Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next

Pull networking updates from David Miller:

 1) Allow setting bluetooth L2CAP modes via socket option, from Luiz
    Augusto von Dentz.

 2) Add GSO partial support to igc, from Sasha Neftin.

 3) Several cleanups and improvements to r8169 from Heiner Kallweit.

 4) Add IF_OPER_TESTING link state and use it when ethtool triggers a
    device self-test. From Andrew Lunn.

 5) Start moving away from custom driver versions, use the globally
    defined kernel version instead, from Leon Romanovsky.

 6) Support GRO vis gro_cells in DSA layer, from Alexander Lobakin.

 7) Allow hard IRQ deferral during NAPI, from Eric Dumazet.

 8) Add sriov and vf support to hinic, from Luo bin.

 9) Support Media Redundancy Protocol (MRP) in the bridging code, from
    Horatiu Vultur.

10) Support netmap in the nft_nat code, from Pablo Neira Ayuso.

11) Allow UDPv6 encapsulation of ESP in the ipsec code, from Sabrina
    Dubroca. Also add ipv6 support for espintcp.

12) Lots of ReST conversions of the networking documentation, from Mauro
    Carvalho Chehab.

13) Support configuration of ethtool rxnfc flows in bcmgenet driver,
    from Doug Berger.

14) Allow to dump cgroup id and filter by it in inet_diag code, from
    Dmitry Yakunin.

15) Add infrastructure to export netlink attribute policies to
    userspace, from Johannes Berg.

16) Several optimizations to sch_fq scheduler, from Eric Dumazet.

17) Fallback to the default qdisc if qdisc init fails because otherwise
    a packet scheduler init failure will make a device inoperative. From
    Jesper Dangaard Brouer.

18) Several RISCV bpf jit optimizations, from Luke Nelson.

19) Correct the return type of the ->ndo_start_xmit() method in several
    drivers, it's netdev_tx_t but many drivers were using
    'int'. From Yunjian Wang.

20) Add an ethtool interface for PHY master/slave config, from Oleksij
    Rempel.

21) Add BPF iterators, from Yonghang Song.

22) Add cable test infrastructure, including ethool interfaces, from
    Andrew Lunn. Marvell PHY driver is the first to support this
    facility.

23) Remove zero-length arrays all over, from Gustavo A. R. Silva.

24) Calculate and maintain an explicit frame size in XDP, from Jesper
    Dangaard Brouer.

25) Add CAP_BPF, from Alexei Starovoitov.

26) Support terse dumps in the packet scheduler, from Vlad Buslov.

27) Support XDP_TX bulking in dpaa2 driver, from Ioana Ciornei.

28) Add devm_register_netdev(), from Bartosz Golaszewski.

29) Minimize qdisc resets, from Cong Wang.

30) Get rid of kernel_getsockopt and kernel_setsockopt in order to
    eliminate set_fs/get_fs calls. From Christoph Hellwig.

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (2517 commits)
  selftests: net: ip_defrag: ignore EPERM
  net_failover: fixed rollback in net_failover_open()
  Revert "tipc: Fix potential tipc_aead refcnt leak in tipc_crypto_rcv"
  Revert "tipc: Fix potential tipc_node refcnt leak in tipc_rcv"
  vmxnet3: allow rx flow hash ops only when rss is enabled
  hinic: add set_channels ethtool_ops support
  selftests/bpf: Add a default $(CXX) value
  tools/bpf: Don't use $(COMPILE.c)
  bpf, selftests: Use bpf_probe_read_kernel
  s390/bpf: Use bcr 0,%0 as tail call nop filler
  s390/bpf: Maintain 8-byte stack alignment
  selftests/bpf: Fix verifier test
  selftests/bpf: Fix sample_cnt shared between two threads
  bpf, selftests: Adapt cls_redirect to call csum_level helper
  bpf: Add csum_level helper for fixing up csum levels
  bpf: Fix up bpf_skb_adjust_room helper's skb csum setting
  sfc: add missing annotation for efx_ef10_try_update_nic_stats_vf()
  crypto/chtls: IPv6 support for inline TLS
  Crypto/chcr: Fixes a coccinile check error
  Crypto/chcr: Fixes compilations warnings
  ...

11 files changed, 449 insertions, 154 deletions

diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index e46a14f44a6f..cf69c6edcfec 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -60,6 +60,7 @@
 
 #define CH_WARN(adap, fmt, ...) dev_warn(adap->pdev_dev, fmt, ## __VA_ARGS__)
 extern struct list_head adapter_list;
+extern struct list_head uld_list;
 extern struct mutex uld_mutex;
 
 /* Suspend an Ethernet Tx queue with fewer available descriptors than this.
@@ -466,8 +467,6 @@ static inline struct mbox_cmd *mbox_cmd_log_entry(struct mbox_cmd_log *log,
 	return &((struct mbox_cmd *)&(log)[1])[entry_idx];
 }
 
-#include "t4fw_api.h"
-
 #define FW_VERSION(chip) ( \
 		FW_HDR_FW_VER_MAJOR_G(chip##FW_VERSION_MAJOR) | \
 		FW_HDR_FW_VER_MINOR_G(chip##FW_VERSION_MINOR) | \
@@ -824,6 +823,13 @@ struct sge_uld_txq_info {
 	u16 ntxq;		/* # of egress uld queues */
 };
 
+/* struct to maintain ULD list to reallocate ULD resources on hotplug */
+struct cxgb4_uld_list {
+	struct cxgb4_uld_info uld_info;
+	struct list_head list_node;
+	enum cxgb4_uld uld_type;
+};
+
 enum sge_eosw_state {
 	CXGB4_EO_STATE_CLOSED = 0, /* Not ready to accept traffic */
 	CXGB4_EO_STATE_FLOWC_OPEN_SEND, /* Send FLOWC open request */
@@ -1093,6 +1099,7 @@ struct adapter {
 
 	/* TC u32 offload */
 	struct cxgb4_tc_u32_table *tc_u32;
+	struct chcr_ktls chcr_ktls;
 	struct chcr_stats_debug chcr_stats;
 
