Closed username223 closed 3 years ago
I've rechecked every value and came up with some misapplied reset values and access modifiers, as well as some registers in GLB that I missed because I assumed that I had done them in PR #4.
The se_trng_0_rosc_en
register is called se_trng_0_rosc_dis
in the SDK, at the same bit position and in the same register. The name has been changed and a note has been added in the description.
Fixed merge conflict from PR https://github.com/sipeed/bl602-pac/pull/5.
Build fails due to warnings
*** WARNING M223: soc602_reg.svd (Line 2)
Input File Name 'soc602_reg' does not match the tag <name> in the <device> section: 'BL602'
*** WARNING M209: soc602_reg.svd
CPU section not set. This is required for CMSIS Headerfile generation and debug support.
Found 0 Error(s) and 2 Warning(s).
Error: Process completed with exit code 1.
Until these are fixed the return code of SVDConv
should be ignored.
Ignoring the return code from program using GHA is painful. I'm going to remove the execution of SVDConv from the workflow for now, and put it in a script or something later so that's easier to test locally
Couldn't you do something like
#! /usr/bin/env bash
./SVDConv soc602_reg.svd
if [ $? == 1 ]; then
exit 0
fi
Since return code 0 is nothing, 1 is warning and 2 is error, only capturing 1 will still fail on errors.
Yep, that's exactly what we need to do
Ah, I see what you meant in your original comment now, Github Actions is indeed a little annoying to use at times.
Corrected duplicated resetValue
s to resetMask
s as was originally intended.
This PR adds reset values and access modifiers directly from the SDK.
The struct fields (specific structs in each commit) parsed looked like
There are some fields with access values as
w1c
andw1p
, although I do not know how to represent this in the SVD so they have not been included. In the future those modifiers should also be added.