[net PATCH 0/9] Macsec fixes for CN10KB

[net PATCH 0/9] Macsec fixes for CN10KB

[Geetha sowjanya]

This patch set has fixes for the issues encountered while
testing macsec on CN10KB silicon. Below is the description
of patches:

Patch 1: For each LMAC two MCSX_MCS_TOP_SLAVE_CHANNEL_CFG registers exist
	 in CN10KB. Bypass has to be disabled in two registers.

Patch 2: Add workaround for errata w.r.t accessing TCAM DATA and MASK registers.

Patch 3: Fixes the parser configuration to allow PTP traffic.

Patch 4: Addresses the IP vector and block level interrupt mask changes.
 
Patch 5: Fix NULL pointer crashes when rebooting

Patch 6: Since MCS is global block shared by all LMACS the TCAM match
	 must include macsec DMAC also to distinguish each macsec interface

Patch 7: Before freeing MCS hardware resource to AF clear the stats also.

Patch 8: Stats which share single counter in hardware are tracked in software.
	 This tracking was based on wrong secy mode params.
	 Use correct secy mode params

Patch 9: When updating secy mode params, PN number was also reset to
	 initial values. Hence do not write to PN value register when
	 updating secy.


Geetha sowjanya (3):
  octeonxt2-af: mcs: Fix per port bypass config
  octeontx2-af: mcs: Config parser to skip 8B header
  octeontx2-af: mcs: Fix MCS block interrupt  

Subbaraya Sundeep (6):
  octeontx2-af: mcs: Write TCAM_DATA and TCAM_MASK registers at once
  octeontx2-pf: mcs: Fix NULL pointer dereferences
  octeontx2-pf: mcs: Match macsec ethertype along with DMAC
  octeontx2-pf: mcs: Clear stats before freeing resource
  octeontx2-pf: mcs: Fix shared counters logic
  octeontx2-pf: mcs: Do not reset PN when updating secy

 .../net/ethernet/marvell/octeontx2/af/mcs.c   | 106 +++++++++---------
 .../net/ethernet/marvell/octeontx2/af/mcs.h   |  26 ++---
 .../marvell/octeontx2/af/mcs_cnf10kb.c        |  63 +++++++++++
 .../ethernet/marvell/octeontx2/af/mcs_reg.h   |   6 +-
 .../marvell/octeontx2/af/mcs_rvu_if.c         |  37 ++++++
 .../net/ethernet/marvell/octeontx2/af/rvu.h   |   1 +
 .../ethernet/marvell/octeontx2/af/rvu_cgx.c   |   2 +
 .../marvell/octeontx2/af/rvu_debugfs.c        |   4 +-
 .../marvell/octeontx2/nic/cn10k_macsec.c      |  54 ++++++---
 .../marvell/octeontx2/nic/otx2_common.h       |   2 +-
 10 files changed, 217 insertions(+), 84 deletions(-)

-- 
2.25.1

[net PATCH 1/9] octeonxt2-af: mcs: Fix per port bypass config

[Geetha sowjanya]

For each lmac port, MCS has two MCS_TOP_SLAVE_CHANNEL_CONFIGX
registers. For CN10KB both register need to be configured for the
port level mcs bypass to work. This patch also sets bitmap
of flowid/secy entry reserved for default bypass so that these
entries can be shown in debugfs.

Fixes: bd69476e86fc ("octeontx2-af: cn10k: mcs: Install a default TCAM for normal traffic")
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/mcs.c       | 11 ++++++++++-
 .../net/ethernet/marvell/octeontx2/af/rvu_debugfs.c   |  5 +++--
 2 files changed, 13 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
index f68a6a0e3aa4..492baa0b594c 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
@@ -494,6 +494,9 @@ int mcs_install_flowid_bypass_entry(struct mcs *mcs)
 
 	/* Flow entry */
 	flow_id = mcs->hw->tcam_entries - MCS_RSRC_RSVD_CNT;
+	__set_bit(flow_id, mcs->rx.flow_ids.bmap);
+	__set_bit(flow_id, mcs->tx.flow_ids.bmap);
+
 	for (reg_id = 0; reg_id < 4; reg_id++) {
 		reg = MCSX_CPM_RX_SLAVE_FLOWID_TCAM_MASKX(reg_id, flow_id);
 		mcs_reg_write(mcs, reg, GENMASK_ULL(63, 0));
@@ -504,6 +507,8 @@ int mcs_install_flowid_bypass_entry(struct mcs *mcs)
 	}
 	/* secy */
 	secy_id = mcs->hw->secy_entries - MCS_RSRC_RSVD_CNT;
+	__set_bit(secy_id, mcs->rx.secy.bmap);
+	__set_bit(secy_id, mcs->tx.secy.bmap);
 
 	/* Set validate frames to NULL and enable control port */
 	plcy = 0x7ull;
@@ -528,6 +533,7 @@ int mcs_install_flowid_bypass_entry(struct mcs *mcs)
 	/* Enable Flowid entry */
 	mcs_ena_dis_flowid_entry(mcs, flow_id, MCS_RX, true);
 	mcs_ena_dis_flowid_entry(mcs, flow_id, MCS_TX, true);
+
 	return 0;
 }
 
@@ -1325,8 +1331,11 @@ void mcs_reset_port(struct mcs *mcs, u8 port_id, u8 reset)
 void mcs_set_lmac_mode(struct mcs *mcs, int lmac_id, u8 mode)
 {
 	u64 reg;
+	int id = lmac_id * 2;
 
-	reg = MCSX_MCS_TOP_SLAVE_CHANNEL_CFG(lmac_id * 2);
+	reg = MCSX_MCS_TOP_SLAVE_CHANNEL_CFG(id);
+	mcs_reg_write(mcs, reg, (u64)mode);
+	reg = MCSX_MCS_TOP_SLAVE_CHANNEL_CFG((id + 1));
 	mcs_reg_write(mcs, reg, (u64)mode);
 }
 
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
index 26cfa501f1a1..9533b1d92960 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
@@ -497,8 +497,9 @@ static int rvu_dbg_mcs_rx_secy_stats_display(struct seq_file *filp, void *unused
 			   stats.octet_validated_cnt);
 		seq_printf(filp, "secy%d: Pkts on disable port: %lld\n", secy_id,
 			   stats.pkt_port_disabled_cnt);
-		seq_printf(filp, "secy%d: Octets validated: %lld\n", secy_id, stats.pkt_badtag_cnt);
-		seq_printf(filp, "secy%d: Octets validated: %lld\n", secy_id, stats.pkt_nosa_cnt);
+		seq_printf(filp, "secy%d: Pkts with badtag: %lld\n", secy_id, stats.pkt_badtag_cnt);
+		seq_printf(filp, "secy%d: Pkts with no SA(sectag.tci.c=0): %lld\n", secy_id,
+			   stats.pkt_nosa_cnt);
 		seq_printf(filp, "secy%d: Pkts with nosaerror: %lld\n", secy_id,
 			   stats.pkt_nosaerror_cnt);
 		seq_printf(filp, "secy%d: Tagged ctrl pkts: %lld\n", secy_id,
-- 
2.25.1

[net PATCH 2/9] octeontx2-af: mcs: Write TCAM_DATA and TCAM_MASK registers at once

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

As per hardware errata on CN10KB, all the four TCAM_DATA
and TCAM_MASK registers has to be written at once otherwise
write to individual registers will fail. Hence write to all
TCAM_DATA registers and then to all TCAM_MASK registers.

Fixes: cfc14181d497 ("octeontx2-af: cn10k: mcs: Manage the MCS block hardware resources")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/af/mcs.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
index 492baa0b594c..148417d633a5 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
@@ -473,6 +473,8 @@ void mcs_flowid_entry_write(struct mcs *mcs, u64 *data, u64 *mask, int flow_id,
 		for (reg_id = 0; reg_id < 4; reg_id++) {
 			reg = MCSX_CPM_RX_SLAVE_FLOWID_TCAM_DATAX(reg_id, flow_id);
 			mcs_reg_write(mcs, reg, data[reg_id]);
+		}
+		for (reg_id = 0; reg_id < 4; reg_id++) {
 			reg = MCSX_CPM_RX_SLAVE_FLOWID_TCAM_MASKX(reg_id, flow_id);
 			mcs_reg_write(mcs, reg, mask[reg_id]);
 		}
@@ -480,6 +482,8 @@ void mcs_flowid_entry_write(struct mcs *mcs, u64 *data, u64 *mask, int flow_id,
 		for (reg_id = 0; reg_id < 4; reg_id++) {
 			reg = MCSX_CPM_TX_SLAVE_FLOWID_TCAM_DATAX(reg_id, flow_id);
 			mcs_reg_write(mcs, reg, data[reg_id]);
+		}
+		for (reg_id = 0; reg_id < 4; reg_id++) {
 			reg = MCSX_CPM_TX_SLAVE_FLOWID_TCAM_MASKX(reg_id, flow_id);
 			mcs_reg_write(mcs, reg, mask[reg_id]);
 		}
-- 
2.25.1

[net PATCH 3/9] octeontx2-af: mcs: Config parser to skip 8B header

[Geetha sowjanya]

When ptp timestamp is enabled in RPM, RPM will append 8B
timestamp header for all RX traffic. MCS need to skip these
8 bytes header while parsing the packet header, so that
correct tcam key is created for lookup. 
This patch fixes the mcs parser configuration to skip this 
8B header for ptp packets.