 	/* TC flower offload */
@@ -1127,19 +1134,20 @@ struct adapter {
  * programmed with various parameters.
  */
 struct ch_sched_params {
-	s8   type;                     /* packet or flow */
+	u8   type;                     /* packet or flow */
 	union {
 		struct {
-			s8   level;    /* scheduler hierarchy level */
-			s8   mode;     /* per-class or per-flow */
-			s8   rateunit; /* bit or packet rate */
-			s8   ratemode; /* %port relative or kbps absolute */
-			s8   channel;  /* scheduler channel [0..N] */
-			s8   class;    /* scheduler class [0..N] */
-			s32  minrate;  /* minimum rate */
-			s32  maxrate;  /* maximum rate */
-			s16  weight;   /* percent weight */
-			s16  pktsize;  /* average packet size */
+			u8   level;    /* scheduler hierarchy level */
+			u8   mode;     /* per-class or per-flow */
+			u8   rateunit; /* bit or packet rate */
+			u8   ratemode; /* %port relative or kbps absolute */
+			u8   channel;  /* scheduler channel [0..N] */
+			u8   class;    /* scheduler class [0..N] */
+			u32  minrate;  /* minimum rate */
+			u32  maxrate;  /* maximum rate */
+			u16  weight;   /* percent weight */
+			u16  pktsize;  /* average packet size */
+			u16  burstsize;  /* burst buffer size */
 		} params;
 	} u;
 };
@@ -1954,9 +1962,10 @@ int t4_sge_ctxt_rd(struct adapter *adap, unsigned int mbox, unsigned int cid,
 		   enum ctxt_type ctype, u32 *data);
 int t4_sge_ctxt_rd_bd(struct adapter *adap, unsigned int cid,
 		      enum ctxt_type ctype, u32 *data);
-int t4_sched_params(struct adapter *adapter, int type, int level, int mode,
-		    int rateunit, int ratemode, int channel, int class,
-		    int minrate, int maxrate, int weight, int pktsize);
+int t4_sched_params(struct adapter *adapter, u8 type, u8 level, u8 mode,
+		    u8 rateunit, u8 ratemode, u8 channel, u8 class,
+		    u32 minrate, u32 maxrate, u16 weight, u16 pktsize,
+		    u16 burstsize);
 void t4_sge_decode_idma_state(struct adapter *adapter, int state);
 void t4_idma_monitor_init(struct adapter *adapter,
 			  struct sge_idma_monitor_state *idma);
@@ -2052,4 +2061,7 @@ int cxgb_open(struct net_device *dev);
 int cxgb_close(struct net_device *dev);
 void cxgb4_enable_rx(struct adapter *adap, struct sge_rspq *q);
 void cxgb4_quiesce_rx(struct sge_rspq *q);
+#ifdef CONFIG_CHELSIO_TLS_DEVICE
+int cxgb4_set_ktls_feature(struct adapter *adap, bool enable);
+#endif
 #endif /* __CXGB4_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
index ebed99f3d4cf..41315712deb8 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
@@ -49,6 +49,7 @@
 #include "cudbg_lib_common.h"
 #include "cudbg_entity.h"
 #include "cudbg_lib.h"
+#include "cxgb4_tc_mqprio.h"
 
 /* generic seq_file support for showing a table of size rows x width. */
 static void *seq_tab_get_idx(struct seq_tab *tb, loff_t pos)
@@ -1812,12 +1813,8 @@ static int mps_tcam_show(struct seq_file *seq, void *v)
 			/* Inner header lookup */
 			if (lookup_type && (lookup_type != DATALKPTYPE_M)) {
 				seq_printf(seq,
-					   "%3u %02x:%02x:%02x:%02x:%02x:%02x "
-					   "%012llx %06x %06x    -    -   %3c"
-					   "      'I'  %4x   "
-					   "%3c   %#x%4u%4d", idx, addr[0],
-					   addr[1], addr[2], addr[3],
-					   addr[4], addr[5],
+					   "%3u %pM %012llx %06x %06x    -    -   %3c      'I'  %4x   %3c   %#x%4u%4d",
+					   idx, addr,
 					   (unsigned long long)mask,
 					   vniy, (vnix | vniy),
 					   dip_hit ? 'Y' : 'N',
@@ -1829,10 +1826,8 @@ static int mps_tcam_show(struct seq_file *seq, void *v)
 					   T6_VF_G(cls_lo) : -1);
 			} else {
 				seq_printf(seq,
-					   "%3u %02x:%02x:%02x:%02x:%02x:%02x "
-					   "%012llx    -       -   ",
-					   idx, addr[0], addr[1], addr[2],
-					   addr[3], addr[4], addr[5],
+					   "%3u %pM %012llx    -       -   ",
+					   idx, addr,
 					   (unsigned long long)mask);
 
 				if (vlan_vld)
@@ -1850,10 +1845,8 @@ static int mps_tcam_show(struct seq_file *seq, void *v)
 					   T6_VF_G(cls_lo) : -1);
 			}
 		} else
-			seq_printf(seq, "%3u %02x:%02x:%02x:%02x:%02x:%02x "
-				   "%012llx%3c   %#x%4u%4d",
-				   idx, addr[0], addr[1], addr[2], addr[3],
-				   addr[4], addr[5], (unsigned long long)mask,
+			seq_printf(seq, "%3u %pM %012llx%3c   %#x%4u%4d",
+				   idx, addr, (unsigned long long)mask,
 				   (cls_lo & SRAM_VLD_F) ? 'Y' : 'N',
 				   PORTMAP_G(cls_hi),
 				   PF_G(cls_lo),
@@ -2657,32 +2650,19 @@ static int sge_qinfo_uld_ciq_entries(const struct adapter *adap, int uld)
 
 static int sge_qinfo_show(struct seq_file *seq, void *v)
 {
-	int eth_entries, ctrl_entries, eo_entries = 0;
+	int eth_entries, ctrl_entries, eohw_entries = 0, eosw_entries = 0;
 	int uld_rxq_entries[CXGB4_ULD_MAX] = { 0 };
 	int uld_ciq_entries[CXGB4_ULD_MAX] = { 0 };
 	int uld_txq_entries[CXGB4_TX_MAX] = { 0 };
 	const struct sge_uld_txq_info *utxq_info;
 	const struct sge_uld_rxq_info *urxq_info;
+	struct cxgb4_tc_port_mqprio *port_mqprio;
 	struct adapter *adap = seq->private;
-	int i, n, r = (uintptr_t)v - 1;
+	int i, j, n, r = (uintptr_t)v - 1;
 	struct sge *s = &adap->sge;
 
 	eth_entries = DIV_ROUND_UP(adap->sge.ethqsets, 4);
 	ctrl_entries = DIV_ROUND_UP(MAX_CTRL_QUEUES, 4);
-	if (adap->sge.eohw_txq)
-		eo_entries = DIV_ROUND_UP(adap->sge.eoqsets, 4);
-
-	mutex_lock(&uld_mutex);
-	if (s->uld_txq_info)
-		for (i = 0; i < ARRAY_SIZE(uld_txq_entries); i++)
-			uld_txq_entries[i] = sge_qinfo_uld_txq_entries(adap, i);
-
-	if (s->uld_rxq_info) {
-		for (i = 0; i < ARRAY_SIZE(uld_rxq_entries); i++) {
-			uld_rxq_entries[i] = sge_qinfo_uld_rxq_entries(adap, i);
-			uld_ciq_entries[i] = sge_qinfo_uld_ciq_entries(adap, i);
-		}
-	}
 
 	if (r)
 		seq_putc(seq, '\n');
@@ -2759,11 +2739,21 @@ do { \
 		RL("FLLow:", fl.low);
 		RL("FLStarving:", fl.starving);
 
