?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
mmc: kconfig: remove EXPERIMENTAL from the DMA selection of atmel-mci
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / drivers / net / myri10ge / myri10ge.c
blobbf84849600ce6e141dc7a6c84579f2ea6c315eaa
1 /*************************************************************************
2 * myri10ge.c: Myricom Myri-10G Ethernet driver.
3 *
4 * Copyright (C) 2005 - 2009 Myricom, Inc.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of Myricom, Inc. nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
20 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
23 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 *
31 *
32 * If the eeprom on your board is not recent enough, you will need to get a
33 * newer firmware image at:
34 * http://www.myri.com/scs/download-Myri10GE.html
35 *
36 * Contact Information:
37 * <help@myri.com>
38 * Myricom, Inc., 325N Santa Anita Avenue, Arcadia, CA 91006
39 *************************************************************************/
40
41 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
42
43 #include <linux/tcp.h>
44 #include <linux/netdevice.h>
45 #include <linux/skbuff.h>
46 #include <linux/string.h>
47 #include <linux/module.h>
48 #include <linux/pci.h>
49 #include <linux/dma-mapping.h>
50 #include <linux/etherdevice.h>
51 #include <linux/if_ether.h>
52 #include <linux/if_vlan.h>
53 #include <linux/inet_lro.h>
54 #include <linux/dca.h>
55 #include <linux/ip.h>
56 #include <linux/inet.h>
57 #include <linux/in.h>
58 #include <linux/ethtool.h>
59 #include <linux/firmware.h>
60 #include <linux/delay.h>
61 #include <linux/timer.h>
62 #include <linux/vmalloc.h>
63 #include <linux/crc32.h>
64 #include <linux/moduleparam.h>
65 #include <linux/io.h>
66 #include <linux/log2.h>
67 #include <linux/slab.h>
68 #include <linux/prefetch.h>
69 #include <net/checksum.h>
70 #include <net/ip.h>
71 #include <net/tcp.h>
72 #include <asm/byteorder.h>
73 #include <asm/io.h>
74 #include <asm/processor.h>
75 #ifdef CONFIG_MTRR
76 #include <asm/mtrr.h>
77 #endif
78
79 #include "myri10ge_mcp.h"
80 #include "myri10ge_mcp_gen_header.h"
81
82 #define MYRI10GE_VERSION_STR "1.5.2-1.459"
83
84 MODULE_DESCRIPTION("Myricom 10G driver (10GbE)");
85 MODULE_AUTHOR("Maintainer: help@myri.com");
86 MODULE_VERSION(MYRI10GE_VERSION_STR);
87 MODULE_LICENSE("Dual BSD/GPL");
88
89 #define MYRI10GE_MAX_ETHER_MTU 9014
90
91 #define MYRI10GE_ETH_STOPPED 0
92 #define MYRI10GE_ETH_STOPPING 1
93 #define MYRI10GE_ETH_STARTING 2
94 #define MYRI10GE_ETH_RUNNING 3
95 #define MYRI10GE_ETH_OPEN_FAILED 4
96
97 #define MYRI10GE_EEPROM_STRINGS_SIZE 256
98 #define MYRI10GE_MAX_SEND_DESC_TSO ((65536 / 2048) * 2)
99 #define MYRI10GE_MAX_LRO_DESCRIPTORS 8
100 #define MYRI10GE_LRO_MAX_PKTS 64
101
102 #define MYRI10GE_NO_CONFIRM_DATA htonl(0xffffffff)
103 #define MYRI10GE_NO_RESPONSE_RESULT 0xffffffff
104
105 #define MYRI10GE_ALLOC_ORDER 0
106 #define MYRI10GE_ALLOC_SIZE ((1 << MYRI10GE_ALLOC_ORDER) * PAGE_SIZE)
107 #define MYRI10GE_MAX_FRAGS_PER_FRAME (MYRI10GE_MAX_ETHER_MTU/MYRI10GE_ALLOC_SIZE + 1)
108
109 #define MYRI10GE_MAX_SLICES 32
110
111 struct myri10ge_rx_buffer_state {
112 struct page *page;
113 int page_offset;
114 DEFINE_DMA_UNMAP_ADDR(bus);
115 DEFINE_DMA_UNMAP_LEN(len);
116 };
117
118 struct myri10ge_tx_buffer_state {
119 struct sk_buff *skb;
120 int last;
121 DEFINE_DMA_UNMAP_ADDR(bus);
122 DEFINE_DMA_UNMAP_LEN(len);
123 };
124
125 struct myri10ge_cmd {
126 u32 data0;
127 u32 data1;
128 u32 data2;
129 };
130
131 struct myri10ge_rx_buf {
132 struct mcp_kreq_ether_recv __iomem *lanai; /* lanai ptr for recv ring */
133 struct mcp_kreq_ether_recv *shadow; /* host shadow of recv ring */
134 struct myri10ge_rx_buffer_state *info;
135 struct page *page;
136 dma_addr_t bus;
137 int page_offset;
138 int cnt;
139 int fill_cnt;
140 int alloc_fail;
141 int mask; /* number of rx slots -1 */
142 int watchdog_needed;
143 };
144
145 struct myri10ge_tx_buf {
146 struct mcp_kreq_ether_send __iomem *lanai; /* lanai ptr for sendq */
147 __be32 __iomem *send_go; /* "go" doorbell ptr */
148 __be32 __iomem *send_stop; /* "stop" doorbell ptr */
149 struct mcp_kreq_ether_send *req_list; /* host shadow of sendq */
150 char *req_bytes;
151 struct myri10ge_tx_buffer_state *info;
152 int mask; /* number of transmit slots -1 */
153 int req ____cacheline_aligned; /* transmit slots submitted */
154 int pkt_start; /* packets started */
155 int stop_queue;
156 int linearized;
157 int done ____cacheline_aligned; /* transmit slots completed */
158 int pkt_done; /* packets completed */
159 int wake_queue;
160 int queue_active;
161 };
162
163 struct myri10ge_rx_done {
164 struct mcp_slot *entry;
165 dma_addr_t bus;
166 int cnt;
167 int idx;
168 struct net_lro_mgr lro_mgr;
169 struct net_lro_desc lro_desc[MYRI10GE_MAX_LRO_DESCRIPTORS];
170 };
171
172 struct myri10ge_slice_netstats {
173 unsigned long rx_packets;
174 unsigned long tx_packets;
175 unsigned long rx_bytes;
176 unsigned long tx_bytes;
177 unsigned long rx_dropped;
178 unsigned long tx_dropped;
179 };
180
181 struct myri10ge_slice_state {
182 struct myri10ge_tx_buf tx; /* transmit ring */
183 struct myri10ge_rx_buf rx_small;
184 struct myri10ge_rx_buf rx_big;
185 struct myri10ge_rx_done rx_done;
186 struct net_device *dev;
187 struct napi_struct napi;
188 struct myri10ge_priv *mgp;