Fixes: ca7f49ff8846 ("octeontx2-af: cn10k: Introduce driver for macsec block.")
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../ethernet/marvell/octeontx2/af/mcs_reg.h   |  1 +
 .../marvell/octeontx2/af/mcs_rvu_if.c         | 37 +++++++++++++++++++
 .../net/ethernet/marvell/octeontx2/af/rvu.h   |  1 +
 .../ethernet/marvell/octeontx2/af/rvu_cgx.c   |  2 +
 4 files changed, 41 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h b/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
index c95a8b8f5eaf..7427e3b1490f 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
@@ -97,6 +97,7 @@
 #define MCSX_PEX_TX_SLAVE_VLAN_CFGX(a)          (0x46f8ull + (a) * 0x8ull)
 #define MCSX_PEX_TX_SLAVE_CUSTOM_TAG_REL_MODE_SEL(a)	(0x788ull + (a) * 0x8ull)
 #define MCSX_PEX_TX_SLAVE_PORT_CONFIG(a)		(0x4738ull + (a) * 0x8ull)
+#define MCSX_PEX_RX_SLAVE_PORT_CFGX(a)		(0x3b98ull + (a) * 0x8ull)
 #define MCSX_PEX_RX_SLAVE_RULE_ETYPE_CFGX(a) ({	\
 	u64 offset;					\
 							\
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
index eb25e458266c..dfd23580e3b8 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
@@ -11,6 +11,7 @@
 
 #include "mcs.h"
 #include "rvu.h"
+#include "mcs_reg.h"
 #include "lmac_common.h"
 
 #define M(_name, _id, _fn_name, _req_type, _rsp_type)			\
@@ -32,6 +33,42 @@ static struct _req_type __maybe_unused					\
 MBOX_UP_MCS_MESSAGES
 #undef M
 
+void rvu_mcs_ptp_cfg(struct rvu *rvu, u8 rpm_id, u8 lmac_id, bool ena)
+{
+	struct mcs *mcs;
+	u64 cfg;
+	u8 port;
+
+	if (!rvu->mcs_blk_cnt)
+		return;
+
+	/* When ptp is enabled, RPM appends 8B header for all
+	 * RX packets. MCS PEX need to configure to skip 8B
+	 * during packet parsing.
+	 */
+
+	/* CNF10K-B */
+	if (rvu->mcs_blk_cnt > 1) {
+		mcs = mcs_get_pdata(rpm_id);
+		cfg = mcs_reg_read(mcs, MCSX_PEX_RX_SLAVE_PEX_CONFIGURATION);
+		if (ena)
+			cfg |= BIT_ULL(lmac_id);
+		else
+			cfg &= ~BIT_ULL(lmac_id);
+		mcs_reg_write(mcs, MCSX_PEX_RX_SLAVE_PEX_CONFIGURATION, cfg);
+		return;
+	}
+	/* CN10KB */
+	mcs = mcs_get_pdata(0);
+	port = (rpm_id * rvu->hw->lmac_per_cgx) + lmac_id;
+	cfg = mcs_reg_read(mcs, MCSX_PEX_RX_SLAVE_PORT_CFGX(port));
+	if (ena)
+		cfg |= BIT_ULL(0);
+	else
+		cfg &= ~BIT_ULL(0);
+	mcs_reg_write(mcs, MCSX_PEX_RX_SLAVE_PORT_CFGX(port), cfg);
+}
+
 int rvu_mbox_handler_mcs_set_lmac_mode(struct rvu *rvu,
 				       struct mcs_set_lmac_mode *req,
 				       struct msg_rsp *rsp)
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
index ef721caeac49..d655bf04a483 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
@@ -920,6 +920,7 @@ int rvu_get_hwvf(struct rvu *rvu, int pcifunc);
 /* CN10K MCS */
 int rvu_mcs_init(struct rvu *rvu);
 int rvu_mcs_flr_handler(struct rvu *rvu, u16 pcifunc);
+void rvu_mcs_ptp_cfg(struct rvu *rvu, u8 rpm_id, u8 lmac_id, bool ena);
 void rvu_mcs_exit(struct rvu *rvu);
 
 #endif /* RVU_H */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
index 438b212fb54a..83b342fa8d75 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
@@ -773,6 +773,8 @@ static int rvu_cgx_ptp_rx_cfg(struct rvu *rvu, u16 pcifunc, bool enable)
 	/* This flag is required to clean up CGX conf if app gets killed */
 	pfvf->hw_rx_tstamp_en = enable;
 
+	/* Inform MCS about 8B RX header */
+	rvu_mcs_ptp_cfg(rvu, cgx_id, lmac_id, enable);
 	return 0;
 }
 
-- 
2.25.1

[net PATCH 4/9] octeontx2-af: mcs: Fix MCS block interrupt

[Geetha sowjanya]

On CN10KB, MCS IP vector number, BBE and PAB interrupt mask
got changed to support more block level interrupts.
To address this changes, this patch fixes the bbe and pab
interrupt handlers.

Fixes: 6c635f78c474 ("octeontx2-af: cn10k: mcs: Handle MCS block interrupts")
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../net/ethernet/marvell/octeontx2/af/mcs.c   | 95 ++++++++-----------
 .../net/ethernet/marvell/octeontx2/af/mcs.h   | 26 +++--
 .../marvell/octeontx2/af/mcs_cnf10kb.c        | 63 ++++++++++++
 .../ethernet/marvell/octeontx2/af/mcs_reg.h   |  5 +-
 4 files changed, 119 insertions(+), 70 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
index 148417d633a5..c43f19dfbd74 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs.c
@@ -936,60 +936,42 @@ static void mcs_tx_misc_intr_handler(struct mcs *mcs, u64 intr)
 	mcs_add_intr_wq_entry(mcs, &event);
 }
 
-static void mcs_bbe_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir)
+void cn10kb_mcs_bbe_intr_handler(struct mcs *mcs, u64 intr,
+				 enum mcs_direction dir)
 {
-	struct mcs_intr_event event = { 0 };
-	int i;
+	u64 val, reg;
+	int lmac;
 
-	if (!(intr & MCS_BBE_INT_MASK))
+	if (!(intr & 0x6ULL))
 		return;
 
-	event.mcs_id = mcs->mcs_id;
-	event.pcifunc = mcs->pf_map[0];
+	if (intr & BIT_ULL(1))
+		reg = (dir == MCS_RX) ? MCSX_BBE_RX_SLAVE_DFIFO_OVERFLOW_0 :
+					MCSX_BBE_TX_SLAVE_DFIFO_OVERFLOW_0;
+	else
+		reg = (dir == MCS_RX) ? MCSX_BBE_RX_SLAVE_PLFIFO_OVERFLOW_0 :
+					MCSX_BBE_TX_SLAVE_PLFIFO_OVERFLOW_0;
+	val = mcs_reg_read(mcs, reg);
 
-	for (i = 0; i < MCS_MAX_BBE_INT; i++) {
-		if (!(intr & BIT_ULL(i)))
+	/* policy/data over flow occurred */
+	for (lmac = 0; lmac < mcs->hw->lmac_cnt; lmac++) {
+		if (!(val & BIT_ULL(lmac)))
 			continue;
-
-		/* Lower nibble denotes data fifo overflow interrupts and
-		 * upper nibble indicates policy fifo overflow interrupts.
-		 */
-		if (intr & 0xFULL)
-			event.intr_mask = (dir == MCS_RX) ?
-					  MCS_BBE_RX_DFIFO_OVERFLOW_INT :
-					  MCS_BBE_TX_DFIFO_OVERFLOW_INT;
-		else
-			event.intr_mask = (dir == MCS_RX) ?
-					  MCS_BBE_RX_PLFIFO_OVERFLOW_INT :
-					  MCS_BBE_TX_PLFIFO_OVERFLOW_INT;
-
-		/* Notify the lmac_id info which ran into BBE fatal error */
-		event.lmac_id = i & 0x3ULL;
-		mcs_add_intr_wq_entry(mcs, &event);
+		dev_warn(mcs->dev, "BEE:Policy or data overflow occurred on lmac:%d\n", lmac);
 	}
 }
 
