[topgen] Generate files for all address spaces #25894

a-will · 2025-01-16T02:35:54Z

Individually generate C and Rust collateral for all address spaces

Fix up alert and irq domains to enable generating collateral where the
alerts and irqs don't necessarily generate from devices in the same
address space as the CPU. Darjeeling has bus bridges that make servicing
those interrupts possible (indirect reachability to the IP).

The autogenerated tests will surely be bonkers, but that is to be fixed
up later.

a-will · 2025-01-16T16:39:09Z

This PR reworks #23122 a bit and adds comments that explain what the functions are trying to do and where there are limitations.

hw/top_darjeeling/sw/autogen/top_darjeeling_soc_dbg.h

hw/top_darjeeling/sw/autogen/top_darjeeling_soc_dbg_memory.ld

Razer6

That looks good. Thanks for improving my works. A quick question, right now some definitions are rendered in all address space. I guess that should only be rendered in the appropriate address space.

hw/top_darjeeling/rtl/autogen/top_darjeeling_soc_dbg_pkg.sv

Razer6

I think there still some output definitions that need to be dropped by adapting the IP detection in the template files.

hw/top_darjeeling/rtl/autogen/top_darjeeling_soc_dbg_pkg.sv

hw/top_darjeeling/rtl/autogen/top_darjeeling_soc_mbx_pkg.sv

hw/top_darjeeling/sw/autogen/top_darjeeling_soc_dbg.h

sw/host/opentitanlib/src/chip/autogen/darjeeling_soc_mbx.rs

sw/host/opentitanlib/src/chip/autogen/top_darjeeling_soc_dbg.rs

sw/host/opentitanlib/src/chip/autogen/top_darjeeling_soc_mbx.rs

a-will · 2025-01-18T15:56:18Z

I think there still some output definitions that need to be dropped by adapting the IP detection in the template files.

Those were not actually focus areas of this PR (which targets just the base addresses, interrupts, and alerts), but I could look at handling them here too.

Edit: ...done. Just a simple application of the same patterns in this PR. A helper function might be in order, but I'll wait for the pass with multiple instances in a design before I bother.

Individually generate C and Rust collateral for all address spaces Fix up alert and irq domains to enable generating collateral where the alerts and irqs don't necessarily generate from devices in the same address space as the CPU. Darjeeling has bus bridges that make servicing those interrupts possible (indirect reachability to the IP). The autogenerated tests will surely be bonkers, but that is to be fixed up later. Co-authored-by: Robert Schilling <[email protected]> Signed-off-by: Alexander Williams <[email protected]>

Razer6

Thanks Alex for removing the unneeded definitions. This looks good to me!

jwnrt · 2025-01-20T16:58:56Z

FYI I think there was some kind of merge skew here with Rust linting:

$ bazel run //quality:rustfmt_fix
Target //quality:rustfmt_fix up-to-date:
  bazel-bin/quality/rustfmt_fix.bash
Error writing files: failed to resolve mod `top_darjeeling_soc_dbg`: opentitan/hw/top_darjeeling/sw/autogen/chip/top_darjeeling_soc_dbg.rs does not exist

I think the last time this went through CI was before 0791e55 was merged

jwnrt · 2025-01-20T17:20:23Z

Can we revert and try again?

It looks like the mod.rs in hw/top_darjeeling/sw/autogen/chip/mod.rs expects to find top_darjeeling_soc_dbg.rs but this file is actually in sw/host/opentitanlib/src/chip/autogen/top_darjeeling_soc_dbg.rs and not included in the mod.rs there.

vogelpi · 2025-01-20T17:50:10Z

I'll give it a try. Thanks @jwnrt for flagging.

This is a follow up of 955bab6 merged with lowRISC#25894. Signed-off-by: Pirmin Vogel <[email protected]>

vogelpi · 2025-01-20T20:32:07Z

util/topgen.py

-                    rsformat_dir / f"{topname}.rs",
-                    helper=rs_helper)
+        # Generate Rust host-side files
+        rsformat_dir = src_tree_top / 'sw/host/opentitanlib/src/chip/autogen'


Note that this change of rsformat_dir inside the address space loop also affects the generation of the following device-side files which is wrong. I've now filed #25953 to fix this and to unblock CI again.

Nice catch! I didn't pay close enough attention to the rust outputs, apparently. That code could be cleaned up a little more...

