Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / net / ethernet / chelsio / cxgb4 / cxgb4_debugfs.c
diff --git a/kernel/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c b/kernel/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
new file mode 100644 (file)
index 0000000..371f75e
--- /dev/null
@@ -0,0 +1,2073 @@
+/*
+ * This file is part of the Chelsio T4 Ethernet driver for Linux.
+ *
+ * Copyright (c) 2003-2014 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+#include <linux/string_helpers.h>
+#include <linux/sort.h>
+#include <linux/ctype.h>
+
+#include "cxgb4.h"
+#include "t4_regs.h"
+#include "t4_values.h"
+#include "t4fw_api.h"
+#include "cxgb4_debugfs.h"
+#include "clip_tbl.h"
+#include "l2t.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)
+{
+       pos -= tb->skip_first;
+       return pos >= tb->rows ? NULL : &tb->data[pos * tb->width];
+}
+
+static void *seq_tab_start(struct seq_file *seq, loff_t *pos)
+{
+       struct seq_tab *tb = seq->private;
+
+       if (tb->skip_first && *pos == 0)
+               return SEQ_START_TOKEN;
+
+       return seq_tab_get_idx(tb, *pos);
+}
+
+static void *seq_tab_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+       v = seq_tab_get_idx(seq->private, *pos + 1);
+       if (v)
+               ++*pos;
+       return v;
+}
+
+static void seq_tab_stop(struct seq_file *seq, void *v)
+{
+}
+
+static int seq_tab_show(struct seq_file *seq, void *v)
+{
+       const struct seq_tab *tb = seq->private;
+
+       return tb->show(seq, v, ((char *)v - tb->data) / tb->width);
+}
+
+static const struct seq_operations seq_tab_ops = {
+       .start = seq_tab_start,
+       .next  = seq_tab_next,
+       .stop  = seq_tab_stop,
+       .show  = seq_tab_show
+};
+
+struct seq_tab *seq_open_tab(struct file *f, unsigned int rows,
+                            unsigned int width, unsigned int have_header,
+                            int (*show)(struct seq_file *seq, void *v, int i))
+{
+       struct seq_tab *p;
+
+       p = __seq_open_private(f, &seq_tab_ops, sizeof(*p) + rows * width);
+       if (p) {
+               p->show = show;
+               p->rows = rows;
+               p->width = width;
+               p->skip_first = have_header != 0;
+       }
+       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)
+               seq_puts(seq, "Status   Data      PC     LS0Stat  LS0Addr "
+                        "            LS0Data\n");
+       else {
+               const u32 *p = v;
+
+               seq_printf(seq,
+                          "  %02x  %x%07x %x%07x %08x %08x %08x%08x%08x%08x\n",
+                          (p[0] >> 4) & 0xff, p[0] & 0xf, p[1] >> 4,
+                          p[1] & 0xf, p[2] >> 4, p[2] & 0xf, p[3], p[4], p[5],
+                          p[6], p[7]);
+       }
+       return 0;
+}
+
+static int cim_la_show_3in1(struct seq_file *seq, void *v, int idx)
+{
+       if (v == SEQ_START_TOKEN) {
+               seq_puts(seq, "Status   Data      PC\n");
+       } else {
+               const u32 *p = v;
+
+               seq_printf(seq, "  %02x   %08x %08x\n", p[5] & 0xff, p[6],
+                          p[7]);
+               seq_printf(seq, "  %02x   %02x%06x %02x%06x\n",
+                          (p[3] >> 8) & 0xff, p[3] & 0xff, p[4] >> 8,
+                          p[4] & 0xff, p[5] >> 8);
+               seq_printf(seq, "  %02x   %x%07x %x%07x\n", (p[0] >> 4) & 0xff,
+                          p[0] & 0xf, p[1] >> 4, p[1] & 0xf, p[2] >> 4);
+       }
+       return 0;
+}
+
+static int cim_la_open(struct inode *inode, struct file *file)
+{
+       int ret;
+       unsigned int cfg;
+       struct seq_tab *p;
+       struct adapter *adap = inode->i_private;
+
+       ret = t4_cim_read(adap, UP_UP_DBG_LA_CFG_A, 1, &cfg);
+       if (ret)
+               return ret;
+
+       p = seq_open_tab(file, adap->params.cim_la_size / 8, 8 * sizeof(u32), 1,
+                        cfg & UPDBGLACAPTPCONLY_F ?
+                        cim_la_show_3in1 : cim_la_show);
+       if (!p)
+               return -ENOMEM;
+
+       ret = t4_cim_read_la(adap, (u32 *)p->data, NULL);
+       if (ret)
+               seq_release_private(inode, file);
+       return ret;
+}
+
+static const struct file_operations cim_la_fops = {
+       .owner   = THIS_MODULE,
+       .open    = cim_la_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private
+};
+
+static int cim_qcfg_show(struct seq_file *seq, void *v)
+{
+       static const char * const qname[] = {
+               "TP0", "TP1", "ULP", "SGE0", "SGE1", "NC-SI",
+               "ULP0", "ULP1", "ULP2", "ULP3", "SGE", "NC-SI",
+               "SGE0-RX", "SGE1-RX"
+       };
+
+       int i;
+       struct adapter *adap = seq->private;
+       u16 base[CIM_NUM_IBQ + CIM_NUM_OBQ_T5];
+       u16 size[CIM_NUM_IBQ + CIM_NUM_OBQ_T5];
+       u32 stat[(4 * (CIM_NUM_IBQ + CIM_NUM_OBQ_T5))];
+       u16 thres[CIM_NUM_IBQ];
+       u32 obq_wr_t4[2 * CIM_NUM_OBQ], *wr;
+       u32 obq_wr_t5[2 * CIM_NUM_OBQ_T5];
+       u32 *p = stat;
+       int cim_num_obq = is_t4(adap->params.chip) ?
+                               CIM_NUM_OBQ : CIM_NUM_OBQ_T5;
+
+       i = t4_cim_read(adap, is_t4(adap->params.chip) ? UP_IBQ_0_RDADDR_A :
+                       UP_IBQ_0_SHADOW_RDADDR_A,
+                       ARRAY_SIZE(stat), stat);
+       if (!i) {
+               if (is_t4(adap->params.chip)) {
+                       i = t4_cim_read(adap, UP_OBQ_0_REALADDR_A,
+                                       ARRAY_SIZE(obq_wr_t4), obq_wr_t4);
+                               wr = obq_wr_t4;
+               } else {
+                       i = t4_cim_read(adap, UP_OBQ_0_SHADOW_REALADDR_A,
+                                       ARRAY_SIZE(obq_wr_t5), obq_wr_t5);
+                               wr = obq_wr_t5;
+               }
+       }
+       if (i)
+               return i;
+
+       t4_read_cimq_cfg(adap, base, size, thres);
+
+       seq_printf(seq,
+                  "  Queue  Base  Size Thres  RdPtr WrPtr  SOP  EOP Avail\n");
+       for (i = 0; i < CIM_NUM_IBQ; i++, p += 4)
+               seq_printf(seq, "%7s %5x %5u %5u %6x  %4x %4u %4u %5u\n",
+                          qname[i], base[i], size[i], thres[i],
+                          IBQRDADDR_G(p[0]), IBQWRADDR_G(p[1]),
+                          QUESOPCNT_G(p[3]), QUEEOPCNT_G(p[3]),
+                          QUEREMFLITS_G(p[2]) * 16);
+       for ( ; i < CIM_NUM_IBQ + cim_num_obq; i++, p += 4, wr += 2)
+               seq_printf(seq, "%7s %5x %5u %12x  %4x %4u %4u %5u\n",
+                          qname[i], base[i], size[i],
+                          QUERDADDR_G(p[0]) & 0x3fff, wr[0] - base[i],
+                          QUESOPCNT_G(p[3]), QUEEOPCNT_G(p[3]),
+                          QUEREMFLITS_G(p[2]) * 16);
+       return 0;
+}
+
+static int cim_qcfg_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, cim_qcfg_show, inode->i_private);
+}
+
+static const struct file_operations cim_qcfg_fops = {
+       .owner   = THIS_MODULE,
