This patchset adds comprehensive support for the CN20K NPC
architecture. CN20K introduces significant changes in MCAM layout,
parser design, KPM/KPU mapping, index management, virtual index handling,
and dynamic rule installation. The patches update the AF, PF/VF, and
common layers to correctly support these new capabilities while
preserving compatibility with previous silicon variants.
MCAM on CN20K differs from older designs: the hardware now contains
two vertical banks of depth 8192, and thirty-two horizontal subbanks of
depth 256. Each subbank can be configured as x2 or x4, enabling
256-bit or 512-bit key storage. Several allocation models are added to
support this layout, including contiguous and non-contiguous allocation
with or without reference ranges and priorities.
Parser and extraction logic are also enhanced. CN20K introduces a new
profile model where up to twenty-four extractors may be configured for
each parsing profile. A new KPM profile scheme is added, grouping
sixteen KPUs into eight KPM profiles, each formed by two KPUs.
Support is added for default index allocation for CN20K-specific
MCAM entry structures, virtual index allocation, improved defragmentation,
and TC rule installation by allowing the AF driver to determine
required x2/x4 rule width during flow install.
Ratheesh Kannoth (7):
octeontx2-af: npc: cn20k: Index management
Add CN20K MCAM allocation support. Implements contiguous and
non-contiguous allocation models with ref, limit, contig,
priority, and count support.
octeontx2-af: npc: cn20k: Allocate default MCAM indexes
Allocate default MCAM entries dynamically in descending index
order during NIX LF attach, reducing MCAM wastage
octeontx2-af: npc: cn20k: Prepare for new SoC
Introduce MCAM metadata structure so low-level functions no
longer receive SoC-specific structures directly.
octeontx2-af: npc: cn20k: virtual index support
Add virtual MCAM index allocation and improve CN20K MCAM
defragmentation handling. Track virtual indexes and restore
statistics correctly.
octeontx2-af: npc: cn20k: Allocate MCAM entry for flow installation
Extend install_flow mailbox so AF can determine rule width and
complete allocation and installation in a single exchange.
octeontx2-af: npc: cn20k: add debugfs support
Debugfs entries to show mcam layout and default mcam entry allocations
Legacy debugfs entries are modified to show hardware priority on cn20k
SoC.
octeontx-af: npc: Use common structures
Low level functions should use maximum mcam size array and modify cam0
and cam1. This is a cleanup patch.
Subbaraya Sundeep (1):
octeontx2-pf: cn20k: Add TC rules support
Add full TC dynamic rule support for CN20K. Handle x2/x4 rule
widths, dynamic allocation, and shifting restrictions when
mixed rule sizes exist.
Suman Ghosh (5):
octeontx2-af: npc: cn20k: KPM profile changes
Add support for CN20K KPM profiles. Sixteen KPUs are grouped
into eight KPM configurations to improve resource usage
octeontx2-af: npc: cn20k: Add default profile
Update mkex profile for CN20K and mark unused objects with
may_be_unused to silence compiler warnings.
ocetontx2-af: npc: cn20k: MKEX profile support
Add support for the new CN20K parser profile. Introduces the
extractor-based model with up to twenty-four extractors per
profile.
octeontx2-af: npc: cn20k: Use common APIs
Update common MCAM APIs for CN20K. Add new register handling,
new access algorithms, and CN20K-specific index management.
octeontx2-af: npc: cn20k: Add new mailboxes for CN20K silicon
Add CN20K-specific MCAM mailbox messages for updated mcam_entry
layout and avoid breaking backward compatibility.
.../ethernet/marvell/octeontx2/af/Makefile | 2 +-
.../marvell/octeontx2/af/cn20k/debugfs.c | 273 +
.../marvell/octeontx2/af/cn20k/debugfs.h | 3 +
.../marvell/octeontx2/af/cn20k/mbox_init.c | 17 +
.../ethernet/marvell/octeontx2/af/cn20k/npc.c | 4421 +++++++++++++++++
.../ethernet/marvell/octeontx2/af/cn20k/npc.h | 340 ++
.../ethernet/marvell/octeontx2/af/cn20k/reg.h | 65 +
.../ethernet/marvell/octeontx2/af/common.h | 4 -
.../net/ethernet/marvell/octeontx2/af/mbox.h | 172 +-
.../net/ethernet/marvell/octeontx2/af/npc.h | 2 +
.../marvell/octeontx2/af/npc_profile.h | 84 +-
.../net/ethernet/marvell/octeontx2/af/rvu.c | 94 +-
.../net/ethernet/marvell/octeontx2/af/rvu.h | 18 +-
.../marvell/octeontx2/af/rvu_debugfs.c | 76 +-
.../marvell/octeontx2/af/rvu_devlink.c | 90 +-
.../ethernet/marvell/octeontx2/af/rvu_nix.c | 1 -
.../ethernet/marvell/octeontx2/af/rvu_npc.c | 580 ++-
.../ethernet/marvell/octeontx2/af/rvu_npc.h | 21 +
.../marvell/octeontx2/af/rvu_npc_fs.c | 693 ++-
.../marvell/octeontx2/af/rvu_npc_fs.h | 14 +-
.../marvell/octeontx2/af/rvu_npc_hash.c | 128 +-
.../marvell/octeontx2/af/rvu_npc_hash.h | 2 +-
.../ethernet/marvell/octeontx2/nic/cn20k.c | 265 +
.../ethernet/marvell/octeontx2/nic/cn20k.h | 13 +
.../marvell/octeontx2/nic/otx2_common.h | 35 +
.../marvell/octeontx2/nic/otx2_flows.c | 267 +-
.../ethernet/marvell/octeontx2/nic/otx2_tc.c | 77 +-
27 files changed, 7286 insertions(+), 471 deletions(-)
create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/cn20k/npc.c
create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/cn20k/npc.h
create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.h
--
v1 -> v2: Addressed comments
v2 -> v3: Resolved build errors, addressed comments
v3 -> v4: Address comments from Jakub
https://lore.kernel.org/netdev/20260109054828.1822307-12-rkannoth@marvell.com/T/#t
v4 -> v5: Addressed comments from Jakub
https://lore.kernel.org/netdev/20260113101658.4144610-6-rkannoth@marvell.com/
v5 -> v6: Addressed Jakub comments.
https://lore.kernel.org/netdev/20260126123254.1000480-11-rkannoth@marvell.com/T/#t
v6 -> v7: Addressed Jakub comments.
https://lore.kernel.org/netdev/20260128071235.3577206-2-rkannoth@marvell.com/T/#t
v7 -> v8: Addressed Jakub comments.
https://lore.kernel.org/netdev/20260129083339.264386-12-rkannoth@marvell.com/T/#t
v8 -> v9: Addressed Jakub comments.
https://lore.kernel.org/netdev/20260204182428.05a77b13@kernel.org/
2.43.0