[PATCH net-next 0/3] octeontx2-af: add support for KPU profile customization

[PATCH net-next 0/3] octeontx2-af: add support for KPU profile customization

[Stanislaw Kardach]

Marvell octeontx2 NPC device contains a configurable Kanguroo Parser Unit
(KPU) and CAM match key data extraction (MKEX). The octeontx2-af driver
configures them both to parse a list of standard protocol headers which
are used by netdev driver and other potential applications (i.e.
userspace through VFIO).
The problem arises when users have some custom protocol headers which
they'd like to use in CAM flow matching. If such protocols are publicly
known, they can be added to the built-in KPU configuration (called
"profile" - in npc_profile.h). If not, then there's more benefit in
keeping such changes local to the user.
For that case a mechanism which would allow users to produce a KPU
profile and load it along with octeontx2-af driver is needed. At the same
time such customization has to take care not to break the netdev driver
operation or other applications (that is be discoverable).

Therefore introduce a mechanism for a limited customization of the
built-in KPU profile via a firmware file (layout and contents described
by struct npc_kpu_profile_fwdata). It allows user modification of only a
limited number of top priority KPU entries, while others are configured
from the built-in KPU profile. Additionally by convention users should
only use NPC_LT_Lx_CUSTOMx LTYPE entries in their profiles to change the
meaning of built-in LTYPEs. This way the baseline protocol support is
always available and the impact of potential user errors is minimized.
As MKEX also needs to be modified to take into account any user
protocols, the KPU profile firmware binary contains also that. Netdev
driver and applications have a way to discover applied MKEX settings by
querying RVU AF device via NPC_GET_KEX_CFG MBOX message.
Finally some users might need to modify hardware packet data alignment
behavior and profile contains settings for that too.

First patch ensures that CUSTOMx LTYPEs are not aliased with meaningful
LTYPEs where possible.

Second patch gathers all KPU profile related data into a single struct
and creates an adapter structure which provides an interface to the KPU
profile for the octeontx2-af driver.

Third patch adds logic for loading the KPU profile through kernel
firmware APIs, filling in the customizable entries in the adapter
structure and programming the MKEX from KPU profile.

Changes from v1:
* Remove unnecessary __packed attributes. All structures in profile are
  naturally aligned and only struct npc_kpu_profile_fwdata is padded
  1B at the end, which is expected.
* Make npc_lt_defaults and npc_mkex_default const as they are read-only.
* Save custom KPU entries in separate struct npc_kpu_profile so that the
  default profile can remain read-only and there's no need to allocate
  and memcpy 25kB of default profile when customizations are present.
  The drawbacks are weaker profile abstraction and slightly more
  complicated programming steps.
  This is a result of:
  4a681bf3456f octeontx2-af: Constify npc_kpu_profile_{action,cam}
* Describe in last commit the reason for using a module parameter
  instead of an arbitrary firmware name.

Stanislaw Kardach (3):
  octeontx2-af: fix LD CUSTOM LTYPE aliasing
  octeontx2-af: prepare for custom KPU profiles
  octeontx2-af: add support for custom KPU entries

 .../net/ethernet/marvell/octeontx2/af/npc.h   |  76 +++-
 .../marvell/octeontx2/af/npc_profile.h        | 244 +++++++++++-
 .../net/ethernet/marvell/octeontx2/af/rvu.c   |   6 +
 .../net/ethernet/marvell/octeontx2/af/rvu.h   |  22 ++
 .../ethernet/marvell/octeontx2/af/rvu_nix.c   |  36 +-
 .../ethernet/marvell/octeontx2/af/rvu_npc.c   | 368 +++++++++++-------
 6 files changed, 587 insertions(+), 165 deletions(-)

-- 
2.20.1

[PATCH net-next v2 1/3] octeontx2-af: fix LD CUSTOM LTYPE aliasing

[Stanislaw Kardach]

Since LD contains LTYPE definitions tweaked toward efficient
NIX_AF_RX_FLOW_KEY_ALG(0..31)_FIELD(0..4) usage, the original location
of NPC_LT_LD_CUSTOM0/1 was aliased with MPLS_IN_* definitions.
Moving custom frame to value 6 and 7 removes the aliasing at the cost of
custom frames being also considered when TCP/UDP RSS algo is configured.

However since the goal of CUSTOM frames is to classify them to a
separate set of RQs, this cost is acceptable.

Change-Id: I7d545e97f2bd652a7da77789091f66378f3018e4
Signed-off-by: Stanislaw Kardach <skardach@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/npc.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc.h b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
index 95c646ae7e23..4f36e5cd8ced 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
@@ -77,6 +77,8 @@ enum npc_kpu_ld_ltype {
 	NPC_LT_LD_ICMP,
 	NPC_LT_LD_SCTP,
 	NPC_LT_LD_ICMP6,
+	NPC_LT_LD_CUSTOM0,
+	NPC_LT_LD_CUSTOM1,
 	NPC_LT_LD_IGMP = 8,
 	NPC_LT_LD_ESP,
 	NPC_LT_LD_AH,
@@ -85,8 +87,6 @@ enum npc_kpu_ld_ltype {
 	NPC_LT_LD_NSH,
 	NPC_LT_LD_TU_MPLS_IN_NSH,
 	NPC_LT_LD_TU_MPLS_IN_IP,
-	NPC_LT_LD_CUSTOM0 = 0xE,
-	NPC_LT_LD_CUSTOM1 = 0xF,
 };
 
 enum npc_kpu_le_ltype {
-- 
2.20.1

[PATCH net-next v2 2/3] octeontx2-af: prepare for custom KPU profiles

[Stanislaw Kardach]

Refactor KPU related NPC code to prepare for upcoming KPU customization
functionality. This requires the following:
* Gathering all KPU profile related data into a single adapter struct.
* Converting the built-in MKEX definition to a structured one to
  streamline the MKEX loading.
* Convert LT default register configuration into a structure which may
  later on be customized.
* Add a single point for KPU profile loading, currently using only
  built-in profile.

Change-Id: Ic643d5ebc1a32dcc52fe59384c7a653177f3285c
Signed-off-by: Stanislaw Kardach <skardach@marvell.com>
---
 .../net/ethernet/marvell/octeontx2/af/npc.h   |  36 ++++
 .../marvell/octeontx2/af/npc_profile.h        | 154 +++++++++++++
 .../net/ethernet/marvell/octeontx2/af/rvu.h   |  17 ++
 .../ethernet/marvell/octeontx2/af/rvu_nix.c   |  36 ++--
 .../ethernet/marvell/octeontx2/af/rvu_npc.c   | 202 ++++++------------
 5 files changed, 301 insertions(+), 144 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc.h b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
index 4f36e5cd8ced..3b069f378b2a 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
@@ -296,6 +296,9 @@ struct nix_rx_action {
 #endif
 };
 
+/* NPC_AF_INTFX_KEX_CFG field masks */
+#define NPC_PARSE_NIBBLE		GENMASK_ULL(30, 0)
+
 /* NIX Receive Vtag Action Structure */
 #define VTAG0_VALID_BIT		BIT_ULL(15)
 #define VTAG0_TYPE_MASK		GENMASK_ULL(14, 12)
@@ -320,4 +323,37 @@ struct npc_mcam_kex {
 	u64 intf_ld_flags[NPC_MAX_INTF][NPC_MAX_LD][NPC_MAX_LFL];
 } __packed;
 
+struct npc_lt_def {
+	u8	ltype_mask;
+	u8	ltype_match;
+	u8	lid;
+};
+
+struct npc_lt_def_ipsec {
+	u8	ltype_mask;
+	u8	ltype_match;
+	u8	lid;
+	u8	spi_offset;
+	u8	spi_nz;
+};
+
+struct npc_lt_def_cfg {
+	struct npc_lt_def	rx_ol2;
+	struct npc_lt_def	rx_oip4;
+	struct npc_lt_def	rx_iip4;
+	struct npc_lt_def	rx_oip6;
+	struct npc_lt_def	rx_iip6;
+	struct npc_lt_def	rx_otcp;
+	struct npc_lt_def	rx_itcp;
+	struct npc_lt_def	rx_oudp;
+	struct npc_lt_def	rx_iudp;
+	struct npc_lt_def	rx_osctp;
+	struct npc_lt_def	rx_isctp;
+	struct npc_lt_def_ipsec	rx_ipsec[2];
+	struct npc_lt_def	pck_ol2;
+	struct npc_lt_def	pck_oip4;
+	struct npc_lt_def	pck_oip6;
+	struct npc_lt_def	pck_iip4;
+};
+
 #endif /* NPC_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h b/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
index b29c6689ace2..a67e9ed718e7 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
@@ -140,6 +140,12 @@
 #define NPC_DSA_EXTEND		0x1000
 #define NPC_DSA_EDSA		0x8000
 
+#define NPC_KEXOF_DMAC	8
+#define MKEX_SIGN	0x19bbfdbd15f /* strtoull of "mkexprof" with base:36 */
+#define KEX_LD_CFG(bytesm1, hdr_ofs, ena, flags_ena, key_ofs)		\
+			(((bytesm1) << 16) | ((hdr_ofs) << 8) | ((ena) << 7) | \
+			 ((flags_ena) << 6) | ((key_ofs) & 0x3F))
+
 enum npc_kpu_parser_state {
 	NPC_S_NA = 0,
 	NPC_S_KPU1_ETHER,
@@ -13114,4 +13120,152 @@ static const struct npc_kpu_profile npc_kpu_profiles[] = {
 	},
 };
 