+       .open    = cim_qcfg_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .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
+};
+
+struct field_desc {
+       const char *name;
+       unsigned int start;
+       unsigned int width;
+};
+
+static void field_desc_show(struct seq_file *seq, u64 v,
+                           const struct field_desc *p)
+{
+       char buf[32];
+       int line_size = 0;
+
+       while (p->name) {
+               u64 mask = (1ULL << p->width) - 1;
+               int len = scnprintf(buf, sizeof(buf), "%s: %llu", p->name,
+                                   ((unsigned long long)v >> p->start) & mask);
+
+               if (line_size + len >= 79) {
+                       line_size = 8;
+                       seq_puts(seq, "\n        ");
+               }
+               seq_printf(seq, "%s ", buf);
+               line_size += len + 1;
+               p++;
+       }
+       seq_putc(seq, '\n');
+}
+
+static struct field_desc tp_la0[] = {
+       { "RcfOpCodeOut", 60, 4 },
+       { "State", 56, 4 },
+       { "WcfState", 52, 4 },
+       { "RcfOpcSrcOut", 50, 2 },
+       { "CRxError", 49, 1 },
+       { "ERxError", 48, 1 },
+       { "SanityFailed", 47, 1 },
+       { "SpuriousMsg", 46, 1 },
+       { "FlushInputMsg", 45, 1 },
+       { "FlushInputCpl", 44, 1 },
+       { "RssUpBit", 43, 1 },
+       { "RssFilterHit", 42, 1 },
+       { "Tid", 32, 10 },
+       { "InitTcb", 31, 1 },
+       { "LineNumber", 24, 7 },
+       { "Emsg", 23, 1 },
+       { "EdataOut", 22, 1 },
+       { "Cmsg", 21, 1 },
+       { "CdataOut", 20, 1 },
+       { "EreadPdu", 19, 1 },
+       { "CreadPdu", 18, 1 },
+       { "TunnelPkt", 17, 1 },
+       { "RcfPeerFin", 16, 1 },
+       { "RcfReasonOut", 12, 4 },
+       { "TxCchannel", 10, 2 },
+       { "RcfTxChannel", 8, 2 },
+       { "RxEchannel", 6, 2 },
+       { "RcfRxChannel", 5, 1 },
+       { "RcfDataOutSrdy", 4, 1 },
+       { "RxDvld", 3, 1 },
+       { "RxOoDvld", 2, 1 },
+       { "RxCongestion", 1, 1 },
+       { "TxCongestion", 0, 1 },
+       { NULL }
+};
+
+static int tp_la_show(struct seq_file *seq, void *v, int idx)
+{
+       const u64 *p = v;
+
+       field_desc_show(seq, *p, tp_la0);
+       return 0;
+}
+
+static int tp_la_show2(struct seq_file *seq, void *v, int idx)
+{
+       const u64 *p = v;
+
+       if (idx)
+               seq_putc(seq, '\n');
+       field_desc_show(seq, p[0], tp_la0);
+       if (idx < (TPLA_SIZE / 2 - 1) || p[1] != ~0ULL)
+               field_desc_show(seq, p[1], tp_la0);
+       return 0;
+}
+
+static int tp_la_show3(struct seq_file *seq, void *v, int idx)
+{
+       static struct field_desc tp_la1[] = {
+               { "CplCmdIn", 56, 8 },
+               { "CplCmdOut", 48, 8 },
+               { "ESynOut", 47, 1 },
+               { "EAckOut", 46, 1 },
+               { "EFinOut", 45, 1 },
+               { "ERstOut", 44, 1 },
+               { "SynIn", 43, 1 },
+               { "AckIn", 42, 1 },
+               { "FinIn", 41, 1 },
+               { "RstIn", 40, 1 },
+               { "DataIn", 39, 1 },
+               { "DataInVld", 38, 1 },
+               { "PadIn", 37, 1 },
+               { "RxBufEmpty", 36, 1 },
+               { "RxDdp", 35, 1 },
+               { "RxFbCongestion", 34, 1 },
+               { "TxFbCongestion", 33, 1 },
+               { "TxPktSumSrdy", 32, 1 },
+               { "RcfUlpType", 28, 4 },
+               { "Eread", 27, 1 },
+               { "Ebypass", 26, 1 },
+               { "Esave", 25, 1 },
+               { "Static0", 24, 1 },
+               { "Cread", 23, 1 },
+               { "Cbypass", 22, 1 },
+               { "Csave", 21, 1 },
+               { "CPktOut", 20, 1 },
+               { "RxPagePoolFull", 18, 2 },
+               { "RxLpbkPkt", 17, 1 },
+               { "TxLpbkPkt", 16, 1 },
+               { "RxVfValid", 15, 1 },
+               { "SynLearned", 14, 1 },
+               { "SetDelEntry", 13, 1 },
+               { "SetInvEntry", 12, 1 },
+               { "CpcmdDvld", 11, 1 },
+               { "CpcmdSave", 10, 1 },
+               { "RxPstructsFull", 8, 2 },
+               { "EpcmdDvld", 7, 1 },
+               { "EpcmdFlush", 6, 1 },
+               { "EpcmdTrimPrefix", 5, 1 },
+               { "EpcmdTrimPostfix", 4, 1 },
+               { "ERssIp4Pkt", 3, 1 },
+               { "ERssIp6Pkt", 2, 1 },
+               { "ERssTcpUdpPkt", 1, 1 },
+               { "ERssFceFipPkt", 0, 1 },
+               { NULL }
+       };
+       static struct field_desc tp_la2[] = {
+               { "CplCmdIn", 56, 8 },
+               { "MpsVfVld", 55, 1 },
+               { "MpsPf", 52, 3 },
+               { "MpsVf", 44, 8 },
+               { "SynIn", 43, 1 },
+               { "AckIn", 42, 1 },
+               { "FinIn", 41, 1 },
+               { "RstIn", 40, 1 },
+               { "DataIn", 39, 1 },
+               { "DataInVld", 38, 1 },
+               { "PadIn", 37, 1 },
+               { "RxBufEmpty", 36, 1 },
+               { "RxDdp", 35, 1 },
+               { "RxFbCongestion", 34, 1 },
+               { "TxFbCongestion", 33, 1 },
+               { "TxPktSumSrdy", 32, 1 },
+               { "RcfUlpType", 28, 4 },
+               { "Eread", 27, 1 },
+               { "Ebypass", 26, 1 },
+               { "Esave", 25, 1 },
+               { "Static0", 24, 1 },
+               { "Cread", 23, 1 },
+               { "Cbypass", 22, 1 },
+               { "Csave", 21, 1 },
+               { "CPktOut", 20, 1 },
+               { "RxPagePoolFull", 18, 2 },
+               { "RxLpbkPkt", 17, 1 },
+               { "TxLpbkPkt", 16, 1 },
+               { "RxVfValid", 15, 1 },
+               { "SynLearned", 14, 1 },
+               { "SetDelEntry", 13, 1 },
+               { "SetInvEntry", 12, 1 },
+               { "CpcmdDvld", 11, 1 },
+               { "CpcmdSave", 10, 1 },
+               { "RxPstructsFull", 8, 2 },
+               { "EpcmdDvld", 7, 1 },
+               { "EpcmdFlush", 6, 1 },
+               { "EpcmdTrimPrefix", 5, 1 },
+               { "EpcmdTrimPostfix", 4, 1 },
+               { "ERssIp4Pkt", 3, 1 },
+               { "ERssIp6Pkt", 2, 1 },
+               { "ERssTcpUdpPkt", 1, 1 },
+               { "ERssFceFipPkt", 0, 1 },
+               { NULL }
+       };
+       const u64 *p = v;
+
+       if (idx)
+               seq_putc(seq, '\n');
+       field_desc_show(seq, p[0], tp_la0);
+       if (idx < (TPLA_SIZE / 2 - 1) || p[1] != ~0ULL)
+               field_desc_show(seq, p[1], (p[0] & BIT(17)) ? tp_la2 : tp_la1);
+       return 0;
+}
+
+static int tp_la_open(struct inode *inode, struct file *file)
+{
+       struct seq_tab *p;
+       struct adapter *adap = inode->i_private;
+
+       switch (DBGLAMODE_G(t4_read_reg(adap, TP_DBG_LA_CONFIG_A))) {
+       case 2:
+               p = seq_open_tab(file, TPLA_SIZE / 2, 2 * sizeof(u64), 0,
+                                tp_la_show2);
+               break;
+       case 3:
+               p = seq_open_tab(file, TPLA_SIZE / 2, 2 * sizeof(u64), 0,
+                                tp_la_show3);
+               break;
+       default:
+               p = seq_open_tab(file, TPLA_SIZE, sizeof(u64), 0, tp_la_show);
+       }
+       if (!p)
+               return -ENOMEM;
+
+       t4_tp_read_la(adap, (u64 *)p->data, NULL);
+       return 0;
+}
+
+static ssize_t tp_la_write(struct file *file, const char __user *buf,
+                          size_t count, loff_t *pos)
+{
+       int err;
+       char s[32];
+       unsigned long val;
+       size_t size = min(sizeof(s) - 1, count);
+       struct adapter *adap = file_inode(file)->i_private;
+
+       if (copy_from_user(s, buf, size))
+               return -EFAULT;
+       s[size] = '\0';
+       err = kstrtoul(s, 0, &val);
+       if (err)
+               return err;
+       if (val > 0xffff)
+               return -EINVAL;
+       adap->params.tp.la_mask = val << 16;
+       t4_set_reg_field(adap, TP_DBG_LA_CONFIG_A, 0xffff0000U,
+                        adap->params.tp.la_mask);
+       return count;
+}
+
+static const struct file_operations tp_la_fops = {
+       .owner   = THIS_MODULE,
+       .open    = tp_la_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private,
+       .write   = tp_la_write
+};
+
+static int ulprx_la_show(struct seq_file *seq, void *v, int idx)
+{
+       const u32 *p = v;
+
+       if (v == SEQ_START_TOKEN)
+               seq_puts(seq, "      Pcmd        Type   Message"
+                        "                Data\n");
+       else
+               seq_printf(seq, "%08x%08x  %4x  %08x  %08x%08x%08x%08x\n",
+                          p[1], p[0], p[2], p[3], p[7], p[6], p[5], p[4]);
+       return 0;
+}
+
+static int ulprx_la_open(struct inode *inode, struct file *file)
+{
+       struct seq_tab *p;
+       struct adapter *adap = inode->i_private;
+
+       p = seq_open_tab(file, ULPRX_LA_SIZE, 8 * sizeof(u32), 1,
+                        ulprx_la_show);
+       if (!p)
+               return -ENOMEM;
+
+       t4_ulprx_read_la(adap, (u32 *)p->data);
+       return 0;
+}
+
+static const struct file_operations ulprx_la_fops = {
+       .owner   = THIS_MODULE,
+       .open    = ulprx_la_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_inode(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
+};
+
+static int cctrl_tbl_show(struct seq_file *seq, void *v)
+{
+       static const char * const dec_fac[] = {
+               "0.5", "0.5625", "0.625", "0.6875", "0.75", "0.8125", "0.875",
+               "0.9375" };
+
+       int i;
+       u16 (*incr)[NCCTRL_WIN];
+       struct adapter *adap = seq->private;
+
+       incr = kmalloc(sizeof(*incr) * NMTUS, GFP_KERNEL);
+       if (!incr)
+               return -ENOMEM;
+
+       t4_read_cong_tbl(adap, incr);
+
+       for (i = 0; i < NCCTRL_WIN; ++i) {
+               seq_printf(seq, "%2d: %4u %4u %4u %4u %4u %4u %4u %4u\n", i,
+                          incr[0][i], incr[1][i], incr[2][i], incr[3][i],
+                          incr[4][i], incr[5][i], incr[6][i], incr[7][i]);
+               seq_printf(seq, "%8u %4u %4u %4u %4u %4u %4u %4u %5u %s\n",
+                          incr[8][i], incr[9][i], incr[10][i], incr[11][i],
+                          incr[12][i], incr[13][i], incr[14][i], incr[15][i],
+                          adap->params.a_wnd[i],
+                          dec_fac[adap->params.b_wnd[i]]);
+       }
+
+       kfree(incr);
+       return 0;
+}
+
+DEFINE_SIMPLE_DEBUGFS_FILE(cctrl_tbl);
+
+/* 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",
+       [FW_DEVLOG_LEVEL_CRIT]          = "CRIT",
+       [FW_DEVLOG_LEVEL_ERR]           = "ERR",
+       [FW_DEVLOG_LEVEL_NOTICE]        = "NOTICE",
+       [FW_DEVLOG_LEVEL_INFO]          = "INFO",
+       [FW_DEVLOG_LEVEL_DEBUG]         = "DEBUG"
+};
+
+static const char * const devlog_facility_strings[] = {
+       [FW_DEVLOG_FACILITY_CORE]       = "CORE",
+       [FW_DEVLOG_FACILITY_SCHED]      = "SCHED",
+       [FW_DEVLOG_FACILITY_TIMER]      = "TIMER",
+       [FW_DEVLOG_FACILITY_RES]        = "RES",
+       [FW_DEVLOG_FACILITY_HW]         = "HW",
+       [FW_DEVLOG_FACILITY_FLR]        = "FLR",
+       [FW_DEVLOG_FACILITY_DMAQ]       = "DMAQ",
+       [FW_DEVLOG_FACILITY_PHY]        = "PHY",
+       [FW_DEVLOG_FACILITY_MAC]        = "MAC",
+       [FW_DEVLOG_FACILITY_PORT]       = "PORT",
+       [FW_DEVLOG_FACILITY_VI]         = "VI",
+       [FW_DEVLOG_FACILITY_FILTER]     = "FILTER",
+       [FW_DEVLOG_FACILITY_ACL]        = "ACL",
+       [FW_DEVLOG_FACILITY_TM]         = "TM",
+       [FW_DEVLOG_FACILITY_QFC]        = "QFC",
+       [FW_DEVLOG_FACILITY_DCB]        = "DCB",
+       [FW_DEVLOG_FACILITY_ETH]        = "ETH",
+       [FW_DEVLOG_FACILITY_OFLD]       = "OFLD",
+       [FW_DEVLOG_FACILITY_RI]         = "RI",
+       [FW_DEVLOG_FACILITY_ISCSI]      = "ISCSI",
+       [FW_DEVLOG_FACILITY_FCOE]       = "FCOE",
+       [FW_DEVLOG_FACILITY_FOISCSI]    = "FOISCSI",
+       [FW_DEVLOG_FACILITY_FOFCOE]     = "FOFCOE"
+};
+
+/* Information gathered by Device Log Open routine for the display routine.
+ */
+struct devlog_info {
+       unsigned int nentries;          /* number of entries in log[] */
+       unsigned int first;             /* first [temporal] entry in log[] */
+       struct fw_devlog_e log[0];      /* Firmware Device Log */
+};
+
+/* Dump a Firmaware Device Log entry.
+ */
+static int devlog_show(struct seq_file *seq, void *v)
+{
+       if (v == SEQ_START_TOKEN)
+               seq_printf(seq, "%10s  %15s  %8s  %8s  %s\n",
+                          "Seq#", "Tstamp", "Level", "Facility", "Message");
+       else {
+               struct devlog_info *dinfo = seq->private;
+               int fidx = (uintptr_t)v - 2;
+               unsigned long index;
+               struct fw_devlog_e *e;
+
+               /* Get a pointer to the log entry to display.  Skip unused log
+                * entries.
+                */
+               index = dinfo->first + fidx;
+               if (index >= dinfo->nentries)
+                       index -= dinfo->nentries;
+               e = &dinfo->log[index];
+               if (e->timestamp == 0)
+                       return 0;
+
+               /* Print the message.  This depends on the firmware using
+                * exactly the same formating strings as the kernel so we may
+                * eventually have to put a format interpreter in here ...
+                */
+               seq_printf(seq, "%10d  %15llu  %8s  %8s  ",
+                          e->seqno, e->timestamp,
+                          (e->level < ARRAY_SIZE(devlog_level_strings)
+                           ? devlog_level_strings[e->level]
+                           : "UNKNOWN"),
+                          (e->facility < ARRAY_SIZE(devlog_facility_strings)
+                           ? devlog_facility_strings[e->facility]
+                           : "UNKNOWN"));
+               seq_printf(seq, e->fmt, e->params[0], e->params[1],
+                          e->params[2], e->params[3], e->params[4],
+                          e->params[5], e->params[6], e->params[7]);
+       }
+       return 0;
+}
+
+/* Sequential File Operations for Device Log.
