[net v2 PATCH 0/5] Octeontx2 AF driver fixes for NPC

[net v2 PATCH 0/5] Octeontx2 AF driver fixes for NPC

[Subbaraya Sundeep]

This patchset includes AF driver fixes wrt packet parser NPC.
Following are the changes:

Patch 1: The parser nibble configuration must be same for
TX and RX interfaces and if not fix up is applied. This fixup was
applied only for default profile currently and it has been fixed
to apply for all profiles.
Patch 2: Firmware image may not be present all times in the kernel image
and default profile is used mostly hence suppress the warning.
Patch 3: Custom profiles may not extract DMAC into the match key
always. Hence fix the driver to allow profiles without DMAC extraction.
Patch 4: This patch fixes a corner case where NIXLF is detached but
without freeing its mcam entries which results in resource leak.
Patch 5: SMAC is overlapped with DMAC mistakenly while installing
rules based on SMAC. This patch fixes that.

v2 changes:
Added the space which was missing between commit hash
and ("octeontx2-af for patch 4.

Thanks,
Sundeep

Harman Kalra (1):
  octeontx2-af: suppress external profile loading warning

Stanislaw Kardach (1):
  octeontx2-af: Apply tx nibble fixup always

Subbaraya Sundeep (2):
  octeontx2-af: Fix mcam entry resource leak
  octeontx2-af: Fix key checking for source mac

Suman Ghosh (1):
  octeontx2-af: Allow mkex profiles without dmac.

 drivers/net/ethernet/marvell/octeontx2/af/npc.h    |  1 +
 drivers/net/ethernet/marvell/octeontx2/af/rvu.c    |  6 ++
 .../ethernet/marvell/octeontx2/af/rvu_debugfs.c    |  6 ++
 .../net/ethernet/marvell/octeontx2/af/rvu_npc.c    | 15 +++-
 .../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 85 +++++++++++++++++-----
 5 files changed, 90 insertions(+), 23 deletions(-)

-- 
2.7.4

[net v2 PATCH 1/5] octeontx2-af: Apply tx nibble fixup always

[Subbaraya Sundeep]

From: Stanislaw Kardach <skardach@marvell.com>

NPC_PARSE_NIBBLE for TX interface has to be equal to the RX one for some
silicon revisions. Mistakenly this fixup was only applied to the default
MKEX profile while it should also be applied to any loaded profile.

Fixes: 1c1935c9945d ("octeontx2-af: Add NIX1 interfaces to NPC")
Signed-off-by: Stanislaw Kardach <skardach@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 3a31fb8..3d99cb9 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -1915,6 +1915,7 @@ static void rvu_npc_hw_init(struct rvu *rvu, int blkaddr)
 
 static void rvu_npc_setup_interfaces(struct rvu *rvu, int blkaddr)
 {
+	struct npc_mcam_kex *mkex = rvu->kpu.mkex;
 	struct npc_mcam *mcam = &rvu->hw->mcam;
 	struct rvu_hwinfo *hw = rvu->hw;
 	u64 nibble_ena, rx_kex, tx_kex;
@@ -1927,15 +1928,15 @@ static void rvu_npc_setup_interfaces(struct rvu *rvu, int blkaddr)
 	mcam->counters.max--;
 	mcam->rx_miss_act_cntr = mcam->counters.max;
 
-	rx_kex = npc_mkex_default.keyx_cfg[NIX_INTF_RX];
-	tx_kex = npc_mkex_default.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);
 
 	nibble_ena = rvu_npc_get_tx_nibble_cfg(rvu, nibble_ena);
 	if (nibble_ena) {
 		tx_kex &= ~NPC_PARSE_NIBBLE;
 		tx_kex |= FIELD_PREP(NPC_PARSE_NIBBLE, nibble_ena);
-		npc_mkex_default.keyx_cfg[NIX_INTF_TX] = tx_kex;
+		mkex->keyx_cfg[NIX_INTF_TX] = tx_kex;
 	}
 
 	/* Configure RX interfaces */
-- 
2.7.4

[net v2 PATCH 2/5] octeontx2-af: suppress external profile loading warning

[Subbaraya Sundeep]

From: Harman Kalra <hkalra@marvell.com>

The packet parser profile supplied as firmware may not
be present all the time and default profile is used mostly.
Hence suppress firmware loading warning from kernel due to
absence of firmware in kernel image.

