ("$reset")
userResetDelay = int(gdb.parse_and_eval ("$delay"))
except gdb.error:
userResetDelay = 0
gdb.execute("py-reset %d" % userResetDelay)
# RCPM_CESRD sticky bit which will get set when you deassert EVT_B0
# REMOVE EXTERNAL DEBUG REQUEST which we issue in ccs on ccs::reset_to_debug
CCSR_BE_M(0x20160704, 0x80000000)
CCSR_BE_M(0x20160708, 0x80000000)
CCSR_BE_M(0x2016070C, 0x80000000)
# Safeguard against reset skid on CMSISDAP
# Clear PC
RM(CORE_CONTEXT, 0x1DA29, 8, 0)
# Clear MMU, caches, aligment checks, WNX in SCTLR_EL3
SCTLR_EL3 = RD(CORE_CONTEXT, 0x1F080, 4)
SCTLR_EL3 = SCTLR_EL3 & 0xFFF7EFF0
RM(CORE_CONTEXT, 0x1F080, 4, SCTLR_EL3)
return
###################################################################
# CCI Initialization
###################################################################
def Init_CCI():
# Set Terminate all barrier transactions bit of CCI-400 Control Register
# (needed for dsb instruction to succeed when doing coherent reads/writes)
# It will be disabled after DDR is initialized
CCSR_LE_M(0x01180000, 0x00000008)
# Enable snoop requests and DVM messages for A53 cluster slave interface 4
CCSR_LE_M(0x01185000, 0x00000003)
return
###################################################################
# Boot Release
###################################################################
def Init_BRR():
# TODO: when we can detect the current context,
# release all cores for SMP, current core for AMP
# Write to BRR to release cores
CCSR_BE_M(0x1EE00E4, 0x0000000F)
return
###################################################################
# DDR Initialization
###################################################################
def Init_DDRC():
# SDRAM_CFG
CCSR_BE_M(0x1080000 + 0x110, 0x450C000C)
# CS0_BNDS
CCSR_BE_M(0x1080000 + 0x000, 0x0000007f)
# CS0_CONFIG
CCSR_BE_M(0x1080000 + 0x080, 0x80010322)
# TIMING_CFG_0
CCSR_BE_M(0x1080000 + 0x104, 0xD0550018)
# TIMING_CFG_1
CCSR_BE_M(0x1080000 + 0x108, 0xC2C68C42)
# TIMING_CFG_2
CCSR_BE_M(0x1080000 + 0x10C, 0x0048C114)
# TIMING_CFG_3
CCSR_BE_M(0x1080000 + 0x100, 0x020C1000)
# TIMING_CFG_4
CCSR_BE_M(0x1080000 + 0x160, 0x00000002)
# TIMING_CFG_5
CCSR_BE_M(0x1080000 + 0x164, 0x04401400)
# TIMING_CFG_7
CCSR_BE_M(0x1080000 + 0x16C, 0x13300000)
# TIMING_CFG_8
CCSR_BE_M(0x1080000 + 0x250, 0x03115600)
# SDRAM_CFG_2
CCSR_BE_M(0x1080000 + 0x114, 0x00401010)
# SDRAM_MODE
CCSR_BE_M(0x1080000 + 0x118, 0x01010214)
# SDRAM_MODE_2
CCSR_BE_M(0x1080000 + 0x11C, 0x0)
# SDRAM_INTERVAL
CCSR_BE_M(0x1080000 + 0x124, 0x18600618)
# DDR_WRLVL_CNTL
CCSR_BE_M(0x1080000 + 0x174, 0x8655F606)
# DDR_WRLVL_CNTL_2
CCSR_BE_M(0x1080000 + 0x190, 0x05070600)
# DDR_WRLVL_CNTL_3
CCSR_BE_M(0x1080000 + 0x194, 0x0)
# DDR_CDR1
CCSR_BE_M(0x1080000 + 0xB28, 0x80040000)
# DDR_CDR2
CCSR_BE_M(0x1080000 + 0xB2C, 0x0000A181)
# SDRAM_CLK_CNTL
CCSR_BE_M(0x1080000 + 0x130, 0x02000000)
# DDR_ZQ_CNTL
CCSR_BE_M(0x1080000 + 0x170, 0x8A090705)
# SDRAM_MODE_9
CCSR_BE_M(0x1080000 + 0x220, 0x00000400)
# SDRAM_MODE_10
CCSR_BE_M(0x1080000 + 0x224, 0x04000000)
# CS0_CONFIG_2
CCSR_BE_M(0x1080000 + 0x0C0, 0)
# SDRAM_CFG
CCSR_BE_M(0x1080000 + 0x110, 0xC50C000C)
# Poll for DDR to be initialized
count = 0
while True:
time.sleep(0.2)
SDRAM_CFG2 = CCSR_BE_D(0x01080114)
D_INIT = SDRAM_CFG2 & 0x10
count = count + 1
if (D_INIT == 0):
break
if (count > 20):
print "DDRC: Time out waiting for D_INIT"
break
ERR_DETECT = CCSR_BE_D(0x01080E40)
if (ERR_DETECT != 0):
print "Memory initialization error on DDRC!"
print format(ERR_DETECT, '08x')
time.sleep(1)
# Clear Terminate all barrier transactions bit of CCI-400 Control Register
CCSR_BE_M(0x01180000, 0x00000000)
return
###################################################################
# IFC Initialization
###################################################################
def Init_IFC():
# Reset IFC
# GCR
CCSR_BE_M(0x0153040C, 0x8000F000)
time.sleep(0.1)
CCSR_BE_M(0x0153040C, 0x0000F000)
# Poll for reset to finish
count = 0
while True:
time.sleep(0.2)
GCR = CCSR_BE_D(0x0153040C)
SOFT_RESET = GCR & 0x80000000
count = count + 1
if (SOFT_RESET == 0):
break
if (count > 20):
print "IFC: Time out waiting for soft reset"
break
# CCR
CCSR_BE_M(0x0153044C, 0x04008000)
# Poll for stable clock
count = 0
while True:
time.sleep(0.2)
CSR = CCSR_BE_D(0x01530450)
count = count + 1
if ((CSR & 0x80000000) != 0):
break
if (count > 20):
print "IFC: Clock not stable"
break
NOR_CS = 0
NAND_CS = 1
# Check IFC_CSPR0. CPLD Chip select is assumed to be always 2.
CSPR0 = CCSR_BE_D(0x1530010)
# Check MSEL to see if NAND is on CS 0, otherwise assume NOR.
if ((CSPR0 & 0x7) == 0x3):
NOR_CS = 1
NAND_CS = 0
# CSPR_EXT
CCSR_BE_M(0x0153000C + NOR_CS * 12, 0x00000000)
# CSPR
CCSR_BE_M(0x01530010 + NOR_CS * 12, 0x60000101)
# AMASK
CCSR_BE_M(0x015300A0 + NOR_CS * 12, 0xF8000000)
# CSOR
CCSR_BE_M(0x01530130 + NOR_CS * 12, 0x0000800C)
# IFC_FTIM0
CCSR_BE_M(0x015301C0 + NOR_CS * 48, 0x1001000C)
# IFC_FTIM1
CCSR_BE_M(0x015301C4 + NOR_CS * 48, 0x1C000B09)
# IFC_FTIM2
CCSR_BE_M(0x015301C8 + NOR_CS * 48, 0x01102010)
# IFC_FTIM3
CCSR_BE_M(0x015301CC + NOR_CS * 48, 0x00000000)
# CSPR_EXT
CCSR_BE_M(0x0153000C + NAND_CS * 12, 0x00000000)
# CSPR
CCSR_BE_M(0x01530010 + NAND_CS * 12, 0x7E800083)
# AMASK
CCSR_BE_M(0x015300A0 + NAND_CS * 12, 0xFFFF0000)
# CSOR
CCSR_BE_M(0x01530130 + NAND_CS * 12, 0x85082100)
# IFC_FTIM0
CCSR_BE_M(0x015301C0 + NAND_CS * 48, 0x0E18070A)
# IFC_FTIM1
CCSR_BE_M(0x015301C4 + NAND_CS * 48, 0x32390E18)
# IFC_FTIM2
CCSR_BE_M(0x015301C8 + NAND_CS * 48, 0x01E0501E)
# IFC_FTIM3
CCSR_BE_M(0x015301CC + NAND_CS * 48, 0x00000000)
return