-		goto unlock;
+		goto out;
 	}
 
 	r -= eth_entries;
-	if (r < eo_entries) {
+	if (!adap->tc_mqprio)
+		goto skip_mqprio;
+
+	mutex_lock(&adap->tc_mqprio->mqprio_mutex);
+	if (!refcount_read(&adap->tc_mqprio->refcnt)) {
+		mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
+		goto skip_mqprio;
+	}
+
+	eohw_entries = DIV_ROUND_UP(adap->sge.eoqsets, 4);
+	if (r < eohw_entries) {
 		int base_qset = r * 4;
 		const struct sge_ofld_rxq *rx = &s->eohw_rxq[base_qset];
 		const struct sge_eohw_txq *tx = &s->eohw_txq[base_qset];
@@ -2808,10 +2798,71 @@ do { \
 		RL("FLLow:", fl.low);
 		RL("FLStarving:", fl.starving);
 
-		goto unlock;
+		mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
+		goto out;
+	}
+
+	r -= eohw_entries;
+	for (j = 0; j < adap->params.nports; j++) {
+		int entries;
+		u8 tc;
+
+		port_mqprio = &adap->tc_mqprio->port_mqprio[j];
+		entries = 0;
+		for (tc = 0; tc < port_mqprio->mqprio.qopt.num_tc; tc++)
+			entries += port_mqprio->mqprio.qopt.count[tc];
+
+		if (!entries)
+			continue;
+
+		eosw_entries = DIV_ROUND_UP(entries, 4);
+		if (r < eosw_entries) {
+			const struct sge_eosw_txq *tx;
+
+			n = min(4, entries - 4 * r);
+			tx = &port_mqprio->eosw_txq[4 * r];
+
+			S("QType:", "EOSW-TXQ");
+			S("Interface:",
+			  adap->port[j] ? adap->port[j]->name : "N/A");
+			T("EOTID:", hwtid);
+			T("HWQID:", hwqid);
+			T("State:", state);
+			T("Size:", ndesc);
+			T("In-Use:", inuse);
+			T("Credits:", cred);
+			T("Compl:", ncompl);
+			T("Last-Compl:", last_compl);
+			T("PIDX:", pidx);
+			T("Last-PIDX:", last_pidx);
+			T("CIDX:", cidx);
+			T("Last-CIDX:", last_cidx);
+			T("FLOWC-IDX:", flowc_idx);
+
+			mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
+			goto out;
+		}
+
+		r -= eosw_entries;
+	}
+	mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
+
+skip_mqprio:
+	if (!is_uld(adap))
+		goto skip_uld;
+
+	mutex_lock(&uld_mutex);
+	if (s->uld_txq_info)
+		for (i = 0; i < ARRAY_SIZE(uld_txq_entries); i++)
+			uld_txq_entries[i] = sge_qinfo_uld_txq_entries(adap, i);
+
+	if (s->uld_rxq_info) {
+		for (i = 0; i < ARRAY_SIZE(uld_rxq_entries); i++) {
+			uld_rxq_entries[i] = sge_qinfo_uld_rxq_entries(adap, i);
+			uld_ciq_entries[i] = sge_qinfo_uld_ciq_entries(adap, i);
+		}
 	}
 
-	r -= eo_entries;
 	if (r < uld_txq_entries[CXGB4_TX_OFLD]) {
 		const struct sge_uld_txq *tx;
 
@@ -2994,6 +3045,9 @@ do { \
 	}
 
 	r -= uld_txq_entries[CXGB4_TX_CRYPTO];
+	mutex_unlock(&uld_mutex);
+
+skip_uld:
 	if (r < ctrl_entries) {
 		const struct sge_ctrl_txq *tx = &s->ctrlq[r * 4];
 
@@ -3008,7 +3062,7 @@ do { \
 		TL("TxQFull:", q.stops);
 		TL("TxQRestarts:", q.restarts);
 
-		goto unlock;
+		goto out;
 	}
 
 	r -= ctrl_entries;
@@ -3026,11 +3080,9 @@ do { \
 		seq_printf(seq, "%-12s %16u\n", "Intr pktcnt:",
 			   s->counter_val[evtq->pktcnt_idx]);
 
-		goto unlock;
+		goto out;
 	}
 
-unlock:
-	mutex_unlock(&uld_mutex);
 #undef R
 #undef RL
 #undef T
@@ -3039,13 +3091,38 @@ unlock:
 #undef R3
 #undef T3
 #undef S3
+out:
+	return 0;
+
+unlock:
+	mutex_unlock(&uld_mutex);
 	return 0;
 }
 
 static int sge_queue_entries(const struct adapter *adap)
 {
-	int tot_uld_entries = 0;
-	int i;
+	int i, tot_uld_entries = 0, eohw_entries = 0, eosw_entries = 0;
+
+	if (adap->tc_mqprio) {
+		struct cxgb4_tc_port_mqprio *port_mqprio;
+		u8 tc;
+
+		mutex_lock(&adap->tc_mqprio->mqprio_mutex);
+		if (adap->sge.eohw_txq)
+			eohw_entries = DIV_ROUND_UP(adap->sge.eoqsets, 4);
+
+		for (i = 0; i < adap->params.nports; i++) {
+			u32 entries = 0;
+
+			port_mqprio = &adap->tc_mqprio->port_mqprio[i];
+			for (tc = 0; tc < port_mqprio->mqprio.qopt.num_tc; tc++)
+				entries += port_mqprio->mqprio.qopt.count[tc];
+
+			if (entries)
+				eosw_entries += DIV_ROUND_UP(entries, 4);
+		}
+		mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
+	}
 
 	if (!is_uld(adap))
 		goto lld_only;
@@ -3062,8 +3139,7 @@ static int sge_queue_entries(const struct adapter *adap)
 
 lld_only:
 	return DIV_ROUND_UP(adap->sge.ethqsets, 4) +
-	       (adap->sge.eohw_txq ? DIV_ROUND_UP(adap->sge.eoqsets, 4) : 0) +
-	       tot_uld_entries +
+	       eohw_entries + eosw_entries + tot_uld_entries +
 	       DIV_ROUND_UP(MAX_CTRL_QUEUES, 4) + 1;
 }
 
