Merge branch 'cxgb4-next'

Hariprasad Shenai says:

====================
Add support to dump cim ibq, obq and qinfo, etc

This patch series adds support to dump cim_ibq, cim_obq, sge_qinfo, pm_stats
and clk debugfs entries.

The patches series is created against 'net-next' tree.
And includes patches on cxgb4 driver.

We have included all the maintainers of respective drivers. Kindly review the
change and let us know in case of any review comments.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

6 files changed, 659 insertions, 2 deletions

diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 24fc162..d98a446 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -1029,6 +1029,8 @@ int cxgb4_t4_bar2_sge_qregs(struct adapter *adapter,
 		      u64 *pbar2_qoffset,
 		      unsigned int *pbar2_qid);
 
+unsigned int qtimer_val(const struct adapter *adap,
+			const struct sge_rspq *q);
 int t4_init_sge_params(struct adapter *adapter);
 int t4_init_tp_params(struct adapter *adap);
 int t4_filter_field_shift(const struct adapter *adap, int filter_sel);
@@ -1052,6 +1054,12 @@ int t4_mc_read(struct adapter *adap, int idx, u32 addr, __be32 *data,
 	       u64 *parity);
 int t4_edc_read(struct adapter *adap, int idx, u32 addr, __be32 *data,
 		u64 *parity);
+void t4_pmtx_get_stats(struct adapter *adap, u32 cnt[], u64 cycles[]);
+void t4_pmrx_get_stats(struct adapter *adap, u32 cnt[], u64 cycles[]);
+int t4_read_cim_ibq(struct adapter *adap, unsigned int qid, u32 *data,
+		    size_t n);
+int t4_read_cim_obq(struct adapter *adap, unsigned int qid, u32 *data,
+		    size_t n);
 int t4_cim_read(struct adapter *adap, unsigned int addr, unsigned int n,
 		unsigned int *valp);
 int t4_cim_write(struct adapter *adap, unsigned int addr, unsigned int n,
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
index 714cc70..47c0869 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
@@ -104,6 +104,17 @@ struct seq_tab *seq_open_tab(struct file *f, unsigned int rows,
 	return p;
 }
 
+/* Trim the size of a seq_tab to the supplied number of rows.  The operation is
+ * irreversible.
+ */
+static int seq_tab_trim(struct seq_tab *p, unsigned int new_rows)
+{
+	if (new_rows > p->rows)
+		return -EINVAL;
+	p->rows = new_rows;
+	return 0;
+}
+
 static int cim_la_show(struct seq_file *seq, void *v, int idx)
 {
 	if (v == SEQ_START_TOKEN)
@@ -239,6 +250,198 @@ static const struct file_operations cim_qcfg_fops = {
 	.release = single_release,
 };
 