+static const struct npc_lt_def_cfg npc_lt_defaults = {
+	.rx_ol2 = {
+		.lid = NPC_LID_LA,
+		.ltype_match = NPC_LT_LA_ETHER,
+		.ltype_mask = 0x0F,
+	},
+	.rx_oip4 = {
+		.lid = NPC_LID_LC,
+		.ltype_match = NPC_LT_LC_IP,
+		.ltype_mask = 0x0E,
+	},
+	.rx_iip4 = {
+		.lid = NPC_LID_LG,
+		.ltype_match = NPC_LT_LG_TU_IP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_oip6 = {
+		.lid = NPC_LID_LC,
+		.ltype_match = NPC_LT_LC_IP6,
+		.ltype_mask = 0x0E,
+	},
+	.rx_iip6 = {
+		.lid = NPC_LID_LG,
+		.ltype_match = NPC_LT_LG_TU_IP6,
+		.ltype_mask = 0x0F,
+	},
+	.rx_otcp = {
+		.lid = NPC_LID_LD,
+		.ltype_match = NPC_LT_LD_TCP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_itcp = {
+		.lid = NPC_LID_LH,
+		.ltype_match = NPC_LT_LH_TU_TCP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_oudp = {
+		.lid = NPC_LID_LD,
+		.ltype_match = NPC_LT_LD_UDP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_iudp = {
+		.lid = NPC_LID_LH,
+		.ltype_match = NPC_LT_LH_TU_UDP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_osctp = {
+		.lid = NPC_LID_LD,
+		.ltype_match = NPC_LT_LD_SCTP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_isctp = {
+		.lid = NPC_LID_LH,
+		.ltype_match = NPC_LT_LH_TU_SCTP,
+		.ltype_mask = 0x0F,
+	},
+	.rx_ipsec = {
+		{
+			.lid = NPC_LID_LD,
+			.ltype_match = NPC_LT_LD_ESP,
+			.ltype_mask = 0x0F,
+		},
+		{
+			.spi_offset = 8,
+			.lid = NPC_LID_LH,
+			.ltype_match = NPC_LT_LH_TU_ESP,
+			.ltype_mask = 0x0F,
+		},
+	},
+	.pck_ol2 = {
+			.lid = NPC_LID_LA,
+			.ltype_match = NPC_LT_LA_ETHER,
+			.ltype_mask = 0x0F,
+	},
+	.pck_oip4 = {
+			.lid = NPC_LID_LC,
+			.ltype_match = NPC_LT_LC_IP,
+			.ltype_mask = 0x0E,
+	},
+	.pck_iip4 = {
+			.lid = NPC_LID_LG,
+			.ltype_match = NPC_LT_LG_TU_IP,
+			.ltype_mask = 0x0F,
+	},
+};
+
+static const struct npc_mcam_kex npc_mkex_default = {
+	.mkex_sign = MKEX_SIGN,
+	.name = "default",
+	.kpu_version = NPC_KPU_PROFILE_VER,
+	.keyx_cfg = {
+		/* nibble: LA..LE (ltype only) + Channel */
+		[NIX_INTF_RX] = ((u64)NPC_MCAM_KEY_X2 << 32) | 0x49247,
+		[NIX_INTF_TX] = ((u64)NPC_MCAM_KEY_X2 << 32) | ((1ULL << 19) - 1),
+	},
+	.intf_lid_lt_ld = {
+	/* Default RX MCAM KEX profile */
+	[NIX_INTF_RX] = {
+		[NPC_LID_LA] = {
+			/* Layer A: Ethernet: */
+			[NPC_LT_LA_ETHER] = {
+				/* DMAC: 6 bytes, KW1[47:0] */
+				KEX_LD_CFG(0x05, 0x0, 0x1, 0x0, NPC_KEXOF_DMAC),
+				/* Ethertype: 2 bytes, KW0[47:32] */
+				KEX_LD_CFG(0x01, 0xc, 0x1, 0x0, 0x4),
+			},
+		},
+		[NPC_LID_LB] = {
+			/* Layer B: Single VLAN (CTAG) */
+			/* CTAG VLAN[2..3] + Ethertype, 4 bytes, KW0[63:32] */
+			[NPC_LT_LB_CTAG] = {
+				KEX_LD_CFG(0x03, 0x0, 0x1, 0x0, 0x4),
+			},
+			/* Layer B: Stacked VLAN (STAG|QinQ) */
+			[NPC_LT_LB_STAG_QINQ] = {
+				/* CTAG VLAN[2..3] + Ethertype, 4 bytes, KW0[63:32] */
+				KEX_LD_CFG(0x03, 0x4, 0x1, 0x0, 0x4),
+			},
+		},
+		[NPC_LID_LC] = {
+			/* Layer C: IPv4 */
+			[NPC_LT_LC_IP] = {
+				/* SIP+DIP: 8 bytes, KW2[63:0] */
+				KEX_LD_CFG(0x07, 0xc, 0x1, 0x0, 0x10),
+				/* TOS: 1 byte, KW1[63:56] */
+				KEX_LD_CFG(0x0, 0x1, 0x1, 0x0, 0xf),
+			},
+		},
+		[NPC_LID_LD] = {
+			/* Layer D:UDP */
+			[NPC_LT_LD_UDP] = {
+				/* SPORT: 2 bytes, KW3[15:0] */
+				KEX_LD_CFG(0x1, 0x0, 0x1, 0x0, 0x18),
+				/* DPORT: 2 bytes, KW3[31:16] */
+				KEX_LD_CFG(0x1, 0x2, 0x1, 0x0, 0x1a),
+			},
+			/* Layer D:TCP */
+			[NPC_LT_LD_TCP] = {
+				/* SPORT: 2 bytes, KW3[15:0] */
+				KEX_LD_CFG(0x1, 0x0, 0x1, 0x0, 0x18),
+				/* DPORT: 2 bytes, KW3[31:16] */
+				KEX_LD_CFG(0x1, 0x2, 0x1, 0x0, 0x1a),
+			},
+		},
+	},
+	},
+};
+
 #endif /* NPC_PROFILE_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index 05da7a91944a..80adf0d2bd1c 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -291,6 +291,20 @@ struct rvu_fwdata {
 
 struct ptp;
 
+/* KPU profile adapter structure which is used to translate between built-in and firmware KPU
+ * profile structures.
+ */
+struct npc_kpu_profile_adapter {
+	const char			*name;
+	u64				version;
+	const struct npc_lt_def_cfg	*lt_def;
+	const struct npc_kpu_profile_action	*ikpu; /* array[pkinds] */
+	const struct npc_kpu_profile	*kpu; /* array[kpus] */
+	const struct npc_mcam_kex	*mkex;
+	size_t				pkinds;
+	size_t				kpus;
+};
+
 struct rvu {
 	void __iomem		*afreg_base;
 	void __iomem		*pfreg_base;
@@ -339,6 +353,9 @@ struct rvu {
 	/* Firmware data */
 	struct rvu_fwdata	*fwdata;
 
+	/* NPC KPU data */
+	struct npc_kpu_profile_adapter kpu;
+
 	struct ptp		*ptp;
 
 #ifdef CONFIG_DEBUG_FS
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
index 08181fc5f5d4..90cc20e334e9 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c
@@ -3104,6 +3104,7 @@ static int nix_aq_init(struct rvu *rvu, struct rvu_block *block)
 
 int rvu_nix_init(struct rvu *rvu)
 {
+	const struct npc_lt_def_cfg *ltdefs;
 	struct rvu_hwinfo *hw = rvu->hw;
 	struct rvu_block *block;
 	int blkaddr, err;
@@ -3134,6 +3135,7 @@ int rvu_nix_init(struct rvu *rvu)
 		rvu_write64(rvu, blkaddr, NIX_AF_SQM_DBG_CTL_STATUS, cfg);
 	}
 