-static void mcs_pab_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir)
+void cn10kb_mcs_pab_intr_handler(struct mcs *mcs, u64 intr,
+				 enum mcs_direction dir)
 {
-	struct mcs_intr_event event = { 0 };
-	int i;
+	int lmac;
 
-	if (!(intr & MCS_PAB_INT_MASK))
+	if (!(intr & 0xFFFFFULL))
 		return;
 
-	event.mcs_id = mcs->mcs_id;
-	event.pcifunc = mcs->pf_map[0];
-
-	for (i = 0; i < MCS_MAX_PAB_INT; i++) {
-		if (!(intr & BIT_ULL(i)))
-			continue;
-
-		event.intr_mask = (dir == MCS_RX) ? MCS_PAB_RX_CHAN_OVERFLOW_INT :
-				  MCS_PAB_TX_CHAN_OVERFLOW_INT;
-
-		/* Notify the lmac_id info which ran into PAB fatal error */
-		event.lmac_id = i;
-		mcs_add_intr_wq_entry(mcs, &event);
+	for (lmac = 0; lmac < mcs->hw->lmac_cnt; lmac++) {
+		if (intr & BIT_ULL(lmac))
+			dev_warn(mcs->dev, "PAB: overflow occurred on lmac:%d\n", lmac);
 	}
 }
 
@@ -998,9 +980,8 @@ static irqreturn_t mcs_ip_intr_handler(int irq, void *mcs_irq)
 	struct mcs *mcs = (struct mcs *)mcs_irq;
 	u64 intr, cpm_intr, bbe_intr, pab_intr;
 
-	/* Disable and clear the interrupt */
+	/* Disable  the interrupt */
 	mcs_reg_write(mcs, MCSX_IP_INT_ENA_W1C, BIT_ULL(0));
-	mcs_reg_write(mcs, MCSX_IP_INT, BIT_ULL(0));
 
 	/* Check which block has interrupt*/
 	intr = mcs_reg_read(mcs, MCSX_TOP_SLAVE_INT_SUM);
@@ -1047,7 +1028,7 @@ static irqreturn_t mcs_ip_intr_handler(int irq, void *mcs_irq)
 	/* BBE RX */
 	if (intr & MCS_BBE_RX_INT_ENA) {
 		bbe_intr = mcs_reg_read(mcs, MCSX_BBE_RX_SLAVE_BBE_INT);
-		mcs_bbe_intr_handler(mcs, bbe_intr, MCS_RX);
+		mcs->mcs_ops->mcs_bbe_intr_handler(mcs, bbe_intr, MCS_RX);
 
 		/* Clear the interrupt */
 		mcs_reg_write(mcs, MCSX_BBE_RX_SLAVE_BBE_INT_INTR_RW, 0);
@@ -1057,7 +1038,7 @@ static irqreturn_t mcs_ip_intr_handler(int irq, void *mcs_irq)
 	/* BBE TX */
 	if (intr & MCS_BBE_TX_INT_ENA) {
 		bbe_intr = mcs_reg_read(mcs, MCSX_BBE_TX_SLAVE_BBE_INT);
-		mcs_bbe_intr_handler(mcs, bbe_intr, MCS_TX);
+		mcs->mcs_ops->mcs_bbe_intr_handler(mcs, bbe_intr, MCS_TX);
 
 		/* Clear the interrupt */
 		mcs_reg_write(mcs, MCSX_BBE_TX_SLAVE_BBE_INT_INTR_RW, 0);
@@ -1067,7 +1048,7 @@ static irqreturn_t mcs_ip_intr_handler(int irq, void *mcs_irq)
 	/* PAB RX */
 	if (intr & MCS_PAB_RX_INT_ENA) {
 		pab_intr = mcs_reg_read(mcs, MCSX_PAB_RX_SLAVE_PAB_INT);
-		mcs_pab_intr_handler(mcs, pab_intr, MCS_RX);
+		mcs->mcs_ops->mcs_pab_intr_handler(mcs, pab_intr, MCS_RX);
 
 		/* Clear the interrupt */
 		mcs_reg_write(mcs, MCSX_PAB_RX_SLAVE_PAB_INT_INTR_RW, 0);
@@ -1077,14 +1058,15 @@ static irqreturn_t mcs_ip_intr_handler(int irq, void *mcs_irq)
 	/* PAB TX */
 	if (intr & MCS_PAB_TX_INT_ENA) {
 		pab_intr = mcs_reg_read(mcs, MCSX_PAB_TX_SLAVE_PAB_INT);
-		mcs_pab_intr_handler(mcs, pab_intr, MCS_TX);
+		mcs->mcs_ops->mcs_pab_intr_handler(mcs, pab_intr, MCS_TX);
 
 		/* Clear the interrupt */
 		mcs_reg_write(mcs, MCSX_PAB_TX_SLAVE_PAB_INT_INTR_RW, 0);
 		mcs_reg_write(mcs, MCSX_PAB_TX_SLAVE_PAB_INT, pab_intr);
 	}
 
-	/* Enable the interrupt */
+	/* Clear and enable the interrupt */
+	mcs_reg_write(mcs, MCSX_IP_INT, BIT_ULL(0));
 	mcs_reg_write(mcs, MCSX_IP_INT_ENA_W1S, BIT_ULL(0));
 
 	return IRQ_HANDLED;
@@ -1166,7 +1148,7 @@ static int mcs_register_interrupts(struct mcs *mcs)
 		return ret;
 	}
 
-	ret = request_irq(pci_irq_vector(mcs->pdev, MCS_INT_VEC_IP),
+	ret = request_irq(pci_irq_vector(mcs->pdev, mcs->hw->ip_vec),
 			  mcs_ip_intr_handler, 0, "MCS_IP", mcs);
 	if (ret) {
 		dev_err(mcs->dev, "MCS IP irq registration failed\n");
@@ -1185,11 +1167,11 @@ static int mcs_register_interrupts(struct mcs *mcs)
 	mcs_reg_write(mcs, MCSX_CPM_TX_SLAVE_TX_INT_ENB, 0x7ULL);
 	mcs_reg_write(mcs, MCSX_CPM_RX_SLAVE_RX_INT_ENB, 0x7FULL);
 
-	mcs_reg_write(mcs, MCSX_BBE_RX_SLAVE_BBE_INT_ENB, 0xff);
-	mcs_reg_write(mcs, MCSX_BBE_TX_SLAVE_BBE_INT_ENB, 0xff);
+	mcs_reg_write(mcs, MCSX_BBE_RX_SLAVE_BBE_INT_ENB, 0xFFULL);
+	mcs_reg_write(mcs, MCSX_BBE_TX_SLAVE_BBE_INT_ENB, 0xFFULL);
 
-	mcs_reg_write(mcs, MCSX_PAB_RX_SLAVE_PAB_INT_ENB, 0xff);
-	mcs_reg_write(mcs, MCSX_PAB_TX_SLAVE_PAB_INT_ENB, 0xff);
+	mcs_reg_write(mcs, MCSX_PAB_RX_SLAVE_PAB_INT_ENB, 0xFFFFFULL);
+	mcs_reg_write(mcs, MCSX_PAB_TX_SLAVE_PAB_INT_ENB, 0xFFFFFULL);
 
 	mcs->tx_sa_active = alloc_mem(mcs, mcs->hw->sc_entries);
 	if (!mcs->tx_sa_active) {
@@ -1200,7 +1182,7 @@ static int mcs_register_interrupts(struct mcs *mcs)
 	return ret;
 
 free_irq:
-	free_irq(pci_irq_vector(mcs->pdev, MCS_INT_VEC_IP), mcs);
+	free_irq(pci_irq_vector(mcs->pdev, mcs->hw->ip_vec), mcs);
 exit:
 	pci_free_irq_vectors(mcs->pdev);
 	mcs->num_vec = 0;
@@ -1497,6 +1479,7 @@ void cn10kb_mcs_set_hw_capabilities(struct mcs *mcs)
 	hw->lmac_cnt = 20;		/* lmacs/ports per mcs block */
 	hw->mcs_x2p_intf = 5;		/* x2p clabration intf */
 	hw->mcs_blks = 1;		/* MCS blocks */
+	hw->ip_vec = MCS_CN10KB_INT_VEC_IP; /* IP vector */
 }
 
 static struct mcs_ops cn10kb_mcs_ops = {
@@ -1505,6 +1488,8 @@ static struct mcs_ops cn10kb_mcs_ops = {
 	.mcs_tx_sa_mem_map_write	= cn10kb_mcs_tx_sa_mem_map_write,
 	.mcs_rx_sa_mem_map_write	= cn10kb_mcs_rx_sa_mem_map_write,
 	.mcs_flowid_secy_map		= cn10kb_mcs_flowid_secy_map,
+	.mcs_bbe_intr_handler		= cn10kb_mcs_bbe_intr_handler,
+	.mcs_pab_intr_handler		= cn10kb_mcs_pab_intr_handler,
 };
 
 static int mcs_probe(struct pci_dev *pdev, const struct pci_device_id *id)
@@ -1605,7 +1590,7 @@ static void mcs_remove(struct pci_dev *pdev)
 