+static int cimq_show(struct seq_file *seq, void *v, int idx)
+{
+	const u32 *p = v;
+
+	seq_printf(seq, "%#06x: %08x %08x %08x %08x\n", idx * 16, p[0], p[1],
+		   p[2], p[3]);
+	return 0;
+}
+
+static int cim_ibq_open(struct inode *inode, struct file *file)
+{
+	int ret;
+	struct seq_tab *p;
+	unsigned int qid = (uintptr_t)inode->i_private & 7;
+	struct adapter *adap = inode->i_private - qid;
+
+	p = seq_open_tab(file, CIM_IBQ_SIZE, 4 * sizeof(u32), 0, cimq_show);
+	if (!p)
+		return -ENOMEM;
+
+	ret = t4_read_cim_ibq(adap, qid, (u32 *)p->data, CIM_IBQ_SIZE * 4);
+	if (ret < 0)
+		seq_release_private(inode, file);
+	else
+		ret = 0;
+	return ret;
+}
+
+static const struct file_operations cim_ibq_fops = {
+	.owner   = THIS_MODULE,
+	.open    = cim_ibq_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release_private
+};
+
+static int cim_obq_open(struct inode *inode, struct file *file)
+{
+	int ret;
+	struct seq_tab *p;
+	unsigned int qid = (uintptr_t)inode->i_private & 7;
+	struct adapter *adap = inode->i_private - qid;
+
+	p = seq_open_tab(file, 6 * CIM_OBQ_SIZE, 4 * sizeof(u32), 0, cimq_show);
+	if (!p)
+		return -ENOMEM;
+
+	ret = t4_read_cim_obq(adap, qid, (u32 *)p->data, 6 * CIM_OBQ_SIZE * 4);
+	if (ret < 0) {
+		seq_release_private(inode, file);
+	} else {
+		seq_tab_trim(p, ret / 4);
+		ret = 0;
+	}
+	return ret;
+}
+
+static const struct file_operations cim_obq_fops = {
+	.owner   = THIS_MODULE,
+	.open    = cim_obq_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release_private
+};
+
+/* Show the PM memory stats.  These stats include:
+ *
+ * TX:
+ *   Read: memory read operation
+ *   Write Bypass: cut-through
+ *   Bypass + mem: cut-through and save copy
+ *
+ * RX:
+ *   Read: memory read
+ *   Write Bypass: cut-through
+ *   Flush: payload trim or drop
+ */
+static int pm_stats_show(struct seq_file *seq, void *v)
+{
+	static const char * const tx_pm_stats[] = {
+		"Read:", "Write bypass:", "Write mem:", "Bypass + mem:"
+	};
+	static const char * const rx_pm_stats[] = {
+		"Read:", "Write bypass:", "Write mem:", "Flush:"
+	};
+
+	int i;
+	u32 tx_cnt[PM_NSTATS], rx_cnt[PM_NSTATS];
+	u64 tx_cyc[PM_NSTATS], rx_cyc[PM_NSTATS];
+	struct adapter *adap = seq->private;
+
+	t4_pmtx_get_stats(adap, tx_cnt, tx_cyc);
+	t4_pmrx_get_stats(adap, rx_cnt, rx_cyc);
+
+	seq_printf(seq, "%13s %10s  %20s\n", " ", "Tx pcmds", "Tx bytes");
+	for (i = 0; i < PM_NSTATS - 1; i++)
+		seq_printf(seq, "%-13s %10u  %20llu\n",
+			   tx_pm_stats[i], tx_cnt[i], tx_cyc[i]);
+
+	seq_printf(seq, "%13s %10s  %20s\n", " ", "Rx pcmds", "Rx bytes");
+	for (i = 0; i < PM_NSTATS - 1; i++)
+		seq_printf(seq, "%-13s %10u  %20llu\n",
+			   rx_pm_stats[i], rx_cnt[i], rx_cyc[i]);
+	return 0;
+}
+
+static int pm_stats_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, pm_stats_show, inode->i_private);
+}
+
+static ssize_t pm_stats_clear(struct file *file, const char __user *buf,
+			      size_t count, loff_t *pos)
+{
+	struct adapter *adap = FILE_DATA(file)->i_private;
+
+	t4_write_reg(adap, PM_RX_STAT_CONFIG_A, 0);
+	t4_write_reg(adap, PM_TX_STAT_CONFIG_A, 0);
+	return count;
+}
+
+static const struct file_operations pm_stats_debugfs_fops = {
+	.owner   = THIS_MODULE,
+	.open    = pm_stats_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = single_release,
+	.write   = pm_stats_clear
+};
+
+/* Format a value in a unit that differs from the value's native unit by the
+ * given factor.
+ */
+static char *unit_conv(char *buf, size_t len, unsigned int val,
+		       unsigned int factor)
+{
+	unsigned int rem = val % factor;
+
+	if (rem == 0) {
+		snprintf(buf, len, "%u", val / factor);
+	} else {
+		while (rem % 10 == 0)
+			rem /= 10;
+		snprintf(buf, len, "%u.%u", val / factor, rem);
+	}
+	return buf;
+}
+
+static int clk_show(struct seq_file *seq, void *v)
+{
+	char buf[32];
+	struct adapter *adap = seq->private;
+	unsigned int cclk_ps = 1000000000 / adap->params.vpd.cclk;  /* in ps */
+	u32 res = t4_read_reg(adap, TP_TIMER_RESOLUTION_A);
+	unsigned int tre = TIMERRESOLUTION_G(res);
+	unsigned int dack_re = DELAYEDACKRESOLUTION_G(res);
+	unsigned long long tp_tick_us = (cclk_ps << tre) / 1000000; /* in us */
+
+	seq_printf(seq, "Core clock period: %s ns\n",
+		   unit_conv(buf, sizeof(buf), cclk_ps, 1000));
+	seq_printf(seq, "TP timer tick: %s us\n",
+		   unit_conv(buf, sizeof(buf), (cclk_ps << tre), 1000000));
+	seq_printf(seq, "TCP timestamp tick: %s us\n",
+		   unit_conv(buf, sizeof(buf),
+			     (cclk_ps << TIMESTAMPRESOLUTION_G(res)), 1000000));
+	seq_printf(seq, "DACK tick: %s us\n",
+		   unit_conv(buf, sizeof(buf), (cclk_ps << dack_re), 1000000));
+	seq_printf(seq, "DACK timer: %u us\n",
+		   ((cclk_ps << dack_re) / 1000000) *
+		   t4_read_reg(adap, TP_DACK_TIMER_A));
+	seq_printf(seq, "Retransmit min: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_RXT_MIN_A));
+	seq_printf(seq, "Retransmit max: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_RXT_MAX_A));
+	seq_printf(seq, "Persist timer min: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_PERS_MIN_A));
+	seq_printf(seq, "Persist timer max: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_PERS_MAX_A));
+	seq_printf(seq, "Keepalive idle timer: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_KEEP_IDLE_A));
+	seq_printf(seq, "Keepalive interval: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_KEEP_INTVL_A));
+	seq_printf(seq, "Initial SRTT: %llu us\n",
+		   tp_tick_us * INITSRTT_G(t4_read_reg(adap, TP_INIT_SRTT_A)));
+	seq_printf(seq, "FINWAIT2 timer: %llu us\n",
+		   tp_tick_us * t4_read_reg(adap, TP_FINWAIT2_TIMER_A));
+
+	return 0;
+}
+
+DEFINE_SIMPLE_DEBUGFS_FILE(clk);
+
 /* Firmware Device Log dump. */
 static const char * const devlog_level_strings[] = {
 	[FW_DEVLOG_LEVEL_EMERG]		= "EMERG",
@@ -1026,6 +1229,216 @@ static const struct file_operations rss_vf_config_debugfs_fops = {
 	.release = seq_release_private
 };
 