@@ -3244,6 +3320,10 @@ static int tid_info_show(struct seq_file *seq, void *v)
 	if (t->nhpftids)
 		seq_printf(seq, "HPFTID range: %u..%u\n", t->hpftid_base,
 			   t->hpftid_base + t->nhpftids - 1);
+	if (t->neotids)
+		seq_printf(seq, "EOTID range: %u..%u, in use: %u\n",
+			   t->eotid_base, t->eotid_base + t->neotids - 1,
+			   atomic_read(&t->eotids_in_use));
 	if (t->ntids)
 		seq_printf(seq, "HW TID usage: %u IP users, %u IPv6 users\n",
 			   t4_read_reg(adap, LE_DB_ACT_CNT_IPV4_A),
@@ -3411,6 +3491,8 @@ static int chcr_stats_show(struct seq_file *seq, void *v)
 		   atomic_read(&adap->chcr_stats.tls_key));
 #ifdef CONFIG_CHELSIO_TLS_DEVICE
 	seq_puts(seq, "\nChelsio KTLS Crypto Accelerator Stats\n");
+	seq_printf(seq, "Tx TLS offload refcount:          %20u\n",
+		   refcount_read(&adap->chcr_ktls.ktls_refcount));
 	seq_printf(seq, "Tx HW offload contexts added:     %20llu\n",
 		   atomic64_read(&adap->chcr_stats.ktls_tx_ctx));
 	seq_printf(seq, "Tx connection created:            %20llu\n",
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index a70018f067aa..854b1717a70d 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -66,6 +66,9 @@
 #include <linux/crash_dump.h>
 #include <net/udp_tunnel.h>
 #include <net/xfrm.h>
+#if defined(CONFIG_CHELSIO_TLS_DEVICE)
+#include <net/tls.h>
+#endif
 
 #include "cxgb4.h"
 #include "cxgb4_filter.h"
@@ -180,6 +183,7 @@ static struct dentry *cxgb4_debugfs_root;
 
 LIST_HEAD(adapter_list);
 DEFINE_MUTEX(uld_mutex);
+LIST_HEAD(uld_list);
 
 static int cfg_queues(struct adapter *adap);
 
@@ -1579,6 +1583,7 @@ static int tid_init(struct tid_info *t)
 	atomic_set(&t->tids_in_use, 0);
 	atomic_set(&t->conns_in_use, 0);
 	atomic_set(&t->hash_tids_in_use, 0);
+	atomic_set(&t->eotids_in_use, 0);
 
 	/* Setup the free list for atid_tab and clear the stid bitmap. */
 	if (natids) {
@@ -3021,7 +3026,7 @@ static int cxgb4_mgmt_set_vf_rate(struct net_device *dev, int vf,
 			      SCHED_CLASS_RATEUNIT_BITS,
 			      SCHED_CLASS_RATEMODE_ABS,
 			      pi->tx_chan, class_id, 0,
-			      max_tx_rate * 1000, 0, pktsize);
+			      max_tx_rate * 1000, 0, pktsize, 0);
 	if (ret) {
 		dev_err(adap->pdev_dev, "Err %d for Traffic Class config\n",
 			ret);
@@ -6062,6 +6067,79 @@ static int cxgb4_iov_configure(struct pci_dev *pdev, int num_vfs)
 }
 #endif /* CONFIG_PCI_IOV */
 
+#if defined(CONFIG_CHELSIO_TLS_DEVICE)
+
+static int cxgb4_ktls_dev_add(struct net_device *netdev, struct sock *sk,
+			      enum tls_offload_ctx_dir direction,
+			      struct tls_crypto_info *crypto_info,
+			      u32 tcp_sn)
+{
+	struct adapter *adap = netdev2adap(netdev);
+	int ret = 0;
+
+	mutex_lock(&uld_mutex);
+	if (!adap->uld[CXGB4_ULD_CRYPTO].handle) {
+		dev_err(adap->pdev_dev, "chcr driver is not loaded\n");
+		ret = -EOPNOTSUPP;
+		goto out_unlock;
+	}
+
+	if (!adap->uld[CXGB4_ULD_CRYPTO].tlsdev_ops) {
+		dev_err(adap->pdev_dev,
+			"chcr driver has no registered tlsdev_ops()\n");
+		ret = -EOPNOTSUPP;
+		goto out_unlock;
+	}
+
+	ret = cxgb4_set_ktls_feature(adap, FW_PARAMS_PARAM_DEV_KTLS_HW_ENABLE);
+	if (ret)
+		goto out_unlock;
+
+	ret = adap->uld[CXGB4_ULD_CRYPTO].tlsdev_ops->tls_dev_add(netdev, sk,
+								  direction,
+								  crypto_info,
+								  tcp_sn);
+	/* if there is a failure, clear the refcount */
+	if (ret)
+		cxgb4_set_ktls_feature(adap,
+				       FW_PARAMS_PARAM_DEV_KTLS_HW_DISABLE);
+out_unlock:
+	mutex_unlock(&uld_mutex);
+	return ret;
+}
+
+static void cxgb4_ktls_dev_del(struct net_device *netdev,
+			       struct tls_context *tls_ctx,
+			       enum tls_offload_ctx_dir direction)
+{
+	struct adapter *adap = netdev2adap(netdev);
+
+	mutex_lock(&uld_mutex);
+	if (!adap->uld[CXGB4_ULD_CRYPTO].handle) {
+		dev_err(adap->pdev_dev, "chcr driver is not loaded\n");
+		goto out_unlock;
+	}
+
+	if (!adap->uld[CXGB4_ULD_CRYPTO].tlsdev_ops) {
+		dev_err(adap->pdev_dev,
+			"chcr driver has no registered tlsdev_ops\n");
+		goto out_unlock;
+	}
+
+	adap->uld[CXGB4_ULD_CRYPTO].tlsdev_ops->tls_dev_del(netdev, tls_ctx,
+							    direction);
+	cxgb4_set_ktls_feature(adap, FW_PARAMS_PARAM_DEV_KTLS_HW_DISABLE);
+
+out_unlock:
+	mutex_unlock(&uld_mutex);
+}
+
+static const struct tlsdev_ops cxgb4_ktls_ops = {
+	.tls_dev_add = cxgb4_ktls_dev_add,
+	.tls_dev_del = cxgb4_ktls_dev_del,
+};
+#endif /* CONFIG_CHELSIO_TLS_DEVICE */
+
 static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	struct net_device *netdev;
@@ -6311,7 +6389,14 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 			netdev->hw_features |= NETIF_F_HIGHDMA;
 		netdev->features |= netdev->hw_features;
 		netdev->vlan_features = netdev->features & VLAN_FEAT;
-
+#if defined(CONFIG_CHELSIO_TLS_DEVICE)
+		if (pi->adapter->params.crypto & FW_CAPS_CONFIG_TLS_HW) {
+			netdev->hw_features |= NETIF_F_HW_TLS_TX;
+			netdev->tlsdev_ops = &cxgb4_ktls_ops;
+			/* initialize the refcount */
+			refcount_set(&pi->adapter->chcr_ktls.ktls_refcount, 0);
+		}
+#endif
 		netdev->priv_flags |= IFF_UNICAST_FLT;
 