 	/* Set MCS to external bypass */
 	mcs_set_external_bypass(mcs, true);
-	free_irq(pci_irq_vector(pdev, MCS_INT_VEC_IP), mcs);
+	free_irq(pci_irq_vector(pdev, mcs->hw->ip_vec), mcs);
 	pci_free_irq_vectors(pdev);
 	pci_release_regions(pdev);
 	pci_disable_device(pdev);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs.h b/drivers/net/ethernet/marvell/octeontx2/af/mcs.h
index 64dc2b80e15d..0f89dcb76465 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs.h
@@ -43,24 +43,15 @@
 /* Reserved resources for default bypass entry */
 #define MCS_RSRC_RSVD_CNT		1
 
-/* MCS Interrupt Vector Enumeration */
-enum mcs_int_vec_e {
-	MCS_INT_VEC_MIL_RX_GBL		= 0x0,
-	MCS_INT_VEC_MIL_RX_LMACX	= 0x1,
-	MCS_INT_VEC_MIL_TX_LMACX	= 0x5,
-	MCS_INT_VEC_HIL_RX_GBL		= 0x9,
-	MCS_INT_VEC_HIL_RX_LMACX	= 0xa,
-	MCS_INT_VEC_HIL_TX_GBL		= 0xe,
-	MCS_INT_VEC_HIL_TX_LMACX	= 0xf,
-	MCS_INT_VEC_IP			= 0x13,
-	MCS_INT_VEC_CNT			= 0x14,
-};
+/* MCS Interrupt Vector */
+#define MCS_CNF10KB_INT_VEC_IP	0x13
+#define MCS_CN10KB_INT_VEC_IP	0x53
 
 #define MCS_MAX_BBE_INT			8ULL
 #define MCS_BBE_INT_MASK		0xFFULL
 
-#define MCS_MAX_PAB_INT			4ULL
-#define MCS_PAB_INT_MASK		0xFULL
+#define MCS_MAX_PAB_INT		8ULL
+#define MCS_PAB_INT_MASK	0xFULL
 
 #define MCS_BBE_RX_INT_ENA		BIT_ULL(0)
 #define MCS_BBE_TX_INT_ENA		BIT_ULL(1)
@@ -137,6 +128,7 @@ struct hwinfo {
 	u8 lmac_cnt;
 	u8 mcs_blks;
 	unsigned long	lmac_bmap; /* bitmap of enabled mcs lmac */
+	u16 ip_vec;
 };
 
 struct mcs {
@@ -165,6 +157,8 @@ struct mcs_ops {
 	void	(*mcs_tx_sa_mem_map_write)(struct mcs *mcs, struct mcs_tx_sc_sa_map *map);
 	void	(*mcs_rx_sa_mem_map_write)(struct mcs *mcs, struct mcs_rx_sc_sa_map *map);
 	void	(*mcs_flowid_secy_map)(struct mcs *mcs, struct secy_mem_map *map, int dir);
+	void	(*mcs_bbe_intr_handler)(struct mcs *mcs, u64 intr, enum mcs_direction dir);
+	void	(*mcs_pab_intr_handler)(struct mcs *mcs, u64 intr, enum mcs_direction dir);
 };
 
 extern struct pci_driver mcs_driver;
@@ -219,6 +213,8 @@ void cn10kb_mcs_tx_sa_mem_map_write(struct mcs *mcs, struct mcs_tx_sc_sa_map *ma
 void cn10kb_mcs_flowid_secy_map(struct mcs *mcs, struct secy_mem_map *map, int dir);
 void cn10kb_mcs_rx_sa_mem_map_write(struct mcs *mcs, struct mcs_rx_sc_sa_map *map);
 void cn10kb_mcs_parser_cfg(struct mcs *mcs);
+void cn10kb_mcs_pab_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir);
+void cn10kb_mcs_bbe_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir);
 
 /* CNF10K-B APIs */
 struct mcs_ops *cnf10kb_get_mac_ops(void);
@@ -229,6 +225,8 @@ void cnf10kb_mcs_rx_sa_mem_map_write(struct mcs *mcs, struct mcs_rx_sc_sa_map *m
 void cnf10kb_mcs_parser_cfg(struct mcs *mcs);
 void cnf10kb_mcs_tx_pn_thresh_reached_handler(struct mcs *mcs);
 void cnf10kb_mcs_tx_pn_wrapped_handler(struct mcs *mcs);
+void cnf10kb_mcs_bbe_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir);
+void cnf10kb_mcs_pab_intr_handler(struct mcs *mcs, u64 intr, enum mcs_direction dir);
 
 /* Stats APIs */
 void mcs_get_sc_stats(struct mcs *mcs, struct mcs_sc_stats *stats, int id, int dir);
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_cnf10kb.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs_cnf10kb.c
index 7b6205414428..9f9b904ab2cd 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_cnf10kb.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_cnf10kb.c
@@ -13,6 +13,8 @@ static struct mcs_ops cnf10kb_mcs_ops   = {
 	.mcs_tx_sa_mem_map_write	= cnf10kb_mcs_tx_sa_mem_map_write,
 	.mcs_rx_sa_mem_map_write	= cnf10kb_mcs_rx_sa_mem_map_write,
 	.mcs_flowid_secy_map		= cnf10kb_mcs_flowid_secy_map,
+	.mcs_bbe_intr_handler		= cnf10kb_mcs_bbe_intr_handler,
+	.mcs_pab_intr_handler		= cnf10kb_mcs_pab_intr_handler,
 };
 
 struct mcs_ops *cnf10kb_get_mac_ops(void)
@@ -31,6 +33,7 @@ void cnf10kb_mcs_set_hw_capabilities(struct mcs *mcs)
 	hw->lmac_cnt = 4;		/* lmacs/ports per mcs block */
 	hw->mcs_x2p_intf = 1;		/* x2p clabration intf */
 	hw->mcs_blks = 7;		/* MCS blocks */
+	hw->ip_vec = MCS_CNF10KB_INT_VEC_IP; /* IP vector */
 }
 
 void cnf10kb_mcs_parser_cfg(struct mcs *mcs)
@@ -212,3 +215,63 @@ void cnf10kb_mcs_tx_pn_wrapped_handler(struct mcs *mcs)
 		mcs_add_intr_wq_entry(mcs, &event);
 	}
 }
+
+void cnf10kb_mcs_bbe_intr_handler(struct mcs *mcs, u64 intr,
+				  enum mcs_direction dir)
+{
+	struct mcs_intr_event event = { 0 };
+	int i;
+
+	if (!(intr & MCS_BBE_INT_MASK))
+		return;
+
+	event.mcs_id = mcs->mcs_id;
+	event.pcifunc = mcs->pf_map[0];
+
+	for (i = 0; i < MCS_MAX_BBE_INT; i++) {
+		if (!(intr & BIT_ULL(i)))
+			continue;
+
+		/* Lower nibble denotes data fifo overflow interrupts and
+		 * upper nibble indicates policy fifo overflow interrupts.
+		 */
+		if (intr & 0xFULL)
+			event.intr_mask = (dir == MCS_RX) ?
+					  MCS_BBE_RX_DFIFO_OVERFLOW_INT :
+					  MCS_BBE_TX_DFIFO_OVERFLOW_INT;
+		else
+			event.intr_mask = (dir == MCS_RX) ?
+					  MCS_BBE_RX_PLFIFO_OVERFLOW_INT :
+					  MCS_BBE_TX_PLFIFO_OVERFLOW_INT;
+
+		/* Notify the lmac_id info which ran into BBE fatal error */
+		event.lmac_id = i & 0x3ULL;
+		mcs_add_intr_wq_entry(mcs, &event);
+	}
+}
+
+void cnf10kb_mcs_pab_intr_handler(struct mcs *mcs, u64 intr,
+				  enum mcs_direction dir)
+{
+	struct mcs_intr_event event = { 0 };
+	int i;
+
+	if (!(intr & MCS_PAB_INT_MASK))
+		return;
+
+	event.mcs_id = mcs->mcs_id;
+	event.pcifunc = mcs->pf_map[0];
+
+	for (i = 0; i < MCS_MAX_PAB_INT; i++) {
+		if (!(intr & BIT_ULL(i)))
+			continue;
+
+		event.intr_mask = (dir == MCS_RX) ?
+				  MCS_PAB_RX_CHAN_OVERFLOW_INT :
+				  MCS_PAB_TX_CHAN_OVERFLOW_INT;
+
+		/* Notify the lmac_id info which ran into PAB fatal error */
+		event.lmac_id = i;
+		mcs_add_intr_wq_entry(mcs, &event);
+	}
+}
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h b/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
index 7427e3b1490f..f3ab01fc363c 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_reg.h
@@ -276,7 +276,10 @@
 #define MCSX_BBE_RX_SLAVE_CAL_ENTRY			0x180ull
 #define MCSX_BBE_RX_SLAVE_CAL_LEN			0x188ull
 #define MCSX_PAB_RX_SLAVE_FIFO_SKID_CFGX(a)		(0x290ull + (a) * 0x40ull)
-
+#define MCSX_BBE_RX_SLAVE_DFIFO_OVERFLOW_0		0xe20
+#define MCSX_BBE_TX_SLAVE_DFIFO_OVERFLOW_0		0x1298
+#define MCSX_BBE_RX_SLAVE_PLFIFO_OVERFLOW_0		0xe40
+#define MCSX_BBE_TX_SLAVE_PLFIFO_OVERFLOW_0		0x12b8
 #define MCSX_BBE_RX_SLAVE_BBE_INT ({	\
 	u64 offset;			\
 					\
-- 
2.25.1

[net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

When system is rebooted after creating macsec interface
below NULL pointer dereference crashes occurred. This
patch fixes those crashes.