+static int sge_qinfo_show(struct seq_file *seq, void *v)
+{
+	struct adapter *adap = seq->private;
+	int eth_entries = DIV_ROUND_UP(adap->sge.ethqsets, 4);
+	int toe_entries = DIV_ROUND_UP(adap->sge.ofldqsets, 4);
+	int rdma_entries = DIV_ROUND_UP(adap->sge.rdmaqs, 4);
+	int ciq_entries = DIV_ROUND_UP(adap->sge.rdmaciqs, 4);
+	int ctrl_entries = DIV_ROUND_UP(MAX_CTRL_QUEUES, 4);
+	int i, r = (uintptr_t)v - 1;
+	int toe_idx = r - eth_entries;
+	int rdma_idx = toe_idx - toe_entries;
+	int ciq_idx = rdma_idx - rdma_entries;
+	int ctrl_idx =  ciq_idx - ciq_entries;
+	int fq_idx =  ctrl_idx - ctrl_entries;
+
+	if (r)
+		seq_putc(seq, '\n');
+
+#define S3(fmt_spec, s, v) \
+do { \
+	seq_printf(seq, "%-12s", s); \
+	for (i = 0; i < n; ++i) \
+		seq_printf(seq, " %16" fmt_spec, v); \
+		seq_putc(seq, '\n'); \
+} while (0)
+#define S(s, v) S3("s", s, v)
+#define T(s, v) S3("u", s, tx[i].v)
+#define R(s, v) S3("u", s, rx[i].v)
+
+	if (r < eth_entries) {
+		int base_qset = r * 4;
+		const struct sge_eth_rxq *rx = &adap->sge.ethrxq[base_qset];
+		const struct sge_eth_txq *tx = &adap->sge.ethtxq[base_qset];
+		int n = min(4, adap->sge.ethqsets - 4 * r);
+
+		S("QType:", "Ethernet");
+		S("Interface:",
+		  rx[i].rspq.netdev ? rx[i].rspq.netdev->name : "N/A");
+		T("TxQ ID:", q.cntxt_id);
+		T("TxQ size:", q.size);
+		T("TxQ inuse:", q.in_use);
+		T("TxQ CIDX:", q.cidx);
+		T("TxQ PIDX:", q.pidx);
+#ifdef CONFIG_CXGB4_DCB
+		T("DCB Prio:", dcb_prio);
+		S3("u", "DCB PGID:",
+		   (ethqset2pinfo(adap, base_qset + i)->dcb.pgid >>
+		    4*(7-tx[i].dcb_prio)) & 0xf);
+		S3("u", "DCB PFC:",
+		   (ethqset2pinfo(adap, base_qset + i)->dcb.pfcen >>
+		    1*(7-tx[i].dcb_prio)) & 0x1);
+#endif
+		R("RspQ ID:", rspq.abs_id);
+		R("RspQ size:", rspq.size);
+		R("RspQE size:", rspq.iqe_len);
+		R("RspQ CIDX:", rspq.cidx);
+		R("RspQ Gen:", rspq.gen);
+		S3("u", "Intr delay:", qtimer_val(adap, &rx[i].rspq));
+		S3("u", "Intr pktcnt:",
+		   adap->sge.counter_val[rx[i].rspq.pktcnt_idx]);
+		R("FL ID:", fl.cntxt_id);
+		R("FL size:", fl.size - 8);
+		R("FL pend:", fl.pend_cred);
+		R("FL avail:", fl.avail);
+		R("FL PIDX:", fl.pidx);
+		R("FL CIDX:", fl.cidx);
+	} else if (toe_idx < toe_entries) {
+		const struct sge_ofld_rxq *rx = &adap->sge.ofldrxq[toe_idx * 4];
+		const struct sge_ofld_txq *tx = &adap->sge.ofldtxq[toe_idx * 4];
+		int n = min(4, adap->sge.ofldqsets - 4 * toe_idx);
+
+		S("QType:", "TOE");
+		T("TxQ ID:", q.cntxt_id);
+		T("TxQ size:", q.size);
+		T("TxQ inuse:", q.in_use);
+		T("TxQ CIDX:", q.cidx);
+		T("TxQ PIDX:", q.pidx);
+		R("RspQ ID:", rspq.abs_id);
+		R("RspQ size:", rspq.size);
+		R("RspQE size:", rspq.iqe_len);
+		R("RspQ CIDX:", rspq.cidx);
+		R("RspQ Gen:", rspq.gen);
+		S3("u", "Intr delay:", qtimer_val(adap, &rx[i].rspq));
+		S3("u", "Intr pktcnt:",
+		   adap->sge.counter_val[rx[i].rspq.pktcnt_idx]);
+		R("FL ID:", fl.cntxt_id);
+		R("FL size:", fl.size - 8);
+		R("FL pend:", fl.pend_cred);
+		R("FL avail:", fl.avail);
+		R("FL PIDX:", fl.pidx);
+		R("FL CIDX:", fl.cidx);
+	} else if (rdma_idx < rdma_entries) {
+		const struct sge_ofld_rxq *rx =
+				&adap->sge.rdmarxq[rdma_idx * 4];
+		int n = min(4, adap->sge.rdmaqs - 4 * rdma_idx);
+
+		S("QType:", "RDMA-CPL");
+		R("RspQ ID:", rspq.abs_id);
+		R("RspQ size:", rspq.size);
+		R("RspQE size:", rspq.iqe_len);
+		R("RspQ CIDX:", rspq.cidx);
+		R("RspQ Gen:", rspq.gen);
+		S3("u", "Intr delay:", qtimer_val(adap, &rx[i].rspq));
+		S3("u", "Intr pktcnt:",
+		   adap->sge.counter_val[rx[i].rspq.pktcnt_idx]);
+		R("FL ID:", fl.cntxt_id);
+		R("FL size:", fl.size - 8);
+		R("FL pend:", fl.pend_cred);
+		R("FL avail:", fl.avail);
+		R("FL PIDX:", fl.pidx);
+		R("FL CIDX:", fl.cidx);
+	} else if (ciq_idx < ciq_entries) {
+		const struct sge_ofld_rxq *rx = &adap->sge.rdmaciq[ciq_idx * 4];
+		int n = min(4, adap->sge.rdmaciqs - 4 * ciq_idx);
+
+		S("QType:", "RDMA-CIQ");
+		R("RspQ ID:", rspq.abs_id);
+		R("RspQ size:", rspq.size);
+		R("RspQE size:", rspq.iqe_len);
+		R("RspQ CIDX:", rspq.cidx);