 		/* MTU range: 81 - 9600 */
@@ -6518,11 +6603,8 @@ fw_attach_fail:
 	/* PCIe EEH recovery on powerpc platforms needs fundamental reset */
 	pdev->needs_freset = 1;
 
-	if (is_uld(adapter)) {
-		mutex_lock(&uld_mutex);
-		list_add_tail(&adapter->list_node, &adapter_list);
-		mutex_unlock(&uld_mutex);
-	}
+	if (is_uld(adapter))
+		cxgb4_uld_enable(adapter);
 
 	if (!is_t4(adapter->params.chip))
 		cxgb4_ptp_init(adapter);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c
index e6af4906d674..ae7123a9de8e 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.c
@@ -342,6 +342,13 @@ static int cxgb4_mqprio_alloc_tc(struct net_device *dev,
 		p.u.params.minrate = div_u64(mqprio->min_rate[i] * 8, 1000);
 		p.u.params.maxrate = div_u64(mqprio->max_rate[i] * 8, 1000);
 
+		/* Request larger burst buffer for smaller MTU, so
+		 * that hardware can work on more data per burst
+		 * cycle.
+		 */
+		if (dev->mtu <= ETH_DATA_LEN)
+			p.u.params.burstsize = 8 * dev->mtu;
+
 		e = cxgb4_sched_class_alloc(dev, &p);
 		if (!e) {
 			ret = -ENOMEM;
@@ -567,6 +574,7 @@ static void cxgb4_mqprio_disable_offload(struct net_device *dev)
 int cxgb4_setup_tc_mqprio(struct net_device *dev,
 			  struct tc_mqprio_qopt_offload *mqprio)
 {
+	struct adapter *adap = netdev2adap(dev);
 	bool needs_bring_up = false;
 	int ret;
 
@@ -574,6 +582,8 @@ int cxgb4_setup_tc_mqprio(struct net_device *dev,
 	if (ret)
 		return ret;
 
+	mutex_lock(&adap->tc_mqprio->mqprio_mutex);
+
 	/* To configure tc params, the current allocated EOTIDs must
 	 * be freed up. However, they can't be freed up if there's
 	 * traffic running on the interface. So, ensure interface is
@@ -609,6 +619,7 @@ out:
 	if (needs_bring_up)
 		cxgb_open(dev);
 
+	mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
 	return ret;
 }
 
@@ -621,6 +632,7 @@ void cxgb4_mqprio_stop_offload(struct adapter *adap)
 	if (!adap->tc_mqprio || !adap->tc_mqprio->port_mqprio)
 		return;
 
+	mutex_lock(&adap->tc_mqprio->mqprio_mutex);
 	for_each_port(adap, i) {
 		dev = adap->port[i];
 		if (!dev)
@@ -632,6 +644,7 @@ void cxgb4_mqprio_stop_offload(struct adapter *adap)
 
 		cxgb4_mqprio_disable_offload(dev);
 	}
+	mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
 }
 
 int cxgb4_init_tc_mqprio(struct adapter *adap)
@@ -653,6 +666,8 @@ int cxgb4_init_tc_mqprio(struct adapter *adap)
 		goto out_free_mqprio;
 	}
 
+	mutex_init(&tc_mqprio->mqprio_mutex);
+
 	tc_mqprio->port_mqprio = tc_port_mqprio;
 	for (i = 0; i < adap->params.nports; i++) {
 		port_mqprio = &tc_mqprio->port_mqprio[i];
@@ -687,6 +702,7 @@ void cxgb4_cleanup_tc_mqprio(struct adapter *adap)
 	u8 i;
 
 	if (adap->tc_mqprio) {
+		mutex_lock(&adap->tc_mqprio->mqprio_mutex);
 		if (adap->tc_mqprio->port_mqprio) {
 			for (i = 0; i < adap->params.nports; i++) {
 				struct net_device *dev = adap->port[i];
@@ -698,6 +714,7 @@ void cxgb4_cleanup_tc_mqprio(struct adapter *adap)
 			}
 			kfree(adap->tc_mqprio->port_mqprio);
 		}
+		mutex_unlock(&adap->tc_mqprio->mqprio_mutex);
 		kfree(adap->tc_mqprio);
 	}
 }
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.h
index ff8794132b22..be96f1dc0372 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_mqprio.h
@@ -33,6 +33,7 @@ struct cxgb4_tc_port_mqprio {
 
 struct cxgb4_tc_mqprio {
 	refcount_t refcnt; /* Refcount for adapter-wide resources */
+	struct mutex mqprio_mutex; /* Lock for accessing MQPRIO info */
 	struct cxgb4_tc_port_mqprio *port_mqprio; /* Per port MQPRIO info */
 };
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.c
index e65b52375dd8..0307e9c69a47 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.c
@@ -174,13 +174,14 @@ static int
 setup_sge_queues_uld(struct adapter *adap, unsigned int uld_type, bool lro)
 {
 	struct sge_uld_rxq_info *rxq_info = adap->sge.uld_rxq_info[uld_type];
-	int i, ret = 0;
+	int i, ret;
 
-	ret = !(!alloc_uld_rxqs(adap, rxq_info, lro));
+	ret = alloc_uld_rxqs(adap, rxq_info, lro);
+	if (ret)
+		return ret;
 
 	/* Tell uP to route control queue completions to rdma rspq */
-	if (adap->flags & CXGB4_FULL_INIT_DONE &&
-	    !ret && uld_type == CXGB4_ULD_RDMA) {
+	if (adap->flags & CXGB4_FULL_INIT_DONE && uld_type == CXGB4_ULD_RDMA) {
 		struct sge *s = &adap->sge;
 		unsigned int cmplqid;
 		u32 param, cmdop;
@@ -662,25 +663,129 @@ static int uld_attach(struct adapter *adap, unsigned int uld)
 	return 0;
 }
 