+ */
+static inline void *devlog_get_idx(struct devlog_info *dinfo, loff_t pos)
+{
+       if (pos > dinfo->nentries)
+               return NULL;
+
+       return (void *)(uintptr_t)(pos + 1);
+}
+
+static void *devlog_start(struct seq_file *seq, loff_t *pos)
+{
+       struct devlog_info *dinfo = seq->private;
+
+       return (*pos
+               ? devlog_get_idx(dinfo, *pos)
+               : SEQ_START_TOKEN);
+}
+
+static void *devlog_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+       struct devlog_info *dinfo = seq->private;
+
+       (*pos)++;
+       return devlog_get_idx(dinfo, *pos);
+}
+
+static void devlog_stop(struct seq_file *seq, void *v)
+{
+}
+
+static const struct seq_operations devlog_seq_ops = {
+       .start = devlog_start,
+       .next  = devlog_next,
+       .stop  = devlog_stop,
+       .show  = devlog_show
+};
+
+/* Set up for reading the firmware's device log.  We read the entire log here
+ * and then display it incrementally in devlog_show().
+ */
+static int devlog_open(struct inode *inode, struct file *file)
+{
+       struct adapter *adap = inode->i_private;
+       struct devlog_params *dparams = &adap->params.devlog;
+       struct devlog_info *dinfo;
+       unsigned int index;
+       u32 fseqno;
+       int ret;
+
+       /* If we don't know where the log is we can't do anything.
+        */
+       if (dparams->start == 0)
+               return -ENXIO;
+
+       /* Allocate the space to read in the firmware's device log and set up
+        * for the iterated call to our display function.
+        */
+       dinfo = __seq_open_private(file, &devlog_seq_ops,
+                                  sizeof(*dinfo) + dparams->size);
+       if (!dinfo)
+               return -ENOMEM;
+
+       /* Record the basic log buffer information and read in the raw log.
+        */
+       dinfo->nentries = (dparams->size / sizeof(struct fw_devlog_e));
+       dinfo->first = 0;
+       spin_lock(&adap->win0_lock);
+       ret = t4_memory_rw(adap, adap->params.drv_memwin, dparams->memtype,
+                          dparams->start, dparams->size, (__be32 *)dinfo->log,
+                          T4_MEMORY_READ);
+       spin_unlock(&adap->win0_lock);
+       if (ret) {
+               seq_release_private(inode, file);
+               return ret;
+       }
+
+       /* Translate log multi-byte integral elements into host native format
+        * and determine where the first entry in the log is.
+        */
+       for (fseqno = ~((u32)0), index = 0; index < dinfo->nentries; index++) {
+               struct fw_devlog_e *e = &dinfo->log[index];
+               int i;
+               __u32 seqno;
+
+               if (e->timestamp == 0)
+                       continue;
+
+               e->timestamp = (__force __be64)be64_to_cpu(e->timestamp);
+               seqno = be32_to_cpu(e->seqno);
+               for (i = 0; i < 8; i++)
+                       e->params[i] =
+                               (__force __be32)be32_to_cpu(e->params[i]);
+
+               if (seqno < fseqno) {
+                       fseqno = seqno;
+                       dinfo->first = index;
+               }
+       }
+       return 0;
+}
+
+static const struct file_operations devlog_fops = {
+       .owner   = THIS_MODULE,
+       .open    = devlog_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private
+};
+
+static int mbox_show(struct seq_file *seq, void *v)
+{
+       static const char * const owner[] = { "none", "FW", "driver",
+                                             "unknown" };
+
+       int i;
+       unsigned int mbox = (uintptr_t)seq->private & 7;
+       struct adapter *adap = seq->private - mbox;
+       void __iomem *addr = adap->regs + PF_REG(mbox, CIM_PF_MAILBOX_DATA_A);
+       unsigned int ctrl_reg = (is_t4(adap->params.chip)
+                                ? CIM_PF_MAILBOX_CTRL_A
+                                : CIM_PF_MAILBOX_CTRL_SHADOW_COPY_A);
+       void __iomem *ctrl = adap->regs + PF_REG(mbox, ctrl_reg);
+
+       i = MBOWNER_G(readl(ctrl));
+       seq_printf(seq, "mailbox owned by %s\n\n", owner[i]);
+
+       for (i = 0; i < MBOX_LEN; i += 8)
+               seq_printf(seq, "%016llx\n",
+                          (unsigned long long)readq(addr + i));
+       return 0;
+}
+
+static int mbox_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, mbox_show, inode->i_private);
+}
+
+static ssize_t mbox_write(struct file *file, const char __user *buf,
+                         size_t count, loff_t *pos)
+{
+       int i;
+       char c = '\n', s[256];
+       unsigned long long data[8];
+       const struct inode *ino;
+       unsigned int mbox;
+       struct adapter *adap;
+       void __iomem *addr;
+       void __iomem *ctrl;
+
+       if (count > sizeof(s) - 1 || !count)
+               return -EINVAL;
+       if (copy_from_user(s, buf, count))
+               return -EFAULT;
+       s[count] = '\0';
+
+       if (sscanf(s, "%llx %llx %llx %llx %llx %llx %llx %llx%c", &data[0],
+                  &data[1], &data[2], &data[3], &data[4], &data[5], &data[6],
+                  &data[7], &c) < 8 || c != '\n')
+               return -EINVAL;
+
+       ino = file_inode(file);
+       mbox = (uintptr_t)ino->i_private & 7;
+       adap = ino->i_private - mbox;
+       addr = adap->regs + PF_REG(mbox, CIM_PF_MAILBOX_DATA_A);
+       ctrl = addr + MBOX_LEN;
+
+       if (MBOWNER_G(readl(ctrl)) != X_MBOWNER_PL)
+               return -EBUSY;
+
+       for (i = 0; i < 8; i++)
+               writeq(data[i], addr + 8 * i);
+
+       writel(MBMSGVALID_F | MBOWNER_V(X_MBOWNER_FW), ctrl);
+       return count;
+}
+
+static const struct file_operations mbox_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = mbox_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = single_release,
+       .write   = mbox_write
+};
+
+static ssize_t flash_read(struct file *file, char __user *buf, size_t count,
+                         loff_t *ppos)
+{
+       loff_t pos = *ppos;
+       loff_t avail = file_inode(file)->i_size;
+       struct adapter *adap = file->private_data;
+
+       if (pos < 0)
+               return -EINVAL;
+       if (pos >= avail)
+               return 0;
+       if (count > avail - pos)
+               count = avail - pos;
+
+       while (count) {
+               size_t len;
+               int ret, ofst;
+               u8 data[256];
+
+               ofst = pos & 3;
+               len = min(count + ofst, sizeof(data));
+               ret = t4_read_flash(adap, pos - ofst, (len + 3) / 4,
+                                   (u32 *)data, 1);
+               if (ret)
+                       return ret;
+
+               len -= ofst;
+               if (copy_to_user(buf, data + ofst, len))
+                       return -EFAULT;
+
+               buf += len;
+               pos += len;
+               count -= len;
+       }
+       count = pos - *ppos;
+       *ppos = pos;
+       return count;
+}
+
+static const struct file_operations flash_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = mem_open,
+       .read    = flash_read,
+};
+
+static inline void tcamxy2valmask(u64 x, u64 y, u8 *addr, u64 *mask)
+{
+       *mask = x | y;
+       y = (__force u64)cpu_to_be64(y);
+       memcpy(addr, (char *)&y + 2, ETH_ALEN);
+}
+
+static int mps_tcam_show(struct seq_file *seq, void *v)
+{
