[scripts/makehtml] Fix displaying of registers with the same addressOffset and parsing of addressOffset #591
Conversation
|
Note about the memory map comparison: |
…ffset and parsing of addressOffset
Co-authored-by: Adam Greig <adam@adamgreig.com>
9a12c78 to
2064371
Compare
scripts/makehtml.py
Outdated
| @@ -146,7 +146,7 @@ def parse_device(svdfile): | |||
| # Bodge to prevent /0 when there are no fields in a register | |||
| if register_fields_total == 0: | |||
| register_fields_total = 1 | |||
| registers[roffset] = {"name": rname, | |||
| registers[rname] = {"name": rname, | |||
There was a problem hiding this comment.
| registers[rname] = {"name": rname, | |
| registers[(roffset, rname)] = {"name": rname, |
I had to check to remember what this is used for, but it's inside makehtml.template.html, where the registers are iterated over by sorted key order, so previously they were in register-offset order but now they'd be in register-alphabetical order. If we make this key (offset, name), we preserve the sort order but allow the overlapping registers to still appear.
I think it would be good to also change makehtml.template.html L121 to something like {% for _, register in peripheral.registers|dictsort %} (changing roffset to _) to make it clear that the key isn't used later, since it's now a tuple.
Finally the indentation of the fields below needs updating since this line changes length.
There was a problem hiding this comment.
Good catch, I thought registers were inserted by their roffset already, but I missed the sorting function.
This pull request fix two bugs in the
makehtml.pyscript:The first bug is fixed by checking if the addressOffset starts with
0xto parse from base16, or parse from base10 by default (when a register is generated from a patch).The second bug is fixed by changing the index used for the list of registers from the addressOffset to the register name.