189 struct myri10ge_slice_netstats stats;
190 __be32 __iomem *irq_claim;
191 struct mcp_irq_data *fw_stats;
192 dma_addr_t fw_stats_bus;
193 int watchdog_tx_done;
194 int watchdog_tx_req;
195 int watchdog_rx_done;
196 #ifdef CONFIG_MYRI10GE_DCA
197 int cached_dca_tag;
198 int cpu;
199 __be32 __iomem *dca_tag;
200 #endif
201 char irq_desc[32];
202 };
203
204 struct myri10ge_priv {
205 struct myri10ge_slice_state *ss;
206 int tx_boundary; /* boundary transmits cannot cross */
207 int num_slices;
208 int running; /* running? */
209 int small_bytes;
210 int big_bytes;
211 int max_intr_slots;
212 struct net_device *dev;
213 spinlock_t stats_lock;
214 u8 __iomem *sram;
215 int sram_size;
216 unsigned long board_span;
217 unsigned long iomem_base;
218 __be32 __iomem *irq_deassert;
219 char *mac_addr_string;
220 struct mcp_cmd_response *cmd;
221 dma_addr_t cmd_bus;
222 struct pci_dev *pdev;
223 int msi_enabled;
224 int msix_enabled;
225 struct msix_entry *msix_vectors;
226 #ifdef CONFIG_MYRI10GE_DCA
227 int dca_enabled;
228 int relaxed_order;
229 #endif
230 u32 link_state;
231 unsigned int rdma_tags_available;
232 int intr_coal_delay;
233 __be32 __iomem *intr_coal_delay_ptr;
234 int mtrr;
235 int wc_enabled;
236 int down_cnt;
237 wait_queue_head_t down_wq;
238 struct work_struct watchdog_work;
239 struct timer_list watchdog_timer;
240 int watchdog_resets;
241 int watchdog_pause;
242 int pause;
243 bool fw_name_allocated;
244 char *fw_name;
245 char eeprom_strings[MYRI10GE_EEPROM_STRINGS_SIZE];
246 char *product_code_string;
247 char fw_version[128];
248 int fw_ver_major;
249 int fw_ver_minor;
250 int fw_ver_tiny;
251 int adopted_rx_filter_bug;
252 u8 mac_addr[6]; /* eeprom mac address */
253 unsigned long serial_number;
254 int vendor_specific_offset;
255 int fw_multicast_support;
256 u32 features;
257 u32 max_tso6;
258 u32 read_dma;
259 u32 write_dma;
260 u32 read_write_dma;
261 u32 link_changes;
262 u32 msg_enable;
263 unsigned int board_number;
264 int rebooted;
265 };
266
267 static char *myri10ge_fw_unaligned = "myri10ge_ethp_z8e.dat";
268 static char *myri10ge_fw_aligned = "myri10ge_eth_z8e.dat";
269 static char *myri10ge_fw_rss_unaligned = "myri10ge_rss_ethp_z8e.dat";
270 static char *myri10ge_fw_rss_aligned = "myri10ge_rss_eth_z8e.dat";
271 MODULE_FIRMWARE("myri10ge_ethp_z8e.dat");
272 MODULE_FIRMWARE("myri10ge_eth_z8e.dat");
273 MODULE_FIRMWARE("myri10ge_rss_ethp_z8e.dat");
274 MODULE_FIRMWARE("myri10ge_rss_eth_z8e.dat");
275
276 /* Careful: must be accessed under kparam_block_sysfs_write */
277 static char *myri10ge_fw_name = NULL;
278 module_param(myri10ge_fw_name, charp, S_IRUGO | S_IWUSR);
279 MODULE_PARM_DESC(myri10ge_fw_name, "Firmware image name");
280
281 #define MYRI10GE_MAX_BOARDS 8
282 static char *myri10ge_fw_names[MYRI10GE_MAX_BOARDS] =
283 {[0 ... (MYRI10GE_MAX_BOARDS - 1)] = NULL };
284 module_param_array_named(myri10ge_fw_names, myri10ge_fw_names, charp, NULL,
285 0444);
286 MODULE_PARM_DESC(myri10ge_fw_name, "Firmware image names per board");
287
288 static int myri10ge_ecrc_enable = 1;
289 module_param(myri10ge_ecrc_enable, int, S_IRUGO);
290 MODULE_PARM_DESC(myri10ge_ecrc_enable, "Enable Extended CRC on PCI-E");
291
292 static int myri10ge_small_bytes = -1; /* -1 == auto */
293 module_param(myri10ge_small_bytes, int, S_IRUGO | S_IWUSR);
294 MODULE_PARM_DESC(myri10ge_small_bytes, "Threshold of small packets");
295
296 static int myri10ge_msi = 1; /* enable msi by default */
297 module_param(myri10ge_msi, int, S_IRUGO | S_IWUSR);
298 MODULE_PARM_DESC(myri10ge_msi, "Enable Message Signalled Interrupts");
299
300 static int myri10ge_intr_coal_delay = 75;
301 module_param(myri10ge_intr_coal_delay, int, S_IRUGO);
302 MODULE_PARM_DESC(myri10ge_intr_coal_delay, "Interrupt coalescing delay");
303
304 static int myri10ge_flow_control = 1;
305 module_param(myri10ge_flow_control, int, S_IRUGO);
306 MODULE_PARM_DESC(myri10ge_flow_control, "Pause parameter");
307
308 static int myri10ge_deassert_wait = 1;
309 module_param(myri10ge_deassert_wait, int, S_IRUGO | S_IWUSR);
310 MODULE_PARM_DESC(myri10ge_deassert_wait,
311 "Wait when deasserting legacy interrupts");
312
313 static int myri10ge_force_firmware = 0;
314 module_param(myri10ge_force_firmware, int, S_IRUGO);
315 MODULE_PARM_DESC(myri10ge_force_firmware,
316 "Force firmware to assume aligned completions");
317
318 static int myri10ge_initial_mtu = MYRI10GE_MAX_ETHER_MTU - ETH_HLEN;
319 module_param(myri10ge_initial_mtu, int, S_IRUGO);
320 MODULE_PARM_DESC(myri10ge_initial_mtu, "Initial MTU");
321
322 static int myri10ge_napi_weight = 64;
323 module_param(myri10ge_napi_weight, int, S_IRUGO);
324 MODULE_PARM_DESC(myri10ge_napi_weight, "Set NAPI weight");
325
326 static int myri10ge_watchdog_timeout = 1;
327 module_param(myri10ge_watchdog_timeout, int, S_IRUGO);
328 MODULE_PARM_DESC(myri10ge_watchdog_timeout, "Set watchdog timeout");
329
330 static int myri10ge_max_irq_loops = 1048576;
331 module_param(myri10ge_max_irq_loops, int, S_IRUGO);
332 MODULE_PARM_DESC(myri10ge_max_irq_loops,
333 "Set stuck legacy IRQ detection threshold");
334
335 #define MYRI10GE_MSG_DEFAULT NETIF_MSG_LINK
336
337 static int myri10ge_debug = -1; /* defaults above */
338 module_param(myri10ge_debug, int, 0);
339 MODULE_PARM_DESC(myri10ge_debug, "Debug level (0=none,...,16=all)");
340
341 static int myri10ge_lro_max_pkts = MYRI10GE_LRO_MAX_PKTS;
