2 * Copyright(c) 2018 Intel Corporation.
4 * This file is provided under a dual BSD/GPLv2 license. When using or
5 * redistributing this file, you may do so under either license.
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of version 2 of the GNU General Public License as
11 * published by the Free Software Foundation.
13 * This program is distributed in the hope that it will be useful, but
14 * WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * General Public License for more details.
20 * Redistribution and use in source and binary forms, with or without
21 * modification, are permitted provided that the following conditions
24 * - Redistributions of source code must retain the above copyright
25 * notice, this list of conditions and the following disclaimer.
26 * - Redistributions in binary form must reproduce the above copyright
27 * notice, this list of conditions and the following disclaimer in
28 * the documentation and/or other materials provided with the
30 * - Neither the name of Intel Corporation nor the names of its
31 * contributors may be used to endorse or promote products derived
32 * from this software without specific prior written permission.
34 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
35 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
36 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
37 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
38 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
39 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
40 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
41 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
42 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
43 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
44 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
47 #include <linux/debugfs.h>
48 #include <linux/seq_file.h>
49 #include <linux/kernel.h>
50 #include <linux/module.h>
51 #include <linux/types.h>
52 #include <linux/bitmap.h>
58 #define HFI1_FAULT_DIR_TX BIT(0)
59 #define HFI1_FAULT_DIR_RX BIT(1)
60 #define HFI1_FAULT_DIR_TXRX (HFI1_FAULT_DIR_TX | HFI1_FAULT_DIR_RX)
62 static void *_fault_stats_seq_start(struct seq_file *s, loff_t *pos)
64 struct hfi1_opcode_stats_perctx *opstats;
66 if (*pos >= ARRAY_SIZE(opstats->stats))
71 static void *_fault_stats_seq_next(struct seq_file *s, void *v, loff_t *pos)
73 struct hfi1_opcode_stats_perctx *opstats;
76 if (*pos >= ARRAY_SIZE(opstats->stats))
81 static void _fault_stats_seq_stop(struct seq_file *s, void *v)
85 static int _fault_stats_seq_show(struct seq_file *s, void *v)
89 u64 n_packets = 0, n_bytes = 0;
90 struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private;
91 struct hfi1_devdata *dd = dd_from_dev(ibd);
92 struct hfi1_ctxtdata *rcd;
94 for (j = 0; j < dd->first_dyn_alloc_ctxt; j++) {
95 rcd = hfi1_rcd_get_by_index(dd, j);
97 n_packets += rcd->opstats->stats[i].n_packets;
98 n_bytes += rcd->opstats->stats[i].n_bytes;
102 for_each_possible_cpu(j) {
103 struct hfi1_opcode_stats_perctx *sp =
104 per_cpu_ptr(dd->tx_opstats, j);
106 n_packets += sp->stats[i].n_packets;
107 n_bytes += sp->stats[i].n_bytes;
109 if (!n_packets && !n_bytes)
111 if (!ibd->fault->n_rxfaults[i] && !ibd->fault->n_txfaults[i])
113 seq_printf(s, "%02llx %llu/%llu (faults rx:%llu faults: tx:%llu)\n", i,
114 (unsigned long long)n_packets,
115 (unsigned long long)n_bytes,
116 (unsigned long long)ibd->fault->n_rxfaults[i],
117 (unsigned long long)ibd->fault->n_txfaults[i]);
121 DEBUGFS_SEQ_FILE_OPS(fault_stats);
122 DEBUGFS_SEQ_FILE_OPEN(fault_stats);
123 DEBUGFS_FILE_OPS(fault_stats);
125 static int fault_opcodes_open(struct inode *inode, struct file *file)
127 file->private_data = inode->i_private;
128 return nonseekable_open(inode, file);
131 static ssize_t fault_opcodes_write(struct file *file, const char __user *buf,
132 size_t len, loff_t *pos)
135 /* 1280 = 256 opcodes * 4 chars/opcode + 255 commas + NULL */
136 size_t copy, datalen = 1280;
137 char *data, *token, *ptr, *end;
138 struct fault *fault = file->private_data;
140 data = kcalloc(datalen, sizeof(*data), GFP_KERNEL);
143 copy = min(len, datalen - 1);
144 if (copy_from_user(data, buf, copy))
147 ret = debugfs_file_get(file->f_path.dentry);
152 for (ptr = data; *ptr; ptr = end + 1, token = ptr) {
154 unsigned long range_start, range_end, i;
156 unsigned long bound = 1U << BITS_PER_BYTE;
158 end = strchr(ptr, ',');
161 if (token[0] == '-') {
