[libre-riscv-dev] daily kan-ban update 09jul2020

Luke Kenneth Casson Leighton lkcl at lkcl.net
Thu Jul 9 18:29:33 BST 2020


yesterday:

* spent considerable time tracking down bugs introduced by the
addition of OP_ATTN, OP_NOP AND OP_ILLEGAL
* added infrastructure that allows attempting to run microwatt hello_world.bin
* discovered that it's little-endian where everything we've been doing
so far is big-endian... sort-of
* discovered some confusion in the marking of what's big-endian and
what's little (set little to get big for example)

today:

* bit of a hairy moment morphing both MUL and DIV pipelines to remove
XER CA/CA32 propagation (neither of them need it)
* added an extra class NextData to Cesar's alu_fsm.py and got Shifter
example looking conformant with the CompALU API.
* re-ran the helloworld fragment example
(simulator/test_helloworld_sim.py) which runs the first few
(disassembled) instructions of microwatt hello_world.bin and found
that it works.

rest of day: resting.  need to review the Qemu code and the ISACaller
simulator and pretty much all testing infrastructure to get them to
switch between big-endian and little-endian.  that's... like...
aaallll unit tests - absolutely all of them.  54 files.

soc$ find . -name "test*.py" | wc
     54      54    1871

---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68



More information about the libre-riscv-dev mailing list