342 module_param(myri10ge_lro_max_pkts, int, S_IRUGO);
343 MODULE_PARM_DESC(myri10ge_lro_max_pkts,
344 "Number of LRO packets to be aggregated");
345
346 static int myri10ge_fill_thresh = 256;
347 module_param(myri10ge_fill_thresh, int, S_IRUGO | S_IWUSR);
348 MODULE_PARM_DESC(myri10ge_fill_thresh, "Number of empty rx slots allowed");
349
350 static int myri10ge_reset_recover = 1;
351
352 static int myri10ge_max_slices = 1;
353 module_param(myri10ge_max_slices, int, S_IRUGO);
354 MODULE_PARM_DESC(myri10ge_max_slices, "Max tx/rx queues");
355
356 static int myri10ge_rss_hash = MXGEFW_RSS_HASH_TYPE_SRC_DST_PORT;
357 module_param(myri10ge_rss_hash, int, S_IRUGO);
358 MODULE_PARM_DESC(myri10ge_rss_hash, "Type of RSS hashing to do");
359
360 static int myri10ge_dca = 1;
361 module_param(myri10ge_dca, int, S_IRUGO);
362 MODULE_PARM_DESC(myri10ge_dca, "Enable DCA if possible");
363
364 #define MYRI10GE_FW_OFFSET 1024*1024
365 #define MYRI10GE_HIGHPART_TO_U32(X) \
366 (sizeof (X) == 8) ? ((u32)((u64)(X) >> 32)) : (0)
367 #define MYRI10GE_LOWPART_TO_U32(X) ((u32)(X))
368
369 #define myri10ge_pio_copy(to,from,size) __iowrite64_copy(to,from,size/8)
370
371 static void myri10ge_set_multicast_list(struct net_device *dev);
372 static netdev_tx_t myri10ge_sw_tso(struct sk_buff *skb,
373 struct net_device *dev);
374
375 static inline void put_be32(__be32 val, __be32 __iomem * p)
376 {
377 __raw_writel((__force __u32) val, (__force void __iomem *)p);
378 }
379
380 static struct net_device_stats *myri10ge_get_stats(struct net_device *dev);
381
382 static void set_fw_name(struct myri10ge_priv *mgp, char *name, bool allocated)
383 {
384 if (mgp->fw_name_allocated)
385 kfree(mgp->fw_name);
386 mgp->fw_name = name;
387 mgp->fw_name_allocated = allocated;
388 }
389
390 static int
391 myri10ge_send_cmd(struct myri10ge_priv *mgp, u32 cmd,
392 struct myri10ge_cmd *data, int atomic)
393 {
394 struct mcp_cmd *buf;
395 char buf_bytes[sizeof(*buf) + 8];
396 struct mcp_cmd_response *response = mgp->cmd;
397 char __iomem *cmd_addr = mgp->sram + MXGEFW_ETH_CMD;
398 u32 dma_low, dma_high, result, value;
399 int sleep_total = 0;
400
401 /* ensure buf is aligned to 8 bytes */
402 buf = (struct mcp_cmd *)ALIGN((unsigned long)buf_bytes, 8);
403
404 buf->data0 = htonl(data->data0);
405 buf->data1 = htonl(data->data1);
406 buf->data2 = htonl(data->data2);
407 buf->cmd = htonl(cmd);
408 dma_low = MYRI10GE_LOWPART_TO_U32(mgp->cmd_bus);
409 dma_high = MYRI10GE_HIGHPART_TO_U32(mgp->cmd_bus);
410
411 buf->response_addr.low = htonl(dma_low);
412 buf->response_addr.high = htonl(dma_high);
413 response->result = htonl(MYRI10GE_NO_RESPONSE_RESULT);
414 mb();
415 myri10ge_pio_copy(cmd_addr, buf, sizeof(*buf));
416
417 /* wait up to 15ms. Longest command is the DMA benchmark,
418 * which is capped at 5ms, but runs from a timeout handler
419 * that runs every 7.8ms. So a 15ms timeout leaves us with
420 * a 2.2ms margin
421 */
422 if (atomic) {
423 /* if atomic is set, do not sleep,
424 * and try to get the completion quickly
425 * (1ms will be enough for those commands) */
426 for (sleep_total = 0;
427 sleep_total < 1000 &&
428 response->result == htonl(MYRI10GE_NO_RESPONSE_RESULT);
429 sleep_total += 10) {
430 udelay(10);
431 mb();
432 }
433 } else {
434 /* use msleep for most command */
435 for (sleep_total = 0;
436 sleep_total < 15 &&
437 response->result == htonl(MYRI10GE_NO_RESPONSE_RESULT);
438 sleep_total++)
439 msleep(1);
440 }
441
442 result = ntohl(response->result);
443 value = ntohl(response->data);
444 if (result != MYRI10GE_NO_RESPONSE_RESULT) {
445 if (result == 0) {
446 data->data0 = value;
447 return 0;
448 } else if (result == MXGEFW_CMD_UNKNOWN) {
449 return -ENOSYS;
450 } else if (result == MXGEFW_CMD_ERROR_UNALIGNED) {
451 return -E2BIG;
452 } else if (result == MXGEFW_CMD_ERROR_RANGE &&
453 cmd == MXGEFW_CMD_ENABLE_RSS_QUEUES &&
454 (data->
455 data1 & MXGEFW_SLICE_ENABLE_MULTIPLE_TX_QUEUES) !=
456 0) {
457 return -ERANGE;
458 } else {
459 dev_err(&mgp->pdev->dev,
460 "command %d failed, result = %d\n",
461 cmd, result);
462 return -ENXIO;
463 }
464 }
465
466 dev_err(&mgp->pdev->dev, "command %d timed out, result = %d\n",
467 cmd, result);
468 return -EAGAIN;
469 }
470
471 /*
472 * The eeprom strings on the lanaiX have the format
473 * SN=x\0
474 * MAC=x:x:x:x:x:x\0
475 * PT:ddd mmm xx xx:xx:xx xx\0
476 * PV:ddd mmm xx xx:xx:xx xx\0
477 */
478 static int myri10ge_read_mac_addr(struct myri10ge_priv *mgp)
479 {
480 char *ptr, *limit;
481 int i;
482
483 ptr = mgp->eeprom_strings;
484 limit = mgp->eeprom_strings + MYRI10GE_EEPROM_STRINGS_SIZE;
485
486 while (*ptr != '\0' && ptr < limit) {
487 if (memcmp(ptr, "MAC=", 4) == 0) {
488 ptr += 4;
489 mgp->mac_addr_string = ptr;
490 for (i = 0; i < 6; i++) {
491 if ((ptr + 2) > limit)
492 goto abort;
493 mgp->mac_addr[i] =
494 simple_strtoul(ptr, &ptr, 16);
495 ptr += 1;
496 }
497 }
498 if (memcmp(ptr, "PC=", 3) == 0) {
499 ptr += 3;
500 mgp->product_code_string = ptr;
501 }
502 if (memcmp((const void *)ptr, "SN=", 3) == 0) {
503 ptr += 3;
504 mgp->serial_number = simple_strtoul(ptr, &ptr, 10);
505 }
506 while (ptr < limit && *ptr++) ;
507 }
508
509 return 0;
510
511 abort:
512 dev_err(&mgp->pdev->dev, "failed to parse eeprom_strings\n");
513 return -ENXIO;
514 }
515
516 /*
517 * Enable or disable periodic RDMAs from the host to make certain