+	ltdefs = rvu->kpu.lt_def;
 	/* Calibrate X2P bus to check if CGX/LBK links are fine */
 	err = nix_calibrate_x2p(rvu, blkaddr);
 	if (err)
@@ -3181,28 +3183,38 @@ int rvu_nix_init(struct rvu *rvu)
 		 * and validate length and checksums.
 		 */
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OL2,
-			    (NPC_LID_LA << 8) | (NPC_LT_LA_ETHER << 4) | 0x0F);
+			    (ltdefs->rx_ol2.lid << 8) | (ltdefs->rx_ol2.ltype_match << 4) |
+			    ltdefs->rx_ol2.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OIP4,
-			    (NPC_LID_LC << 8) | (NPC_LT_LC_IP << 4) | 0x0F);
+			    (ltdefs->rx_oip4.lid << 8) | (ltdefs->rx_oip4.ltype_match << 4) |
+			    ltdefs->rx_oip4.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_IIP4,
-			    (NPC_LID_LG << 8) | (NPC_LT_LG_TU_IP << 4) | 0x0F);
+			    (ltdefs->rx_iip4.lid << 8) | (ltdefs->rx_iip4.ltype_match << 4) |
+			    ltdefs->rx_iip4.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OIP6,
-			    (NPC_LID_LC << 8) | (NPC_LT_LC_IP6 << 4) | 0x0F);
+			    (ltdefs->rx_oip6.lid << 8) | (ltdefs->rx_oip6.ltype_match << 4) |
+			    ltdefs->rx_oip6.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_IIP6,
-			    (NPC_LID_LG << 8) | (NPC_LT_LG_TU_IP6 << 4) | 0x0F);
+			    (ltdefs->rx_iip6.lid << 8) | (ltdefs->rx_iip6.ltype_match << 4) |
+			    ltdefs->rx_iip6.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OTCP,
-			    (NPC_LID_LD << 8) | (NPC_LT_LD_TCP << 4) | 0x0F);
+			    (ltdefs->rx_otcp.lid << 8) | (ltdefs->rx_otcp.ltype_match << 4) |
+			    ltdefs->rx_otcp.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_ITCP,
-			    (NPC_LID_LH << 8) | (NPC_LT_LH_TU_TCP << 4) | 0x0F);
+			    (ltdefs->rx_itcp.lid << 8) | (ltdefs->rx_itcp.ltype_match << 4) |
+			    ltdefs->rx_itcp.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OUDP,
-			    (NPC_LID_LD << 8) | (NPC_LT_LD_UDP << 4) | 0x0F);
+			    (ltdefs->rx_oudp.lid << 8) | (ltdefs->rx_oudp.ltype_match << 4) |
+			    ltdefs->rx_oudp.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_IUDP,
-			    (NPC_LID_LH << 8) | (NPC_LT_LH_TU_UDP << 4) | 0x0F);
+			    (ltdefs->rx_iudp.lid << 8) | (ltdefs->rx_iudp.ltype_match << 4) |
+			    ltdefs->rx_iudp.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_OSCTP,
-			    (NPC_LID_LD << 8) | (NPC_LT_LD_SCTP << 4) | 0x0F);
+			    (ltdefs->rx_osctp.lid << 8) | (ltdefs->rx_osctp.ltype_match << 4) |
+			    ltdefs->rx_osctp.ltype_mask);
 		rvu_write64(rvu, blkaddr, NIX_AF_RX_DEF_ISCTP,
-			    (NPC_LID_LH << 8) | (NPC_LT_LH_TU_SCTP << 4) |
-			    0x0F);
+			    (ltdefs->rx_isctp.lid << 8) | (ltdefs->rx_isctp.ltype_match << 4) |
+			    ltdefs->rx_isctp.ltype_mask);
 
 		err = nix_rx_flowkey_alg_cfg(rvu, blkaddr);
 		if (err)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 6a1ddb2da1a5..0abe5fd12131 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -29,6 +29,8 @@
 #define NPC_PARSE_RESULT_DMAC_OFFSET	8
 #define NPC_HW_TSTAMP_OFFSET		8
 
+static const char def_pfl_name[] = "default";
+
 static void npc_mcam_free_all_entries(struct rvu *rvu, struct npc_mcam *mcam,
 				      int blkaddr, u16 pcifunc);
 static void npc_mcam_free_all_counters(struct rvu *rvu, struct npc_mcam *mcam,
@@ -448,7 +450,7 @@ void rvu_npc_install_promisc_entry(struct rvu *rvu, u16 pcifunc,
 	entry.kw_mask[0] = 0xFFFULL;
 
 	if (allmulti) {
-		kwi = NPC_PARSE_RESULT_DMAC_OFFSET / sizeof(u64);
+		kwi = NPC_KEXOF_DMAC / sizeof(u64);
 		entry.kw[kwi] = BIT_ULL(40); /* LSB bit of 1st byte in DMAC */
 		entry.kw_mask[kwi] = BIT_ULL(40);
 	}
@@ -718,88 +720,8 @@ void rvu_npc_disable_mcam_entries(struct rvu *rvu, u16 pcifunc, int nixlf)
 	rvu_write64(rvu, blkaddr,			\
 		NPC_AF_INTFX_LDATAX_FLAGSX_CFG(intf, ld, flags), cfg)
 
-#define KEX_LD_CFG(bytesm1, hdr_ofs, ena, flags_ena, key_ofs)		\
-			(((bytesm1) << 16) | ((hdr_ofs) << 8) | ((ena) << 7) | \
-			 ((flags_ena) << 6) | ((key_ofs) & 0x3F))
-
-static void npc_config_ldata_extract(struct rvu *rvu, int blkaddr)
-{
-	struct npc_mcam *mcam = &rvu->hw->mcam;
-	int lid, ltype;
-	int lid_count;
-	u64 cfg;
-
-	cfg = rvu_read64(rvu, blkaddr, NPC_AF_CONST);
-	lid_count = (cfg >> 4) & 0xF;
-
-	/* First clear any existing config i.e
-	 * disable LDATA and FLAGS extraction.
-	 */
-	for (lid = 0; lid < lid_count; lid++) {
-		for (ltype = 0; ltype < 16; ltype++) {
-			SET_KEX_LD(NIX_INTF_RX, lid, ltype, 0, 0ULL);
-			SET_KEX_LD(NIX_INTF_RX, lid, ltype, 1, 0ULL);
-			SET_KEX_LD(NIX_INTF_TX, lid, ltype, 0, 0ULL);
-			SET_KEX_LD(NIX_INTF_TX, lid, ltype, 1, 0ULL);
-
-			SET_KEX_LDFLAGS(NIX_INTF_RX, 0, ltype, 0ULL);
-			SET_KEX_LDFLAGS(NIX_INTF_RX, 1, ltype, 0ULL);
-			SET_KEX_LDFLAGS(NIX_INTF_TX, 0, ltype, 0ULL);
-			SET_KEX_LDFLAGS(NIX_INTF_TX, 1, ltype, 0ULL);
-		}
-	}
-
-	if (mcam->keysize != NPC_MCAM_KEY_X2)
-		return;
-
-	/* Default MCAM KEX profile */
-	/* Layer A: Ethernet: */
-
-	/* DMAC: 6 bytes, KW1[47:0] */
-	cfg = KEX_LD_CFG(0x05, 0x0, 0x1, 0x0, NPC_PARSE_RESULT_DMAC_OFFSET);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LA, NPC_LT_LA_ETHER, 0, cfg);
-
-	/* Ethertype: 2 bytes, KW0[47:32] */
-	cfg = KEX_LD_CFG(0x01, 0xc, 0x1, 0x0, 0x4);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LA, NPC_LT_LA_ETHER, 1, cfg);
-
-	/* Layer B: Single VLAN (CTAG) */
-	/* CTAG VLAN[2..3] + Ethertype, 4 bytes, KW0[63:32] */
-	cfg = KEX_LD_CFG(0x03, 0x0, 0x1, 0x0, 0x4);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LB, NPC_LT_LB_CTAG, 0, cfg);
-
-	/* Layer B: Stacked VLAN (STAG|QinQ) */
-	/* CTAG VLAN[2..3] + Ethertype, 4 bytes, KW0[63:32] */
-	cfg = KEX_LD_CFG(0x03, 0x4, 0x1, 0x0, 0x4);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LB, NPC_LT_LB_STAG_QINQ, 0, cfg);
-
-	/* Layer C: IPv4 */
-	/* SIP+DIP: 8 bytes, KW2[63:0] */
-	cfg = KEX_LD_CFG(0x07, 0xc, 0x1, 0x0, 0x10);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LC, NPC_LT_LC_IP, 0, cfg);
-	/* TOS: 1 byte, KW1[63:56] */
-	cfg = KEX_LD_CFG(0x0, 0x1, 0x1, 0x0, 0xf);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LC, NPC_LT_LC_IP, 1, cfg);
-
-	/* Layer D:UDP */
-	/* SPORT: 2 bytes, KW3[15:0] */
-	cfg = KEX_LD_CFG(0x1, 0x0, 0x1, 0x0, 0x18);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LD, NPC_LT_LD_UDP, 0, cfg);
-	/* DPORT: 2 bytes, KW3[31:16] */
-	cfg = KEX_LD_CFG(0x1, 0x2, 0x1, 0x0, 0x1a);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LD, NPC_LT_LD_UDP, 1, cfg);
-
-	/* Layer D:TCP */
-	/* SPORT: 2 bytes, KW3[15:0] */
-	cfg = KEX_LD_CFG(0x1, 0x0, 0x1, 0x0, 0x18);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LD, NPC_LT_LD_TCP, 0, cfg);
-	/* DPORT: 2 bytes, KW3[31:16] */
-	cfg = KEX_LD_CFG(0x1, 0x2, 0x1, 0x0, 0x1a);
-	SET_KEX_LD(NIX_INTF_RX, NPC_LID_LD, NPC_LT_LD_TCP, 1, cfg);
-}
-
 static void npc_program_mkex_profile(struct rvu *rvu, int blkaddr,
-				     struct npc_mcam_kex *mkex)
+				     const struct npc_mcam_kex *mkex)
 {
 	int lid, lt, ld, fl;
 