+		R("RspQ Gen:", rspq.gen);
+		S3("u", "Intr delay:", qtimer_val(adap, &rx[i].rspq));
+		S3("u", "Intr pktcnt:",
+		   adap->sge.counter_val[rx[i].rspq.pktcnt_idx]);
+	} else if (ctrl_idx < ctrl_entries) {
+		const struct sge_ctrl_txq *tx = &adap->sge.ctrlq[ctrl_idx * 4];
+		int n = min(4, adap->params.nports - 4 * ctrl_idx);
+
+		S("QType:", "Control");
+		T("TxQ ID:", q.cntxt_id);
+		T("TxQ size:", q.size);
+		T("TxQ inuse:", q.in_use);
+		T("TxQ CIDX:", q.cidx);
+		T("TxQ PIDX:", q.pidx);
+	} else if (fq_idx == 0) {
+		const struct sge_rspq *evtq = &adap->sge.fw_evtq;
+
+		seq_printf(seq, "%-12s %16s\n", "QType:", "FW event queue");
+		seq_printf(seq, "%-12s %16u\n", "RspQ ID:", evtq->abs_id);
+		seq_printf(seq, "%-12s %16u\n", "RspQ size:", evtq->size);
+		seq_printf(seq, "%-12s %16u\n", "RspQE size:", evtq->iqe_len);
+		seq_printf(seq, "%-12s %16u\n", "RspQ CIDX:", evtq->cidx);
+		seq_printf(seq, "%-12s %16u\n", "RspQ Gen:", evtq->gen);
+		seq_printf(seq, "%-12s %16u\n", "Intr delay:",
+			   qtimer_val(adap, evtq));
+		seq_printf(seq, "%-12s %16u\n", "Intr pktcnt:",
+			   adap->sge.counter_val[evtq->pktcnt_idx]);
+	}
+#undef R
+#undef T
+#undef S
+#undef S3
+return 0;
+}
+
+static int sge_queue_entries(const struct adapter *adap)
+{
+	return DIV_ROUND_UP(adap->sge.ethqsets, 4) +
+	       DIV_ROUND_UP(adap->sge.ofldqsets, 4) +
+	       DIV_ROUND_UP(adap->sge.rdmaqs, 4) +
+	       DIV_ROUND_UP(adap->sge.rdmaciqs, 4) +
+	       DIV_ROUND_UP(MAX_CTRL_QUEUES, 4) + 1;
+}
+
+static void *sge_queue_start(struct seq_file *seq, loff_t *pos)
+{
+	int entries = sge_queue_entries(seq->private);
+
+	return *pos < entries ? (void *)((uintptr_t)*pos + 1) : NULL;
+}
+
+static void sge_queue_stop(struct seq_file *seq, void *v)
+{
+}
+
+static void *sge_queue_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	int entries = sge_queue_entries(seq->private);
+
+	++*pos;
+	return *pos < entries ? (void *)((uintptr_t)*pos + 1) : NULL;
+}
+
+static const struct seq_operations sge_qinfo_seq_ops = {
+	.start = sge_queue_start,
+	.next  = sge_queue_next,
+	.stop  = sge_queue_stop,
+	.show  = sge_qinfo_show
+};
+
+static int sge_qinfo_open(struct inode *inode, struct file *file)
+{
+	int res = seq_open(file, &sge_qinfo_seq_ops);
+
+	if (!res) {
+		struct seq_file *seq = file->private_data;
+
+		seq->private = inode->i_private;
+	}
+	return res;
+}
+
+static const struct file_operations sge_qinfo_debugfs_fops = {
+	.owner   = THIS_MODULE,
+	.open    = sge_qinfo_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release,
+};
+
 int mem_open(struct inode *inode, struct file *file)
 {
 	unsigned int mem;
@@ -1127,6 +1540,7 @@ int t4_setup_debugfs(struct adapter *adap)
 	static struct t4_debugfs_entry t4_debugfs_files[] = {
 		{ "cim_la", &cim_la_fops, S_IRUSR, 0 },
 		{ "cim_qcfg", &cim_qcfg_fops, S_IRUSR, 0 },
+		{ "clk", &clk_debugfs_fops, S_IRUSR, 0 },
 		{ "devlog", &devlog_fops, S_IRUSR, 0 },
 		{ "l2t", &t4_l2t_fops, S_IRUSR, 0},
 		{ "mps_tcam", &mps_tcam_debugfs_fops, S_IRUSR, 0 },
@@ -1135,14 +1549,39 @@ int t4_setup_debugfs(struct adapter *adap)
 		{ "rss_key", &rss_key_debugfs_fops, S_IRUSR, 0 },
 		{ "rss_pf_config", &rss_pf_config_debugfs_fops, S_IRUSR, 0 },
 		{ "rss_vf_config", &rss_vf_config_debugfs_fops, S_IRUSR, 0 },
+		{ "sge_qinfo", &sge_qinfo_debugfs_fops, S_IRUSR, 0 },
+		{ "ibq_tp0",  &cim_ibq_fops, S_IRUSR, 0 },
+		{ "ibq_tp1",  &cim_ibq_fops, S_IRUSR, 1 },
+		{ "ibq_ulp",  &cim_ibq_fops, S_IRUSR, 2 },
+		{ "ibq_sge0", &cim_ibq_fops, S_IRUSR, 3 },
+		{ "ibq_sge1", &cim_ibq_fops, S_IRUSR, 4 },
+		{ "ibq_ncsi", &cim_ibq_fops, S_IRUSR, 5 },
+		{ "obq_ulp0", &cim_obq_fops, S_IRUSR, 0 },
+		{ "obq_ulp1", &cim_obq_fops, S_IRUSR, 1 },
+		{ "obq_ulp2", &cim_obq_fops, S_IRUSR, 2 },
+		{ "obq_ulp3", &cim_obq_fops, S_IRUSR, 3 },
+		{ "obq_sge",  &cim_obq_fops, S_IRUSR, 4 },
+		{ "obq_ncsi", &cim_obq_fops, S_IRUSR, 5 },
+		{ "pm_stats", &pm_stats_debugfs_fops, S_IRUSR, 0 },
 #if IS_ENABLED(CONFIG_IPV6)
 		{ "clip_tbl", &clip_tbl_debugfs_fops, S_IRUSR, 0 },
 #endif
 	};
 