+static bool cxgb4_uld_in_use(struct adapter *adap)
+{
+	const struct tid_info *t = &adap->tids;
+
+	return (atomic_read(&t->conns_in_use) || t->stids_in_use);
+}
+
 #ifdef CONFIG_CHELSIO_TLS_DEVICE
 /* cxgb4_set_ktls_feature: request FW to enable/disable ktls settings.
  * @adap: adapter info
  * @enable: 1 to enable / 0 to disable ktls settings.
  */
-static void cxgb4_set_ktls_feature(struct adapter *adap, bool enable)
+int cxgb4_set_ktls_feature(struct adapter *adap, bool enable)
 {
-	u32 params = (FW_PARAMS_MNEM_V(FW_PARAMS_MNEM_DEV) |
-		      FW_PARAMS_PARAM_X_V(FW_PARAMS_PARAM_DEV_KTLS_TX_HW) |
-		      FW_PARAMS_PARAM_Y_V(enable));
 	int ret = 0;
+	u32 params =
+		FW_PARAMS_MNEM_V(FW_PARAMS_MNEM_DEV) |
+		FW_PARAMS_PARAM_X_V(FW_PARAMS_PARAM_DEV_KTLS_HW) |
+		FW_PARAMS_PARAM_Y_V(enable) |
+		FW_PARAMS_PARAM_Z_V(FW_PARAMS_PARAM_DEV_KTLS_HW_USER_ENABLE);
+
+	if (enable) {
+		if (!refcount_read(&adap->chcr_ktls.ktls_refcount)) {
+			/* At this moment if ULD connection are up means, other
+			 * ULD is/are already active, return failure.
+			 */
+			if (cxgb4_uld_in_use(adap)) {
+				dev_warn(adap->pdev_dev,
+					 "ULD connections (tid/stid) active. Can't enable kTLS\n");
+				return -EINVAL;
+			}
+			ret = t4_set_params(adap, adap->mbox, adap->pf,
+					    0, 1, &params, &params);
+			if (ret)
+				return ret;
+			refcount_set(&adap->chcr_ktls.ktls_refcount, 1);
+			pr_info("kTLS has been enabled. Restrictions placed on ULD support\n");
+		} else {
+			/* ktls settings already up, just increment refcount. */
+			refcount_inc(&adap->chcr_ktls.ktls_refcount);
+		}
+	} else {
+		/* return failure if refcount is already 0. */
+		if (!refcount_read(&adap->chcr_ktls.ktls_refcount))
+			return -EINVAL;
+		/* decrement refcount and test, if 0, disable ktls feature,
+		 * else return command success.
+		 */
+		if (refcount_dec_and_test(&adap->chcr_ktls.ktls_refcount)) {
+			ret = t4_set_params(adap, adap->mbox, adap->pf,
+					    0, 1, &params, &params);
+			if (ret)
+				return ret;
+			pr_info("kTLS is disabled. Restrictions on ULD support removed\n");
+		}
+	}
 
-	ret = t4_set_params(adap, adap->mbox, adap->pf, 0, 1, &params, &params);
-	/* if fw returns failure, clear the ktls flag */
-	if (ret)
-		adap->params.crypto &= ~ULP_CRYPTO_KTLS_INLINE;
+	return ret;
 }
 #endif
 
+static void cxgb4_uld_alloc_resources(struct adapter *adap,
+				      enum cxgb4_uld type,
+				      const struct cxgb4_uld_info *p)
+{
+	int ret = 0;
+
+	if ((type == CXGB4_ULD_CRYPTO && !is_pci_uld(adap)) ||
+	    (type != CXGB4_ULD_CRYPTO && !is_offload(adap)))
+		return;
+	if (type == CXGB4_ULD_ISCSIT && is_t4(adap->params.chip))
+		return;
+	ret = cfg_queues_uld(adap, type, p);
+	if (ret)
+		goto out;
+	ret = setup_sge_queues_uld(adap, type, p->lro);
+	if (ret)
+		goto free_queues;
+	if (adap->flags & CXGB4_USING_MSIX) {
+		ret = request_msix_queue_irqs_uld(adap, type);
+		if (ret)
+			goto free_rxq;
+	}
+	if (adap->flags & CXGB4_FULL_INIT_DONE)
+		enable_rx_uld(adap, type);
+	if (adap->uld[type].add)
+		goto free_irq;
+	ret = setup_sge_txq_uld(adap, type, p);
+	if (ret)
+		goto free_irq;
+	adap->uld[type] = *p;
+	ret = uld_attach(adap, type);
+	if (ret)
+		goto free_txq;
+	return;
+free_txq:
+	release_sge_txq_uld(adap, type);
+free_irq:
+	if (adap->flags & CXGB4_FULL_INIT_DONE)
+		quiesce_rx_uld(adap, type);
+	if (adap->flags & CXGB4_USING_MSIX)
+		free_msix_queue_irqs_uld(adap, type);
+free_rxq:
+	free_sge_queues_uld(adap, type);
+free_queues:
+	free_queues_uld(adap, type);
+out:
+	dev_warn(adap->pdev_dev,
+		 "ULD registration failed for uld type %d\n", type);
+}
+
+void cxgb4_uld_enable(struct adapter *adap)
+{
+	struct cxgb4_uld_list *uld_entry;
+
+	mutex_lock(&uld_mutex);
+	list_add_tail(&adap->list_node, &adapter_list);
+	list_for_each_entry(uld_entry, &uld_list, list_node)
+		cxgb4_uld_alloc_resources(adap, uld_entry->uld_type,
+					  &uld_entry->uld_info);
+	mutex_unlock(&uld_mutex);
+}
+
 /* cxgb4_register_uld - register an upper-layer driver
  * @type: the ULD type
  * @p: the ULD methods
@@ -691,63 +796,23 @@ static void cxgb4_set_ktls_feature(struct adapter *adap, bool enable)
 void cxgb4_register_uld(enum cxgb4_uld type,
 			const struct cxgb4_uld_info *p)
 {
+	struct cxgb4_uld_list *uld_entry;
 	struct adapter *adap;
-	int ret = 0;
 
 	if (type >= CXGB4_ULD_MAX)
 		return;
 