@@ -839,34 +761,31 @@ static void npc_program_mkex_profile(struct rvu *rvu, int blkaddr,
 	}
 }
 
-/* strtoull of "mkexprof" with base:36 */
-#define MKEX_SIGN      0x19bbfdbd15f
 #define MKEX_END_SIGN  0xdeadbeef
 
-static void npc_load_mkex_profile(struct rvu *rvu, int blkaddr)
+static void npc_load_mkex_profile(struct rvu *rvu, int blkaddr,
+				  const char *mkex_profile)
 {
-	const char *mkex_profile = rvu->mkex_pfl_name;
 	struct device *dev = &rvu->pdev->dev;
-	void __iomem *mkex_prfl_addr = NULL;
 	struct npc_mcam_kex *mcam_kex;
-	u64 prfl_addr;
-	u64 prfl_sz;
+	void *mkex_prfl_addr = NULL;
+	u64 prfl_addr, prfl_sz;
 
 	/* If user not selected mkex profile */
-	if (!strncmp(mkex_profile, "default", MKEX_NAME_LEN))
-		goto load_default;
+	if (!strncmp(mkex_profile, def_pfl_name, MKEX_NAME_LEN))
+		goto program_mkex;
 
 	if (!rvu->fwdata)
-		goto load_default;
+		goto program_mkex;
 	prfl_addr = rvu->fwdata->mcam_addr;
 	prfl_sz = rvu->fwdata->mcam_sz;
 
 	if (!prfl_addr || !prfl_sz)
-		goto load_default;
+		goto program_mkex;
 
-	mkex_prfl_addr = ioremap_wc(prfl_addr, prfl_sz);
+	mkex_prfl_addr = memremap(prfl_addr, prfl_sz, MEMREMAP_WC);
 	if (!mkex_prfl_addr)
-		goto load_default;
+		goto program_mkex;
 
 	mcam_kex = (struct npc_mcam_kex *)mkex_prfl_addr;
 
@@ -878,31 +797,23 @@ static void npc_load_mkex_profile(struct rvu *rvu, int blkaddr)
 			 * parse nibble enable configuration has to be
 			 * identical for both Rx and Tx interfaces.
 			 */
-			if (is_rvu_96xx_B0(rvu) &&
-			    mcam_kex->keyx_cfg[NIX_INTF_RX] !=
-			    mcam_kex->keyx_cfg[NIX_INTF_TX])
-				goto load_default;
-
-			/* Program selected mkex profile */
-			npc_program_mkex_profile(rvu, blkaddr, mcam_kex);
-
-			goto unmap;
+			if (!is_rvu_96xx_B0(rvu) ||
+			    mcam_kex->keyx_cfg[NIX_INTF_RX] == mcam_kex->keyx_cfg[NIX_INTF_TX])
+				rvu->kpu.mkex = mcam_kex;
+			goto program_mkex;
 		}
 
 		mcam_kex++;
 		prfl_sz -= sizeof(struct npc_mcam_kex);
 	}
-	dev_warn(dev, "Failed to load requested profile: %s\n",
-		 rvu->mkex_pfl_name);
+	dev_warn(dev, "Failed to load requested profile: %s\n", mkex_profile);
 
-load_default:
-	dev_info(rvu->dev, "Using default mkex profile\n");
-	/* Config packet data and flags extraction into PARSE result */
-	npc_config_ldata_extract(rvu, blkaddr);
-
-unmap:
+program_mkex:
+	dev_info(rvu->dev, "Using %s mkex profile\n", rvu->kpu.mkex->name);
+	/* Program selected mkex profile */
+	npc_program_mkex_profile(rvu, blkaddr, rvu->kpu.mkex);
 	if (mkex_prfl_addr)
-		iounmap(mkex_prfl_addr);
+		memunmap(mkex_prfl_addr);
 }
 
 static void npc_config_kpuaction(struct rvu *rvu, int blkaddr,
@@ -1014,6 +925,27 @@ static void npc_program_kpu_profile(struct rvu *rvu, int blkaddr, int kpu,
 	rvu_write64(rvu, blkaddr, NPC_AF_KPUX_CFG(kpu), 0x01);
 }
 
+static int npc_prepare_default_kpu(struct npc_kpu_profile_adapter *profile)
+{
+	profile->name = def_pfl_name;
+	profile->version = NPC_KPU_PROFILE_VER;
+	profile->ikpu = ikpu_action_entries;
+	profile->pkinds = ARRAY_SIZE(ikpu_action_entries);
+	profile->kpu = npc_kpu_profiles;
+	profile->kpus = ARRAY_SIZE(npc_kpu_profiles);
+	profile->lt_def = &npc_lt_defaults;
+	profile->mkex = &npc_mkex_default;
+
+	return 0;
+}
+
+static void npc_load_kpu_profile(struct rvu *rvu)
+{
+	struct npc_kpu_profile_adapter *profile = &rvu->kpu;
+
+	npc_prepare_default_kpu(profile);
+}
+
 static void npc_parser_profile_init(struct rvu *rvu, int blkaddr)
 {
 	struct rvu_hwinfo *hw = rvu->hw;
@@ -1032,25 +964,26 @@ static void npc_parser_profile_init(struct rvu *rvu, int blkaddr)
 		rvu_write64(rvu, blkaddr, NPC_AF_KPUX_CFG(idx), 0x00);
 	}
 
+	/* Load and customize KPU profile. */
+	npc_load_kpu_profile(rvu);
+
 	/* First program IKPU profile i.e PKIND configs.
 	 * Check HW max count to avoid configuring junk or
 	 * writing to unsupported CSR addresses.
 	 */
 	pkind = &hw->pkind;
-	num_pkinds = ARRAY_SIZE(ikpu_action_entries);
+	num_pkinds = rvu->kpu.pkinds;
 	num_pkinds = min_t(int, pkind->rsrc.max, num_pkinds);
 
 	for (idx = 0; idx < num_pkinds; idx++)
-		npc_config_kpuaction(rvu, blkaddr,
-				     &ikpu_action_entries[idx], 0, idx, true);
+		npc_config_kpuaction(rvu, blkaddr, &rvu->kpu.ikpu[idx], 0, idx, true);
 
 	/* Program KPU CAM and Action profiles */
-	num_kpus = ARRAY_SIZE(npc_kpu_profiles);
+	num_kpus = rvu->kpu.kpus;
 	num_kpus = min_t(int, hw->npc_kpus, num_kpus);
 
 	for (idx = 0; idx < num_kpus; idx++)
-		npc_program_kpu_profile(rvu, blkaddr,
-					idx, &npc_kpu_profiles[idx]);
+		npc_program_kpu_profile(rvu, blkaddr, idx, &rvu->kpu.kpu[idx]);
 }
 
 static int npc_mcam_rsrcs_init(struct rvu *rvu, int blkaddr)
@@ -1175,11 +1108,11 @@ static int npc_mcam_rsrcs_init(struct rvu *rvu, int blkaddr)
 
 int rvu_npc_init(struct rvu *rvu)
 {
+	struct npc_kpu_profile_adapter *kpu = &rvu->kpu;
 	struct npc_pkind *pkind = &rvu->hw->pkind;
 	struct npc_mcam *mcam = &rvu->hw->mcam;
-	u64 keyz = NPC_MCAM_KEY_X2;
+	u64 cfg, nibble_ena, rx_kex, tx_kex;
 	int blkaddr, entry, bank, err;
-	u64 cfg, nibble_ena;
 
 	blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0);
 	if (blkaddr < 0) {
@@ -1213,13 +1146,16 @@ int rvu_npc_init(struct rvu *rvu)
 
 	/* Config Outer L2, IPv4's NPC layer info */
 	rvu_write64(rvu, blkaddr, NPC_AF_PCK_DEF_OL2,
-		    (NPC_LID_LA << 8) | (NPC_LT_LA_ETHER << 4) | 0x0F);
+		    (kpu->lt_def->pck_ol2.lid << 8) | (kpu->lt_def->pck_ol2.ltype_match << 4) |
+		    kpu->lt_def->pck_ol2.ltype_mask);
 	rvu_write64(rvu, blkaddr, NPC_AF_PCK_DEF_OIP4,
-		    (NPC_LID_LC << 8) | (NPC_LT_LC_IP << 4) | 0x0F);
+		    (kpu->lt_def->pck_oip4.lid << 8) | (kpu->lt_def->pck_oip4.ltype_match << 4) |
+		    kpu->lt_def->pck_oip4.ltype_mask);
 
 	/* Config Inner IPV4 NPC layer info */
 	rvu_write64(rvu, blkaddr, NPC_AF_PCK_DEF_IIP4,
-		    (NPC_LID_LG << 8) | (NPC_LT_LG_TU_IP << 4) | 0x0F);
+		    (kpu->lt_def->pck_iip4.lid << 8) | (kpu->lt_def->pck_iip4.ltype_match << 4) |
+		    kpu->lt_def->pck_iip4.ltype_mask);
 