Fixes: 3a7244152f9c ("octeontx2-af: add support for custom KPU entries")
Signed-off-by: Harman Kalra <hkalra@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 3d99cb9..9404f86 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -1650,7 +1650,7 @@ static void npc_load_kpu_profile(struct rvu *rvu)
 	 * Firmware database method.
 	 * Default KPU profile.
 	 */
-	if (!request_firmware(&fw, kpu_profile, rvu->dev)) {
+	if (!firmware_request_nowarn(&fw, kpu_profile, rvu->dev)) {
 		dev_info(rvu->dev, "Loading KPU profile from firmware: %s\n",
 			 kpu_profile);
 		rvu->kpu_fwdata = kzalloc(fw->size, GFP_KERNEL);
-- 
2.7.4

[net v2 PATCH 3/5] octeontx2-af: Allow mkex profiles without dmac.

[Subbaraya Sundeep]

From: Suman Ghosh <sumang@marvell.com>

It is possible to have custom mkex profiles which do not extract
DMAC into the key to free up space in the key and use it for L3
or L4 packet fields. Current code bails out if DMAC extraction is
not present in the key. This patch fixes it by allowing profiles
without DMAC and also supports installing rules based on L2MB bit
set by hardware for multicast and broadcast packets.

This patch also adds debugging prints needed to identify profiles
with wrong configuration.

Fixes: 9b179a960a96 ("octeontx2-af: Generate key field bit mask from KEX profile")
Signed-off-by: Suman Ghosh <sumang@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/npc.h    |  1 +
 .../ethernet/marvell/octeontx2/af/rvu_debugfs.c    |  6 ++
 .../net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 82 +++++++++++++++++-----
 3 files changed, 71 insertions(+), 18 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc.h b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
index 9b6e587..2d9c767 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
@@ -595,6 +595,7 @@ struct rvu_npc_mcam_rule {
 	bool vfvlan_cfg;
 	u16 chan;
 	u16 chan_mask;
+	u8 lxmb;
 };
 
 #endif /* NPC_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
index 2ad73b1..7cd386b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
@@ -2410,6 +2410,12 @@ static void rvu_dbg_npc_mcam_show_flows(struct seq_file *s,
 	for_each_set_bit(bit, (unsigned long *)&rule->features, 64) {
 		seq_printf(s, "\t%s  ", npc_get_field_name(bit));
 		switch (bit) {
+		case NPC_LXMB:
+			if (rule->lxmb == 1)
+				seq_puts(s, "\tL2M nibble is set\n");
+			else
+				seq_puts(s, "\tL2B nibble is set\n");
+			break;
 		case NPC_DMAC:
 			seq_printf(s, "%pM ", rule->packet.dmac);
 			seq_printf(s, "mask %pM\n", rule->mask.dmac);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index 19c53e5..0a163fa 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -41,6 +41,7 @@ static const char * const npc_flow_names[] = {
 	[NPC_DPORT_UDP]	= "udp destination port",
 	[NPC_SPORT_SCTP] = "sctp source port",
 	[NPC_DPORT_SCTP] = "sctp destination port",
+	[NPC_LXMB]	= "Mcast/Bcast header ",
 	[NPC_UNKNOWN]	= "unknown",
 };
 
@@ -318,8 +319,10 @@ static void npc_handle_multi_layer_fields(struct rvu *rvu, int blkaddr, u8 intf)
 	vlan_tag2 = &key_fields[NPC_VLAN_TAG2];
 
 	/* if key profile programmed does not extract Ethertype at all */
-	if (!etype_ether->nr_kws && !etype_tag1->nr_kws && !etype_tag2->nr_kws)
+	if (!etype_ether->nr_kws && !etype_tag1->nr_kws && !etype_tag2->nr_kws) {
+		dev_err(rvu->dev, "mkex: Ethertype is not extracted.\n");
 		goto vlan_tci;
+	}
 