518 * chipsets resend dropped PCIe messages
519 */
520
521 static void myri10ge_dummy_rdma(struct myri10ge_priv *mgp, int enable)
522 {
523 char __iomem *submit;
524 __be32 buf[16] __attribute__ ((__aligned__(8)));
525 u32 dma_low, dma_high;
526 int i;
527
528 /* clear confirmation addr */
529 mgp->cmd->data = 0;
530 mb();
531
532 /* send a rdma command to the PCIe engine, and wait for the
533 * response in the confirmation address. The firmware should
534 * write a -1 there to indicate it is alive and well
535 */
536 dma_low = MYRI10GE_LOWPART_TO_U32(mgp->cmd_bus);
537 dma_high = MYRI10GE_HIGHPART_TO_U32(mgp->cmd_bus);
538
539 buf[0] = htonl(dma_high); /* confirm addr MSW */
540 buf[1] = htonl(dma_low); /* confirm addr LSW */
541 buf[2] = MYRI10GE_NO_CONFIRM_DATA; /* confirm data */
542 buf[3] = htonl(dma_high); /* dummy addr MSW */
543 buf[4] = htonl(dma_low); /* dummy addr LSW */
544 buf[5] = htonl(enable); /* enable? */
545
546 submit = mgp->sram + MXGEFW_BOOT_DUMMY_RDMA;
547
548 myri10ge_pio_copy(submit, &buf, sizeof(buf));
549 for (i = 0; mgp->cmd->data != MYRI10GE_NO_CONFIRM_DATA && i < 20; i++)
550 msleep(1);
551 if (mgp->cmd->data != MYRI10GE_NO_CONFIRM_DATA)
552 dev_err(&mgp->pdev->dev, "dummy rdma %s failed\n",
553 (enable ? "enable" : "disable"));
554 }
555
556 static int
557 myri10ge_validate_firmware(struct myri10ge_priv *mgp,
558 struct mcp_gen_header *hdr)
559 {
560 struct device *dev = &mgp->pdev->dev;
561
562 /* check firmware type */
563 if (ntohl(hdr->mcp_type) != MCP_TYPE_ETH) {
564 dev_err(dev, "Bad firmware type: 0x%x\n", ntohl(hdr->mcp_type));
565 return -EINVAL;
566 }
567
568 /* save firmware version for ethtool */
569 strncpy(mgp->fw_version, hdr->version, sizeof(mgp->fw_version));
570
571 sscanf(mgp->fw_version, "%d.%d.%d", &mgp->fw_ver_major,
572 &mgp->fw_ver_minor, &mgp->fw_ver_tiny);
573
574 if (!(mgp->fw_ver_major == MXGEFW_VERSION_MAJOR &&
575 mgp->fw_ver_minor == MXGEFW_VERSION_MINOR)) {
576 dev_err(dev, "Found firmware version %s\n", mgp->fw_version);
577 dev_err(dev, "Driver needs %d.%d\n", MXGEFW_VERSION_MAJOR,
578 MXGEFW_VERSION_MINOR);
579 return -EINVAL;
580 }
581 return 0;
582 }
583
584 static int myri10ge_load_hotplug_firmware(struct myri10ge_priv *mgp, u32 * size)
585 {
586 unsigned crc, reread_crc;
587 const struct firmware *fw;
588 struct device *dev = &mgp->pdev->dev;
589 unsigned char *fw_readback;
590 struct mcp_gen_header *hdr;
591 size_t hdr_offset;
592 int status;
593 unsigned i;
594
595 if ((status = request_firmware(&fw, mgp->fw_name, dev)) < 0) {
596 dev_err(dev, "Unable to load %s firmware image via hotplug\n",
597 mgp->fw_name);
598 status = -EINVAL;
599 goto abort_with_nothing;
600 }
601
602 /* check size */
603
604 if (fw->size >= mgp->sram_size - MYRI10GE_FW_OFFSET ||
605 fw->size < MCP_HEADER_PTR_OFFSET + 4) {
606 dev_err(dev, "Firmware size invalid:%d\n", (int)fw->size);
607 status = -EINVAL;
608 goto abort_with_fw;
609 }
610
611 /* check id */
612 hdr_offset = ntohl(*(__be32 *) (fw->data + MCP_HEADER_PTR_OFFSET));
613 if ((hdr_offset & 3) || hdr_offset + sizeof(*hdr) > fw->size) {
614 dev_err(dev, "Bad firmware file\n");
615 status = -EINVAL;
616 goto abort_with_fw;
617 }
618 hdr = (void *)(fw->data + hdr_offset);
619
620 status = myri10ge_validate_firmware(mgp, hdr);
621 if (status != 0)
622 goto abort_with_fw;
623
624 crc = crc32(~0, fw->data, fw->size);
625 for (i = 0; i < fw->size; i += 256) {
626 myri10ge_pio_copy(mgp->sram + MYRI10GE_FW_OFFSET + i,
627 fw->data + i,
628 min(256U, (unsigned)(fw->size - i)));
629 mb();
630 readb(mgp->sram);
631 }
632 fw_readback = vmalloc(fw->size);
633 if (!fw_readback) {
634 status = -ENOMEM;
635 goto abort_with_fw;
636 }
637 /* corruption checking is good for parity recovery and buggy chipset */
638 memcpy_fromio(fw_readback, mgp->sram + MYRI10GE_FW_OFFSET, fw->size);
639 reread_crc = crc32(~0, fw_readback, fw->size);
640 vfree(fw_readback);
641 if (crc != reread_crc) {
642 dev_err(dev, "CRC failed(fw-len=%u), got 0x%x (expect 0x%x)\n",
643 (unsigned)fw->size, reread_crc, crc);
644 status = -EIO;
645 goto abort_with_fw;
646 }
647 *size = (u32) fw->size;
648
649 abort_with_fw:
650 release_firmware(fw);
651
652 abort_with_nothing:
653 return status;
654 }
655
656 static int myri10ge_adopt_running_firmware(struct myri10ge_priv *mgp)
657 {
658 struct mcp_gen_header *hdr;
659 struct device *dev = &mgp->pdev->dev;
660 const size_t bytes = sizeof(struct mcp_gen_header);
661 size_t hdr_offset;
662 int status;
663
664 /* find running firmware header */
665 hdr_offset = swab32(readl(mgp->sram + MCP_HEADER_PTR_OFFSET));
666
667 if ((hdr_offset & 3) || hdr_offset + sizeof(*hdr) > mgp->sram_size) {
668 dev_err(dev, "Running firmware has bad header offset (%d)\n",
669 (int)hdr_offset);
670 return -EIO;
671 }
672
673 /* copy header of running firmware from SRAM to host memory to
674 * validate firmware */
675 hdr = kmalloc(bytes, GFP_KERNEL);
676 if (hdr == NULL) {
677 dev_err(dev, "could not malloc firmware hdr\n");
678 return -ENOMEM;
679 }
680 memcpy_fromio(hdr, mgp->sram + hdr_offset, bytes);
681 status = myri10ge_validate_firmware(mgp, hdr);
682 kfree(hdr);
683
684 /* check to see if adopted firmware has bug where adopting
685 * it will cause broadcasts to be filtered unless the NIC
686 * is kept in ALLMULTI mode */