[ 3324.406942] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
[ 3324.415726] Mem abort info:
[ 3324.418510]   ESR = 0x96000006
[ 3324.421557]   EC = 0x25: DABT (current EL), IL = 32 bits
[ 3324.426865]   SET = 0, FnV = 0
[ 3324.429913]   EA = 0, S1PTW = 0
[ 3324.433047] Data abort info:
[ 3324.435921]   ISV = 0, ISS = 0x00000006
[ 3324.439748]   CM = 0, WnR = 0
....
[ 3324.575915] Call trace:
[ 3324.578353]  cn10k_mdo_del_secy+0x24/0x180
[ 3324.582440]  macsec_common_dellink+0xec/0x120
[ 3324.586788]  macsec_notify+0x17c/0x1c0
[ 3324.590529]  raw_notifier_call_chain+0x50/0x70
[ 3324.594965]  call_netdevice_notifiers_info+0x34/0x7c
[ 3324.599921]  rollback_registered_many+0x354/0x5bc
[ 3324.604616]  unregister_netdevice_queue+0x88/0x10c
[ 3324.609399]  unregister_netdev+0x20/0x30
[ 3324.613313]  otx2_remove+0x8c/0x310
[ 3324.616794]  pci_device_shutdown+0x30/0x70
[ 3324.620882]  device_shutdown+0x11c/0x204

[  966.664930] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
[  966.673712] Mem abort info:
[  966.676497]   ESR = 0x96000006
[  966.679543]   EC = 0x25: DABT (current EL), IL = 32 bits
[  966.684848]   SET = 0, FnV = 0
[  966.687895]   EA = 0, S1PTW = 0
[  966.691028] Data abort info:
[  966.693900]   ISV = 0, ISS = 0x00000006
[  966.697729]   CM = 0, WnR = 0
....
[  966.833467] Call trace:
[  966.835904]  cn10k_mdo_stop+0x20/0xa0
[  966.839557]  macsec_dev_stop+0xe8/0x11c
[  966.843384]  __dev_close_many+0xbc/0x140
[  966.847298]  dev_close_many+0x84/0x120
[  966.851039]  rollback_registered_many+0x114/0x5bc
[  966.855735]  unregister_netdevice_many.part.0+0x14/0xa0
[  966.860952]  unregister_netdevice_many+0x18/0x24
[  966.865560]  macsec_notify+0x1ac/0x1c0
[  966.869303]  raw_notifier_call_chain+0x50/0x70
[  966.873738]  call_netdevice_notifiers_info+0x34/0x7c
[  966.878694]  rollback_registered_many+0x354/0x5bc
[  966.883390]  unregister_netdevice_queue+0x88/0x10c
[  966.888173]  unregister_netdev+0x20/0x30
[  966.892090]  otx2_remove+0x8c/0x310
[  966.895571]  pci_device_shutdown+0x30/0x70
[  966.899660]  device_shutdown+0x11c/0x204
[  966.903574]  __do_sys_reboot+0x208/0x290
[  966.907487]  __arm64_sys_reboot+0x20/0x30
[  966.911489]  el0_svc_handler+0x80/0x1c0
[  966.915316]  el0_svc+0x8/0x180
[  966.918362] Code: f9400000 f9400a64 91220014 f94b3403 (f9400060)
[  966.924448] ---[ end trace 341778e799c3d8d7 ]---

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
index 9ec5f38d38a8..5f4402f7b03e 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
@@ -1065,6 +1065,9 @@ static int cn10k_mdo_stop(struct macsec_context *ctx)
 	struct cn10k_mcs_txsc *txsc;
 	int err;
 
+	if (!cfg)
+		return 0;
+
 	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
 	if (!txsc)
 		return -ENOENT;
@@ -1146,6 +1149,9 @@ static int cn10k_mdo_del_secy(struct macsec_context *ctx)
 	struct cn10k_mcs_cfg *cfg = pfvf->macsec_cfg;
 	struct cn10k_mcs_txsc *txsc;
 
+	if (!cfg)
+		return 0;
+
 	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
 	if (!txsc)
 		return -ENOENT;
-- 
2.25.1

[net PATCH 6/9] octeontx2-pf: mcs: Match macsec ethertype along with DMAC

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

On CN10KB silicon a single hardware macsec block is
present and offloads macsec operations for all the
ethernet LMACs. TCAM match with macsec ethertype 0x88e5
alone at RX side is not sufficient to distinguish all the
macsec interfaces created on top of netdevs. Hence append
the DMAC of the macsec interface too. Otherwise the first
created macsec interface only receives all the macsec traffic.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c  | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
index 5f4402f7b03e..f3140f466b37 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
@@ -9,6 +9,7 @@
 #include <net/macsec.h>
 #include "otx2_common.h"
 
+#define MCS_TCAM0_MAC_DA_MASK		GENMASK_ULL(47, 0)
 #define MCS_TCAM0_MAC_SA_MASK		GENMASK_ULL(63, 48)
 #define MCS_TCAM1_MAC_SA_MASK		GENMASK_ULL(31, 0)
 #define MCS_TCAM1_ETYPE_MASK		GENMASK_ULL(47, 32)
@@ -237,8 +238,10 @@ static int cn10k_mcs_write_rx_flowid(struct otx2_nic *pfvf,
 				     struct cn10k_mcs_rxsc *rxsc, u8 hw_secy_id)
 {
 	struct macsec_rx_sc *sw_rx_sc = rxsc->sw_rxsc;
+	struct macsec_secy *secy = rxsc->sw_secy;
 	struct mcs_flowid_entry_write_req *req;
 	struct mbox *mbox = &pfvf->mbox;
+	u64 mac_da;
 	int ret;
 
 	mutex_lock(&mbox->lock);
@@ -249,11 +252,16 @@ static int cn10k_mcs_write_rx_flowid(struct otx2_nic *pfvf,
 		goto fail;
 	}
 
+	mac_da = ether_addr_to_u64(secy->netdev->dev_addr);
+
+	req->data[0] = FIELD_PREP(MCS_TCAM0_MAC_DA_MASK, mac_da);
+	req->mask[0] = ~0ULL;
+	req->mask[0] = ~MCS_TCAM0_MAC_DA_MASK;
+
 	req->data[1] = FIELD_PREP(MCS_TCAM1_ETYPE_MASK, ETH_P_MACSEC);
 	req->mask[1] = ~0ULL;
 	req->mask[1] &= ~MCS_TCAM1_ETYPE_MASK;
 
-	req->mask[0] = ~0ULL;
 	req->mask[2] = ~0ULL;
 	req->mask[3] = ~0ULL;
 
-- 
2.25.1

[net PATCH 7/9] octeontx2-pf: mcs: Clear stats before freeing resource

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

When freeing MCS hardware resources like SecY, SC and
SA the corresponding stats needs to be cleared. Otherwise
previous stats are shown in newly created macsec interfaces.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c    | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
index f3140f466b37..a108f986be50 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
@@ -150,11 +150,20 @@ static void cn10k_mcs_free_rsrc(struct otx2_nic *pfvf, enum mcs_direction dir,
 				enum mcs_rsrc_type type, u16 hw_rsrc_id,
 				bool all)
 {
+	struct mcs_clear_stats *clear_req;
 	struct mbox *mbox = &pfvf->mbox;
 	struct mcs_free_rsrc_req *req;
 
 	mutex_lock(&mbox->lock);
 
+	clear_req = otx2_mbox_alloc_msg_mcs_clear_stats(mbox);
+	if (!clear_req)
+		goto fail;
+
+	clear_req->id = hw_rsrc_id;
+	clear_req->type = type;
+	clear_req->dir = dir;
+
 	req = otx2_mbox_alloc_msg_mcs_free_resources(mbox);
 	if (!req)
 		goto fail;
-- 
2.25.1