 	/* if key profile programmed extracts Ethertype from one layer */
 	if (etype_ether->nr_kws && !etype_tag1->nr_kws && !etype_tag2->nr_kws)
@@ -332,35 +335,45 @@ static void npc_handle_multi_layer_fields(struct rvu *rvu, int blkaddr, u8 intf)
 	/* if key profile programmed extracts Ethertype from multiple layers */
 	if (etype_ether->nr_kws && etype_tag1->nr_kws) {
 		for (i = 0; i < NPC_MAX_KWS_IN_KEY; i++) {
-			if (etype_ether->kw_mask[i] != etype_tag1->kw_mask[i])
+			if (etype_ether->kw_mask[i] != etype_tag1->kw_mask[i]) {
+				dev_err(rvu->dev, "mkex: Etype pos is different for untagged and tagged pkts.\n");
 				goto vlan_tci;
+			}
 		}
 		key_fields[NPC_ETYPE] = *etype_tag1;
 	}
 	if (etype_ether->nr_kws && etype_tag2->nr_kws) {
 		for (i = 0; i < NPC_MAX_KWS_IN_KEY; i++) {
-			if (etype_ether->kw_mask[i] != etype_tag2->kw_mask[i])
+			if (etype_ether->kw_mask[i] != etype_tag2->kw_mask[i]) {
+				dev_err(rvu->dev, "mkex: Etype pos is different for untagged and double tagged pkts.\n");
 				goto vlan_tci;
+			}
 		}
 		key_fields[NPC_ETYPE] = *etype_tag2;
 	}
 	if (etype_tag1->nr_kws && etype_tag2->nr_kws) {
 		for (i = 0; i < NPC_MAX_KWS_IN_KEY; i++) {
-			if (etype_tag1->kw_mask[i] != etype_tag2->kw_mask[i])
+			if (etype_tag1->kw_mask[i] != etype_tag2->kw_mask[i]) {
+				dev_err(rvu->dev, "mkex: Etype pos is different for tagged and double tagged pkts.\n");
 				goto vlan_tci;
+			}
 		}
 		key_fields[NPC_ETYPE] = *etype_tag2;
 	}
 
 	/* check none of higher layers overwrite Ethertype */
 	start_lid = key_fields[NPC_ETYPE].layer_mdata.lid + 1;
-	if (npc_check_overlap(rvu, blkaddr, NPC_ETYPE, start_lid, intf))
+	if (npc_check_overlap(rvu, blkaddr, NPC_ETYPE, start_lid, intf)) {
+		dev_err(rvu->dev, "mkex: Ethertype is overwritten by higher layers.\n");
 		goto vlan_tci;
+	}
 	*features |= BIT_ULL(NPC_ETYPE);
 vlan_tci:
 	/* if key profile does not extract outer vlan tci at all */
-	if (!vlan_tag1->nr_kws && !vlan_tag2->nr_kws)
+	if (!vlan_tag1->nr_kws && !vlan_tag2->nr_kws) {
+		dev_err(rvu->dev, "mkex: Outer vlan tci is not extracted.\n");
 		goto done;
+	}
 
 	/* if key profile extracts outer vlan tci from one layer */
 	if (vlan_tag1->nr_kws && !vlan_tag2->nr_kws)
@@ -371,15 +384,19 @@ static void npc_handle_multi_layer_fields(struct rvu *rvu, int blkaddr, u8 intf)
 	/* if key profile extracts outer vlan tci from multiple layers */
 	if (vlan_tag1->nr_kws && vlan_tag2->nr_kws) {
 		for (i = 0; i < NPC_MAX_KWS_IN_KEY; i++) {
-			if (vlan_tag1->kw_mask[i] != vlan_tag2->kw_mask[i])
+			if (vlan_tag1->kw_mask[i] != vlan_tag2->kw_mask[i]) {
+				dev_err(rvu->dev, "mkex: Out vlan tci pos is different for tagged and double tagged pkts.\n");
 				goto done;
+			}
 		}
 		key_fields[NPC_OUTER_VID] = *vlan_tag2;
 	}
 	/* check none of higher layers overwrite outer vlan tci */
 	start_lid = key_fields[NPC_OUTER_VID].layer_mdata.lid + 1;
-	if (npc_check_overlap(rvu, blkaddr, NPC_OUTER_VID, start_lid, intf))
+	if (npc_check_overlap(rvu, blkaddr, NPC_OUTER_VID, start_lid, intf)) {
+		dev_err(rvu->dev, "mkex: Outer vlan tci is overwritten by higher layers.\n");
 		goto done;
+	}
 	*features |= BIT_ULL(NPC_OUTER_VID);
 done:
 	return;
@@ -499,6 +516,10 @@ static void npc_set_features(struct rvu *rvu, int blkaddr, u8 intf)
 	if (npc_check_field(rvu, blkaddr, NPC_LB, intf))
 		*features |= BIT_ULL(NPC_VLAN_ETYPE_CTAG) |
 			     BIT_ULL(NPC_VLAN_ETYPE_STAG);
+
+	/* for L2M/L2B/L3M/L3B, check if the type is present in the key */
+	if (npc_check_field(rvu, blkaddr, NPC_LXMB, intf))
+		*features |= BIT_ULL(NPC_LXMB);
 }
 