+       if (v == SEQ_START_TOKEN)
+               seq_puts(seq, "Idx  Ethernet address     Mask     Vld Ports PF"
+                        "  VF              Replication             "
+                        "P0 P1 P2 P3  ML\n");
+       else {
+               u64 mask;
+               u8 addr[ETH_ALEN];
+               struct adapter *adap = seq->private;
+               unsigned int idx = (uintptr_t)v - 2;
+               u64 tcamy = t4_read_reg64(adap, MPS_CLS_TCAM_Y_L(idx));
+               u64 tcamx = t4_read_reg64(adap, MPS_CLS_TCAM_X_L(idx));
+               u32 cls_lo = t4_read_reg(adap, MPS_CLS_SRAM_L(idx));
+               u32 cls_hi = t4_read_reg(adap, MPS_CLS_SRAM_H(idx));
+               u32 rplc[4] = {0, 0, 0, 0};
+
+               if (tcamx & tcamy) {
+                       seq_printf(seq, "%3u         -\n", idx);
+                       goto out;
+               }
+
+               if (cls_lo & REPLICATE_F) {
+                       struct fw_ldst_cmd ldst_cmd;
+                       int ret;
+
+                       memset(&ldst_cmd, 0, sizeof(ldst_cmd));
+                       ldst_cmd.op_to_addrspace =
+                               htonl(FW_CMD_OP_V(FW_LDST_CMD) |
+                                     FW_CMD_REQUEST_F |
+                                     FW_CMD_READ_F |
+                                     FW_LDST_CMD_ADDRSPACE_V(
+                                             FW_LDST_ADDRSPC_MPS));
+                       ldst_cmd.cycles_to_len16 = htonl(FW_LEN16(ldst_cmd));
+                       ldst_cmd.u.mps.fid_ctl =
+                               htons(FW_LDST_CMD_FID_V(FW_LDST_MPS_RPLC) |
+                                     FW_LDST_CMD_CTL_V(idx));
+                       ret = t4_wr_mbox(adap, adap->mbox, &ldst_cmd,
+                                        sizeof(ldst_cmd), &ldst_cmd);
+                       if (ret)
+                               dev_warn(adap->pdev_dev, "Can't read MPS "
+                                        "replication map for idx %d: %d\n",
+                                        idx, -ret);
+                       else {
+                               rplc[0] = ntohl(ldst_cmd.u.mps.rplc31_0);
+                               rplc[1] = ntohl(ldst_cmd.u.mps.rplc63_32);
+                               rplc[2] = ntohl(ldst_cmd.u.mps.rplc95_64);
+                               rplc[3] = ntohl(ldst_cmd.u.mps.rplc127_96);
+                       }
+               }
+
+               tcamxy2valmask(tcamx, tcamy, addr, &mask);
+               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,
+                          (cls_lo & SRAM_VLD_F) ? 'Y' : 'N', PORTMAP_G(cls_hi),
+                          PF_G(cls_lo),
+                          (cls_lo & VF_VALID_F) ? VF_G(cls_lo) : -1);
+               if (cls_lo & REPLICATE_F)
+                       seq_printf(seq, " %08x %08x %08x %08x",
+                                  rplc[3], rplc[2], rplc[1], rplc[0]);
+               else
+                       seq_printf(seq, "%36c", ' ');
+               seq_printf(seq, "%4u%3u%3u%3u %#x\n",
+                          SRAM_PRIO0_G(cls_lo), SRAM_PRIO1_G(cls_lo),
+                          SRAM_PRIO2_G(cls_lo), SRAM_PRIO3_G(cls_lo),
+                          (cls_lo >> MULTILISTEN0_S) & 0xf);
+       }
+out:   return 0;
+}
+
+static inline void *mps_tcam_get_idx(struct seq_file *seq, loff_t pos)
+{
+       struct adapter *adap = seq->private;
+       int max_mac_addr = is_t4(adap->params.chip) ?
+                               NUM_MPS_CLS_SRAM_L_INSTANCES :
+                               NUM_MPS_T5_CLS_SRAM_L_INSTANCES;
+       return ((pos <= max_mac_addr) ? (void *)(uintptr_t)(pos + 1) : NULL);
+}
+
+static void *mps_tcam_start(struct seq_file *seq, loff_t *pos)
+{
+       return *pos ? mps_tcam_get_idx(seq, *pos) : SEQ_START_TOKEN;
+}
+
+static void *mps_tcam_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+       ++*pos;
+       return mps_tcam_get_idx(seq, *pos);
+}
+
+static void mps_tcam_stop(struct seq_file *seq, void *v)
+{
+}
+
+static const struct seq_operations mps_tcam_seq_ops = {
+       .start = mps_tcam_start,
+       .next  = mps_tcam_next,
+       .stop  = mps_tcam_stop,
+       .show  = mps_tcam_show
+};
+
+static int mps_tcam_open(struct inode *inode, struct file *file)
+{
+       int res = seq_open(file, &mps_tcam_seq_ops);
+
+       if (!res) {
+               struct seq_file *seq = file->private_data;
+
+               seq->private = inode->i_private;
+       }
+       return res;
+}
+
+static const struct file_operations mps_tcam_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = mps_tcam_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release,
+};
+
+/* Display various sensor information.
+ */
+static int sensors_show(struct seq_file *seq, void *v)
+{
+       struct adapter *adap = seq->private;
+       u32 param[7], val[7];
+       int ret;
+
+       /* Note that if the sensors haven't been initialized and turned on
+        * we'll get values of 0, so treat those as "<unknown>" ...
+        */
+       param[0] = (FW_PARAMS_MNEM_V(FW_PARAMS_MNEM_DEV) |
+                   FW_PARAMS_PARAM_X_V(FW_PARAMS_PARAM_DEV_DIAG) |
+                   FW_PARAMS_PARAM_Y_V(FW_PARAM_DEV_DIAG_TMP));
+       param[1] = (FW_PARAMS_MNEM_V(FW_PARAMS_MNEM_DEV) |
+                   FW_PARAMS_PARAM_X_V(FW_PARAMS_PARAM_DEV_DIAG) |
+                   FW_PARAMS_PARAM_Y_V(FW_PARAM_DEV_DIAG_VDD));
+       ret = t4_query_params(adap, adap->mbox, adap->fn, 0, 2,
+                             param, val);
+
+       if (ret < 0 || val[0] == 0)
+               seq_puts(seq, "Temperature: <unknown>\n");
+       else
+               seq_printf(seq, "Temperature: %dC\n", val[0]);
+
+       if (ret < 0 || val[1] == 0)
+               seq_puts(seq, "Core VDD:    <unknown>\n");
+       else
+               seq_printf(seq, "Core VDD:    %dmV\n", val[1]);
+
+       return 0;
+}
+
+DEFINE_SIMPLE_DEBUGFS_FILE(sensors);
+
+#if IS_ENABLED(CONFIG_IPV6)
+static int clip_tbl_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, clip_tbl_show, inode->i_private);
+}
+
+static const struct file_operations clip_tbl_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = clip_tbl_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = single_release
+};
+#endif
+
+/*RSS Table.
+ */
+
+static int rss_show(struct seq_file *seq, void *v, int idx)
+{
+       u16 *entry = v;
+
+       seq_printf(seq, "%4d:  %4u  %4u  %4u  %4u  %4u  %4u  %4u  %4u\n",
+                  idx * 8, entry[0], entry[1], entry[2], entry[3], entry[4],
+                  entry[5], entry[6], entry[7]);
+       return 0;
+}
+
+static int rss_open(struct inode *inode, struct file *file)
+{
+       int ret;
+       struct seq_tab *p;
+       struct adapter *adap = inode->i_private;
+
+       p = seq_open_tab(file, RSS_NENTRIES / 8, 8 * sizeof(u16), 0, rss_show);
+       if (!p)
+               return -ENOMEM;
+
+       ret = t4_read_rss(adap, (u16 *)p->data);
+       if (ret)
+               seq_release_private(inode, file);
+
+       return ret;
+}
+
+static const struct file_operations rss_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = rss_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private
+};
+
+/* RSS Configuration.
+ */
+
+/* Small utility function to return the strings "yes" or "no" if the supplied
+ * argument is non-zero.
