[zeeshanlakhani]

Previously, SoftNPU didn't interact with multicast. mc_group_create and mc_port_add returned a stubbed OperationUnsupported, and port remove/destroy were no-ops.

With the merge of bitmap-based replication in oxidecomputer/p4rs#240, SoftNPU can now replicate off of port bitmaps carried in the replication table entry itself. On Tofino, membership subscriptions live in the PRE and change independently of match-action state. On SoftNPU, membership is captured at write time, and a membership mutation involves rewriting these affected replication entries.

Includes:

• asic/SoftNPU:
  • Live replication entries are tracked per group and rewritten when members are added/removed. The replication tracking lock covers both the snapshot and table updates.
  • Source filter, mcast NAT (dst-only, per [multicast] Bitmap-based multicast replication for sidecar-lite sidecar-lite#152), egress decap, and mcast src MAC tables now map onto their sidecar-lite tables.
    • egress.asic_id_to_port maps ASIC port IDs to logical port numbers for indexing the decap bitmap (same as on Tofino).
  • TableRemove is sent before TableAdd because p4rs drops an add whose key already exists and keeps the older entry (action and params).
• dpd/mcast:
  • mcast::reset deletes each known group's entries before the table clears. SoftNPU's Table::clear is still a stub.
  • del_entries_where skips the stale-entry table sweep when the backend cannot list entries (e.g., in SoftNPU).
• deps
  • softnpu-lib + p4rs have been updated on main ([multicast] softnpu with bitmap-based replication softnpu#183, [multicast] Bitmap-based replication and bit-slice assignment for softnpu/a4x2 p4#240). SoftNPU must still loads [multicast] Bitmap-based multicast replication for sidecar-lite sidecar-lite#152 at runtime where used (until its merged).

[cfzimmerman]

Voxel exercises softnpu, right? Can you run a voxel test and post some validation results? I'll stamp once we know it works.

[cfzimmerman]

Out of curiosity, how did you learn this? Are there docs somewhere that explain? Or is this from reading tofino-backend compiler source?

[zeeshanlakhani]

I started from the TNA doc: §5.7.3 "Packet Replication Engine" (pp. 26-27), Ingress only hands the TM (traffic manager) a group id (mcast_grp_a/b, rid, exclusion ids; see §5.4 "Destinations"), and the control plane configures each group's L1 nodes / L2 ports separately (§5.7.3.1 "Multicast group configuration").

The fields we use are here: https://github.com/barefootnetworks/Open-Tofino/blob/5b6ef19873698fc7d9c49cb33fd54c5fca2ecadd/share/p4c/p4include/tofino1_base.p4#L154-L181. And, we interact with them here:

dendrite/asic/src/tofino_asic/mcast.rs

Lines 65 to 140 in 5076359

	bf_mc_mgrp_create(mcast_hdl, dev_id, domain, &mut mgrp_hdl)
	.check_error("creating multicast group")?;
	}
	Ok(mgrp_hdl)
	}
	fn mgrp_destroy(
	mcast_hdl: bf_mc_session_hdl_t,
	dev_id: bf_dev_id_t,
	mgrp_hdl: bf_mc_mgrp_hdl_t,
	) -> AsicResult<()> {
	unsafe {
	bf_mc_mgrp_destroy(mcast_hdl, dev_id, mgrp_hdl)
	.check_error("destroying multicast group")?;
	}
	Ok(())
	}
	fn mgrp_get_count(
	mcast_hdl: &Handle,
	dev_id: bf_dev_id_t,
	mut count: u32,
	) -> AsicResult<usize> {
	unsafe {
	bf_mc_mgrp_get_count(mcast_hdl.bf_get().mcast_hdl, dev_id, &mut count)
	.check_error("getting total count of multicast groups")?;
	}
	Ok(count as usize)
	}
	fn associate_node(
	mcast_hdl: bf_mc_session_hdl_t,
	dev_id: bf_dev_id_t,
	mgrp_hdl: bf_mc_mgrp_hdl_t,
	node_hdl: bf_mc_node_hdl_t,
	exclusion_id: u16,
	) -> AsicResult<()> {
	unsafe {
	bf_mc_associate_node(
	mcast_hdl,
	dev_id,
	mgrp_hdl,
	node_hdl,
	exclusion_id != 0,
	exclusion_id,
	)
	.check_error("associating multicast node")?;
	}
	Ok(())
	}
	fn dissociate_node(
	mcast_hdl: bf_mc_session_hdl_t,
	dev_id: bf_dev_id_t,
	mgrp_hdl: bf_mc_mgrp_hdl_t,
	node_hdl: bf_mc_node_hdl_t,
	) -> AsicResult<()> {
	unsafe {
	bf_mc_dissociate_node(mcast_hdl, dev_id, mgrp_hdl, node_hdl)
	.check_error("dissociating multicast node from group")?;
	}
	Ok(())
	}
	fn node_create(
	mcast_hdl: bf_mc_session_hdl_t,
	dev_id: bf_dev_id_t,
	repl_id: u16,
	portmap: &mut [u8],
	lagmap: &mut [u8],
	) -> AsicResult<u32> {
	let mut node_hdl = 0u32;
	unsafe {
	bf_mc_node_create(

So, yeah, on the sidecar, a port join never rewrites the table entry. SoftNPU, on the other hand, has no PRE, which is why we insert the bitmap into the entry and rewrite it on membership updates.

[zeeshanlakhani]

Voxel exercises softnpu, right? Can you run a voxel test and post some validation results? I'll stamp once we know it works.

Incoming with this line of work. Via oxidecomputer/voxel#75 and past iterations, I demo'ed the prototype set of branches working.

[cfzimmerman]

Left a reply on the table.rs diff. Otherwise looks good to me once we're confident in e2e voxel testing.