[axi_sub] Remove the duplicated axi_sub library#209
Conversation
|
|
The lib was previously duplicated from the original caliptra version only to introduce a duplicate parameter (AG with the same function as BW). Additionally, axi_sub would not support different data widths for the axi and component side. Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
There was a problem hiding this comment.
Sorry to change my review. I realized that the underlying lint problem we need to resolve also requires changes here:
https://github.com/glaserf/i3c-core/blob/41a7968399d318445bece2750a1eac93a8b37103/src/hci/axi_adapter.sv#L142
The axi_if adds 4 new AR and AW signals in newer versions of caliptra-rtl, so you will need to update the submodule. Then the new AxPROT,AxCACHE, AxREGION, AxQOS signals should be routed from the top level down into axi_sub.
In fact, I'd actually recommend getting rid of all the local signal definitions entirely, and just routing the axi_if from the top i3c module all the way down to axi_adapter/axi_sub. Then it will be portable and compatible with any future changes without having to modify the i3c repo.
Hi @calebofearth |
|
Apologies for the longer wait time. |
This allows for easier updates of the bus infra in caliptra-ss Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
This is mainly done to incorporate the added signals of the axi_if, but also to get rid of the vastly outdated submodule version. Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
|
@calebofearth I updated the |
It looks like the only relevant updates from caliptra-rtl are for skidbuffer, axi_if, axi_sub, and caliptra_prim modules (fifo_sync and flop_2sync). The AXI changes are necessary, and the caliptra_prim changes are innocuous IMO. Do you see any other modules affected by the update? If not, I think this is fine. |
|
@calebofearth Sorry I totally forgot to reply. I agree that the submodule update should be unproblematic. |
3 participants
The lib was previously duplicated from the original Caliptra version, solely to introduce a duplicate parameter (
AGwith the same function asBW).Additionally,
axi_subdoes anyways not support different data widths for the axi and component side, which the additional parameter seems to have been targeted at.