+	/* Debug FS nodes common to all T5 and later adapters.
+	 */
+	static struct t4_debugfs_entry t5_debugfs_files[] = {
+		{ "obq_sge_rx_q0", &cim_obq_fops, S_IRUSR, 6 },
+		{ "obq_sge_rx_q1", &cim_obq_fops, S_IRUSR, 7 },
+	};
+
 	add_debugfs_files(adap,
 			  t4_debugfs_files,
 			  ARRAY_SIZE(t4_debugfs_files));
+	if (!is_t4(adap->params.chip))
+		add_debugfs_files(adap,
+				  t5_debugfs_files,
+				  ARRAY_SIZE(t5_debugfs_files));
 
 	i = t4_read_reg(adap, MA_TARGET_MEM_ENABLE_A);
 	if (i & EDRAM0_ENABLE_F) {
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index 12c1a3f..c27dcd9 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -2500,8 +2500,8 @@ static int closest_thres(const struct sge *s, int thres)
 /*
  * Return a queue's interrupt hold-off time in us.  0 means no timer.
  */
-static unsigned int qtimer_val(const struct adapter *adap,
-			       const struct sge_rspq *q)
+unsigned int qtimer_val(const struct adapter *adap,
+			const struct sge_rspq *q)
 {
 	unsigned int idx = q->intr_params >> 1;
 