+ */
+static const char *yesno(int x)
+{
+       static const char *yes = "yes";
+       static const char *no = "no";
+
+       return x ? yes : no;
+}
+
+static int rss_config_show(struct seq_file *seq, void *v)
+{
+       struct adapter *adapter = seq->private;
+       static const char * const keymode[] = {
+               "global",
+               "global and per-VF scramble",
+               "per-PF and per-VF scramble",
+               "per-VF and per-VF scramble",
+       };
+       u32 rssconf;
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_A);
+       seq_printf(seq, "TP_RSS_CONFIG: %#x\n", rssconf);
+       seq_printf(seq, "  Tnl4TupEnIpv6: %3s\n", yesno(rssconf &
+                                                       TNL4TUPENIPV6_F));
+       seq_printf(seq, "  Tnl2TupEnIpv6: %3s\n", yesno(rssconf &
+                                                       TNL2TUPENIPV6_F));
+       seq_printf(seq, "  Tnl4TupEnIpv4: %3s\n", yesno(rssconf &
+                                                       TNL4TUPENIPV4_F));
+       seq_printf(seq, "  Tnl2TupEnIpv4: %3s\n", yesno(rssconf &
+                                                       TNL2TUPENIPV4_F));
+       seq_printf(seq, "  TnlTcpSel:     %3s\n", yesno(rssconf & TNLTCPSEL_F));
+       seq_printf(seq, "  TnlIp6Sel:     %3s\n", yesno(rssconf & TNLIP6SEL_F));
+       seq_printf(seq, "  TnlVrtSel:     %3s\n", yesno(rssconf & TNLVRTSEL_F));
+       seq_printf(seq, "  TnlMapEn:      %3s\n", yesno(rssconf & TNLMAPEN_F));
+       seq_printf(seq, "  OfdHashSave:   %3s\n", yesno(rssconf &
+                                                       OFDHASHSAVE_F));
+       seq_printf(seq, "  OfdVrtSel:     %3s\n", yesno(rssconf & OFDVRTSEL_F));
+       seq_printf(seq, "  OfdMapEn:      %3s\n", yesno(rssconf & OFDMAPEN_F));
+       seq_printf(seq, "  OfdLkpEn:      %3s\n", yesno(rssconf & OFDLKPEN_F));
+       seq_printf(seq, "  Syn4TupEnIpv6: %3s\n", yesno(rssconf &
+                                                       SYN4TUPENIPV6_F));
+       seq_printf(seq, "  Syn2TupEnIpv6: %3s\n", yesno(rssconf &
+                                                       SYN2TUPENIPV6_F));
+       seq_printf(seq, "  Syn4TupEnIpv4: %3s\n", yesno(rssconf &
+                                                       SYN4TUPENIPV4_F));
+       seq_printf(seq, "  Syn2TupEnIpv4: %3s\n", yesno(rssconf &
+                                                       SYN2TUPENIPV4_F));
+       seq_printf(seq, "  Syn4TupEnIpv6: %3s\n", yesno(rssconf &
+                                                       SYN4TUPENIPV6_F));
+       seq_printf(seq, "  SynIp6Sel:     %3s\n", yesno(rssconf & SYNIP6SEL_F));
+       seq_printf(seq, "  SynVrt6Sel:    %3s\n", yesno(rssconf & SYNVRTSEL_F));
+       seq_printf(seq, "  SynMapEn:      %3s\n", yesno(rssconf & SYNMAPEN_F));
+       seq_printf(seq, "  SynLkpEn:      %3s\n", yesno(rssconf & SYNLKPEN_F));
+       seq_printf(seq, "  ChnEn:         %3s\n", yesno(rssconf &
+                                                       CHANNELENABLE_F));
+       seq_printf(seq, "  PrtEn:         %3s\n", yesno(rssconf &
+                                                       PORTENABLE_F));
+       seq_printf(seq, "  TnlAllLkp:     %3s\n", yesno(rssconf &
+                                                       TNLALLLOOKUP_F));
+       seq_printf(seq, "  VrtEn:         %3s\n", yesno(rssconf &
+                                                       VIRTENABLE_F));
+       seq_printf(seq, "  CngEn:         %3s\n", yesno(rssconf &
+                                                       CONGESTIONENABLE_F));
+       seq_printf(seq, "  HashToeplitz:  %3s\n", yesno(rssconf &
+                                                       HASHTOEPLITZ_F));
+       seq_printf(seq, "  Udp4En:        %3s\n", yesno(rssconf & UDPENABLE_F));
+       seq_printf(seq, "  Disable:       %3s\n", yesno(rssconf & DISABLE_F));
+
+       seq_puts(seq, "\n");
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_TNL_A);
+       seq_printf(seq, "TP_RSS_CONFIG_TNL: %#x\n", rssconf);
+       seq_printf(seq, "  MaskSize:      %3d\n", MASKSIZE_G(rssconf));
+       seq_printf(seq, "  MaskFilter:    %3d\n", MASKFILTER_G(rssconf));
+       if (CHELSIO_CHIP_VERSION(adapter->params.chip) > CHELSIO_T5) {
+               seq_printf(seq, "  HashAll:     %3s\n",
+                          yesno(rssconf & HASHALL_F));
+               seq_printf(seq, "  HashEth:     %3s\n",
+                          yesno(rssconf & HASHETH_F));
+       }
+       seq_printf(seq, "  UseWireCh:     %3s\n", yesno(rssconf & USEWIRECH_F));
+
+       seq_puts(seq, "\n");
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_OFD_A);
+       seq_printf(seq, "TP_RSS_CONFIG_OFD: %#x\n", rssconf);
+       seq_printf(seq, "  MaskSize:      %3d\n", MASKSIZE_G(rssconf));
+       seq_printf(seq, "  RRCplMapEn:    %3s\n", yesno(rssconf &
+                                                       RRCPLMAPEN_F));
+       seq_printf(seq, "  RRCplQueWidth: %3d\n", RRCPLQUEWIDTH_G(rssconf));
+
+       seq_puts(seq, "\n");
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_SYN_A);
+       seq_printf(seq, "TP_RSS_CONFIG_SYN: %#x\n", rssconf);
+       seq_printf(seq, "  MaskSize:      %3d\n", MASKSIZE_G(rssconf));
+       seq_printf(seq, "  UseWireCh:     %3s\n", yesno(rssconf & USEWIRECH_F));
+
+       seq_puts(seq, "\n");
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_VRT_A);
+       seq_printf(seq, "TP_RSS_CONFIG_VRT: %#x\n", rssconf);
+       if (CHELSIO_CHIP_VERSION(adapter->params.chip) > CHELSIO_T5) {
+               seq_printf(seq, "  KeyWrAddrX:     %3d\n",
+                          KEYWRADDRX_G(rssconf));
+               seq_printf(seq, "  KeyExtend:      %3s\n",
+                          yesno(rssconf & KEYEXTEND_F));
+       }
+       seq_printf(seq, "  VfRdRg:        %3s\n", yesno(rssconf & VFRDRG_F));
+       seq_printf(seq, "  VfRdEn:        %3s\n", yesno(rssconf & VFRDEN_F));
+       seq_printf(seq, "  VfPerrEn:      %3s\n", yesno(rssconf & VFPERREN_F));
+       seq_printf(seq, "  KeyPerrEn:     %3s\n", yesno(rssconf & KEYPERREN_F));
+       seq_printf(seq, "  DisVfVlan:     %3s\n", yesno(rssconf &
+                                                       DISABLEVLAN_F));
+       seq_printf(seq, "  EnUpSwt:       %3s\n", yesno(rssconf & ENABLEUP0_F));
+       seq_printf(seq, "  HashDelay:     %3d\n", HASHDELAY_G(rssconf));
+       if (CHELSIO_CHIP_VERSION(adapter->params.chip) <= CHELSIO_T5)
+               seq_printf(seq, "  VfWrAddr:      %3d\n", VFWRADDR_G(rssconf));
+       seq_printf(seq, "  KeyMode:       %s\n", keymode[KEYMODE_G(rssconf)]);
+       seq_printf(seq, "  VfWrEn:        %3s\n", yesno(rssconf & VFWREN_F));
+       seq_printf(seq, "  KeyWrEn:       %3s\n", yesno(rssconf & KEYWREN_F));
+       seq_printf(seq, "  KeyWrAddr:     %3d\n", KEYWRADDR_G(rssconf));
+
+       seq_puts(seq, "\n");
+
+       rssconf = t4_read_reg(adapter, TP_RSS_CONFIG_CNG_A);
+       seq_printf(seq, "TP_RSS_CONFIG_CNG: %#x\n", rssconf);
+       seq_printf(seq, "  ChnCount3:     %3s\n", yesno(rssconf & CHNCOUNT3_F));
+       seq_printf(seq, "  ChnCount2:     %3s\n", yesno(rssconf & CHNCOUNT2_F));
+       seq_printf(seq, "  ChnCount1:     %3s\n", yesno(rssconf & CHNCOUNT1_F));
+       seq_printf(seq, "  ChnCount0:     %3s\n", yesno(rssconf & CHNCOUNT0_F));
+       seq_printf(seq, "  ChnUndFlow3:   %3s\n", yesno(rssconf &
+                                                       CHNUNDFLOW3_F));
+       seq_printf(seq, "  ChnUndFlow2:   %3s\n", yesno(rssconf &
+                                                       CHNUNDFLOW2_F));
+       seq_printf(seq, "  ChnUndFlow1:   %3s\n", yesno(rssconf &
+                                                       CHNUNDFLOW1_F));
+       seq_printf(seq, "  ChnUndFlow0:   %3s\n", yesno(rssconf &
+                                                       CHNUNDFLOW0_F));
+       seq_printf(seq, "  RstChn3:       %3s\n", yesno(rssconf & RSTCHN3_F));
+       seq_printf(seq, "  RstChn2:       %3s\n", yesno(rssconf & RSTCHN2_F));
+       seq_printf(seq, "  RstChn1:       %3s\n", yesno(rssconf & RSTCHN1_F));
+       seq_printf(seq, "  RstChn0:       %3s\n", yesno(rssconf & RSTCHN0_F));
+       seq_printf(seq, "  UpdVld:        %3s\n", yesno(rssconf & UPDVLD_F));
+       seq_printf(seq, "  Xoff:          %3s\n", yesno(rssconf & XOFF_F));
+       seq_printf(seq, "  UpdChn3:       %3s\n", yesno(rssconf & UPDCHN3_F));
+       seq_printf(seq, "  UpdChn2:       %3s\n", yesno(rssconf & UPDCHN2_F));
+       seq_printf(seq, "  UpdChn1:       %3s\n", yesno(rssconf & UPDCHN1_F));
+       seq_printf(seq, "  UpdChn0:       %3s\n", yesno(rssconf & UPDCHN0_F));
+       seq_printf(seq, "  Queue:         %3d\n", QUEUE_G(rssconf));
+
+       return 0;
+}
+
+DEFINE_SIMPLE_DEBUGFS_FILE(rss_config);
+
+/* RSS Secret Key.