 	/* Enable below for Rx pkts.
 	 * - Outer IPv4 header checksum validation.
@@ -1235,23 +1171,25 @@ int rvu_npc_init(struct rvu *rvu)
 	/* Set RX and TX side MCAM search key size.
 	 * LA..LD (ltype only) + Channel
 	 */
-	nibble_ena = 0x49247;
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX),
-			((keyz & 0x3) << 32) | nibble_ena);
+	rx_kex = npc_mkex_default.keyx_cfg[NIX_INTF_RX];
+	tx_kex = npc_mkex_default.keyx_cfg[NIX_INTF_TX];
+	nibble_ena = FIELD_GET(NPC_PARSE_NIBBLE, rx_kex);
+	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX), rx_kex);
 	/* Due to an errata (35786) in A0 pass silicon, parse nibble enable
 	 * configuration has to be identical for both Rx and Tx interfaces.
 	 */
-	if (!is_rvu_96xx_B0(rvu))
-		nibble_ena = (1ULL << 19) - 1;
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_TX),
-			((keyz & 0x3) << 32) | nibble_ena);
+	if (is_rvu_96xx_B0(rvu)) {
+		tx_kex &= ~NPC_PARSE_NIBBLE;
+		tx_kex |= FIELD_PREP(NPC_PARSE_NIBBLE, nibble_ena);
+	}
+	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_TX), tx_kex);
 
 	err = npc_mcam_rsrcs_init(rvu, blkaddr);
 	if (err)
 		return err;
 
 	/* Configure MKEX profile */
-	npc_load_mkex_profile(rvu, blkaddr);
+	npc_load_mkex_profile(rvu, blkaddr, rvu->mkex_pfl_name);
 
 	/* Set TX miss action to UCAST_DEFAULT i.e
 	 * transmit the packet on NIX LF SQ's default channel.
-- 
2.20.1

[PATCH net-next v2 3/3] octeontx2-af: add support for custom KPU entries

[Stanislaw Kardach]

Add ability to load a set of custom KPU entries via firmware APIs. This
allows for flexible support for custom protocol parsing and CAM matching.

The firmware file name is specified by a module parameter (kpu_profile)
to allow re-using the same kernel and initramfs package on nodes in
different parts of the network where support for different protocols is
required.

AF driver will attempt to load the profile from the firmware file and
verify if it can fit hardware capabilities. If not, it will revert to
the built-in profile.

Next it will read the maximum first KPU_MAX_CST_LT (2) custom entries
from the firmware image. Those will be later programmed at the top of
each KPU after the built-in profile entries have been programmed.
The built-in profile is amended to always contain KPU_MAX_CSR_LT first
no-match entries and AF driver will disable those in the KPU unless
custom profile is loaded.

By convention the custom entries should only utilize NPC_LT_Lx_CUSTOMy
LTYPEs to maintain interoperability with netdev driver.

In relation to MKEX profile, the order of load priority is as follows:

1. Profile in loaded KPU profile.
2. Profile defined by mkex_profile parameter.
3. Built-in MKEX profile.

Firmware image contains also a list of default protocol overrides to
allow for custom protocols to be used there. This allows to apply some
packet alignment fixups for custom protocols at the cost of HW protocol
checks.

Change-Id: I5e25485c9ac40edce4bf342740c712688fb4f266
Signed-off-by: Stanislaw Kardach <skardach@marvell.com>
---
 .../net/ethernet/marvell/octeontx2/af/npc.h   |  36 ++++
 .../marvell/octeontx2/af/npc_profile.h        |  90 +++++++-
 .../net/ethernet/marvell/octeontx2/af/rvu.c   |   6 +
 .../net/ethernet/marvell/octeontx2/af/rvu.h   |   5 +
 .../ethernet/marvell/octeontx2/af/rvu_npc.c   | 198 +++++++++++++++---
 5 files changed, 300 insertions(+), 35 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc.h b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
index 3b069f378b2a..8aeef8715504 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
@@ -323,6 +323,15 @@ struct npc_mcam_kex {
 	u64 intf_ld_flags[NPC_MAX_INTF][NPC_MAX_LD][NPC_MAX_LFL];
 } __packed;
 
+struct npc_kpu_fwdata {
+	int	entries;
+	/* What follows is:
+	 * struct npc_kpu_profile_cam[entries];
+	 * struct npc_kpu_profile_action[entries];
+	 */
+	u8	data[0];
+};
+
 struct npc_lt_def {
 	u8	ltype_mask;
 	u8	ltype_match;
@@ -356,4 +365,31 @@ struct npc_lt_def_cfg {
 	struct npc_lt_def	pck_iip4;
 };
 
+/* Loadable KPU profile firmware data */
+struct npc_kpu_profile_fwdata {
+/* strtoull of "kpuprof" with base:36 */
+#define KPU_SIGN	0x00666f727075706b
+#define KPU_NAME_LEN	32
+/** Maximum number of custom KPU entries supported by the built-in profile. */
+#define KPU_MAX_CST_ENT	2
+	/* KPU Profle Header */
+	__le64	signature; /* "kpuprof\0" (8 bytes/ASCII characters) */
+	u8	name[KPU_NAME_LEN]; /* KPU Profile name */
+	__le64	version; /* KPU profile version */
+	u8	kpus;
+	u8	reserved[7];
+
+	/* Default MKEX profile to be used with this KPU profile. Format is same as for the MKEX
+	 * profile to streamline processing.
+	 */
+	struct npc_mcam_kex	mkex;
+	/* LTYPE values for specific HW offloaded protocols. */
+	struct npc_lt_def_cfg	lt_def;
+	/* Dynamically sized data:
+	 *  Custom KPU CAM and ACTION configuration entries.
+	 * struct npc_kpu_fwdata kpu[kpus];
+	 */
+	u8	data[0];
+};
+
 #endif /* NPC_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h b/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
index a67e9ed718e7..32720838122d 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc_profile.h
@@ -11,7 +11,10 @@
 #ifndef NPC_PROFILE_H
 #define NPC_PROFILE_H
 
-#define NPC_KPU_PROFILE_VER    0x0000000100050000
+#define NPC_KPU_PROFILE_VER	0x0000000100050000
+#define NPC_KPU_VER_MAJ(ver)	(u16)(((ver) >> 32) & 0xFFFF)
+#define NPC_KPU_VER_MIN(ver)	(u16)(((ver) >> 16) & 0xFFFF)
+#define NPC_KPU_VER_PATCH(ver)	(u16)((ver) & 0xFFFF)
 
 #define NPC_IH_W		0x8000
 #define NPC_IH_UTAG		0x2000
@@ -424,6 +427,27 @@ enum NPC_ERRLEV_E {
 	NPC_ERRLEV_ENUM_LAST = 16,
 };
 