165 dash = strchr(token, '-');
168 if (kstrtoul(token, 0, &range_start))
172 if (kstrtoul(token, 0, &range_end))
175 range_end = range_start;
177 if (range_start == range_end && range_start == -1UL) {
178 bitmap_zero(fault->opcodes, sizeof(fault->opcodes) *
182 /* Check the inputs */
183 if (range_start >= bound || range_end >= bound)
186 for (i = range_start; i <= range_end; i++) {
188 clear_bit(i, fault->opcodes);
190 set_bit(i, fault->opcodes);
197 debugfs_file_put(file->f_path.dentry);
202 static ssize_t fault_opcodes_read(struct file *file, char __user *buf,
203 size_t len, loff_t *pos)
207 size_t datalen = 1280, size = 0; /* see fault_opcodes_write() */
208 unsigned long bit = 0, zero = 0;
209 struct fault *fault = file->private_data;
210 size_t bitsize = sizeof(fault->opcodes) * BITS_PER_BYTE;
212 data = kcalloc(datalen, sizeof(*data), GFP_KERNEL);
215 ret = debugfs_file_get(file->f_path.dentry);
218 bit = find_first_bit(fault->opcodes, bitsize);
219 while (bit < bitsize) {
220 zero = find_next_zero_bit(fault->opcodes, bitsize, bit);
222 size += snprintf(data + size,
224 "0x%lx-0x%lx,", bit, zero - 1);
226 size += snprintf(data + size,
227 datalen - size - 1, "0x%lx,",
229 bit = find_next_bit(fault->opcodes, bitsize, zero);
231 debugfs_file_put(file->f_path.dentry);
232 data[size - 1] = '\n';
234 ret = simple_read_from_buffer(buf, len, pos, data, size);
239 static const struct file_operations __fault_opcodes_fops = {
240 .owner = THIS_MODULE,
241 .open = fault_opcodes_open,
242 .read = fault_opcodes_read,
243 .write = fault_opcodes_write,
247 void hfi1_fault_exit_debugfs(struct hfi1_ibdev *ibd)
250 debugfs_remove_recursive(ibd->fault->dir);
255 int hfi1_fault_init_debugfs(struct hfi1_ibdev *ibd)
257 struct dentry *parent = ibd->hfi1_ibdev_dbg;
258 struct dentry *fault_dir;
260 ibd->fault = kzalloc(sizeof(*ibd->fault), GFP_KERNEL);
264 ibd->fault->attr.interval = 1;
265 ibd->fault->attr.require_end = ULONG_MAX;
266 ibd->fault->attr.stacktrace_depth = 32;
267 ibd->fault->attr.dname = NULL;
268 ibd->fault->attr.verbose = 0;
269 ibd->fault->enable = false;
270 ibd->fault->opcode = false;
271 ibd->fault->fault_skip = 0;
272 ibd->fault->skip = 0;
273 ibd->fault->direction = HFI1_FAULT_DIR_TXRX;
274 ibd->fault->suppress_err = false;
275 bitmap_zero(ibd->fault->opcodes,
276 sizeof(ibd->fault->opcodes) * BITS_PER_BYTE);
279 fault_create_debugfs_attr("fault", parent, &ibd->fault->attr);
280 if (IS_ERR(fault_dir)) {
285 ibd->fault->dir = fault_dir;
287 debugfs_create_file("fault_stats", 0444, fault_dir, ibd,
288 &_fault_stats_file_ops);
289 debugfs_create_bool("enable", 0600, fault_dir, &ibd->fault->enable);
290 debugfs_create_bool("suppress_err", 0600, fault_dir,
291 &ibd->fault->suppress_err);
292 debugfs_create_bool("opcode_mode", 0600, fault_dir,
293 &ibd->fault->opcode);
294 debugfs_create_file("opcodes", 0600, fault_dir, ibd->fault,
295 &__fault_opcodes_fops);
296 debugfs_create_u64("skip_pkts", 0600, fault_dir,
297 &ibd->fault->fault_skip);
298 debugfs_create_u64("skip_usec", 0600, fault_dir,
299 &ibd->fault->fault_skip_usec);
300 debugfs_create_u8("direction", 0600, fault_dir, &ibd->fault->direction);
305 bool hfi1_dbg_fault_suppress_err(struct hfi1_ibdev *ibd)
308 return ibd->fault->suppress_err;
312 static bool __hfi1_should_fault(struct hfi1_ibdev *ibd, u32 opcode,
317 if (!ibd->fault || !ibd->fault->enable)
319 if (!(ibd->fault->direction & direction))
321 if (ibd->fault->opcode) {
322 if (bitmap_empty(ibd->fault->opcodes,
323 (sizeof(ibd->fault->opcodes) *
326 if (!(test_bit(opcode, ibd->fault->opcodes)))
329 if (ibd->fault->fault_skip_usec &&
330 time_before(jiffies, ibd->fault->skip_usec))
332 if (ibd->fault->fault_skip && ibd->fault->skip) {
336 ret = should_fail(&ibd->fault->attr, 1);
338 ibd->fault->skip = ibd->fault->fault_skip;
339 ibd->fault->skip_usec = jiffies +
340 usecs_to_jiffies(ibd->fault->fault_skip_usec);
345 bool hfi1_dbg_should_fault_tx(struct rvt_qp *qp, u32 opcode)
347 struct hfi1_ibdev *ibd = to_idev(qp->ibqp.device);
349 if (__hfi1_should_fault(ibd, opcode, HFI1_FAULT_DIR_TX)) {
350 trace_hfi1_fault_opcode(qp, opcode);
351 ibd->fault->n_txfaults[opcode]++;
357 bool hfi1_dbg_should_fault_rx(struct hfi1_packet *packet)
359 struct hfi1_ibdev *ibd = &packet->rcd->dd->verbs_dev;
361 if (__hfi1_should_fault(ibd, packet->opcode, HFI1_FAULT_DIR_RX)) {
362 trace_hfi1_fault_packet(packet);
363 ibd->fault->n_rxfaults[packet->opcode]++;