+ */
+
+static int rss_key_show(struct seq_file *seq, void *v)
+{
+       u32 key[10];
+
+       t4_read_rss_key(seq->private, key);
+       seq_printf(seq, "%08x%08x%08x%08x%08x%08x%08x%08x%08x%08x\n",
+                  key[9], key[8], key[7], key[6], key[5], key[4], key[3],
+                  key[2], key[1], key[0]);
+       return 0;
+}
+
+static int rss_key_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, rss_key_show, inode->i_private);
+}
+
+static ssize_t rss_key_write(struct file *file, const char __user *buf,
+                            size_t count, loff_t *pos)
+{
+       int i, j;
+       u32 key[10];
+       char s[100], *p;
+       struct adapter *adap = file_inode(file)->i_private;
+
+       if (count > sizeof(s) - 1)
+               return -EINVAL;
+       if (copy_from_user(s, buf, count))
+               return -EFAULT;
+       for (i = count; i > 0 && isspace(s[i - 1]); i--)
+               ;
+       s[i] = '\0';
+
+       for (p = s, i = 9; i >= 0; i--) {
+               key[i] = 0;
+               for (j = 0; j < 8; j++, p++) {
+                       if (!isxdigit(*p))
+                               return -EINVAL;
+                       key[i] = (key[i] << 4) | hex2val(*p);
+               }
+       }
+
+       t4_write_rss_key(adap, key, -1);
+       return count;
+}
+
+static const struct file_operations rss_key_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = rss_key_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = single_release,
+       .write   = rss_key_write
+};
+
+/* PF RSS Configuration.
+ */
+
+struct rss_pf_conf {
+       u32 rss_pf_map;
+       u32 rss_pf_mask;
+       u32 rss_pf_config;
+};
+
+static int rss_pf_config_show(struct seq_file *seq, void *v, int idx)
+{
+       struct rss_pf_conf *pfconf;
+
+       if (v == SEQ_START_TOKEN) {
+               /* use the 0th entry to dump the PF Map Index Size */
+               pfconf = seq->private + offsetof(struct seq_tab, data);
+               seq_printf(seq, "PF Map Index Size = %d\n\n",
+                          LKPIDXSIZE_G(pfconf->rss_pf_map));
+
+               seq_puts(seq, "     RSS              PF   VF    Hash Tuple Enable         Default\n");
+               seq_puts(seq, "     Enable       IPF Mask Mask  IPv6      IPv4      UDP   Queue\n");
+               seq_puts(seq, " PF  Map Chn Prt  Map Size Size  Four Two  Four Two  Four  Ch1  Ch0\n");
+       } else {
+               #define G_PFnLKPIDX(map, n) \
+                       (((map) >> PF1LKPIDX_S*(n)) & PF0LKPIDX_M)
+               #define G_PFnMSKSIZE(mask, n) \
+                       (((mask) >> PF1MSKSIZE_S*(n)) & PF1MSKSIZE_M)
+
+               pfconf = v;
+               seq_printf(seq, "%3d  %3s %3s %3s  %3d  %3d  %3d   %3s %3s   %3s %3s   %3s  %3d  %3d\n",
+                          idx,
+                          yesno(pfconf->rss_pf_config & MAPENABLE_F),
+                          yesno(pfconf->rss_pf_config & CHNENABLE_F),
+                          yesno(pfconf->rss_pf_config & PRTENABLE_F),
+                          G_PFnLKPIDX(pfconf->rss_pf_map, idx),
+                          G_PFnMSKSIZE(pfconf->rss_pf_mask, idx),
+                          IVFWIDTH_G(pfconf->rss_pf_config),
+                          yesno(pfconf->rss_pf_config & IP6FOURTUPEN_F),
+                          yesno(pfconf->rss_pf_config & IP6TWOTUPEN_F),
+                          yesno(pfconf->rss_pf_config & IP4FOURTUPEN_F),
+                          yesno(pfconf->rss_pf_config & IP4TWOTUPEN_F),
+                          yesno(pfconf->rss_pf_config & UDPFOURTUPEN_F),
+                          CH1DEFAULTQUEUE_G(pfconf->rss_pf_config),
+                          CH0DEFAULTQUEUE_G(pfconf->rss_pf_config));
+
+               #undef G_PFnLKPIDX
+               #undef G_PFnMSKSIZE
+       }
+       return 0;
+}
+
+static int rss_pf_config_open(struct inode *inode, struct file *file)
+{
+       struct adapter *adapter = inode->i_private;
+       struct seq_tab *p;
+       u32 rss_pf_map, rss_pf_mask;
+       struct rss_pf_conf *pfconf;
+       int pf;
+
+       p = seq_open_tab(file, 8, sizeof(*pfconf), 1, rss_pf_config_show);
+       if (!p)
+               return -ENOMEM;
+
+       pfconf = (struct rss_pf_conf *)p->data;
+       rss_pf_map = t4_read_rss_pf_map(adapter);
+       rss_pf_mask = t4_read_rss_pf_mask(adapter);
+       for (pf = 0; pf < 8; pf++) {
+               pfconf[pf].rss_pf_map = rss_pf_map;
+               pfconf[pf].rss_pf_mask = rss_pf_mask;
+               t4_read_rss_pf_config(adapter, pf, &pfconf[pf].rss_pf_config);
+       }
+       return 0;
+}
+
+static const struct file_operations rss_pf_config_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = rss_pf_config_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private
+};
+
+/* VF RSS Configuration.
