[earlgrey,dv] Remove bogus dmi device from TopEarlgrey#29616
[earlgrey,dv] Remove bogus dmi device from TopEarlgrey#29616rswarbrick wants to merge 1 commit intolowRISC:masterfrom
Conversation
rswarbrick
force-pushed
the
no-bogus-dmi
branch
from
fddaec3 to
9a29699
Compare
vogelpi
left a comment
vogelpi
left a comment
There was a problem hiding this comment.
Thanks @rswarbrick for the PR and the clear description in the code. LGTM!
|
CHANGE AUTHORIZED: hw/top_earlgrey/data/top_earlgrey.hjson This PR removes the DMI interface which was not used and not functional in Earlgrey anyway. |
|
CHANGE AUTHORIZED: hw/top_earlgrey/data/top_earlgrey.hjson This change will not have any effect on the Earlgrey design. Any potential change would just be removing an accidental post- change made in order to allow something for Darjeeling. |
This was added for Darjeeling and it appears the engineers concerned didn't think about the fact this would convince various bits of tooling that the phantom interface exists for Earlgrey too. The bulk of this commit is tidying up the logic in topgen's lib.py, merge.py and top_uvm_reg.sv.tpl to allow the possibility that an instance of a block doesn't connect up all the interfaces that the block defines. To do this properly, it also fixes some incorrect type annotations in the code (which seem to have been introduced when first adding multiple address maps). The Earlgrey-specific change is tiny: it's just remove the "dmi" line in hw/top_earlgrey/data/top_earlgrey.hjson, together with the comment that shows the author realises it didn't make much sense... Signed-off-by: Rupert Swarbrick <rswarbrick@lowrisc.org>
rswarbrick
force-pushed
the
no-bogus-dmi
branch
from
9a29699 to
4596313
Compare
pamaury
left a comment
pamaury
left a comment
There was a problem hiding this comment.
I had a look mostly at the topgen/dt library changes and they look good to me (though I do find this part of topgen to be hard to follow), it's great that you added a lot of comments!
3 participants
This was added for Darjeeling and it appears the engineers concerned didn't think about the fact this would convince various bits of tooling that the phantom interface exists for Earlgrey too.
The bulk of this commit is tidying up the logic in topgen's lib.py, merge.py and top_uvm_reg.sv.tpl to allow the possibility that an instance of a block doesn't connect up all the interfaces that the block defines. To do this properly, it also fixes some incorrect type annotations in the code (which seem to have been introduced when first adding multiple address maps).
The Earlgrey-specific change is tiny: it's just remove the "dmi" line in hw/top_earlgrey/data/top_earlgrey.hjson, together with the comment that shows the author realises it didn't make much sense...