Skip to content

[top] Restructure chiplevel hierarchies#30387

Draft
glaserf wants to merge 16 commits into
lowRISC:masterfrom
glaserf:toplevel-hier
Draft

[top] Restructure chiplevel hierarchies#30387
glaserf wants to merge 16 commits into
lowRISC:masterfrom
glaserf:toplevel-hier

Conversation

@glaserf

@glaserf glaserf commented Jun 12, 2026

Copy link
Copy Markdown
Contributor

No description provided.

@glaserf glaserf force-pushed the toplevel-hier branch 9 times, most recently from d4538a9 to a15a63b Compare June 19, 2026 14:18
@glaserf glaserf force-pushed the toplevel-hier branch 5 times, most recently from d9b1045 to bc8e2da Compare June 23, 2026 14:15
@glaserf glaserf added the CI:Rerun Rerun failed CI jobs label Jun 23, 2026
@github-actions github-actions Bot removed the CI:Rerun Rerun failed CI jobs label Jun 23, 2026
glaserf added 14 commits June 23, 2026 19:51
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
This commit moves the dummy padring for Verilator from the testbench to
the chiplevel, and replaces the connections to/from the DPI modules with
hierarchical module references (XMRs).

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
This commit adds the missing englishbreakfast chiplevel for Verilator,
and adapts the tb to instantiate it.

Like for earlgrey in the preceeding commits, this includes a
Verilator-specific padring and a few additional branches in the shared
chiplevel template.

Since the combined tb and chiplevel did not include the real ast up
until this point, the respective lint collateral for ast is linked to
earlgrey.

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
This commit is analogous to the preceeding ones but applies the changes
to darjeeling.

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
!!REMOVE this commit before PR!!

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
In accordance with the power domain reorganization RFC, this commit
consistently names the toplevel power domain HDL modules
`{topname}_pd_{domain}`. This mainly means that all the `top_{topname}`
modules are now called `{topname}_pd_main`, i.e., the default power
domain no longer has a special name for the corresponding HDL module.

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
After the recent power domain reorganization (lowRISC#30076), the HDL paths in
the generated RAL collateral require knowledge about the power domain of
each IP. This issue so far hasn't shown itself since so far the path to
the main power domain wasn't changed, and the chip level dv env did not
do any RAL-based accesses to IPs in the Aon domain.

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
With `top_{topname}` being renamed to `{topname}_pd_{domain}` in the
preceeding commit, this one reintroduces `top_{topname}` as an
additional level of hierarchy that wraps all "core" logic:

- All power domains (or more precisely, their wrappers)
- AST - split/absorption into the above still pending
- Any FPGA glue logic

The chiplevels now only instantiate the module described above and any
hard macros, such as pad cells or e.g., the USB differential receiver.

Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
glaserf added 2 commits June 23, 2026 19:52
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
Signed-off-by: Florian Glaser <glaserf@lowrisc.org>
@glaserf glaserf force-pushed the toplevel-hier branch 2 times, most recently from fb4df85 to cb50597 Compare June 23, 2026 17:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant