A python tool for register-based chip verification and validation
"Cheap Pie" is a python tool for register-based chip verification and validation. The name is a translitteration of "chip py" for obvious reasons.
Given an input description file for the chip, it provides a register-level and bitfield-level read/write access, through a generic transport layer.
Currently the implemented description input modes are:
- CMSIS-SVD (https://www.keil.com/pack/doc/CMSIS/SVD/html/svd_Format_pg.html)
- IP-XACT ( https://www.accellera.org/downloads/standards/ip-xact )
- SystemRDL (https://www.accellera.org/activities/working-groups/systemrdl)
but it should be relatively easy to add different chip description formats.
Although tested on few real chips (NXP QN9080, I.MX RT1010, K64F), cheap_pie parser already supports dozen of devices, listed in the CMSIS-SVD repository https://github.com/posborne/cmsis-svd .
Currently the supported transport layers are jlink and pyocd, but it should be really easy to add support for different transport layers, like for instance openSDA, CMSIS-DAP, Total Phase Cheetah, GDB or any other.
Experimental support for pyverilator transport allows to run interactive simulation of register blocks generated from SystemRDL source.
Author: Marco Merlin Tested on ipython3 (python 3.8.5) on ubuntu 20.04
%run cheap_pie
inval = "0xFFFFFFFF"
hal.regs.ADC_ANA_CTRL.setreg(inval)
retval = hex(hal.regs.ADC_ANA_CTRL.getreg())
assert(literal_eval(inval) == literal_eval(retval))
# decimal assignement
inval = 2
hal.regs.ADC_ANA_CTRL.setreg(inval)
retval = hal.regs.ADC_ANA_CTRL.getreg()
assert(inval == retval)
hal.regs.ADC_ANA_CTRL
hal.regs.ADC_ANA_CTRL.display()
print('Test bitfield methods...')
hal.regs.ADC_ANA_CTRL.bitfields.ADC_BM
hal.regs.ADC_ANA_CTRL.bitfields.ADC_BM.display()
hal.regs.ADC_ANA_CTRL.bitfields.ADC_BM.display(2)
hal.regs.ADC_ANA_CTRL.bitfields.ADC_BM.setbit(inval)
retval = hal.regs.ADC_ANA_CTRL.bitfields.ADC_BM.getbit()
assert(inval == retval)
# subscriptable register access
hal[0]
# subscriptable bitfield access
hal[0][0]
# subscriptable as a dictionary
hal['SYSCON_RST_SW_SET']
hal['ADC_ANA_CTRL']['ADC_BM']
# assignement
hal['ADC_ANA_CTRL'] = 1
hal['ADC_ANA_CTRL']['ADC_BM'] = 2
# dict-based assignement in single register write
hal['ADC_ANA_CTRL'] = {'DITHER_EN': 1, 'CHOP_EN': 1, 'INV_CLK': 1}
# help
hal.regs.ADC_ANA_CTRL.help()
ADC core and reference setting regsiter
ADC_BM:
: ADC bias current selection.
ADC_ORDER:
: 1 to enable SD ADC 2 order mode selection
DITHER_EN:
: 1 to enable SD ADC PN Sequence in chopper mode
CHOP_EN:
: 1 to enable SD ADC chopper
INV_CLK:
: 1 to invert SD ADC Output Clock
VREF_BM:
: SD ADC Reference Driver bias current selection.
VREF_BM_X3:
: SD ADC Reference Driver bias current triple.
VINN_IN_BM:
: PGA VlNN Input Driver bias current selection.
VINN_OUT_BM:
: PGA VlNN Output Driver bias current selection.
VINN_OUT_BM_X3:
: PGA VlNN Output Driver bias current triple.
ADC_BM_DIV2:
: SD ADC bias current half.
# load RT1010 from local svd file under ./devices/
# automatically calls ipython and cheap_pie initialization
./cheap_pie.sh -rf MIMXRT1011.svd -t jlink
# load K64 from CMSIS-SVD
# need to specify vendor for svd not in ./devices/
./cheap_pie.sh -rf MK64F12.svd -ve Freescale -t jlink
# calls QN9080 with dummy transport layer
# useful to explore device registers
./cheap_pie.sh -t dummy
# calls QN9080 device with dummy transport layer
./cfgs/cp_qn9080_dummy.sh
# calls RT1010 device with jlink transport layer
./cfgs/cp_rt1010_jlink.sh
# calls K20 device with dummy transport layer
./cfgs/cp_k20_dummy.sh
./tools/rdl2verilog.py -f ./devices/rdl/basic.rdl
./cheap_pie.sh -dd ./devices/rdl -rf basic.rdl -fmt rdl -t verilator -topv ./devices/rdl/basic/basic_rf.sv
pip3 install cheap_pie
pip3 install git+https://github.com/bat52/cheap_pie.git@master
# for XML parsing (used by legacy svd parser and IP-XACT parser)
pip3 install untangle
# for exporting XML info into a human-readable document
pip3 install python-docx
# for dumping registers
pip3 install hickle
# CMSIS-SVD python parser including many svd files https://github.com/posborne/cmsis-svd
pip3 install cmsis-svd
# SPIRIT IP-XACT parser through ipyxact https://github.com/olofk/ipyxact
pip3 install ipyxact
# SystemRDL to register-file verilog
https://github.com/hughjackson/PeakRDL-verilog
# SystemRDL to IP-XACT
https://github.com/SystemRDL/PeakRDL-ipxact
# for JLINK
pip3 install pylink-square
# pyOCD for CMSIS-DAP and JLINK support (only tested in python-venv)
pip3 install pyocd
# esptool for Espressif devices (not yet functional)
pip3 install esptool
# verilator
https://www.veripool.org/verilator/
# pyverilator (python verilator wrapper)
https://github.com/csail-csg/pyverilator
# gtkwave
http://gtkwave.sourceforge.net/
regtool from opentitan project seems similar, using JSON to represent chip/IP structure, and I2C transport https://docs.opentitan.org/doc/rm/register_tool/
custom input, output: verilog, VHDL, YAML, JSON, TOML, Spreadsheet (XLSX, XLS, OSD, CSV) https://github.com/rggen/rggen
convert ipxact register file description into verilog register bank https://github.com/oddball/ipxact2systemverilog
In conjunction with pyVISA (https://pyvisa.readthedocs.io/en/master/), used for instument control, it provides a simple and fully python-contained environment for silicon validation.
Graphical Render of bitfield structures https://github.com/wavedrom/bitfield
C++ register/bitfields access (including generation from svd) https://github.com/thanks4opensource/regbits
STM C++ regbits implementation https://github.com/thanks4opensource/regbits_stm
a barebone embedded library generator https://modm.io/
hardware descriptions for AVR and STM32 devices https://github.com/modm-io/modm-devices
STM32 Peripheral Access Crates (from svd) https://github.com/stm32-rs/stm32-rs
Banner created with pyfiglet https://www.devdungeon.com/content/create-ascii-art-text-banners-python#install_pyfiglet
Cheap Pie is modeled after an original Octave/Matlab implementation that cannot be shared due to licensing reasons. The original code was converted to python using SMOP ( https://github.com/ripple-neuro/smop ).