[net PATCH 8/9] octeontx2-pf: mcs: Fix shared counters logic

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

Macsec stats like InPktsLate and InPktsDelayed share
same counter in hardware. If SecY replay_protect is true
then counter represents InPktsLate otherwise InPktsDelayed.
This mode change was tracked based on protect_frames
instead of replay_protect mistakenly. Similarly InPktsUnchecked
and InPktsOk share same counter and mode change was tracked
based on validate_check instead of validate_disabled.
This patch fixes those problems.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../ethernet/marvell/octeontx2/nic/cn10k_macsec.c  | 14 +++++++-------
 .../ethernet/marvell/octeontx2/nic/otx2_common.h   |  2 +-
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
index a108f986be50..0ef44faa1ee8 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
@@ -1014,7 +1014,7 @@ static void cn10k_mcs_sync_stats(struct otx2_nic *pfvf, struct macsec_secy *secy
 
 	/* Check if sync is really needed */
 	if (secy->validate_frames == txsc->last_validate_frames &&
-	    secy->protect_frames == txsc->last_protect_frames)
+	    secy->replay_protect == txsc->last_replay_protect)
 		return;
 
 	cn10k_mcs_secy_stats(pfvf, txsc->hw_secy_id_rx, &rx_rsp, MCS_RX, true);
@@ -1036,19 +1036,19 @@ static void cn10k_mcs_sync_stats(struct otx2_nic *pfvf, struct macsec_secy *secy
 		rxsc->stats.InPktsInvalid += sc_rsp.pkt_invalid_cnt;
 		rxsc->stats.InPktsNotValid += sc_rsp.pkt_notvalid_cnt;
 
-		if (txsc->last_protect_frames)
+		if (txsc->last_replay_protect)
 			rxsc->stats.InPktsLate += sc_rsp.pkt_late_cnt;
 		else
 			rxsc->stats.InPktsDelayed += sc_rsp.pkt_late_cnt;
 
-		if (txsc->last_validate_frames == MACSEC_VALIDATE_CHECK)
+		if (txsc->last_validate_frames == MACSEC_VALIDATE_DISABLED)
 			rxsc->stats.InPktsUnchecked += sc_rsp.pkt_unchecked_cnt;
 		else
 			rxsc->stats.InPktsOK += sc_rsp.pkt_unchecked_cnt;
 	}
 
 	txsc->last_validate_frames = secy->validate_frames;
-	txsc->last_protect_frames = secy->protect_frames;
+	txsc->last_replay_protect = secy->replay_protect;
 }
 
 static int cn10k_mdo_open(struct macsec_context *ctx)
@@ -1120,7 +1120,7 @@ static int cn10k_mdo_add_secy(struct macsec_context *ctx)
 	txsc->sw_secy = secy;
 	txsc->encoding_sa = secy->tx_sc.encoding_sa;
 	txsc->last_validate_frames = secy->validate_frames;
-	txsc->last_protect_frames = secy->protect_frames;
+	txsc->last_replay_protect = secy->replay_protect;
 
 	list_add(&txsc->entry, &cfg->txsc_list);
 
@@ -1544,12 +1544,12 @@ static int cn10k_mdo_get_rx_sc_stats(struct macsec_context *ctx)
 	rxsc->stats.InPktsInvalid += rsp.pkt_invalid_cnt;
 	rxsc->stats.InPktsNotValid += rsp.pkt_notvalid_cnt;
 
-	if (secy->protect_frames)
+	if (secy->replay_protect)
 		rxsc->stats.InPktsLate += rsp.pkt_late_cnt;
 	else
 		rxsc->stats.InPktsDelayed += rsp.pkt_late_cnt;
 
-	if (secy->validate_frames == MACSEC_VALIDATE_CHECK)
+	if (secy->validate_frames == MACSEC_VALIDATE_DISABLED)
 		rxsc->stats.InPktsUnchecked += rsp.pkt_unchecked_cnt;
 	else
 		rxsc->stats.InPktsOK += rsp.pkt_unchecked_cnt;
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
index 3d22cc6a2804..f42b2b65bfd7 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h
@@ -389,7 +389,7 @@ struct cn10k_mcs_txsc {
 	struct cn10k_txsc_stats stats;
 	struct list_head entry;
 	enum macsec_validation_type last_validate_frames;
-	bool last_protect_frames;
+	bool last_replay_protect;
 	u16 hw_secy_id_tx;
 	u16 hw_secy_id_rx;
 	u16 hw_flow_id;
-- 
2.25.1

[net PATCH 9/9] octeontx2-pf: mcs: Do not reset PN while updating secy

[Geetha sowjanya]

From: Subbaraya Sundeep <sbhatta@marvell.com>

After creating SecYs, SCs and SAs a SecY can be modified
to change attributes like validation mode, protect frames
mode etc. During this SecY update, packet number is reset to
initial user given value by mistake. Hence do not reset
PN when updating SecY parameters.

Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
---
 .../ethernet/marvell/octeontx2/nic/cn10k_macsec.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
index 0ef44faa1ee8..9ef70980f786 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
@@ -1137,6 +1137,7 @@ static int cn10k_mdo_upd_secy(struct macsec_context *ctx)
 	struct macsec_secy *secy = ctx->secy;
 	struct macsec_tx_sa *sw_tx_sa;
 	struct cn10k_mcs_txsc *txsc;
+	bool active;
 	u8 sa_num;
 	int err;
 
@@ -1144,15 +1145,19 @@ static int cn10k_mdo_upd_secy(struct macsec_context *ctx)
 	if (!txsc)
 		return -ENOENT;
 
-	txsc->encoding_sa = secy->tx_sc.encoding_sa;
-
-	sa_num = txsc->encoding_sa;
-	sw_tx_sa = rcu_dereference_bh(secy->tx_sc.sa[sa_num]);
+	/* Encoding SA got changed */
+	if (txsc->encoding_sa != secy->tx_sc.encoding_sa) {
+		txsc->encoding_sa = secy->tx_sc.encoding_sa;
+		sa_num = txsc->encoding_sa;
+		sw_tx_sa = rcu_dereference_bh(secy->tx_sc.sa[sa_num]);
+		active = sw_tx_sa ? sw_tx_sa->active : false;
+		cn10k_mcs_link_tx_sa2sc(pfvf, secy, txsc, sa_num, active);
+	}
 
 	if (netif_running(secy->netdev)) {
 		cn10k_mcs_sync_stats(pfvf, secy, txsc);
 
-		err = cn10k_mcs_secy_tx_cfg(pfvf, secy, txsc, sw_tx_sa, sa_num);
+		err = cn10k_mcs_secy_tx_cfg(pfvf, secy, txsc, NULL, 0);
 		if (err)
 			return err;
 	}
-- 
2.25.1

Re: [net PATCH 2/9] octeontx2-af: mcs: Write TCAM_DATA and TCAM_MASK registers at once

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:47PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> As per hardware errata on CN10KB, all the four TCAM_DATA
> and TCAM_MASK registers has to be written at once otherwise
> write to individual registers will fail. Hence write to all
> TCAM_DATA registers and then to all TCAM_MASK registers.
> 
> Fixes: cfc14181d497 ("octeontx2-af: cn10k: mcs: Manage the MCS block hardware resources")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  drivers/net/ethernet/marvell/octeontx2/af/mcs.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 3/9] octeontx2-af: mcs: Config parser to skip 8B header

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:48PM +0530, Geetha sowjanya wrote:
> When ptp timestamp is enabled in RPM, RPM will append 8B
> timestamp header for all RX traffic. MCS need to skip these
> 8 bytes header while parsing the packet header, so that
> correct tcam key is created for lookup. 
> This patch fixes the mcs parser configuration to skip this 
> 8B header for ptp packets.
> 
> Fixes: ca7f49ff8846 ("octeontx2-af: cn10k: Introduce driver for macsec block.")
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../ethernet/marvell/octeontx2/af/mcs_reg.h   |  1 +
>  .../marvell/octeontx2/af/mcs_rvu_if.c         | 37 +++++++++++++++++++
>  .../net/ethernet/marvell/octeontx2/af/rvu.h   |  1 +
>  .../ethernet/marvell/octeontx2/af/rvu_cgx.c   |  2 +
>  4 files changed, 41 insertions(+)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 1/9] octeonxt2-af: mcs: Fix per port bypass config

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:46PM +0530, Geetha sowjanya wrote:
> For each lmac port, MCS has two MCS_TOP_SLAVE_CHANNEL_CONFIGX
> registers. For CN10KB both register need to be configured for the
> port level mcs bypass to work. This patch also sets bitmap
> of flowid/secy entry reserved for default bypass so that these
> entries can be shown in debugfs.
> 
> Fixes: bd69476e86fc ("octeontx2-af: cn10k: mcs: Install a default TCAM for normal traffic")
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> ---
>  drivers/net/ethernet/marvell/octeontx2/af/mcs.c       | 11 ++++++++++-
>  .../net/ethernet/marvell/octeontx2/af/rvu_debugfs.c   |  5 +++--
>  2 files changed, 13 insertions(+), 3 deletions(-)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 4/9] octeontx2-af: mcs: Fix MCS block interrupt