+	uld_entry = kzalloc(sizeof(*uld_entry), GFP_KERNEL);
+	if (!uld_entry)
+		return;
+
+	memcpy(&uld_entry->uld_info, p, sizeof(struct cxgb4_uld_info));
 	mutex_lock(&uld_mutex);
-	list_for_each_entry(adap, &adapter_list, list_node) {
-		if ((type == CXGB4_ULD_CRYPTO && !is_pci_uld(adap)) ||
-		    (type != CXGB4_ULD_CRYPTO && !is_offload(adap)))
-			continue;
-		if (type == CXGB4_ULD_ISCSIT && is_t4(adap->params.chip))
-			continue;
-		ret = cfg_queues_uld(adap, type, p);
-		if (ret)
-			goto out;
-		ret = setup_sge_queues_uld(adap, type, p->lro);
-		if (ret)
-			goto free_queues;
-		if (adap->flags & CXGB4_USING_MSIX) {
-			ret = request_msix_queue_irqs_uld(adap, type);
-			if (ret)
-				goto free_rxq;
-		}
-		if (adap->flags & CXGB4_FULL_INIT_DONE)
-			enable_rx_uld(adap, type);
-#ifdef CONFIG_CHELSIO_TLS_DEVICE
-		/* send mbox to enable ktls related settings. */
-		if (type == CXGB4_ULD_CRYPTO &&
-		    (adap->params.crypto & FW_CAPS_CONFIG_TX_TLS_HW))
-			cxgb4_set_ktls_feature(adap, 1);
-#endif
-		if (adap->uld[type].add)
-			goto free_irq;
-		ret = setup_sge_txq_uld(adap, type, p);
-		if (ret)
-			goto free_irq;
-		adap->uld[type] = *p;
-		ret = uld_attach(adap, type);
-		if (ret)
-			goto free_txq;
-		continue;
-free_txq:
-		release_sge_txq_uld(adap, type);
-free_irq:
-		if (adap->flags & CXGB4_FULL_INIT_DONE)
-			quiesce_rx_uld(adap, type);
-		if (adap->flags & CXGB4_USING_MSIX)
-			free_msix_queue_irqs_uld(adap, type);
-free_rxq:
-		free_sge_queues_uld(adap, type);
-free_queues:
-		free_queues_uld(adap, type);
-out:
-		dev_warn(adap->pdev_dev,
-			 "ULD registration failed for uld type %d\n", type);
-	}
+	list_for_each_entry(adap, &adapter_list, list_node)
+		cxgb4_uld_alloc_resources(adap, type, p);
+
+	uld_entry->uld_type = type;
+	list_add_tail(&uld_entry->list_node, &uld_list);
 	mutex_unlock(&uld_mutex);
 	return;
 }
@@ -761,6 +826,7 @@ EXPORT_SYMBOL(cxgb4_register_uld);
  */
 int cxgb4_unregister_uld(enum cxgb4_uld type)
 {
+	struct cxgb4_uld_list *uld_entry, *tmp;
 	struct adapter *adap;
 
 	if (type >= CXGB4_ULD_MAX)
@@ -775,13 +841,13 @@ int cxgb4_unregister_uld(enum cxgb4_uld type)
 			continue;
 
 		cxgb4_shutdown_uld_adapter(adap, type);
+	}
 
-#ifdef CONFIG_CHELSIO_TLS_DEVICE
-		/* send mbox to disable ktls related settings. */
-		if (type == CXGB4_ULD_CRYPTO &&
-		    (adap->params.crypto & FW_CAPS_CONFIG_TX_TLS_HW))
-			cxgb4_set_ktls_feature(adap, 0);
-#endif
+	list_for_each_entry_safe(uld_entry, tmp, &uld_list, list_node) {
+		if (uld_entry->uld_type == type) {
+			list_del(&uld_entry->list_node);
+			kfree(uld_entry);
+		}
 	}
 	mutex_unlock(&uld_mutex);
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
index be831317520a..dbce99b209d6 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h
@@ -147,6 +147,9 @@ struct tid_info {
 	/* TIDs in the HASH */
 	atomic_t hash_tids_in_use;
 	atomic_t conns_in_use;
+	/* ETHOFLD TIDs used for rate limiting */
+	atomic_t eotids_in_use;
+
 	/* lock for setting/clearing filter bitmap */
 	spinlock_t ftid_lock;
 
@@ -221,12 +224,14 @@ static inline void cxgb4_alloc_eotid(struct tid_info *t, u32 eotid, void *data)
 {
 	set_bit(eotid, t->eotid_bmap);
 	t->eotid_tab[eotid].data = data;
+	atomic_inc(&t->eotids_in_use);
 }
 
 static inline void cxgb4_free_eotid(struct tid_info *t, u32 eotid)
 {
 	clear_bit(eotid, t->eotid_bmap);
 	t->eotid_tab[eotid].data = NULL;
+	atomic_dec(&t->eotids_in_use);
 }
 
 int cxgb4_alloc_atid(struct tid_info *t, void *data);
@@ -263,6 +268,10 @@ struct filter_ctx {
 	u32 tid;			/* to store tid */
 };
 
+struct chcr_ktls {
+	refcount_t ktls_refcount;
+};
+
 struct ch_filter_specification;
 
 int cxgb4_get_free_ftid(struct net_device *dev, u8 family, bool hash_en,
@@ -322,6 +331,7 @@ enum cxgb4_control {
 	CXGB4_CONTROL_DB_DROP,
 };
 
+struct adapter;
 struct pci_dev;
 struct l2t_data;
 struct net_device;
@@ -458,8 +468,12 @@ struct cxgb4_uld_info {
 			      struct napi_struct *napi);
 	void (*lro_flush)(struct t4_lro_mgr *);
 	int (*tx_handler)(struct sk_buff *skb, struct net_device *dev);
+#if IS_ENABLED(CONFIG_TLS_DEVICE)
+	const struct tlsdev_ops *tlsdev_ops;
+#endif
 };
 
+void cxgb4_uld_enable(struct adapter *adap);
 void cxgb4_register_uld(enum cxgb4_uld type, const struct cxgb4_uld_info *p);
 int cxgb4_unregister_uld(enum cxgb4_uld type);
 int cxgb4_ofld_send(struct net_device *dev, struct sk_buff *skb);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sched.c b/drivers/net/ethernet/chelsio/cxgb4/sched.c
index cebe1412d960..fde93c50cfec 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sched.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sched.c
@@ -57,7 +57,8 @@ static int t4_sched_class_fw_cmd(struct port_info *pi,
 				      p->u.params.ratemode,
 				      p->u.params.channel, e->idx,
 				      p->u.params.minrate, p->u.params.maxrate,
-				      p->u.params.weight, p->u.params.pktsize);
+				      p->u.params.weight, p->u.params.pktsize,
+				      p->u.params.burstsize);
 		break;
 	default:
 		err = -ENOTSUPP;
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c
index 6516c45864b3..1359158652b7 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c
@@ -2091,10 +2091,9 @@ static inline u8 ethofld_calc_tx_flits(struct adapter *adap,
 	return flits + nsgl;
 }
 
-static inline void *write_eo_wr(struct adapter *adap,
-				struct sge_eosw_txq *eosw_txq,
-				struct sk_buff *skb, struct fw_eth_tx_eo_wr *wr,
-				u32 hdr_len, u32 wrlen)
+static void *write_eo_wr(struct adapter *adap, struct sge_eosw_txq *eosw_txq,
+			 struct sk_buff *skb, struct fw_eth_tx_eo_wr *wr,
+			 u32 hdr_len, u32 wrlen)
 {
 	const struct skb_shared_info *ssi = skb_shinfo(skb);
 	struct cpl_tx_pkt_core *cpl;
@@ -2113,7 +2112,8 @@ static inline void *write_eo_wr(struct adapter *adap,
 	immd_len += hdr_len;
 
 	if (!eosw_txq->ncompl ||
-	    eosw_txq->last_compl >= adap->params.ofldq_wr_cred / 2) {
+	    (eosw_txq->last_compl + wrlen16) >=
+	    (adap->params.ofldq_wr_cred / 2)) {
 		compl = true;
 		eosw_txq->ncompl++;
 		eosw_txq->last_compl = 0;
@@ -2153,8 +2153,8 @@ static inline void *write_eo_wr(struct adapter *adap,
 	return cpl;
 }
 