 /* Scan key extraction profile and record how fields of our interest
@@ -564,16 +585,6 @@ static int npc_scan_verify_kex(struct rvu *rvu, int blkaddr)
 		dev_err(rvu->dev, "Channel cannot be overwritten\n");
 		return -EINVAL;
 	}
-	/* DMAC should be present in key for unicast filter to work */
-	if (!npc_is_field_present(rvu, NPC_DMAC, NIX_INTF_RX)) {
-		dev_err(rvu->dev, "DMAC not present in Key\n");
-		return -EINVAL;
-	}
-	/* check that none of the fields overwrite DMAC */
-	if (npc_check_overlap(rvu, blkaddr, NPC_DMAC, 0, NIX_INTF_RX)) {
-		dev_err(rvu->dev, "DMAC cannot be overwritten\n");
-		return -EINVAL;
-	}
 
 	npc_set_features(rvu, blkaddr, NIX_INTF_TX);
 	npc_set_features(rvu, blkaddr, NIX_INTF_RX);
@@ -817,6 +828,11 @@ static void npc_update_flow(struct rvu *rvu, struct mcam_entry *entry,
 		npc_update_entry(rvu, NPC_LE, entry, NPC_LT_LE_ESP,
 				 0, ~0ULL, 0, intf);
 
+	if (features & BIT_ULL(NPC_LXMB)) {
+		output->lxmb = is_broadcast_ether_addr(pkt->dmac) ? 2 : 1;
+		npc_update_entry(rvu, NPC_LXMB, entry, output->lxmb, 0,
+				 output->lxmb, 0, intf);
+	}
 #define NPC_WRITE_FLOW(field, member, val_lo, val_hi, mask_lo, mask_hi)	      \
 do {									      \
 	if (features & BIT_ULL((field))) {				      \
@@ -1114,6 +1130,7 @@ static int npc_install_flow(struct rvu *rvu, int blkaddr, u16 target,
 	rule->chan_mask = write_req.entry_data.kw_mask[0] & NPC_KEX_CHAN_MASK;
 	rule->chan = write_req.entry_data.kw[0] & NPC_KEX_CHAN_MASK;
 	rule->chan &= rule->chan_mask;
+	rule->lxmb = dummy.lxmb;
 	if (is_npc_intf_tx(req->intf))
 		rule->intf = pfvf->nix_tx_intf;
 	else
@@ -1176,6 +1193,35 @@ int rvu_mbox_handler_npc_install_flow(struct rvu *rvu,
 	if (!is_npc_interface_valid(rvu, req->intf))
 		return NPC_FLOW_INTF_INVALID;
 
+	/* If DMAC is not extracted in MKEX, rules installed by AF
+	 * can rely on L2MB bit set by hardware protocol checker for
+	 * broadcast and multicast addresses.
+	 */
+	if (npc_check_field(rvu, blkaddr, NPC_DMAC, req->intf))
+		goto process_flow;
+
+	if (is_pffunc_af(req->hdr.pcifunc) &&
+	    req->features & BIT_ULL(NPC_DMAC)) {
+		if (is_unicast_ether_addr(req->packet.dmac)) {
+			dev_err(rvu->dev,
+				"%s: mkex profile does not support ucast flow\n",
+				__func__);
+			return NPC_FLOW_NOT_SUPPORTED;
+		}
+
+		if (!npc_is_field_present(rvu, NPC_LXMB, req->intf)) {
+			dev_err(rvu->dev,
+				"%s: mkex profile does not support bcast/mcast flow",
+				__func__);
+			return NPC_FLOW_NOT_SUPPORTED;
+		}
+
+		/* Modify feature to use LXMB instead of DMAC */
+		req->features &= ~BIT_ULL(NPC_DMAC);
+		req->features |= BIT_ULL(NPC_LXMB);
+	}
+
+process_flow:
 	if (from_vf && req->default_rule)
 		return NPC_FLOW_VF_PERM_DENIED;
 