+ */
+
+struct rss_vf_conf {
+       u32 rss_vf_vfl;
+       u32 rss_vf_vfh;
+};
+
+static int rss_vf_config_show(struct seq_file *seq, void *v, int idx)
+{
+       if (v == SEQ_START_TOKEN) {
+               seq_puts(seq, "     RSS                     Hash Tuple Enable\n");
+               seq_puts(seq, "     Enable   IVF  Dis  Enb  IPv6      IPv4      UDP    Def  Secret Key\n");
+               seq_puts(seq, " VF  Chn Prt  Map  VLAN  uP  Four Two  Four Two  Four   Que  Idx       Hash\n");
+       } else {
+               struct rss_vf_conf *vfconf = v;
+
+               seq_printf(seq, "%3d  %3s %3s  %3d   %3s %3s   %3s %3s   %3s  %3s   %3s  %4d  %3d %#10x\n",
+                          idx,
+                          yesno(vfconf->rss_vf_vfh & VFCHNEN_F),
+                          yesno(vfconf->rss_vf_vfh & VFPRTEN_F),
+                          VFLKPIDX_G(vfconf->rss_vf_vfh),
+                          yesno(vfconf->rss_vf_vfh & VFVLNEX_F),
+                          yesno(vfconf->rss_vf_vfh & VFUPEN_F),
+                          yesno(vfconf->rss_vf_vfh & VFIP4FOURTUPEN_F),
+                          yesno(vfconf->rss_vf_vfh & VFIP6TWOTUPEN_F),
+                          yesno(vfconf->rss_vf_vfh & VFIP4FOURTUPEN_F),
+                          yesno(vfconf->rss_vf_vfh & VFIP4TWOTUPEN_F),
+                          yesno(vfconf->rss_vf_vfh & ENABLEUDPHASH_F),
+                          DEFAULTQUEUE_G(vfconf->rss_vf_vfh),
+                          KEYINDEX_G(vfconf->rss_vf_vfh),
+                          vfconf->rss_vf_vfl);
+       }
+       return 0;
+}
+
+static int rss_vf_config_open(struct inode *inode, struct file *file)
+{
+       struct adapter *adapter = inode->i_private;
+       struct seq_tab *p;
+       struct rss_vf_conf *vfconf;
+       int vf;
+
+       p = seq_open_tab(file, 128, sizeof(*vfconf), 1, rss_vf_config_show);
+       if (!p)
+               return -ENOMEM;
+
+       vfconf = (struct rss_vf_conf *)p->data;
+       for (vf = 0; vf < 128; vf++) {
+               t4_read_rss_vf_config(adapter, vf, &vfconf[vf].rss_vf_vfl,
+                                     &vfconf[vf].rss_vf_vfh);
+       }
+       return 0;
+}
+
+static const struct file_operations rss_vf_config_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = rss_vf_config_open,
+       .read    = seq_read,
+       .llseek  = seq_lseek,
+       .release = seq_release_private
+};
+
+/**
+ * ethqset2pinfo - return port_info of an Ethernet Queue Set
+ * @adap: the adapter
+ * @qset: Ethernet Queue Set
+ */
+static inline struct port_info *ethqset2pinfo(struct adapter *adap, int qset)
+{
+       int pidx;
+
+       for_each_port(adap, pidx) {
+               struct port_info *pi = adap2pinfo(adap, pidx);
+
+               if (qset >= pi->first_qset &&
+                   qset < pi->first_qset + pi->nqsets)
+                       return pi;
+       }
+
+       /* should never happen! */
+       BUG_ON(1);
+       return NULL;
+}
+
+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_CHELSIO_T4_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");
+               S("Interface:",
+                 rx[i].rspq.netdev ? rx[i].rspq.netdev->name : "N/A");
+               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");
+               S("Interface:",
+                 rx[i].rspq.netdev ? rx[i].rspq.netdev->name : "N/A");
+               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;
+       struct adapter *adap;
+
+       file->private_data = inode->i_private;
+
+       mem = (uintptr_t)file->private_data & 0x3;
+       adap = file->private_data - mem;
+
+       (void)t4_fwcache(adap, FW_PARAM_DEV_FWCACHE_FLUSH);
+
+       return 0;
+}
+
+static ssize_t mem_read(struct file *file, char __user *buf, size_t count,
+                       loff_t *ppos)
+{
+       loff_t pos = *ppos;
+       loff_t avail = file_inode(file)->i_size;
+       unsigned int mem = (uintptr_t)file->private_data & 3;
+       struct adapter *adap = file->private_data - mem;
+       __be32 *data;
+       int ret;
+
+       if (pos < 0)
+               return -EINVAL;
+       if (pos >= avail)
+               return 0;
+       if (count > avail - pos)
+               count = avail - pos;
+
+       data = t4_alloc_mem(count);
+       if (!data)
+               return -ENOMEM;
+
+       spin_lock(&adap->win0_lock);
+       ret = t4_memory_rw(adap, 0, mem, pos, count, data, T4_MEMORY_READ);
+       spin_unlock(&adap->win0_lock);
+       if (ret) {
+               t4_free_mem(data);
+               return ret;
+       }
+       ret = copy_to_user(buf, data, count);
+
+       t4_free_mem(data);
+       if (ret)
+               return -EFAULT;
+
+       *ppos = pos + count;
+       return count;
+}
+static const struct file_operations mem_debugfs_fops = {
+       .owner   = THIS_MODULE,
+       .open    = simple_open,
+       .read    = mem_read,
+       .llseek  = default_llseek,
+};
+
+static void add_debugfs_mem(struct adapter *adap, const char *name,
+                           unsigned int idx, unsigned int size_mb)
+{
+       debugfs_create_file_size(name, S_IRUSR, adap->debugfs_root,
+                                (void *)adap + idx, &mem_debugfs_fops,
+                                size_mb << 20);
+}
+
+/* Add an array of Debug FS files.
+ */
+void add_debugfs_files(struct adapter *adap,
+                      struct t4_debugfs_entry *files,
+                      unsigned int nfiles)
+{
+       int i;
+
+       /* debugfs support is best effort */
+       for (i = 0; i < nfiles; i++)
+               debugfs_create_file(files[i].name, files[i].mode,
+                                   adap->debugfs_root,
+                                   (void *)adap + files[i].data,
+                                   files[i].ops);
+}
+
+int t4_setup_debugfs(struct adapter *adap)
+{
+       int i;
+       u32 size;
+       struct dentry *de;
+
+       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 },
+               { "mbox0", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 0 },
+               { "mbox1", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 1 },
+               { "mbox2", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 2 },
+               { "mbox3", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 3 },
+               { "mbox4", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 4 },
+               { "mbox5", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 5 },
+               { "mbox6", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 6 },
+               { "mbox7", &mbox_debugfs_fops, S_IRUSR | S_IWUSR, 7 },
+               { "l2t", &t4_l2t_fops, S_IRUSR, 0},
+               { "mps_tcam", &mps_tcam_debugfs_fops, S_IRUSR, 0 },
+               { "rss", &rss_debugfs_fops, S_IRUSR, 0 },
+               { "rss_config", &rss_config_debugfs_fops, S_IRUSR, 0 },
+               { "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 },
+               { "tp_la", &tp_la_fops, S_IRUSR, 0 },
+               { "ulprx_la", &ulprx_la_fops, S_IRUSR, 0 },
+               { "sensors", &sensors_debugfs_fops, S_IRUSR, 0 },
+               { "pm_stats", &pm_stats_debugfs_fops, S_IRUSR, 0 },
+               { "cctrl", &cctrl_tbl_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) {
+               size = t4_read_reg(adap, MA_EDRAM0_BAR_A);
+               add_debugfs_mem(adap, "edc0", MEM_EDC0, EDRAM0_SIZE_G(size));
+       }
+       if (i & EDRAM1_ENABLE_F) {
+               size = t4_read_reg(adap, MA_EDRAM1_BAR_A);
+               add_debugfs_mem(adap, "edc1", MEM_EDC1, EDRAM1_SIZE_G(size));
+       }
+       if (is_t4(adap->params.chip)) {
+               size = t4_read_reg(adap, MA_EXT_MEMORY_BAR_A);
+               if (i & EXT_MEM_ENABLE_F)
+                       add_debugfs_mem(adap, "mc", MEM_MC,
+                                       EXT_MEM_SIZE_G(size));
+       } else {
+               if (i & EXT_MEM0_ENABLE_F) {
+                       size = t4_read_reg(adap, MA_EXT_MEMORY0_BAR_A);
+                       add_debugfs_mem(adap, "mc0", MEM_MC0,
+                                       EXT_MEM0_SIZE_G(size));
+               }
+               if (i & EXT_MEM1_ENABLE_F) {
+                       size = t4_read_reg(adap, MA_EXT_MEMORY1_BAR_A);
+                       add_debugfs_mem(adap, "mc1", MEM_MC1,
+                                       EXT_MEM1_SIZE_G(size));
+               }
+       }
+
+       de = debugfs_create_file_size("flash", S_IRUSR, adap->debugfs_root, adap,
+                                     &flash_debugfs_fops, adap->params.sf_size);
+
+       return 0;
+}