-static void ethofld_hard_xmit(struct net_device *dev,
-			      struct sge_eosw_txq *eosw_txq)
+static int ethofld_hard_xmit(struct net_device *dev,
+			     struct sge_eosw_txq *eosw_txq)
 {
 	struct port_info *pi = netdev2pinfo(dev);
 	struct adapter *adap = netdev2adap(dev);
@@ -2167,8 +2167,8 @@ static void ethofld_hard_xmit(struct net_device *dev,
 	bool skip_eotx_wr = false;
 	struct tx_sw_desc *d;
 	struct sk_buff *skb;
+	int left, ret = 0;
 	u8 flits, ndesc;
-	int left;
 
 	eohw_txq = &adap->sge.eohw_txq[eosw_txq->hwqid];
 	spin_lock(&eohw_txq->lock);
@@ -2198,11 +2198,19 @@ static void ethofld_hard_xmit(struct net_device *dev,
 	wrlen = flits * 8;
 	wrlen16 = DIV_ROUND_UP(wrlen, 16);
 
-	/* If there are no CPL credits, then wait for credits
-	 * to come back and retry again
+	left = txq_avail(&eohw_txq->q) - ndesc;
+
+	/* If there are no descriptors left in hardware queues or no
+	 * CPL credits left in software queues, then wait for them
+	 * to come back and retry again. Note that we always request
+	 * for credits update via interrupt for every half credits
+	 * consumed. So, the interrupt will eventually restore the
+	 * credits and invoke the Tx path again.
 	 */
-	if (unlikely(wrlen16 > eosw_txq->cred))
+	if (unlikely(left < 0 || wrlen16 > eosw_txq->cred)) {
+		ret = -ENOMEM;
 		goto out_unlock;
+	}
 
 	if (unlikely(skip_eotx_wr)) {
 		start = (u64 *)wr;
@@ -2231,7 +2239,8 @@ write_wr_headers:
 	sgl = (u64 *)inline_tx_skb_header(skb, &eohw_txq->q, (void *)start,
 					  hdr_len);
 	if (data_len) {
-		if (unlikely(cxgb4_map_skb(adap->pdev_dev, skb, d->addr))) {
+		ret = cxgb4_map_skb(adap->pdev_dev, skb, d->addr);
+		if (unlikely(ret)) {
 			memset(d->addr, 0, sizeof(d->addr));
 			eohw_txq->mapping_err++;
 			goto out_unlock;
@@ -2277,12 +2286,13 @@ write_wr_headers:
 
 out_unlock:
 	spin_unlock(&eohw_txq->lock);
+	return ret;
 }
 
 static void ethofld_xmit(struct net_device *dev, struct sge_eosw_txq *eosw_txq)
 {
 	struct sk_buff *skb;
-	int pktcount;
+	int pktcount, ret;
 
 	switch (eosw_txq->state) {
 	case CXGB4_EO_STATE_ACTIVE:
@@ -2307,7 +2317,9 @@ static void ethofld_xmit(struct net_device *dev, struct sge_eosw_txq *eosw_txq)
 			continue;
 		}
 
-		ethofld_hard_xmit(dev, eosw_txq);
+		ret = ethofld_hard_xmit(dev, eosw_txq);
+		if (ret)
+			break;
 	}
 }
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index 2a3480fc1d91..1c8068c02728 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -10361,9 +10361,10 @@ int t4_sge_ctxt_rd_bd(struct adapter *adap, unsigned int cid,
 	return ret;
 }
 
-int t4_sched_params(struct adapter *adapter, int type, int level, int mode,
-		    int rateunit, int ratemode, int channel, int class,
-		    int minrate, int maxrate, int weight, int pktsize)
+int t4_sched_params(struct adapter *adapter, u8 type, u8 level, u8 mode,
+		    u8 rateunit, u8 ratemode, u8 channel, u8 class,
+		    u32 minrate, u32 maxrate, u16 weight, u16 pktsize,
+		    u16 burstsize)
 {
 	struct fw_sched_cmd cmd;
 
@@ -10385,6 +10386,7 @@ int t4_sched_params(struct adapter *adapter, int type, int level, int mode,
 	cmd.u.params.max = cpu_to_be32(maxrate);
 	cmd.u.params.weight = cpu_to_be16(weight);
 	cmd.u.params.pktsize = cpu_to_be16(pktsize);
+	cmd.u.params.burstsize = cpu_to_be16(burstsize);
 
 	return t4_wr_mbox_meat(adapter, adapter->mbox, &cmd, sizeof(cmd),
 			       NULL, 1);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
index 68fe734b9b37..0a326c054707 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
@@ -1205,7 +1205,7 @@ enum fw_caps_config_crypto {
 	FW_CAPS_CONFIG_CRYPTO_LOOKASIDE = 0x00000001,
 	FW_CAPS_CONFIG_TLS_INLINE = 0x00000002,
 	FW_CAPS_CONFIG_IPSEC_INLINE = 0x00000004,
-	FW_CAPS_CONFIG_TX_TLS_HW = 0x00000008,
+	FW_CAPS_CONFIG_TLS_HW = 0x00000008,
 };
 
 enum fw_caps_config_fcoe {
@@ -1329,7 +1329,7 @@ enum fw_params_param_dev {
 	FW_PARAMS_PARAM_DEV_DBQ_TIMERTICK = 0x2A,
 	FW_PARAMS_PARAM_DEV_NUM_TM_CLASS = 0x2B,
 	FW_PARAMS_PARAM_DEV_FILTER = 0x2E,
-	FW_PARAMS_PARAM_DEV_KTLS_TX_HW = 0x31,
+	FW_PARAMS_PARAM_DEV_KTLS_HW = 0x31,
 };
 
 /*
@@ -1412,6 +1412,12 @@ enum fw_params_param_dmaq {
 	FW_PARAMS_PARAM_DMAQ_CONM_CTXT = 0x20,
 };
 
+enum fw_params_param_dev_ktls_hw {
+	FW_PARAMS_PARAM_DEV_KTLS_HW_DISABLE      = 0x00,
+	FW_PARAMS_PARAM_DEV_KTLS_HW_ENABLE       = 0x01,
+	FW_PARAMS_PARAM_DEV_KTLS_HW_USER_ENABLE  = 0x01,
+};
+
 enum fw_params_param_dev_phyfw {
 	FW_PARAMS_PARAM_DEV_PHYFW_DOWNLOAD = 0x00,
 	FW_PARAMS_PARAM_DEV_PHYFW_VERSION = 0x01,