-- 
2.7.4

[net v2 PATCH 4/5] octeontx2-af: Fix mcam entry resource leak

[Subbaraya Sundeep]

The teardown sequence in FLR handler returns if no NIX LF
is attached to PF/VF because it indicates that graceful
shutdown of resources already happened. But there is a
chance of all allocated MCAM entries not being freed by
PF/VF. Hence free mcam entries even in case of detached LF.

Fixes: c554f9c1574e ("octeontx2-af: Teardown NPA, NIX LF upon receiving FLR")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/rvu.c     | 6 ++++++
 drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c | 6 ++++++
 2 files changed, 12 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
index 54e1b27..1484d33 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.c
@@ -2564,6 +2564,12 @@ static void __rvu_flr_handler(struct rvu *rvu, u16 pcifunc)
 	rvu_blklf_teardown(rvu, pcifunc, BLKADDR_NPA);
 	rvu_reset_lmt_map_tbl(rvu, pcifunc);
 	rvu_detach_rsrcs(rvu, NULL, pcifunc);
+	/* In scenarios where PF/VF drivers detach NIXLF without freeing MCAM
+	 * entries, check and free the MCAM entries explicitly to avoid leak.
+	 * Since LF is detached use LF number as -1.
+	 */
+	rvu_npc_free_mcam_entries(rvu, pcifunc, -1);
+
 	mutex_unlock(&rvu->flr_lock);
 }
 
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
index 9404f86..4b39e13 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c
@@ -1096,6 +1096,9 @@ static void npc_enadis_default_entries(struct rvu *rvu, u16 pcifunc,
 
 void rvu_npc_disable_default_entries(struct rvu *rvu, u16 pcifunc, int nixlf)
 {
+	if (nixlf < 0)
+		return;
+
 	npc_enadis_default_entries(rvu, pcifunc, nixlf, false);
 
 	/* Delete multicast and promisc MCAM entries */
@@ -1107,6 +1110,9 @@ void rvu_npc_disable_default_entries(struct rvu *rvu, u16 pcifunc, int nixlf)
 
 void rvu_npc_enable_default_entries(struct rvu *rvu, u16 pcifunc, int nixlf)
 {
+	if (nixlf < 0)
+		return;
+
 	/* Enables only broadcast match entry. Promisc/Allmulti are enabled
 	 * in set_rx_mode mbox handler.
 	 */
-- 
2.7.4

[net v2 PATCH 5/5] octeontx2-af: Fix key checking for source mac

[Subbaraya Sundeep]

Given a field with its location/offset in input packet,
the key checking logic verifies whether extracting the
field can be supported or not based on the mkex profile
loaded in hardware. This logic is wrong wrt source mac
and this patch fixes that.

Fixes: 9b179a960a96 ("octeontx2-af: Generate key field bit mask from KEX profile")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index 0a163fa..fc744cf 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -462,7 +462,8 @@ do {									       \
 	NPC_SCAN_HDR(NPC_VLAN_TAG1, NPC_LID_LB, NPC_LT_LB_CTAG, 2, 2);
 	NPC_SCAN_HDR(NPC_VLAN_TAG2, NPC_LID_LB, NPC_LT_LB_STAG_QINQ, 2, 2);
 	NPC_SCAN_HDR(NPC_DMAC, NPC_LID_LA, la_ltype, la_start, 6);
-	NPC_SCAN_HDR(NPC_SMAC, NPC_LID_LA, la_ltype, la_start, 6);
+	/* SMAC follows the DMAC(which is 6 bytes) */
+	NPC_SCAN_HDR(NPC_SMAC, NPC_LID_LA, la_ltype, la_start + 6, 6);
 	/* PF_FUNC is 2 bytes at 0th byte of NPC_LT_LA_IH_NIX_ETHER */
 	NPC_SCAN_HDR(NPC_PF_FUNC, NPC_LID_LA, NPC_LT_LA_IH_NIX_ETHER, 0, 2);
 }
-- 
2.7.4