687 if (mgp->fw_ver_major == 1 && mgp->fw_ver_minor == 4 &&
688 mgp->fw_ver_tiny >= 4 && mgp->fw_ver_tiny <= 11) {
689 mgp->adopted_rx_filter_bug = 1;
690 dev_warn(dev, "Adopting fw %d.%d.%d: "
691 "working around rx filter bug\n",
692 mgp->fw_ver_major, mgp->fw_ver_minor,
693 mgp->fw_ver_tiny);
694 }
695 return status;
696 }
697
698 static int myri10ge_get_firmware_capabilities(struct myri10ge_priv *mgp)
699 {
700 struct myri10ge_cmd cmd;
701 int status;
702
703 /* probe for IPv6 TSO support */
704 mgp->features = NETIF_F_SG | NETIF_F_HW_CSUM | NETIF_F_TSO;
705 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_MAX_TSO6_HDR_SIZE,
706 &cmd, 0);
707 if (status == 0) {
708 mgp->max_tso6 = cmd.data0;
709 mgp->features |= NETIF_F_TSO6;
710 }
711
712 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_RX_RING_SIZE, &cmd, 0);
713 if (status != 0) {
714 dev_err(&mgp->pdev->dev,
715 "failed MXGEFW_CMD_GET_RX_RING_SIZE\n");
716 return -ENXIO;
717 }
718
719 mgp->max_intr_slots = 2 * (cmd.data0 / sizeof(struct mcp_dma_addr));
720
721 return 0;
722 }
723
724 static int myri10ge_load_firmware(struct myri10ge_priv *mgp, int adopt)
725 {
726 char __iomem *submit;
727 __be32 buf[16] __attribute__ ((__aligned__(8)));
728 u32 dma_low, dma_high, size;
729 int status, i;
730
731 size = 0;
732 status = myri10ge_load_hotplug_firmware(mgp, &size);
733 if (status) {
734 if (!adopt)
735 return status;
736 dev_warn(&mgp->pdev->dev, "hotplug firmware loading failed\n");
737
738 /* Do not attempt to adopt firmware if there
739 * was a bad crc */
740 if (status == -EIO)
741 return status;
742
743 status = myri10ge_adopt_running_firmware(mgp);
744 if (status != 0) {
745 dev_err(&mgp->pdev->dev,
746 "failed to adopt running firmware\n");
747 return status;
748 }
749 dev_info(&mgp->pdev->dev,
750 "Successfully adopted running firmware\n");
751 if (mgp->tx_boundary == 4096) {
752 dev_warn(&mgp->pdev->dev,
753 "Using firmware currently running on NIC"
754 ". For optimal\n");
755 dev_warn(&mgp->pdev->dev,
756 "performance consider loading optimized "
757 "firmware\n");
758 dev_warn(&mgp->pdev->dev, "via hotplug\n");
759 }
760
761 set_fw_name(mgp, "adopted", false);
762 mgp->tx_boundary = 2048;
763 myri10ge_dummy_rdma(mgp, 1);
764 status = myri10ge_get_firmware_capabilities(mgp);
765 return status;
766 }
767
768 /* clear confirmation addr */
769 mgp->cmd->data = 0;
770 mb();
771
772 /* send a reload command to the bootstrap MCP, and wait for the
773 * response in the confirmation address. The firmware should
774 * write a -1 there to indicate it is alive and well
775 */
776 dma_low = MYRI10GE_LOWPART_TO_U32(mgp->cmd_bus);
777 dma_high = MYRI10GE_HIGHPART_TO_U32(mgp->cmd_bus);
778
779 buf[0] = htonl(dma_high); /* confirm addr MSW */
780 buf[1] = htonl(dma_low); /* confirm addr LSW */
781 buf[2] = MYRI10GE_NO_CONFIRM_DATA; /* confirm data */
782
783 /* FIX: All newest firmware should un-protect the bottom of
784 * the sram before handoff. However, the very first interfaces
785 * do not. Therefore the handoff copy must skip the first 8 bytes
786 */
787 buf[3] = htonl(MYRI10GE_FW_OFFSET + 8); /* where the code starts */
788 buf[4] = htonl(size - 8); /* length of code */
789 buf[5] = htonl(8); /* where to copy to */
790 buf[6] = htonl(0); /* where to jump to */
791
792 submit = mgp->sram + MXGEFW_BOOT_HANDOFF;
793
794 myri10ge_pio_copy(submit, &buf, sizeof(buf));
795 mb();
796 msleep(1);
797 mb();
798 i = 0;
799 while (mgp->cmd->data != MYRI10GE_NO_CONFIRM_DATA && i < 9) {
800 msleep(1 << i);
801 i++;
802 }
803 if (mgp->cmd->data != MYRI10GE_NO_CONFIRM_DATA) {
804 dev_err(&mgp->pdev->dev, "handoff failed\n");
805 return -ENXIO;
806 }
807 myri10ge_dummy_rdma(mgp, 1);
808 status = myri10ge_get_firmware_capabilities(mgp);
809
810 return status;
811 }
812
813 static int myri10ge_update_mac_address(struct myri10ge_priv *mgp, u8 * addr)
814 {
815 struct myri10ge_cmd cmd;
816 int status;
817
818 cmd.data0 = ((addr[0] << 24) | (addr[1] << 16)
819 | (addr[2] << 8) | addr[3]);
820
821 cmd.data1 = ((addr[4] << 8) | (addr[5]));
822
823 status = myri10ge_send_cmd(mgp, MXGEFW_SET_MAC_ADDRESS, &cmd, 0);
824 return status;
825 }
826
827 static int myri10ge_change_pause(struct myri10ge_priv *mgp, int pause)
828 {
829 struct myri10ge_cmd cmd;
830 int status, ctl;
831
832 ctl = pause ? MXGEFW_ENABLE_FLOW_CONTROL : MXGEFW_DISABLE_FLOW_CONTROL;
833 status = myri10ge_send_cmd(mgp, ctl, &cmd, 0);
834
835 if (status) {
836 netdev_err(mgp->dev, "Failed to set flow control mode\n");
837 return status;
838 }
839 mgp->pause = pause;
840 return 0;
841 }
842
843 static void
844 myri10ge_change_promisc(struct myri10ge_priv *mgp, int promisc, int atomic)
845 {
846 struct myri10ge_cmd cmd;
847 int status, ctl;
848
849 ctl = promisc ? MXGEFW_ENABLE_PROMISC : MXGEFW_DISABLE_PROMISC;
850 status = myri10ge_send_cmd(mgp, ctl, &cmd, atomic);
851 if (status)
852 netdev_err(mgp->dev, "Failed to set promisc mode\n");
853 }
854
855 static int myri10ge_dma_test(struct myri10ge_priv *mgp, int test_type)
856 {
857 struct myri10ge_cmd cmd;
858 int status;
859 u32 len;
860 struct page *dmatest_page;
861 dma_addr_t dmatest_bus;
862 char *test = " ";
863
864 dmatest_page = alloc_page(GFP_KERNEL);
865 if (!dmatest_page)
866 return -ENOMEM;
867 dmatest_bus = pci_map_page(mgp->pdev, dmatest_page, 0, PAGE_SIZE,
868 DMA_BIDIRECTIONAL);
869
870 /* Run a small DMA test.
