HackRF-Treasure-Chest/Software/portapack-mayhem/hackrf/docs/source/LPC43XX_Debugging.rst

207 lines
6.2 KiB
ReStructuredText
Raw Normal View History

2022-09-22 18:26:57 +02:00
================================================
LPC43xx Debugging
================================================
Various debugger options for the LPC43xx exist.
Black Magic Probe
~~~~~~~~~~~~~~~~~
`https://github.com/blacksphere/blackmagic <https://github.com/blacksphere/blackmagic>`__
An example of using gdb with the Black Magic Probe:
.. code-block :: sh
arm-none-eabi-gdb -n blinky.elf
target extended-remote /dev/ttyACM0
monitor swdp_scan
attach 1
set {int}0x40043100 = 0x10000000
load
cont
It is possible to attach to the M0 instead of the M4 if you use jtag_scan instead of swdp_scan, but the Black Magic Probe had some bugs when trying to work with the M0 the last time I tried it.
LPC-Link
~~~~~~~~
(included with LPCXpresso boards)
TitanMKD has had some success. See the tutorial in hackrf/doc/LPCXPresso_Flash_Debug_Tutorial.pdf or .odt (PDF and OpenOffice document) Doc Link [https://github.com/mossmann/hackrf/tree/master/doc)
ST-LINK/V2
~~~~~~~~~~
Hardware Configuration
^^^^^^^^^^^^^^^^^^^^^^
Start with an STM32F4-Discovery board. Remove the jumpers from CN3. Connect the target's SWD interface to CN2 "SWD" connector.
Software Configuration
^^^^^^^^^^^^^^^^^^^^^^
I'm using libusb-1.0.9.
Install OpenOCD-0.6.0 dev
+++++++++++++++++++++++++
.. code-block:: sh
# Cloned at hash a21affa42906f55311ec047782a427fcbcb98994
git clone git://openocd.git.sourceforge.net/gitroot/openocd/openocd
cd openocd
./bootstrap
./configure --enable-stlink --enable-buspirate --enable-jlink --enable-maintainer-mode
make
sudo make install
OpenOCD configuration files
+++++++++++++++++++++++++++
openocd.cfg
.. code-block:: sh
#debug_level 3
source [find interface/stlink-v2.cfg]
source ./lpc4350.cfg
lpc4350.cfg
.. code-block :: sh
set _CHIPNAME lpc4350
set _M0_CPUTAPID 0x4ba00477
set _M4_SWDTAPID 0x2ba01477
set _M0_TAPID 0x0BA01477
set _TRANSPORT stlink_swd
transport select $_TRANSPORT
stlink newtap $_CHIPNAME m4 -expected-id $_M4_SWDTAPID
stlink newtap $_CHIPNAME m0 -expected-id $_M0_TAPID
target create $_CHIPNAME.m4 stm32_stlink -chain-position $_CHIPNAME.m4
#target create $_CHIPNAME.m0 stm32_stlink -chain-position $_CHIPNAME.m0
target.xml, nabbed from an OpenOCD mailing list thread, to fix a communication problem between GDB and newer OpenOCD builds.
.. code-block:: sh
<?xml version="1.0"?>
<!DOCTYPE target SYSTEM "gdb-target.dtd">
<target>
<feature name="org.gnu.gdb.arm.core">
<reg name="r0" bitsize="32" type="uint32"/>
<reg name="r1" bitsize="32" type="uint32"/>
<reg name="r2" bitsize="32" type="uint32"/>
<reg name="r3" bitsize="32" type="uint32"/>
<reg name="r4" bitsize="32" type="uint32"/>
<reg name="r5" bitsize="32" type="uint32"/>
<reg name="r6" bitsize="32" type="uint32"/>
<reg name="r7" bitsize="32" type="uint32"/>
<reg name="r8" bitsize="32" type="uint32"/>
<reg name="r9" bitsize="32" type="uint32"/>
<reg name="r10" bitsize="32" type="uint32"/>
<reg name="r11" bitsize="32" type="uint32"/>
<reg name="r12" bitsize="32" type="uint32"/>
<reg name="sp" bitsize="32" type="data_ptr"/>
<reg name="lr" bitsize="32"/>
<reg name="pc" bitsize="32" type="code_ptr"/>
<reg name="cpsr" bitsize="32" regnum="25"/>
</feature>
<feature name="org.gnu.gdb.arm.fpa">
<reg name="f0" bitsize="96" type="arm_fpa_ext" regnum="16"/>
<reg name="f1" bitsize="96" type="arm_fpa_ext"/>
<reg name="f2" bitsize="96" type="arm_fpa_ext"/>
<reg name="f3" bitsize="96" type="arm_fpa_ext"/>
<reg name="f4" bitsize="96" type="arm_fpa_ext"/>
<reg name="f5" bitsize="96" type="arm_fpa_ext"/>
<reg name="f6" bitsize="96" type="arm_fpa_ext"/>
<reg name="f7" bitsize="96" type="arm_fpa_ext"/>
<reg name="fps" bitsize="32"/>
</feature>
</target>
Run ARM GDB
~~~~~~~~~~~
Soon, I should dump this stuff into a .gdbinit file.
.. code-block:: sh
arm-none-eabi-gdb -n
target extended-remote localhost:3333
set tdesc filename target.xml
monitor reset init
monitor mww 0x40043100 0x10000000
monitor mdw 0x40043100 # Verify 0x0 shadow register is set properly.
file lpc4350-test.axf # This is an ELF file.
load # Place image into RAM.
monitor reset init
break main # Set a breakpoint.
continue # Run to breakpoint.
continue # To continue from the breakpoint.
step # Step-execute the next source line.
stepi # Step-execute the next processor instruction.
info reg # Show processor registers.
More GDB tips for the GDB-unfamiliar:
.. code-block:: sh
# Write the variable "buffer" (an array) to file "buffer.u8".
dump binary value buffer.u8 buffer
# Display the first 32 values in buffer whenever you halt
# execution.
display/32xh buffer
# Print the contents of a range of registers (in this case the
# CGU peripheral, starting at 0x40050014, for 46 words):
x/46 0x40050014
And still more, for debugging ARM Cortex-M4 Hard Faults:
.. code-block :: sh
# Assuming you have a hard-fault handler wired in:
display/8xw args
# Examine fault-related registers:
# Configurable Fault Status Register (CFSR) contains:
# CFSR[15:8]: BusFault Status Register (BFSR)
# "Shows the status of bus errors resulting from instruction
# prefetches and data accesses."
# BFSR[7]: BFARVALID: BFSR contents valid.
# BFSR[5]: LSPERR: fault during FP lazy state preservation.
# BFSR[4]: STKERR: derived bus fault on exception entry.
# BFSR[3]: UNSTKERR: derived bus fault on exception return.
# BFSR[2]: IMPRECISERR: imprecise data access error.
# BFSR[1]: PRECISERR: precise data access error, faulting
# address in BFAR.
# BFSR[0]: IBUSERR: bus fault on instruction prefetch. Occurs
# only if instruction is issued.
display/xw 0xE000ED28
# BusFault Address Register (BFAR)
# "Shows the address associated with a precise data access fault."
# "This is the location addressed by an attempted data access that
# was faulted. The BFSR shows the reason for the fault and whether
# BFAR_ADDRESS is valid..."
# "For unaligned access faults, the value returned is the address
# requested by the instruction. This might not be the address that
# faulted."
display/xw 0xE000ED38