+#define NPC_KPU_NOP_CAM		\
+	{			\
+		NPC_S_NA, 0xff,	\
+		0x0000,		\
+		0x0000,		\
+		0x0000,		\
+		0x0000,		\
+		0x0000,		\
+		0x0000,		\
+	}
+
+#define NPC_KPU_NOP_ACTION			\
+	{					\
+		NPC_ERRLEV_RE, NPC_EC_NOERR,	\
+		0, 0, 0, 0, 0,			\
+		NPC_S_NA, 0, 0,			\
+		NPC_LID_LA, NPC_LT_NA,		\
+		0,				\
+		0, 0, 0, 0,			\
+	}
+
 static const struct npc_kpu_profile_action ikpu_action_entries[] = {
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
@@ -1004,6 +1028,8 @@ static const struct npc_kpu_profile_action ikpu_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu1_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU1_ETHER, 0xff,
 		NPC_ETYPE_IP,
@@ -1673,6 +1699,8 @@ static const struct npc_kpu_profile_cam kpu1_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu2_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU2_CTAG, 0xff,
 		NPC_ETYPE_IP,
@@ -2801,6 +2829,8 @@ static const struct npc_kpu_profile_cam kpu2_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu3_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU3_CTAG, 0xff,
 		NPC_ETYPE_IP,
@@ -3920,6 +3950,8 @@ static const struct npc_kpu_profile_cam kpu3_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu4_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU4_MPLS, 0xff,
 		NPC_MPLS_S,
@@ -4013,6 +4045,8 @@ static const struct npc_kpu_profile_cam kpu4_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu5_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU5_IP, 0xff,
 		0x0000,
@@ -4583,6 +4617,8 @@ static const struct npc_kpu_profile_cam kpu5_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu6_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU6_IP6_EXT, 0xff,
 		0x0000,
@@ -4928,6 +4964,8 @@ static const struct npc_kpu_profile_cam kpu6_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu7_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU7_IP6_EXT, 0xff,
 		0x0000,
@@ -5147,6 +5185,8 @@ static const struct npc_kpu_profile_cam kpu7_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu8_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU8_TCP, 0xff,
 		0x0000,
@@ -5879,6 +5919,8 @@ static const struct npc_kpu_profile_cam kpu8_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu9_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU9_TU_MPLS_IN_GRE, 0xff,
 		NPC_MPLS_S,
@@ -6341,6 +6383,8 @@ static const struct npc_kpu_profile_cam kpu9_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu10_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU10_TU_MPLS, 0xff,
 		NPC_MPLS_S,
@@ -6506,6 +6550,8 @@ static const struct npc_kpu_profile_cam kpu10_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu11_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU11_TU_ETHER, 0xff,
 		NPC_ETYPE_IP,
@@ -6815,6 +6861,8 @@ static const struct npc_kpu_profile_cam kpu11_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu12_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU12_TU_IP, 0xff,
 		NPC_IPNH_TCP,
@@ -7070,6 +7118,8 @@ static const struct npc_kpu_profile_cam kpu12_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu13_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU13_TU_IP6_EXT, 0xff,
 		0x0000,
@@ -7082,6 +7132,8 @@ static const struct npc_kpu_profile_cam kpu13_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu14_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU14_TU_IP6_EXT, 0xff,
 		0x0000,
@@ -7094,6 +7146,8 @@ static const struct npc_kpu_profile_cam kpu14_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu15_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU15_TU_TCP, 0xff,
 		0x0000,
@@ -7295,6 +7349,8 @@ static const struct npc_kpu_profile_cam kpu15_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_cam kpu16_cam_entries[] = {
+	NPC_KPU_NOP_CAM,
+	NPC_KPU_NOP_CAM,
 	{
 		NPC_S_KPU16_TCP_DATA, 0xff,
 		0x0000,
@@ -7352,6 +7408,8 @@ static const struct npc_kpu_profile_cam kpu16_cam_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu1_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		8, 0, 6, 3, 0,
@@ -7969,6 +8027,8 @@ static const struct npc_kpu_profile_action kpu1_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu2_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		8, 0, 6, 2, 0,
@@ -8972,6 +9032,8 @@ static const struct npc_kpu_profile_action kpu2_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu3_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		8, 0, 6, 1, 0,
@@ -9967,6 +10029,8 @@ static const struct npc_kpu_profile_action kpu3_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu4_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 0,
@@ -10050,6 +10114,8 @@ static const struct npc_kpu_profile_action kpu4_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu5_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_LC, NPC_EC_IP_TTL_0,
 		0, 0, 0, 0, 1,
@@ -10557,6 +10623,8 @@ static const struct npc_kpu_profile_action kpu5_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu6_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 1,
@@ -10864,6 +10932,8 @@ static const struct npc_kpu_profile_action kpu6_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu7_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 1,
@@ -11059,6 +11129,8 @@ static const struct npc_kpu_profile_action kpu7_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu8_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_LD, NPC_EC_TCP_FLAGS_FIN_ONLY,
 		0, 0, 0, 0, 1,
@@ -11710,6 +11782,8 @@ static const struct npc_kpu_profile_action kpu8_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu9_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 0,
@@ -12121,6 +12195,8 @@ static const struct npc_kpu_profile_action kpu9_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu10_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		8, 0, 6, 1, 0,
@@ -12268,6 +12344,8 @@ static const struct npc_kpu_profile_action kpu10_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu11_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		8, 0, 6, 0, 0,
@@ -12543,6 +12621,8 @@ static const struct npc_kpu_profile_action kpu11_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu12_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		2, 12, 0, 2, 0,
@@ -12770,6 +12850,8 @@ static const struct npc_kpu_profile_action kpu12_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu13_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 1,
@@ -12781,6 +12863,8 @@ static const struct npc_kpu_profile_action kpu13_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu14_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 1,
@@ -12792,6 +12876,8 @@ static const struct npc_kpu_profile_action kpu14_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu15_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_LG, NPC_EC_TCP_FLAGS_FIN_ONLY,
 		0, 0, 0, 0, 1,
@@ -12971,6 +13057,8 @@ static const struct npc_kpu_profile_action kpu15_action_entries[] = {
 };
 
 static const struct npc_kpu_profile_action kpu16_action_entries[] = {
+	NPC_KPU_NOP_ACTION,
+	NPC_KPU_NOP_ACTION,
 	{
 		NPC_ERRLEV_RE, NPC_EC_NOERR,
 		0, 0, 0, 0, 1,
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
index c3ef73ae782c..a737f61662dd 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
@@ -55,6 +55,10 @@ static char *mkex_profile; /* MKEX profile name */
 module_param(mkex_profile, charp, 0000);
 MODULE_PARM_DESC(mkex_profile, "MKEX profile name string");
 
+static char *kpu_profile; /* KPU profile name */
+module_param(kpu_profile, charp, 0000);
+MODULE_PARM_DESC(kpu_profile, "KPU profile name string");
+
 static void rvu_setup_hw_capabilities(struct rvu *rvu)
 {
 	struct rvu_hwinfo *hw = rvu->hw;
@@ -2524,6 +2528,8 @@ static void rvu_update_module_params(struct rvu *rvu)
 
 	strscpy(rvu->mkex_pfl_name,
 		mkex_profile ? mkex_profile : default_pfl_name, MKEX_NAME_LEN);
+	strscpy(rvu->kpu_pfl_name,
+		kpu_profile ? kpu_profile : default_pfl_name, KPU_NAME_LEN);
 }
 
 static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index 80adf0d2bd1c..95e3b5131c0e 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -15,6 +15,7 @@
 #include "rvu_struct.h"
 #include "common.h"
 #include "mbox.h"
+#include "npc.h"
 
 /* PCI device IDs */
 #define	PCI_DEVID_OCTEONTX2_RVU_AF		0xA065
@@ -300,6 +301,7 @@ struct npc_kpu_profile_adapter {
 	const struct npc_lt_def_cfg	*lt_def;
 	const struct npc_kpu_profile_action	*ikpu; /* array[pkinds] */
 	const struct npc_kpu_profile	*kpu; /* array[kpus] */
+	struct npc_kpu_profile		*cst_kpu; /* array[kpus] */
 	const struct npc_mcam_kex	*mkex;
 	size_t				pkinds;
 	size_t				kpus;
@@ -349,9 +351,12 @@ struct rvu {
 	struct mutex		cgx_cfg_lock; /* serialize cgx configuration */
 
 	char mkex_pfl_name[MKEX_NAME_LEN]; /* Configured MKEX profile name */
+	char kpu_pfl_name[KPU_NAME_LEN]; /* Configured KPU profile name */
 
 	/* Firmware data */
 	struct rvu_fwdata	*fwdata;
+	void			*kpu_fwdata;
+	size_t			kpu_fwdata_sz;
 
 	/* NPC KPU data */
 	struct npc_kpu_profile_adapter kpu;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 0abe5fd12131..1ee390991846 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -11,6 +11,8 @@
 #include <linux/bitfield.h>
 #include <linux/module.h>
 #include <linux/pci.h>
+#include <linux/firmware.h>
+#include <linux/stddef.h>
 
 #include "rvu_struct.h"
 #include "rvu_reg.h"
@@ -723,12 +725,22 @@ void rvu_npc_disable_mcam_entries(struct rvu *rvu, u16 pcifunc, int nixlf)
 static void npc_program_mkex_profile(struct rvu *rvu, int blkaddr,
 				     const struct npc_mcam_kex *mkex)
 {
+	u64 nibble_ena, rx_kex, tx_kex;
 	int lid, lt, ld, fl;
 
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX),
-		    mkex->keyx_cfg[NIX_INTF_RX]);
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_TX),
-		    mkex->keyx_cfg[NIX_INTF_TX]);
+	rx_kex = mkex->keyx_cfg[NIX_INTF_RX];
+	tx_kex = mkex->keyx_cfg[NIX_INTF_TX];
+	nibble_ena = FIELD_GET(NPC_PARSE_NIBBLE, rx_kex);
+	/* Due to an errata (35786) in A0/B0 pass silicon, parse nibble enable
+	 * configuration has to be identical for both Rx and Tx interfaces.
+	 */
+	if (is_rvu_96xx_B0(rvu)) {
+		tx_kex &= ~NPC_PARSE_NIBBLE;
+		tx_kex |= FIELD_PREP(NPC_PARSE_NIBBLE, nibble_ena);
+	}
+
+	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX), rx_kex);
+	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_TX), tx_kex);
 
 	for (ld = 0; ld < NPC_MAX_LD; ld++)
 		rvu_write64(rvu, blkaddr, NPC_AF_KEX_LDATAX_FLAGS_CFG(ld),
@@ -771,8 +783,14 @@ static void npc_load_mkex_profile(struct rvu *rvu, int blkaddr,
 	void *mkex_prfl_addr = NULL;
 	u64 prfl_addr, prfl_sz;
 
+	/* Order of precedence (high to low):
+	 * 1. Embedded in custom KPU profile firmware via kpu_profile param.
+	 * 2. Via mkex_profile, loaded from ATF if KPU profile wasn't loaded.
+	 * 3. Built-in KEX profile from npc_mkex_default.
+	 */
 	/* If user not selected mkex profile */
-	if (!strncmp(mkex_profile, def_pfl_name, MKEX_NAME_LEN))
+	if (rvu->kpu_fwdata_sz ||
+	    !strncmp(mkex_profile, def_pfl_name, MKEX_NAME_LEN))
 		goto program_mkex;
 
 	if (!rvu->fwdata)
@@ -793,13 +811,7 @@ static void npc_load_mkex_profile(struct rvu *rvu, int blkaddr,
 		/* Compare with mkex mod_param name string */
 		if (mcam_kex->mkex_sign == MKEX_SIGN &&
 		    !strncmp(mcam_kex->name, mkex_profile, MKEX_NAME_LEN)) {
-			/* Due to an errata (35786) in A0/B0 pass silicon,
-			 * parse nibble enable configuration has to be
-			 * identical for both Rx and Tx interfaces.
-			 */
-			if (!is_rvu_96xx_B0(rvu) ||
-			    mcam_kex->keyx_cfg[NIX_INTF_RX] == mcam_kex->keyx_cfg[NIX_INTF_TX])
-				rvu->kpu.mkex = mcam_kex;
+			rvu->kpu.mkex = mcam_kex;
 			goto program_mkex;
 		}
 