871 * The magic multipliers to the length tell the firmware
872 * to do DMA read, write, or read+write tests. The
873 * results are returned in cmd.data0. The upper 16
874 * bits or the return is the number of transfers completed.
875 * The lower 16 bits is the time in 0.5us ticks that the
876 * transfers took to complete.
877 */
878
879 len = mgp->tx_boundary;
880
881 cmd.data0 = MYRI10GE_LOWPART_TO_U32(dmatest_bus);
882 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(dmatest_bus);
883 cmd.data2 = len * 0x10000;
884 status = myri10ge_send_cmd(mgp, test_type, &cmd, 0);
885 if (status != 0) {
886 test = "read";
887 goto abort;
888 }
889 mgp->read_dma = ((cmd.data0 >> 16) * len * 2) / (cmd.data0 & 0xffff);
890 cmd.data0 = MYRI10GE_LOWPART_TO_U32(dmatest_bus);
891 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(dmatest_bus);
892 cmd.data2 = len * 0x1;
893 status = myri10ge_send_cmd(mgp, test_type, &cmd, 0);
894 if (status != 0) {
895 test = "write";
896 goto abort;
897 }
898 mgp->write_dma = ((cmd.data0 >> 16) * len * 2) / (cmd.data0 & 0xffff);
899
900 cmd.data0 = MYRI10GE_LOWPART_TO_U32(dmatest_bus);
901 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(dmatest_bus);
902 cmd.data2 = len * 0x10001;
903 status = myri10ge_send_cmd(mgp, test_type, &cmd, 0);
904 if (status != 0) {
905 test = "read/write";
906 goto abort;
907 }
908 mgp->read_write_dma = ((cmd.data0 >> 16) * len * 2 * 2) /
909 (cmd.data0 & 0xffff);
910
911 abort:
912 pci_unmap_page(mgp->pdev, dmatest_bus, PAGE_SIZE, DMA_BIDIRECTIONAL);
913 put_page(dmatest_page);
914
915 if (status != 0 && test_type != MXGEFW_CMD_UNALIGNED_TEST)
916 dev_warn(&mgp->pdev->dev, "DMA %s benchmark failed: %d\n",
917 test, status);
918
919 return status;
920 }
921
922 static int myri10ge_reset(struct myri10ge_priv *mgp)
923 {
924 struct myri10ge_cmd cmd;
925 struct myri10ge_slice_state *ss;
926 int i, status;
927 size_t bytes;
928 #ifdef CONFIG_MYRI10GE_DCA
929 unsigned long dca_tag_off;
930 #endif
931
932 /* try to send a reset command to the card to see if it
933 * is alive */
934 memset(&cmd, 0, sizeof(cmd));
935 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_RESET, &cmd, 0);
936 if (status != 0) {
937 dev_err(&mgp->pdev->dev, "failed reset\n");
938 return -ENXIO;
939 }
940
941 (void)myri10ge_dma_test(mgp, MXGEFW_DMA_TEST);
942 /*
943 * Use non-ndis mcp_slot (eg, 4 bytes total,
944 * no toeplitz hash value returned. Older firmware will
945 * not understand this command, but will use the correct
946 * sized mcp_slot, so we ignore error returns
947 */
948 cmd.data0 = MXGEFW_RSS_MCP_SLOT_TYPE_MIN;
949 (void)myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_RSS_MCP_SLOT_TYPE, &cmd, 0);
950
951 /* Now exchange information about interrupts */
952
953 bytes = mgp->max_intr_slots * sizeof(*mgp->ss[0].rx_done.entry);
954 cmd.data0 = (u32) bytes;
955 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_INTRQ_SIZE, &cmd, 0);
956
957 /*
958 * Even though we already know how many slices are supported
959 * via myri10ge_probe_slices() MXGEFW_CMD_GET_MAX_RSS_QUEUES
960 * has magic side effects, and must be called after a reset.
961 * It must be called prior to calling any RSS related cmds,
962 * including assigning an interrupt queue for anything but
963 * slice 0. It must also be called *after*
964 * MXGEFW_CMD_SET_INTRQ_SIZE, since the intrq size is used by
965 * the firmware to compute offsets.