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:49PM +0530, Geetha sowjanya wrote:
> On CN10KB, MCS IP vector number, BBE and PAB interrupt mask
> got changed to support more block level interrupts.
> To address this changes, this patch fixes the bbe and pab
> interrupt handlers.
> 
> Fixes: 6c635f78c474 ("octeontx2-af: cn10k: mcs: Handle MCS block interrupts")
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../net/ethernet/marvell/octeontx2/af/mcs.c   | 95 ++++++++-----------
>  .../net/ethernet/marvell/octeontx2/af/mcs.h   | 26 +++--
>  .../marvell/octeontx2/af/mcs_cnf10kb.c        | 63 ++++++++++++
>  .../ethernet/marvell/octeontx2/af/mcs_reg.h   |  5 +-
>  4 files changed, 119 insertions(+), 70 deletions(-)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:50PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> When system is rebooted after creating macsec interface
> below NULL pointer dereference crashes occurred. This
> patch fixes those crashes.
> 
> [ 3324.406942] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
> [ 3324.415726] Mem abort info:
> [ 3324.418510]   ESR = 0x96000006
> [ 3324.421557]   EC = 0x25: DABT (current EL), IL = 32 bits
> [ 3324.426865]   SET = 0, FnV = 0
> [ 3324.429913]   EA = 0, S1PTW = 0
> [ 3324.433047] Data abort info:
> [ 3324.435921]   ISV = 0, ISS = 0x00000006
> [ 3324.439748]   CM = 0, WnR = 0
> ....
> [ 3324.575915] Call trace:
> [ 3324.578353]  cn10k_mdo_del_secy+0x24/0x180
> [ 3324.582440]  macsec_common_dellink+0xec/0x120
> [ 3324.586788]  macsec_notify+0x17c/0x1c0
> [ 3324.590529]  raw_notifier_call_chain+0x50/0x70
> [ 3324.594965]  call_netdevice_notifiers_info+0x34/0x7c
> [ 3324.599921]  rollback_registered_many+0x354/0x5bc
> [ 3324.604616]  unregister_netdevice_queue+0x88/0x10c
> [ 3324.609399]  unregister_netdev+0x20/0x30
> [ 3324.613313]  otx2_remove+0x8c/0x310
> [ 3324.616794]  pci_device_shutdown+0x30/0x70
> [ 3324.620882]  device_shutdown+0x11c/0x204
> 
> [  966.664930] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
> [  966.673712] Mem abort info:
> [  966.676497]   ESR = 0x96000006
> [  966.679543]   EC = 0x25: DABT (current EL), IL = 32 bits
> [  966.684848]   SET = 0, FnV = 0
> [  966.687895]   EA = 0, S1PTW = 0
> [  966.691028] Data abort info:
> [  966.693900]   ISV = 0, ISS = 0x00000006
> [  966.697729]   CM = 0, WnR = 0
> ....
> [  966.833467] Call trace:
> [  966.835904]  cn10k_mdo_stop+0x20/0xa0
> [  966.839557]  macsec_dev_stop+0xe8/0x11c
> [  966.843384]  __dev_close_many+0xbc/0x140
> [  966.847298]  dev_close_many+0x84/0x120
> [  966.851039]  rollback_registered_many+0x114/0x5bc
> [  966.855735]  unregister_netdevice_many.part.0+0x14/0xa0
> [  966.860952]  unregister_netdevice_many+0x18/0x24
> [  966.865560]  macsec_notify+0x1ac/0x1c0
> [  966.869303]  raw_notifier_call_chain+0x50/0x70
> [  966.873738]  call_netdevice_notifiers_info+0x34/0x7c
> [  966.878694]  rollback_registered_many+0x354/0x5bc
> [  966.883390]  unregister_netdevice_queue+0x88/0x10c
> [  966.888173]  unregister_netdev+0x20/0x30
> [  966.892090]  otx2_remove+0x8c/0x310
> [  966.895571]  pci_device_shutdown+0x30/0x70
> [  966.899660]  device_shutdown+0x11c/0x204
> [  966.903574]  __do_sys_reboot+0x208/0x290
> [  966.907487]  __arm64_sys_reboot+0x20/0x30
> [  966.911489]  el0_svc_handler+0x80/0x1c0
> [  966.915316]  el0_svc+0x8/0x180
> [  966.918362] Code: f9400000 f9400a64 91220014 f94b3403 (f9400060)
> [  966.924448] ---[ end trace 341778e799c3d8d7 ]---
> 
> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
> index 9ec5f38d38a8..5f4402f7b03e 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
> @@ -1065,6 +1065,9 @@ static int cn10k_mdo_stop(struct macsec_context *ctx)
>  	struct cn10k_mcs_txsc *txsc;
>  	int err;
>  
> +	if (!cfg)
> +		return 0;
> +
>  	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
>  	if (!txsc)
>  		return -ENOENT;
> @@ -1146,6 +1149,9 @@ static int cn10k_mdo_del_secy(struct macsec_context *ctx)
>  	struct cn10k_mcs_cfg *cfg = pfvf->macsec_cfg;
>  	struct cn10k_mcs_txsc *txsc;
>  
> +	if (!cfg)
> +		return 0;

How did you get call to .mdo_del_secy if you didn't add any secy?

Thanks

> +
>  	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
>  	if (!txsc)
>  		return -ENOENT;
> -- 
> 2.25.1
> 

Re: [net PATCH 6/9] octeontx2-pf: mcs: Match macsec ethertype along with DMAC

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:51PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> On CN10KB silicon a single hardware macsec block is
> present and offloads macsec operations for all the
> ethernet LMACs. TCAM match with macsec ethertype 0x88e5
> alone at RX side is not sufficient to distinguish all the
> macsec interfaces created on top of netdevs. Hence append
> the DMAC of the macsec interface too. Otherwise the first
> created macsec interface only receives all the macsec traffic.
> 
> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c  | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 7/9] octeontx2-pf: mcs: Clear stats before freeing resource

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:52PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> When freeing MCS hardware resources like SecY, SC and
> SA the corresponding stats needs to be cleared. Otherwise
> previous stats are shown in newly created macsec interfaces.
> 
> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c    | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 8/9] octeontx2-pf: mcs: Fix shared counters logic

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:53PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> Macsec stats like InPktsLate and InPktsDelayed share
> same counter in hardware. If SecY replay_protect is true
> then counter represents InPktsLate otherwise InPktsDelayed.
> This mode change was tracked based on protect_frames
> instead of replay_protect mistakenly. Similarly InPktsUnchecked
> and InPktsOk share same counter and mode change was tracked
> based on validate_check instead of validate_disabled.
> This patch fixes those problems.
> 
> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../ethernet/marvell/octeontx2/nic/cn10k_macsec.c  | 14 +++++++-------
>  .../ethernet/marvell/octeontx2/nic/otx2_common.h   |  2 +-
>  2 files changed, 8 insertions(+), 8 deletions(-)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

Re: [net PATCH 9/9] octeontx2-pf: mcs: Do not reset PN while updating secy

[Leon Romanovsky]

On Sun, Apr 23, 2023 at 03:24:54PM +0530, Geetha sowjanya wrote:
> From: Subbaraya Sundeep <sbhatta@marvell.com>
> 
> After creating SecYs, SCs and SAs a SecY can be modified
> to change attributes like validation mode, protect frames
> mode etc. During this SecY update, packet number is reset to
> initial user given value by mistake. Hence do not reset
> PN when updating SecY parameters.
> 
> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
> ---
>  .../ethernet/marvell/octeontx2/nic/cn10k_macsec.c | 15 ++++++++++-----
>  1 file changed, 10 insertions(+), 5 deletions(-)
> 

Thanks,
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>

RE: [EXT] Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Subbaraya Sundeep Bhatta]