Re: [net v2 PATCH 2/5] octeontx2-af: suppress external profile loading warning

[Jakub Kicinski]

On Tue, 26 Jul 2022 19:41:19 +0530 Subbaraya Sundeep wrote:
> -	if (!request_firmware(&fw, kpu_profile, rvu->dev)) {
> +	if (!firmware_request_nowarn(&fw, kpu_profile, rvu->dev)) {

Consider switching to request_firmware_direct() in net-next.
I doubt you need the sysfs fallback, I think udev dropped 
the support for it.

Re: [net v2 PATCH 3/5] octeontx2-af: Allow mkex profiles without dmac.

[Jakub Kicinski]

On Tue, 26 Jul 2022 19:41:20 +0530 Subbaraya Sundeep wrote:
> From: Suman Ghosh <sumang@marvell.com>
> 
> It is possible to have custom mkex profiles which do not extract
> DMAC into the key to free up space in the key and use it for L3
> or L4 packet fields. Current code bails out if DMAC extraction is
> not present in the key. This patch fixes it by allowing profiles
> without DMAC and also supports installing rules based on L2MB bit
> set by hardware for multicast and broadcast packets.

This sounds half way between a feature and a fix. Can you make it
clearer why it's a fix and not an optimization?

> This patch also adds debugging prints needed to identify profiles
> with wrong configuration.

All those prints make the patch even less acceptable as a fix.
We merge fixes into net-next every week, you can send a minimal
fix and extend the code in net-next soon after that.

> Fixes: 9b179a960a96 ("octeontx2-af: Generate key field bit mask from KEX profile")

> +	/* If DMAC is not extracted in MKEX, rules installed by AF
> +	 * can rely on L2MB bit set by hardware protocol checker for
> +	 * broadcast and multicast addresses.
> +	 */
> +	if (npc_check_field(rvu, blkaddr, NPC_DMAC, req->intf))
> +		goto process_flow;

Merge this condition with the condition below, goto should be avoided
but for error paths.

> +	if (is_pffunc_af(req->hdr.pcifunc) &&
> +	    req->features & BIT_ULL(NPC_DMAC)) {
> +		if (is_unicast_ether_addr(req->packet.dmac)) {
> +			dev_err(rvu->dev,
> +				"%s: mkex profile does not support ucast flow\n",
> +				__func__);
> +			return NPC_FLOW_NOT_SUPPORTED;
> +		}
> +
> +		if (!npc_is_field_present(rvu, NPC_LXMB, req->intf)) {
> +			dev_err(rvu->dev,
> +				"%s: mkex profile does not support bcast/mcast flow",
> +				__func__);
> +			return NPC_FLOW_NOT_SUPPORTED;
> +		}
> +
> +		/* Modify feature to use LXMB instead of DMAC */
> +		req->features &= ~BIT_ULL(NPC_DMAC);
> +		req->features |= BIT_ULL(NPC_LXMB);
> +	}
> +
> +process_flow:
>  	if (from_vf && req->default_rule)
>  		return NPC_FLOW_VF_PERM_DENIED;
>  

Re: [net v2 PATCH 2/5] octeontx2-af: suppress external profile loading warning

[Jakub Kicinski]

On Wed, 27 Jul 2022 19:41:20 -0700 Jakub Kicinski wrote:
> On Tue, 26 Jul 2022 19:41:19 +0530 Subbaraya Sundeep wrote:
> > -	if (!request_firmware(&fw, kpu_profile, rvu->dev)) {
> > +	if (!firmware_request_nowarn(&fw, kpu_profile, rvu->dev)) {  
> 
> Consider switching to request_firmware_direct() in net-next.
> I doubt you need the sysfs fallback, I think udev dropped 
> the support for it.

Well, the next patch needs work, so perhaps do it in v3?