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index 8f99878..ea16c62 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -2512,6 +2512,60 @@ void t4_load_mtus(struct adapter *adap, const unsigned short *mtus,
 }
 
 /**
+ *	t4_pmtx_get_stats - returns the HW stats from PMTX
+ *	@adap: the adapter
+ *	@cnt: where to store the count statistics
+ *	@cycles: where to store the cycle statistics
+ *
+ *	Returns performance statistics from PMTX.
+ */
+void t4_pmtx_get_stats(struct adapter *adap, u32 cnt[], u64 cycles[])
+{
+	int i;
+	u32 data[2];
+
+	for (i = 0; i < PM_NSTATS; i++) {
+		t4_write_reg(adap, PM_TX_STAT_CONFIG_A, i + 1);
+		cnt[i] = t4_read_reg(adap, PM_TX_STAT_COUNT_A);
+		if (is_t4(adap->params.chip)) {
+			cycles[i] = t4_read_reg64(adap, PM_TX_STAT_LSB_A);
+		} else {
+			t4_read_indirect(adap, PM_TX_DBG_CTRL_A,
+					 PM_TX_DBG_DATA_A, data, 2,
+					 PM_TX_DBG_STAT_MSB_A);
+			cycles[i] = (((u64)data[0] << 32) | data[1]);
+		}
+	}
+}
+
+/**
+ *	t4_pmrx_get_stats - returns the HW stats from PMRX
+ *	@adap: the adapter
+ *	@cnt: where to store the count statistics
+ *	@cycles: where to store the cycle statistics
+ *
+ *	Returns performance statistics from PMRX.
+ */
+void t4_pmrx_get_stats(struct adapter *adap, u32 cnt[], u64 cycles[])
+{
+	int i;
+	u32 data[2];
+
+	for (i = 0; i < PM_NSTATS; i++) {
+		t4_write_reg(adap, PM_RX_STAT_CONFIG_A, i + 1);
+		cnt[i] = t4_read_reg(adap, PM_RX_STAT_COUNT_A);
+		if (is_t4(adap->params.chip)) {
+			cycles[i] = t4_read_reg64(adap, PM_RX_STAT_LSB_A);
+		} else {
+			t4_read_indirect(adap, PM_RX_DBG_CTRL_A,
+					 PM_RX_DBG_DATA_A, data, 2,
+					 PM_RX_DBG_STAT_MSB_A);
+			cycles[i] = (((u64)data[0] << 32) | data[1]);
+		}
+	}
+}
+
+/**
  *	get_mps_bg_map - return the buffer groups associated with a port
  *	@adap: the adapter
  *	@idx: the port index
@@ -4526,6 +4580,91 @@ void t4_read_cimq_cfg(struct adapter *adap, u16 *base, u16 *size, u16 *thres)
 }
 