966 */
967
968 if (mgp->num_slices > 1) {
969
970 /* ask the maximum number of slices it supports */
971 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_MAX_RSS_QUEUES,
972 &cmd, 0);
973 if (status != 0) {
974 dev_err(&mgp->pdev->dev,
975 "failed to get number of slices\n");
976 }
977
978 /*
979 * MXGEFW_CMD_ENABLE_RSS_QUEUES must be called prior
980 * to setting up the interrupt queue DMA
981 */
982
983 cmd.data0 = mgp->num_slices;
984 cmd.data1 = MXGEFW_SLICE_INTR_MODE_ONE_PER_SLICE;
985 if (mgp->dev->real_num_tx_queues > 1)
986 cmd.data1 |= MXGEFW_SLICE_ENABLE_MULTIPLE_TX_QUEUES;
987 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_ENABLE_RSS_QUEUES,
988 &cmd, 0);
989
990 /* Firmware older than 1.4.32 only supports multiple
991 * RX queues, so if we get an error, first retry using a
992 * single TX queue before giving up */
993 if (status != 0 && mgp->dev->real_num_tx_queues > 1) {
994 netif_set_real_num_tx_queues(mgp->dev, 1);
995 cmd.data0 = mgp->num_slices;
996 cmd.data1 = MXGEFW_SLICE_INTR_MODE_ONE_PER_SLICE;
997 status = myri10ge_send_cmd(mgp,
998 MXGEFW_CMD_ENABLE_RSS_QUEUES,
999 &cmd, 0);
1000 }
1001
1002 if (status != 0) {
1003 dev_err(&mgp->pdev->dev,
1004 "failed to set number of slices\n");
1005
1006 return status;
1007 }
1008 }
1009 for (i = 0; i < mgp->num_slices; i++) {
1010 ss = &mgp->ss[i];
1011 cmd.data0 = MYRI10GE_LOWPART_TO_U32(ss->rx_done.bus);
1012 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(ss->rx_done.bus);
1013 cmd.data2 = i;
1014 status |= myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_INTRQ_DMA,
1015 &cmd, 0);
1016 };
1017
1018 status |=
1019 myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_IRQ_ACK_OFFSET, &cmd, 0);
1020 for (i = 0; i < mgp->num_slices; i++) {
1021 ss = &mgp->ss[i];
1022 ss->irq_claim =
1023 (__iomem __be32 *) (mgp->sram + cmd.data0 + 8 * i);
1024 }
1025 status |= myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_IRQ_DEASSERT_OFFSET,
1026 &cmd, 0);
1027 mgp->irq_deassert = (__iomem __be32 *) (mgp->sram + cmd.data0);
1028
1029 status |= myri10ge_send_cmd
1030 (mgp, MXGEFW_CMD_GET_INTR_COAL_DELAY_OFFSET, &cmd, 0);
1031 mgp->intr_coal_delay_ptr = (__iomem __be32 *) (mgp->sram + cmd.data0);
1032 if (status != 0) {
1033 dev_err(&mgp->pdev->dev, "failed set interrupt parameters\n");
1034 return status;
1035 }
1036 put_be32(htonl(mgp->intr_coal_delay), mgp->intr_coal_delay_ptr);
1037
1038 #ifdef CONFIG_MYRI10GE_DCA
1039 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_DCA_OFFSET, &cmd, 0);
1040 dca_tag_off = cmd.data0;
1041 for (i = 0; i < mgp->num_slices; i++) {
1042 ss = &mgp->ss[i];
1043 if (status == 0) {
1044 ss->dca_tag = (__iomem __be32 *)
1045 (mgp->sram + dca_tag_off + 4 * i);
1046 } else {
1047 ss->dca_tag = NULL;
1048 }
1049 }
1050 #endif /* CONFIG_MYRI10GE_DCA */
1051
1052 /* reset mcp/driver shared state back to 0 */
1053
1054 mgp->link_changes = 0;
1055 for (i = 0; i < mgp->num_slices; i++) {
1056 ss = &mgp->ss[i];
1057
1058 memset(ss->rx_done.entry, 0, bytes);
1059 ss->tx.req = 0;
1060 ss->tx.done = 0;
1061 ss->tx.pkt_start = 0;
1062 ss->tx.pkt_done = 0;
1063 ss->rx_big.cnt = 0;
1064 ss->rx_small.cnt = 0;
1065 ss->rx_done.idx = 0;
1066 ss->rx_done.cnt = 0;
1067 ss->tx.wake_queue = 0;
1068 ss->tx.stop_queue = 0;
1069 }
1070
1071 status = myri10ge_update_mac_address(mgp, mgp->dev->dev_addr);
1072 myri10ge_change_pause(mgp, mgp->pause);
1073 myri10ge_set_multicast_list(mgp->dev);
1074 return status;
1075 }
1076
1077 #ifdef CONFIG_MYRI10GE_DCA
1078 static int myri10ge_toggle_relaxed(struct pci_dev *pdev, int on)
1079 {
1080 int ret, cap, err;
1081 u16 ctl;
1082
1083 cap = pci_find_capability(pdev, PCI_CAP_ID_EXP);
1084 if (!cap)
1085 return 0;
1086
1087 err = pci_read_config_word(pdev, cap + PCI_EXP_DEVCTL, &ctl);
1088 ret = (ctl & PCI_EXP_DEVCTL_RELAX_EN) >> 4;
1089 if (ret != on) {
1090 ctl &= ~PCI_EXP_DEVCTL_RELAX_EN;
1091 ctl |= (on << 4);
1092 pci_write_config_word(pdev, cap + PCI_EXP_DEVCTL, ctl);
1093 }
1094 return ret;
1095 }
1096
1097 static void
1098 myri10ge_write_dca(struct myri10ge_slice_state *ss, int cpu, int tag)
1099 {
1100 ss->cached_dca_tag = tag;
1101 put_be32(htonl(tag), ss->dca_tag);
1102 }
1103
1104 static inline void myri10ge_update_dca(struct myri10ge_slice_state *ss)
1105 {
1106 int cpu = get_cpu();
1107 int tag;
1108
1109 if (cpu != ss->cpu) {
1110 tag = dca3_get_tag(&ss->mgp->pdev->dev, cpu);
1111 if (ss->cached_dca_tag != tag)
1112 myri10ge_write_dca(ss, cpu, tag);
1113 ss->cpu = cpu;
1114 }
1115 put_cpu();
1116 }
1117
1118 static void myri10ge_setup_dca(struct myri10ge_priv *mgp)
1119 {
1120 int err, i;
1121 struct pci_dev *pdev = mgp->pdev;
1122
1123 if (mgp->ss[0].dca_tag == NULL || mgp->dca_enabled)
1124 return;
1125 if (!myri10ge_dca) {
1126 dev_err(&pdev->dev, "dca disabled by administrator\n");
1127 return;
1128 }
1129 err = dca_add_requester(&pdev->dev);
1130 if (err) {
1131 if (err != -ENODEV)
1132 dev_err(&pdev->dev,
1133 "dca_add_requester() failed, err=%d\n", err);
1134 return;
1135 }
1136 mgp->relaxed_order = myri10ge_toggle_relaxed(pdev, 0);
1137 mgp->dca_enabled = 1;
1138 for (i = 0; i < mgp->num_slices; i++) {
1139 mgp->ss[i].cpu = -1;
1140 mgp->ss[i].cached_dca_tag = -1;
1141 myri10ge_update_dca(&mgp->ss[i]);
1142 }
1143 }
1144
1145 static void myri10ge_teardown_dca(struct myri10ge_priv *mgp)
1146 {
1147 struct pci_dev *pdev = mgp->pdev;
1148 int err;
1149
1150 if (!mgp->dca_enabled)
1151 return;
1152 mgp->dca_enabled = 0;
1153 if (mgp->relaxed_order)
1154 myri10ge_toggle_relaxed(pdev, 1);
1155 err = dca_remove_requester(&pdev->dev);
1156 }
1157
1158 static int myri10ge_notify_dca_device(struct device *dev, void *data)