Don't worry, it's working now :-)

This is a follow up of 955bab6 merged with #25894. Signed-off-by: Pirmin Vogel <[email protected]>

a-will force-pushed the addr-space-output-redo branch from fb78b14 to 062ed30 Compare January 16, 2025 15:28

a-will changed the title ~~WIP Start handling indirectly reachable IPs on same alert / irq domains~~ [topgen] Generate files for all address spaces Jan 16, 2025

a-will force-pushed the addr-space-output-redo branch 2 times, most recently from a7dae8e to 6bcfc4a Compare January 16, 2025 16:36

a-will force-pushed the addr-space-output-redo branch from 6bcfc4a to aaa22be Compare January 16, 2025 17:26

a-will marked this pull request as ready for review January 16, 2025 17:27

a-will requested review from msfschaffner and a team as code owners January 16, 2025 17:27

a-will requested review from cfrantz, pamaury, Razer6, matutem and qmn and removed request for a team January 16, 2025 17:27

pamaury reviewed Jan 17, 2025

View reviewed changes

hw/top_darjeeling/sw/autogen/top_darjeeling_soc_dbg.h Outdated Show resolved Hide resolved

pamaury reviewed Jan 17, 2025

View reviewed changes

hw/top_darjeeling/sw/autogen/top_darjeeling_soc_dbg_memory.ld Outdated Show resolved Hide resolved

Razer6 suggested changes Jan 17, 2025

View reviewed changes

hw/top_darjeeling/rtl/autogen/top_darjeeling_soc_dbg_pkg.sv Outdated Show resolved Hide resolved

a-will force-pushed the addr-space-output-redo branch 2 times, most recently from 1116d0d to 7c16f7a Compare January 18, 2025 00:37

a-will requested review from Razer6 and pamaury January 18, 2025 00:42

Razer6 reviewed Jan 18, 2025

View reviewed changes

a-will force-pushed the addr-space-output-redo branch from 7c16f7a to e4692ec Compare January 18, 2025 16:08

a-will force-pushed the addr-space-output-redo branch from e4692ec to 31a1a0a Compare January 18, 2025 16:12

a-will requested a review from Razer6 January 18, 2025 16:14

Razer6 approved these changes Jan 18, 2025

View reviewed changes

pamaury approved these changes Jan 20, 2025

View reviewed changes

vogelpi added the CI:Rerun Rerun failed CI jobs label Jan 20, 2025

github-actions bot removed the CI:Rerun Rerun failed CI jobs label Jan 20, 2025

vogelpi merged commit 955bab6 into lowRISC:master Jan 20, 2025
40 checks passed

This was referenced Jan 20, 2025

[topgen] Improve support for multiple address spaces #23122

Closed

[top_darjeeling] Instantiate AC Range checks #25862

Open

a-will deleted the addr-space-output-redo branch January 20, 2025 16:32

vogelpi added a commit to vogelpi/opentitan that referenced this pull request Jan 20, 2025

[topgen] Fix path for top_darjeeling_soc_*.rs files

8056be3

This is a follow up of 955bab6 merged with lowRISC#25894. Signed-off-by: Pirmin Vogel <[email protected]>

vogelpi mentioned this pull request Jan 20, 2025

[topgen] Fix path for top_darjeeling_soc_*.rs files #25953

Merged

vogelpi reviewed Jan 20, 2025

View reviewed changes

vogelpi added a commit that referenced this pull request Jan 20, 2025

[topgen] Fix path for top_darjeeling_soc_*.rs files

f7e5178

This is a follow up of 955bab6 merged with #25894. Signed-off-by: Pirmin Vogel <[email protected]>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[topgen] Generate files for all address spaces #25894

[topgen] Generate files for all address spaces #25894

a-will commented Jan 16, 2025 •

edited

Loading

a-will commented Jan 16, 2025

Razer6 left a comment

Razer6 left a comment

a-will commented Jan 18, 2025 •

edited

Loading

Razer6 left a comment

jwnrt commented Jan 20, 2025 •

edited

Loading

jwnrt commented Jan 20, 2025

vogelpi commented Jan 20, 2025

vogelpi Jan 20, 2025

a-will Jan 20, 2025

vogelpi Jan 20, 2025

[topgen] Generate files for all address spaces #25894

[topgen] Generate files for all address spaces #25894

Conversation

a-will commented Jan 16, 2025 • edited Loading

a-will commented Jan 16, 2025

Razer6 left a comment

Choose a reason for hiding this comment

Razer6 left a comment

Choose a reason for hiding this comment

a-will commented Jan 18, 2025 • edited Loading

Razer6 left a comment

Choose a reason for hiding this comment

jwnrt commented Jan 20, 2025 • edited Loading

jwnrt commented Jan 20, 2025

vogelpi commented Jan 20, 2025

vogelpi Jan 20, 2025

Choose a reason for hiding this comment

a-will Jan 20, 2025

Choose a reason for hiding this comment

vogelpi Jan 20, 2025

Choose a reason for hiding this comment

a-will commented Jan 16, 2025 •

edited

Loading

a-will commented Jan 18, 2025 •

edited

Loading

jwnrt commented Jan 20, 2025 •

edited

Loading