 /**
+ *	t4_read_cim_ibq - read the contents of a CIM inbound queue
+ *	@adap: the adapter
+ *	@qid: the queue index
+ *	@data: where to store the queue contents
+ *	@n: capacity of @data in 32-bit words
+ *
+ *	Reads the contents of the selected CIM queue starting at address 0 up
+ *	to the capacity of @data.  @n must be a multiple of 4.  Returns < 0 on
+ *	error and the number of 32-bit words actually read on success.
+ */
+int t4_read_cim_ibq(struct adapter *adap, unsigned int qid, u32 *data, size_t n)
+{
+	int i, err, attempts;
+	unsigned int addr;
+	const unsigned int nwords = CIM_IBQ_SIZE * 4;
+
+	if (qid > 5 || (n & 3))
+		return -EINVAL;
+
+	addr = qid * nwords;
+	if (n > nwords)
+		n = nwords;
+
+	/* It might take 3-10ms before the IBQ debug read access is allowed.
+	 * Wait for 1 Sec with a delay of 1 usec.
+	 */
+	attempts = 1000000;
+
+	for (i = 0; i < n; i++, addr++) {
+		t4_write_reg(adap, CIM_IBQ_DBG_CFG_A, IBQDBGADDR_V(addr) |
+			     IBQDBGEN_F);
+		err = t4_wait_op_done(adap, CIM_IBQ_DBG_CFG_A, IBQDBGBUSY_F, 0,
+				      attempts, 1);
+		if (err)
+			return err;
+		*data++ = t4_read_reg(adap, CIM_IBQ_DBG_DATA_A);
+	}
+	t4_write_reg(adap, CIM_IBQ_DBG_CFG_A, 0);
+	return i;
+}
+
+/**
+ *	t4_read_cim_obq - read the contents of a CIM outbound queue
+ *	@adap: the adapter
+ *	@qid: the queue index
+ *	@data: where to store the queue contents
+ *	@n: capacity of @data in 32-bit words
+ *
+ *	Reads the contents of the selected CIM queue starting at address 0 up
+ *	to the capacity of @data.  @n must be a multiple of 4.  Returns < 0 on
+ *	error and the number of 32-bit words actually read on success.
+ */
+int t4_read_cim_obq(struct adapter *adap, unsigned int qid, u32 *data, size_t n)
+{
+	int i, err;
+	unsigned int addr, v, nwords;
+	int cim_num_obq = is_t4(adap->params.chip) ?
+				CIM_NUM_OBQ : CIM_NUM_OBQ_T5;
+
+	if ((qid > (cim_num_obq - 1)) || (n & 3))
+		return -EINVAL;
+
+	t4_write_reg(adap, CIM_QUEUE_CONFIG_REF_A, OBQSELECT_F |
+		     QUENUMSELECT_V(qid));
+	v = t4_read_reg(adap, CIM_QUEUE_CONFIG_CTRL_A);
+
+	addr = CIMQBASE_G(v) * 64;    /* muliple of 256 -> muliple of 4 */
+	nwords = CIMQSIZE_G(v) * 64;  /* same */
+	if (n > nwords)
+		n = nwords;
+
+	for (i = 0; i < n; i++, addr++) {
+		t4_write_reg(adap, CIM_OBQ_DBG_CFG_A, OBQDBGADDR_V(addr) |
+			     OBQDBGEN_F);
+		err = t4_wait_op_done(adap, CIM_OBQ_DBG_CFG_A, OBQDBGBUSY_F, 0,
+				      2, 1);
+		if (err)
+			return err;
+		*data++ = t4_read_reg(adap, CIM_OBQ_DBG_DATA_A);
+	}
+	t4_write_reg(adap, CIM_OBQ_DBG_CFG_A, 0);
+	return i;
+}
+
+/**
  *	t4_cim_read - read a block from CIM internal address space
  *	@adap: the adapter
  *	@addr: the start address within the CIM address space
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
index f6b82da..664375f 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.h
@@ -48,6 +48,7 @@ enum {
 	NMTUS          = 16,    /* size of MTU table */
 	NCCTRL_WIN     = 32,    /* # of congestion control windows */
 	L2T_SIZE       = 4096,  /* # of L2T entries */
+	PM_NSTATS      = 5,     /* # of PM stats */
 	MBOX_LEN       = 64,    /* mailbox size in bytes */
 	TRACE_LEN      = 112,   /* length of trace data and mask */
 	FILTER_OPT_LEN = 36,    /* filter tuple width for optional components */
@@ -60,6 +61,8 @@ enum {
 	CIM_NUM_OBQ    = 6,     /* # of CIM OBQs */
 	CIM_NUM_OBQ_T5 = 8,     /* # of CIM OBQs for T5 adapter */
 	CIMLA_SIZE     = 2048,  /* # of 32-bit words in CIM LA */
+	CIM_IBQ_SIZE   = 128,   /* # of 128-bit words in a CIM IBQ */
+	CIM_OBQ_SIZE   = 128,   /* # of 128-bit words in a CIM OBQ */
 };
 
 enum {
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
index e036b56..940b56c 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
@@ -1203,12 +1203,35 @@
 #define TIMERRESOLUTION_M    0xffU
 #define TIMERRESOLUTION_G(x) (((x) >> TIMERRESOLUTION_S) & TIMERRESOLUTION_M)
 
+#define TIMESTAMPRESOLUTION_S    8
+#define TIMESTAMPRESOLUTION_M    0xffU
+#define TIMESTAMPRESOLUTION_G(x) \
+	(((x) >> TIMESTAMPRESOLUTION_S) & TIMESTAMPRESOLUTION_M)
+
 #define DELAYEDACKRESOLUTION_S    0
 #define DELAYEDACKRESOLUTION_M    0xffU
 #define DELAYEDACKRESOLUTION_G(x) \
 	(((x) >> DELAYEDACKRESOLUTION_S) & DELAYEDACKRESOLUTION_M)
 