Hi,

>-----Original Message-----
>From: Leon Romanovsky <leon@kernel.org>
>Sent: Sunday, April 23, 2023 10:22 PM
>To: Geethasowjanya Akula <gakula@marvell.com>
>Cc: netdev@vger.kernel.org; linux-kernel@vger.kernel.org; kuba@kernel.org;
>davem@davemloft.net; edumazet@google.com; pabeni@redhat.com;
>richardcochran@gmail.com; Sunil Kovvuri Goutham
><sgoutham@marvell.com>; Subbaraya Sundeep Bhatta
><sbhatta@marvell.com>; Hariprasad Kelam <hkelam@marvell.com>
>Subject: [EXT] Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer
>dereferences
>
>On Sun, Apr 23, 2023 at 03:24:50PM +0530, Geetha sowjanya wrote:
>> From: Subbaraya Sundeep <sbhatta@marvell.com>
>>
>> When system is rebooted after creating macsec interface below NULL
>> pointer dereference crashes occurred. This patch fixes those crashes.
>>
>> [ 3324.406942] Unable to handle kernel NULL pointer dereference at
>> virtual address 0000000000000000 [ 3324.415726] Mem abort info:
>> [ 3324.418510]   ESR = 0x96000006
>> [ 3324.421557]   EC = 0x25: DABT (current EL), IL = 32 bits
>> [ 3324.426865]   SET = 0, FnV = 0
>> [ 3324.429913]   EA = 0, S1PTW = 0
>> [ 3324.433047] Data abort info:
>> [ 3324.435921]   ISV = 0, ISS = 0x00000006
>> [ 3324.439748]   CM = 0, WnR = 0
>> ....
>> [ 3324.575915] Call trace:
>> [ 3324.578353]  cn10k_mdo_del_secy+0x24/0x180 [ 3324.582440]
>> macsec_common_dellink+0xec/0x120 [ 3324.586788]
>> macsec_notify+0x17c/0x1c0 [ 3324.590529]
>> raw_notifier_call_chain+0x50/0x70 [ 3324.594965]
>> call_netdevice_notifiers_info+0x34/0x7c
>> [ 3324.599921]  rollback_registered_many+0x354/0x5bc
>> [ 3324.604616]  unregister_netdevice_queue+0x88/0x10c
>> [ 3324.609399]  unregister_netdev+0x20/0x30 [ 3324.613313]
>> otx2_remove+0x8c/0x310 [ 3324.616794]  pci_device_shutdown+0x30/0x70
>[
>> 3324.620882]  device_shutdown+0x11c/0x204
>>
>> [  966.664930] Unable to handle kernel NULL pointer dereference at
>> virtual address 0000000000000000 [  966.673712] Mem abort info:
>> [  966.676497]   ESR = 0x96000006
>> [  966.679543]   EC = 0x25: DABT (current EL), IL = 32 bits
>> [  966.684848]   SET = 0, FnV = 0
>> [  966.687895]   EA = 0, S1PTW = 0
>> [  966.691028] Data abort info:
>> [  966.693900]   ISV = 0, ISS = 0x00000006
>> [  966.697729]   CM = 0, WnR = 0
>> ....
>> [  966.833467] Call trace:
>> [  966.835904]  cn10k_mdo_stop+0x20/0xa0 [  966.839557]
>> macsec_dev_stop+0xe8/0x11c [  966.843384]
>__dev_close_many+0xbc/0x140
>> [  966.847298]  dev_close_many+0x84/0x120 [  966.851039]
>> rollback_registered_many+0x114/0x5bc
>> [  966.855735]  unregister_netdevice_many.part.0+0x14/0xa0
>> [  966.860952]  unregister_netdevice_many+0x18/0x24
>> [  966.865560]  macsec_notify+0x1ac/0x1c0 [  966.869303]
>> raw_notifier_call_chain+0x50/0x70 [  966.873738]
>> call_netdevice_notifiers_info+0x34/0x7c
>> [  966.878694]  rollback_registered_many+0x354/0x5bc
>> [  966.883390]  unregister_netdevice_queue+0x88/0x10c
>> [  966.888173]  unregister_netdev+0x20/0x30 [  966.892090]
>> otx2_remove+0x8c/0x310 [  966.895571]  pci_device_shutdown+0x30/0x70 [
>> 966.899660]  device_shutdown+0x11c/0x204 [  966.903574]
>> __do_sys_reboot+0x208/0x290 [  966.907487]
>> __arm64_sys_reboot+0x20/0x30 [  966.911489]
>> el0_svc_handler+0x80/0x1c0 [  966.915316]  el0_svc+0x8/0x180 [
>> 966.918362] Code: f9400000 f9400a64 91220014 f94b3403 (f9400060) [
>> 966.924448] ---[ end trace 341778e799c3d8d7 ]---
>>
>> Fixes: c54ffc73601c ("octeontx2-pf: mcs: Introduce MACSEC hardware
>> offloading")
>> Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
>> Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
>> Signed-off-by: Geetha sowjanya <gakula@marvell.com>
>> ---
>>  drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c | 6 ++++++
>>  1 file changed, 6 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
>> b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
>> index 9ec5f38d38a8..5f4402f7b03e 100644
>> --- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
>> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_macsec.c
>> @@ -1065,6 +1065,9 @@ static int cn10k_mdo_stop(struct macsec_context
>*ctx)
>>  	struct cn10k_mcs_txsc *txsc;
>>  	int err;
>>
>> +	if (!cfg)
>> +		return 0;
>> +
>>  	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
>>  	if (!txsc)
>>  		return -ENOENT;
>> @@ -1146,6 +1149,9 @@ static int cn10k_mdo_del_secy(struct
>macsec_context *ctx)
>>  	struct cn10k_mcs_cfg *cfg = pfvf->macsec_cfg;
>>  	struct cn10k_mcs_txsc *txsc;
>>
>> +	if (!cfg)
>> +		return 0;
>
>How did you get call to .mdo_del_secy if you didn't add any secy?
>
>Thanks
>
It is because of the order of teardown in otx2_remove:
        cn10k_mcs_free(pf);
        unregister_netdev(netdev);

cn10k_mcs_free free the resources and makes cfg as NULL.
Later unregister_netdev calls mdo_del_secy and finds cfg as NULL.
Thanks for the review I will change the order and submit next version.

Sundeep

>> +
>>  	txsc = cn10k_mcs_get_txsc(cfg, ctx->secy);
>>  	if (!txsc)
>>  		return -ENOENT;
>> --
>> 2.25.1
>>

Re: [EXT] Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Jakub Kicinski]

On Mon, 24 Apr 2023 10:29:02 +0000 Subbaraya Sundeep Bhatta wrote:
> >How did you get call to .mdo_del_secy if you didn't add any secy?
> >
> >Thanks
> >  
> It is because of the order of teardown in otx2_remove:
>         cn10k_mcs_free(pf);
>         unregister_netdev(netdev);
> 
> cn10k_mcs_free free the resources and makes cfg as NULL.
> Later unregister_netdev calls mdo_del_secy and finds cfg as NULL.
> Thanks for the review I will change the order and submit next version.

Leon, ack? Looks like the patches got "changes requested" but I see 
no other complaint.

Re: [EXT] Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Leon Romanovsky]

On Tue, Apr 25, 2023 at 08:51:40AM -0700, Jakub Kicinski wrote:
> On Mon, 24 Apr 2023 10:29:02 +0000 Subbaraya Sundeep Bhatta wrote:
> > >How did you get call to .mdo_del_secy if you didn't add any secy?
> > >
> > >Thanks
> > >  
> > It is because of the order of teardown in otx2_remove:
> >         cn10k_mcs_free(pf);
> >         unregister_netdev(netdev);
> > 
> > cn10k_mcs_free free the resources and makes cfg as NULL.
> > Later unregister_netdev calls mdo_del_secy and finds cfg as NULL.
> > Thanks for the review I will change the order and submit next version.
> 
> Leon, ack? Looks like the patches got "changes requested" but I see 
> no other complaint.

Honestly, I was confused and didn't know what to answer, so decided to
see next version.

From one side Subbaraya said that it is possible (which was not
convincing to me, but ok, most time I'm wrong :)), from another
he said that he will submit next version.

Thanks

Re: [EXT] Re: [net PATCH 5/9] octeontx2-pf: mcs: Fix NULL pointer dereferences

[Leon Romanovsky]

On Wed, Apr 26, 2023 at 09:16:54AM +0300, Leon Romanovsky wrote:
> On Tue, Apr 25, 2023 at 08:51:40AM -0700, Jakub Kicinski wrote:
> > On Mon, 24 Apr 2023 10:29:02 +0000 Subbaraya Sundeep Bhatta wrote:
> > > >How did you get call to .mdo_del_secy if you didn't add any secy?
> > > >
> > > >Thanks
> > > >  
> > > It is because of the order of teardown in otx2_remove:
> > >         cn10k_mcs_free(pf);
> > >         unregister_netdev(netdev);
> > > 
> > > cn10k_mcs_free free the resources and makes cfg as NULL.
> > > Later unregister_netdev calls mdo_del_secy and finds cfg as NULL.
> > > Thanks for the review I will change the order and submit next version.
> > 
> > Leon, ack? Looks like the patches got "changes requested" but I see 
> > no other complaint.
> 
> Honestly, I was confused and didn't know what to answer, so decided to
> see next version.
> 
> From one side Subbaraya said that it is possible (which was not
> convincing to me, but ok, most time I'm wrong :)), from another
> he said that he will submit next version.

Jakub,

v2 is perfectly fine.

Thanks

> 
> Thanks