@@ -910,11 +922,21 @@ static void npc_program_kpu_profile(struct rvu *rvu, int blkaddr, int kpu,
 	for (entry = 0; entry < num_entries; entry++)
 		npc_config_kpuaction(rvu, blkaddr, &profile->action[entry],
 				     kpu, entry, false);
+}
+
+static void npc_enable_kpu_profile(struct rvu *rvu, int blkaddr, int kpu,
+				   const struct npc_kpu_profile *profile)
+{
+	int num_entries;
+	u64 entry_mask;
 
-	/* Enable all programmed entries */
 	num_entries = min_t(int, profile->action_entries, profile->cam_entries);
+	entry_mask = enable_mask(num_entries);
+	/* Disable first KPU_MAX_CST_ENT entries for built-in profile */
+	if (!rvu->kpu.cst_kpu)
+		entry_mask |= GENMASK_ULL(KPU_MAX_CST_ENT - 1, 0);
 	rvu_write64(rvu, blkaddr,
-		    NPC_AF_KPUX_ENTRY_DISX(kpu, 0), enable_mask(num_entries));
+		    NPC_AF_KPUX_ENTRY_DISX(kpu, 0), entry_mask);
 	if (num_entries > 64) {
 		rvu_write64(rvu, blkaddr,
 			    NPC_AF_KPUX_ENTRY_DISX(kpu, 1),
@@ -927,6 +949,7 @@ static void npc_program_kpu_profile(struct rvu *rvu, int blkaddr, int kpu,
 
 static int npc_prepare_default_kpu(struct npc_kpu_profile_adapter *profile)
 {
+	profile->cst_kpu = NULL;
 	profile->name = def_pfl_name;
 	profile->version = NPC_KPU_PROFILE_VER;
 	profile->ikpu = ikpu_action_entries;
@@ -939,10 +962,126 @@ static int npc_prepare_default_kpu(struct npc_kpu_profile_adapter *profile)
 	return 0;
 }
 
+static int npc_apply_custom_kpu(struct rvu *rvu, struct npc_kpu_profile_adapter *profile)
+{
+	size_t hdr_sz = sizeof(struct npc_kpu_profile_fwdata), offset = 0;
+	struct npc_kpu_profile_fwdata *fw = rvu->kpu_fwdata;
+	struct npc_kpu_profile_action *action;
+	struct npc_kpu_profile_cam *cam;
+	struct npc_kpu_fwdata *fw_kpu;
+	u16 kpu, entry;
+	int entries;
+
+	if (rvu->kpu_fwdata_sz < hdr_sz) {
+		dev_warn(rvu->dev, "Invalid KPU profile size\n");
+		return -EINVAL;
+	}
+	if (le64_to_cpu(fw->signature) != KPU_SIGN) {
+		dev_warn(rvu->dev, "Invalid KPU profile signature %llx\n",
+			 le64_to_cpu(fw->signature));
+		return -EINVAL;
+	}
+	profile->name = fw->name;
+	profile->version = le64_to_cpu(fw->version);
+	profile->mkex = &fw->mkex;
+
+	/* Verify if the using known profile structure */
+	if (NPC_KPU_VER_MAJ(profile->version) >
+	    NPC_KPU_VER_MAJ(NPC_KPU_PROFILE_VER)) {
+		dev_warn(rvu->dev, "Not supported Major version: %d > %d\n",
+			 NPC_KPU_VER_MAJ(profile->version), NPC_KPU_VER_MAJ(NPC_KPU_PROFILE_VER));
+		return -EINVAL;
+	}
+	/* Verify if profile fits the HW */
+	if (fw->kpus > profile->kpus) {
+		dev_warn(rvu->dev, "Not enough KPUs: %d > %ld\n", fw->kpus, profile->kpus);
+		return -EINVAL;
+	}
+	/* Update adapter structure and ensure endianness where needed. */
+	profile->lt_def = &fw->lt_def;
+
+	for (kpu = 0; kpu < fw->kpus; kpu++) {
+		if (rvu->kpu_fwdata_sz < hdr_sz + offset) {
+			dev_warn(rvu->dev, "Profile size mismatch on KPU%i parsing.\n", kpu + 1);
+			return -EINVAL;
+		}
+		fw_kpu = (struct npc_kpu_fwdata *)(fw->data + offset);
+		if (fw_kpu->entries > KPU_MAX_CST_ENT)
+			dev_warn(rvu->dev, "Too many custom entries on KPU%d: %d > %d\n", kpu,
+				 fw_kpu->entries, KPU_MAX_CST_ENT);
+		entries = min(fw_kpu->entries, KPU_MAX_CST_ENT);
+		cam = (struct npc_kpu_profile_cam *)fw_kpu->data;
+		offset += sizeof(*fw_kpu) + fw_kpu->entries * sizeof(*cam);
+		action = (struct npc_kpu_profile_action *)(fw->data + offset);
+		offset += fw_kpu->entries * sizeof(*action);
+		/* Fix endianness */
+		for (entry = 0; entry < entries; entry++) {
+			cam[entry].dp0 = le16_to_cpu((__force __le16)cam[entry].dp0);
+			cam[entry].dp0_mask = le16_to_cpu((__force __le16)cam[entry].dp0_mask);
+			cam[entry].dp1 = le16_to_cpu((__force __le16)cam[entry].dp1);
+			cam[entry].dp1_mask = le16_to_cpu((__force __le16)cam[entry].dp1_mask);
+			cam[entry].dp2 = le16_to_cpu((__force __le16)cam[entry].dp2);
+			cam[entry].dp2_mask = le16_to_cpu((__force __le16)cam[entry].dp2_mask);
+		}
+	}
+
+	profile->cst_kpu = kcalloc(fw->kpus, sizeof(struct npc_kpu_profile), GFP_KERNEL);
+	if (!profile->cst_kpu)
+		return -ENOMEM;
+	/* Update the adapter after all data is validated and ready */
+	for (kpu = 0; kpu < fw->kpus; kpu++) {
+		fw_kpu = (struct npc_kpu_fwdata *)(fw->data + offset);
+		entries = min(fw_kpu->entries, KPU_MAX_CST_ENT);
+		cam = (struct npc_kpu_profile_cam *)fw_kpu->data;
+		offset += sizeof(*fw_kpu) + fw_kpu->entries * sizeof(*cam);
+		action = (struct npc_kpu_profile_action *)(fw->data + offset);
+		offset += fw_kpu->entries * sizeof(*action);
+		profile->cst_kpu[kpu].cam_entries = entries;
+		profile->cst_kpu[kpu].action_entries = entries;
+		profile->cst_kpu[kpu].cam = cam;
+		profile->cst_kpu[kpu].action = action;
+	}
+
+	return 0;
+}
+
 static void npc_load_kpu_profile(struct rvu *rvu)
 {
 	struct npc_kpu_profile_adapter *profile = &rvu->kpu;
+	const char *kpu_profile = rvu->kpu_pfl_name;
+	const struct firmware *fw = NULL;
+
+	/* If user not specified profile customization */
+	if (!strncmp(kpu_profile, def_pfl_name, KPU_NAME_LEN))
+		goto revert_to_default;
+	/* First prepare default KPU, then we'll customize top entries. */
+	npc_prepare_default_kpu(profile);
+
+	dev_info(rvu->dev, "Loading KPU profile from firmware: %s\n", kpu_profile);
+	if (!request_firmware(&fw, kpu_profile, rvu->dev)) {
+		rvu->kpu_fwdata = kzalloc(fw->size, GFP_KERNEL);
+		if (rvu->kpu_fwdata) {
+			memcpy(rvu->kpu_fwdata, fw->data, fw->size);
+			rvu->kpu_fwdata_sz = fw->size;
+		}
+	}
+	release_firmware(fw);
+
+	/* Apply profile customization if firmware was loaded. */
+	if (!rvu->kpu_fwdata_sz || npc_apply_custom_kpu(rvu, profile)) {
+		dev_warn(rvu->dev, "Can't load KPU profile %s. Using default.\n", kpu_profile);
+		kfree(rvu->kpu_fwdata);
+		rvu->kpu_fwdata = NULL;
+		goto revert_to_default;
+	}
+
+	dev_info(rvu->dev, "Using custom profile '%s', version %d.%d.%d\n", profile->name,
+		 NPC_KPU_VER_MAJ(profile->version), NPC_KPU_VER_MIN(profile->version),
+		 NPC_KPU_VER_PATCH(profile->version));
 
+	return;
+
+revert_to_default:
 	npc_prepare_default_kpu(profile);
 }
 
@@ -982,8 +1121,12 @@ static void npc_parser_profile_init(struct rvu *rvu, int blkaddr)
 	num_kpus = rvu->kpu.kpus;
 	num_kpus = min_t(int, hw->npc_kpus, num_kpus);
 