###################################################################
# Adds Flash devices for this board
###################################################################
def Config_Flash_Devices():
# Add NOR device
gdb.execute("monitor flash device alias nor name MT28EW01 address 0x60000000 ws_address 0x10000000 ws_size 0x1FFFF geometry 16x1 controller IFC")
# Add NAND device
gdb.execute("monitor flash device alias nand name MT29F4G08ABBDAv1 address 0x7E800000 ws_address 0x10000000 ws_size 0x1FFFF geometry 8x1 controller IFC")
# Add SD/eMMC device
gdb.execute("monitor flash device alias sd name SDSP16GB_LSCH2 address 0x00000000 ws_address 0x80000000 ws_size 0x1FFFF geometry 8x1 controller IFC")
#set nor as current device
gdb.execute("monitor flash current nor")
return
###################################################################
# Main
###################################################################
# Uncomment to enable RCW override
useSafeRCW = True
Reset(useSafeRCW)
Init_CCI()
Init_IFC()
Init_DDRC()
Init_BRR()
Config_Flash_Devices()
end

Re: CWTAP connection failed when using CW to debug LS1043A

ufedor — Wed, 12 Apr 2017 06:14:10 GMT

In the future please attach big texts instead of "inline".

Is the board capable to boot?

Re: CWTAP connection failed when using CW to debug LS1043A

uncoldice — Wed, 12 Apr 2017 06:41:33 GMT

Thank you for your reminder ufedor‌.

Root cause is that I didn't stop the U-Boot count down, and the board goes to a 32-bit Linux kernel before my debug.