 #define TP_SHIFT_CNT_A 0x7dc0
+#define TP_RXT_MIN_A 0x7d98
+#define TP_RXT_MAX_A 0x7d9c
+#define TP_PERS_MIN_A 0x7da0
+#define TP_PERS_MAX_A 0x7da4
+#define TP_KEEP_IDLE_A 0x7da8
+#define TP_KEEP_INTVL_A 0x7dac
+#define TP_INIT_SRTT_A 0x7db0
+#define TP_DACK_TIMER_A 0x7db4
+#define TP_FINWAIT2_TIMER_A 0x7db8
+
+#define INITSRTT_S    0
+#define INITSRTT_M    0xffffU
+#define INITSRTT_G(x) (((x) >> INITSRTT_S) & INITSRTT_M)
+
+#define PERSMAX_S    0
+#define PERSMAX_M    0x3fffffffU
+#define PERSMAX_V(x) ((x) << PERSMAX_S)
+#define PERSMAX_G(x) (((x) >> PERSMAX_S) & PERSMAX_M)
 
 #define SYNSHIFTMAX_S    24
 #define SYNSHIFTMAX_M    0xffU
@@ -1380,6 +1403,12 @@
 #define PBL_BOUND_ERR_CH0_F    PBL_BOUND_ERR_CH0_V(1U)
 
 #define PM_RX_INT_CAUSE_A	0x8fdc
+#define PM_RX_STAT_CONFIG_A 0x8fc8
+#define PM_RX_STAT_COUNT_A 0x8fcc
+#define PM_RX_STAT_LSB_A 0x8fd0
+#define PM_RX_DBG_CTRL_A 0x8fd0
+#define PM_RX_DBG_DATA_A 0x8fd4
+#define PM_RX_DBG_STAT_MSB_A 0x10013
 
 #define PMRX_FRAMING_ERROR_F	0x003ffff0U
 
@@ -1404,6 +1433,12 @@
 #define PMRX_E_PCMD_PAR_ERROR_F    PMRX_E_PCMD_PAR_ERROR_V(1U)
 
 #define PM_TX_INT_CAUSE_A	0x8ffc
+#define PM_TX_STAT_CONFIG_A 0x8fe8
+#define PM_TX_STAT_COUNT_A 0x8fec
+#define PM_TX_STAT_LSB_A 0x8ff0
+#define PM_TX_DBG_CTRL_A 0x8ff0
+#define PM_TX_DBG_DATA_A 0x8ff4
+#define PM_TX_DBG_STAT_MSB_A 0x1001a
 
 #define PCMD_LEN_OVFL0_S    31
 #define PCMD_LEN_OVFL0_V(x) ((x) << PCMD_LEN_OVFL0_S)
@@ -2538,6 +2573,39 @@
 #define HOSTWRITE_V(x)	((x) << HOSTWRITE_S)
 #define HOSTWRITE_F	HOSTWRITE_V(1U)
 
+#define CIM_IBQ_DBG_CFG_A 0x7b60
+
+#define IBQDBGADDR_S    16
+#define IBQDBGADDR_M    0xfffU
+#define IBQDBGADDR_V(x) ((x) << IBQDBGADDR_S)
+#define IBQDBGADDR_G(x) (((x) >> IBQDBGADDR_S) & IBQDBGADDR_M)
+
+#define IBQDBGBUSY_S    1
+#define IBQDBGBUSY_V(x) ((x) << IBQDBGBUSY_S)
+#define IBQDBGBUSY_F    IBQDBGBUSY_V(1U)
+
+#define IBQDBGEN_S    0
+#define IBQDBGEN_V(x) ((x) << IBQDBGEN_S)
+#define IBQDBGEN_F    IBQDBGEN_V(1U)
+
+#define CIM_OBQ_DBG_CFG_A 0x7b64
+
+#define OBQDBGADDR_S    16
+#define OBQDBGADDR_M    0xfffU
+#define OBQDBGADDR_V(x) ((x) << OBQDBGADDR_S)
+#define OBQDBGADDR_G(x) (((x) >> OBQDBGADDR_S) & OBQDBGADDR_M)
+
+#define OBQDBGBUSY_S    1
+#define OBQDBGBUSY_V(x) ((x) << OBQDBGBUSY_S)
+#define OBQDBGBUSY_F    OBQDBGBUSY_V(1U)
+
+#define OBQDBGEN_S    0
+#define OBQDBGEN_V(x) ((x) << OBQDBGEN_S)
+#define OBQDBGEN_F    OBQDBGEN_V(1U)
+
+#define CIM_IBQ_DBG_DATA_A 0x7b68
+#define CIM_OBQ_DBG_DATA_A 0x7b6c
+
 #define UPDBGLARDEN_S		1
 #define UPDBGLARDEN_V(x)	((x) << UPDBGLARDEN_S)
 #define UPDBGLARDEN_F		UPDBGLARDEN_V(1U)