-	for (idx = 0; idx < num_kpus; idx++)
+	for (idx = 0; idx < num_kpus; idx++) {
 		npc_program_kpu_profile(rvu, blkaddr, idx, &rvu->kpu.kpu[idx]);
+		if (rvu->kpu.cst_kpu)
+			npc_program_kpu_profile(rvu, blkaddr, idx, &rvu->kpu.cst_kpu[idx]);
+		npc_enable_kpu_profile(rvu, blkaddr, idx, &rvu->kpu.kpu[idx]);
+	}
 }
 
 static int npc_mcam_rsrcs_init(struct rvu *rvu, int blkaddr)
@@ -1111,8 +1254,8 @@ int rvu_npc_init(struct rvu *rvu)
 	struct npc_kpu_profile_adapter *kpu = &rvu->kpu;
 	struct npc_pkind *pkind = &rvu->hw->pkind;
 	struct npc_mcam *mcam = &rvu->hw->mcam;
-	u64 cfg, nibble_ena, rx_kex, tx_kex;
 	int blkaddr, entry, bank, err;
+	u64 cfg;
 
 	blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, 0);
 	if (blkaddr < 0) {
@@ -1168,29 +1311,14 @@ int rvu_npc_init(struct rvu *rvu)
 		    BIT_ULL(32) | BIT_ULL(24) | BIT_ULL(6) |
 		    BIT_ULL(2) | BIT_ULL(1));
 
-	/* Set RX and TX side MCAM search key size.
-	 * LA..LD (ltype only) + Channel
-	 */
-	rx_kex = npc_mkex_default.keyx_cfg[NIX_INTF_RX];
-	tx_kex = npc_mkex_default.keyx_cfg[NIX_INTF_TX];
-	nibble_ena = FIELD_GET(NPC_PARSE_NIBBLE, rx_kex);
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_RX), rx_kex);
-	/* Due to an errata (35786) in A0 pass silicon, parse nibble enable
-	 * configuration has to be identical for both Rx and Tx interfaces.
-	 */
-	if (is_rvu_96xx_B0(rvu)) {
-		tx_kex &= ~NPC_PARSE_NIBBLE;
-		tx_kex |= FIELD_PREP(NPC_PARSE_NIBBLE, nibble_ena);
-	}
-	rvu_write64(rvu, blkaddr, NPC_AF_INTFX_KEX_CFG(NIX_INTF_TX), tx_kex);
+	/* Configure MKEX profile */
+	npc_load_mkex_profile(rvu, blkaddr, rvu->mkex_pfl_name);
 
+	/* Setup MCAM based on MKEX configuration. */
 	err = npc_mcam_rsrcs_init(rvu, blkaddr);
 	if (err)
 		return err;
 
-	/* Configure MKEX profile */
-	npc_load_mkex_profile(rvu, blkaddr, rvu->mkex_pfl_name);
-
 	/* Set TX miss action to UCAST_DEFAULT i.e
 	 * transmit the packet on NIX LF SQ's default channel.
 	 */
@@ -1215,6 +1343,8 @@ void rvu_npc_freemem(struct rvu *rvu)
 
 	kfree(pkind->rsrc.bmap);
 	kfree(mcam->counters.bmap);
+	kfree(rvu->kpu.cst_kpu);
+	kfree(rvu->kpu_fwdata);
 	mutex_destroy(&mcam->lock);
 }
 
-- 
2.20.1

Re: [PATCH net-next v2 3/3] octeontx2-af: add support for custom KPU entries

[Jakub Kicinski]

On Mon, 21 Sep 2020 19:54:42 +0200 Stanislaw Kardach wrote:
> Add ability to load a set of custom KPU entries via firmware APIs. This
> allows for flexible support for custom protocol parsing and CAM matching.
> 
> The firmware file name is specified by a module parameter (kpu_profile)
> to allow re-using the same kernel and initramfs package on nodes in
> different parts of the network where support for different protocols is
> required.
> 
> AF driver will attempt to load the profile from the firmware file and
> verify if it can fit hardware capabilities. If not, it will revert to
> the built-in profile.
> 
> Next it will read the maximum first KPU_MAX_CST_LT (2) custom entries
> from the firmware image. Those will be later programmed at the top of
> each KPU after the built-in profile entries have been programmed.
> The built-in profile is amended to always contain KPU_MAX_CSR_LT first
> no-match entries and AF driver will disable those in the KPU unless
> custom profile is loaded.

So the driver loads the firmware contents, interprets them and programs
the device appropriately?

Real firmware files are not usually interpreted or parsed by the driver.

Re: [PATCH net-next v2 3/3] octeontx2-af: add support for custom KPU entries

[Stanislaw Kardach]

On Mon, Sep 21, 2020 at 04:26:43PM -0700, Jakub Kicinski wrote:
> Date: Mon, 21 Sep 2020 16:26:43 -0700
> From: Jakub Kicinski <kuba@kernel.org>
> To: Stanislaw Kardach <skardach@marvell.com>
> Cc: davem@davemloft.net, sgoutham@marvell.com, netdev@vger.kernel.org,
>  kda@semihalf.com
> Subject: Re: [PATCH net-next v2 3/3] octeontx2-af: add support for custom
>  KPU entries
> 
> On Mon, 21 Sep 2020 19:54:42 +0200 Stanislaw Kardach wrote:
> > Add ability to load a set of custom KPU entries via firmware APIs. This
> > allows for flexible support for custom protocol parsing and CAM matching.
> > 
> > The firmware file name is specified by a module parameter (kpu_profile)
> > to allow re-using the same kernel and initramfs package on nodes in
> > different parts of the network where support for different protocols is
> > required.
> > 
> > AF driver will attempt to load the profile from the firmware file and
> > verify if it can fit hardware capabilities. If not, it will revert to
> > the built-in profile.
> > 
> > Next it will read the maximum first KPU_MAX_CST_LT (2) custom entries
> > from the firmware image. Those will be later programmed at the top of
> > each KPU after the built-in profile entries have been programmed.
> > The built-in profile is amended to always contain KPU_MAX_CSR_LT first
> > no-match entries and AF driver will disable those in the KPU unless
> > custom profile is loaded.
> 
> So the driver loads the firmware contents, interprets them and programs
> the device appropriately?
> 
> Real firmware files are not usually interpreted or parsed by the driver.

Correct. I'm using the firmware file as a delivery method for a custom
configuration. There are several reasons why I chose it:

1. The parsing engine (KPU) has to be configured fully at RVU AF device
   probe, before any networking part of that or other RVU devices is
   configured. So I think this rules out devlink, ioctl or sysfs.
2. The configuration is rather extensive so cramping it into module
   parameters doesn't seem right.
3. Adding it to Device Tree in form of custom nodes makes update process
   risky to some users (as opposed to switching firmware file on a
   filesystem).
4. The request_firmware API provides a nice abstraction for the blob data
   source so I thought it might as well be used for fetching data of a
   known structure. Especially that the full layout is visible in the
   kernel and users might create those files themselves by hand.

That said all above might be because I'm unaware of a better interface to
use in such situation. If there is, I would be obliged if you could point
me in the right direction.

Stanislaw Kardach

Re: [PATCH net-next v2 3/3] octeontx2-af: add support for custom KPU entries

[Jakub Kicinski]

On Tue, 22 Sep 2020 13:40:15 +0200 Stanislaw Kardach wrote:
> > So the driver loads the firmware contents, interprets them and programs
> > the device appropriately?
> > 
> > Real firmware files are not usually interpreted or parsed by the driver.  
> 
> Correct. I'm using the firmware file as a delivery method for a custom
> configuration. There are several reasons why I chose it:
> 
> 1. The parsing engine (KPU) has to be configured fully at RVU AF device
>    probe, before any networking part of that or other RVU devices is
>    configured. So I think this rules out devlink, ioctl or sysfs.
> 2. The configuration is rather extensive so cramping it into module
>    parameters doesn't seem right.
> 3. Adding it to Device Tree in form of custom nodes makes update process
>    risky to some users (as opposed to switching firmware file on a
>    filesystem).
> 4. The request_firmware API provides a nice abstraction for the blob data
>    source so I thought it might as well be used for fetching data of a
>    known structure. Especially that the full layout is visible in the
>    kernel and users might create those files themselves by hand.
> 
> That said all above might be because I'm unaware of a better interface to
> use in such situation. If there is, I would be obliged if you could point
> me in the right direction.

Sadly I don't think such interface exists today. You'd need to create
one. Parser configuration is something that has been coming up in
recent years but nobody done the work.

We try to push back on workarounds like this one to force people to
create proper abstract interfaces which can be used by multiple vendors.

Re: [PATCH net-next 0/3] octeontx2-af: add support for KPU profile customization

[David Miller]

Please create a proper abstraction for uploading configuration information
from a file into a device.

Otherwise every driver will have their own custom weird stuff, unique
module parameter names (module parameters are absolutely rejected for
networking drivers for this reason), etc.  Therefore the user experience
will be awful.

This is non-negotiable.  We are long overdue for such a facility, and
if I were to allow a workaround like this for you, I have to allow it
for everyone else too.

Thank you.