1159 {
1160 struct myri10ge_priv *mgp;
1161 unsigned long event;
1162
1163 mgp = dev_get_drvdata(dev);
1164 event = *(unsigned long *)data;
1165
1166 if (event == DCA_PROVIDER_ADD)
1167 myri10ge_setup_dca(mgp);
1168 else if (event == DCA_PROVIDER_REMOVE)
1169 myri10ge_teardown_dca(mgp);
1170 return 0;
1171 }
1172 #endif /* CONFIG_MYRI10GE_DCA */
1173
1174 static inline void
1175 myri10ge_submit_8rx(struct mcp_kreq_ether_recv __iomem * dst,
1176 struct mcp_kreq_ether_recv *src)
1177 {
1178 __be32 low;
1179
1180 low = src->addr_low;
1181 src->addr_low = htonl(DMA_BIT_MASK(32));
1182 myri10ge_pio_copy(dst, src, 4 * sizeof(*src));
1183 mb();
1184 myri10ge_pio_copy(dst + 4, src + 4, 4 * sizeof(*src));
1185 mb();
1186 src->addr_low = low;
1187 put_be32(low, &dst->addr_low);
1188 mb();
1189 }
1190
1191 static inline void myri10ge_vlan_ip_csum(struct sk_buff *skb, __wsum hw_csum)
1192 {
1193 struct vlan_hdr *vh = (struct vlan_hdr *)(skb->data);
1194
1195 if ((skb->protocol == htons(ETH_P_8021Q)) &&
1196 (vh->h_vlan_encapsulated_proto == htons(ETH_P_IP) ||
1197 vh->h_vlan_encapsulated_proto == htons(ETH_P_IPV6))) {
1198 skb->csum = hw_csum;
1199 skb->ip_summed = CHECKSUM_COMPLETE;
1200 }
1201 }
1202
1203 static inline void
1204 myri10ge_rx_skb_build(struct sk_buff *skb, u8 * va,
1205 struct skb_frag_struct *rx_frags, int len, int hlen)
1206 {
1207 struct skb_frag_struct *skb_frags;
1208
1209 skb->len = skb->data_len = len;
1210 skb->truesize = len + sizeof(struct sk_buff);
1211 /* attach the page(s) */
1212
1213 skb_frags = skb_shinfo(skb)->frags;
1214 while (len > 0) {
1215 memcpy(skb_frags, rx_frags, sizeof(*skb_frags));
1216 len -= rx_frags->size;
1217 skb_frags++;
1218 rx_frags++;
1219 skb_shinfo(skb)->nr_frags++;
1220 }
1221
1222 /* pskb_may_pull is not available in irq context, but
1223 * skb_pull() (for ether_pad and eth_type_trans()) requires
1224 * the beginning of the packet in skb_headlen(), move it
1225 * manually */
1226 skb_copy_to_linear_data(skb, va, hlen);
1227 skb_shinfo(skb)->frags[0].page_offset += hlen;
1228 skb_shinfo(skb)->frags[0].size -= hlen;
1229 skb->data_len -= hlen;
1230 skb->tail += hlen;
1231 skb_pull(skb, MXGEFW_PAD);
1232 }
1233
1234 static void
1235 myri10ge_alloc_rx_pages(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx,
1236 int bytes, int watchdog)
1237 {
1238 struct page *page;
1239 int idx;
1240 #if MYRI10GE_ALLOC_SIZE > 4096
1241 int end_offset;
1242 #endif
1243
1244 if (unlikely(rx->watchdog_needed && !watchdog))
1245 return;
1246
1247 /* try to refill entire ring */
1248 while (rx->fill_cnt != (rx->cnt + rx->mask + 1)) {
1249 idx = rx->fill_cnt & rx->mask;
1250 if (rx->page_offset + bytes <= MYRI10GE_ALLOC_SIZE) {
1251 /* we can use part of previous page */
1252 get_page(rx->page);
1253 } else {
1254 /* we need a new page */
1255 page =
1256 alloc_pages(GFP_ATOMIC | __GFP_COMP,
1257 MYRI10GE_ALLOC_ORDER);
1258 if (unlikely(page == NULL)) {
1259 if (rx->fill_cnt - rx->cnt < 16)
1260 rx->watchdog_needed = 1;
1261 return;
1262 }
1263 rx->page = page;
1264 rx->page_offset = 0;
1265 rx->bus = pci_map_page(mgp->pdev, page, 0,
1266 MYRI10GE_ALLOC_SIZE,
1267 PCI_DMA_FROMDEVICE);
1268 }
1269 rx->info[idx].page = rx->page;
1270 rx->info[idx].page_offset = rx->page_offset;
1271 /* note that this is the address of the start of the
1272 * page */
1273 dma_unmap_addr_set(&rx->info[idx], bus, rx->bus);
1274 rx->shadow[idx].addr_low =
1275 htonl(MYRI10GE_LOWPART_TO_U32(rx->bus) + rx->page_offset);
1276 rx->shadow[idx].addr_high =
1277 htonl(MYRI10GE_HIGHPART_TO_U32(rx->bus));
1278
1279 /* start next packet on a cacheline boundary */
1280 rx->page_offset += SKB_DATA_ALIGN(bytes);
1281
1282 #if MYRI10GE_ALLOC_SIZE > 4096
1283 /* don't cross a 4KB boundary */
1284 end_offset = rx->page_offset + bytes - 1;
1285 if ((unsigned)(rx->page_offset ^ end_offset) > 4095)
1286 rx->page_offset = end_offset & ~4095;
1287 #endif
1288 rx->fill_cnt++;
1289
1290 /* copy 8 descriptors to the firmware at a time */
1291 if ((idx & 7) == 7) {
1292 myri10ge_submit_8rx(&rx->lanai[idx - 7],
1293 &rx->shadow[idx - 7]);
1294 }
1295 }
1296 }
1297
1298 static inline void
1299 myri10ge_unmap_rx_page(struct pci_dev *pdev,
1300 struct myri10ge_rx_buffer_state *info, int bytes)
1301 {
1302 /* unmap the recvd page if we're the only or last user of it */
1303 if (bytes >= MYRI10GE_ALLOC_SIZE / 2 ||
1304 (info->page_offset + 2 * bytes) > MYRI10GE_ALLOC_SIZE) {
1305 pci_unmap_page(pdev, (dma_unmap_addr(info, bus)
1306 & ~(MYRI10GE_ALLOC_SIZE - 1)),
1307 MYRI10GE_ALLOC_SIZE, PCI_DMA_FROMDEVICE);
1308 }
1309 }
1310
1311 #define MYRI10GE_HLEN 64 /* The number of bytes to copy from a
1312 * page into an skb */
1313
1314 static inline int
1315 myri10ge_rx_done(struct myri10ge_slice_state *ss, int len, __wsum csum,
1316 int lro_enabled)
1317 {
1318 struct myri10ge_priv *mgp = ss->mgp;
1319 struct sk_buff *skb;
1320 struct skb_frag_struct rx_frags[MYRI10GE_MAX_FRAGS_PER_FRAME];
1321 struct myri10ge_rx_buf *rx;
1322 int i, idx, hlen, remainder, bytes;
1323 struct pci_dev *pdev = mgp->pdev;
1324 struct net_device *dev = mgp->dev;
1325 u8 *va;
1326
1327 if (len <= mgp->small_bytes) {
1328 rx = &ss->rx_small;
1329 bytes = mgp->small_bytes;
1330 } else {
1331 rx = &ss->rx_big;
1332 bytes = mgp->big_bytes;
1333 }
1334
1335 len += MXGEFW_PAD;
1336 idx = rx->cnt & rx->mask;
1337 va = page_address(rx->info[idx].page) + rx->info[idx].page_offset;
1338 prefetch(va);
1339 /* Fill skb_frag_struct(s) with data from our receive */
1340 for (i = 0, remainder = len; remainder > 0; i++) {
1341 myri10ge_unmap_rx_page(pdev, &rx->info[idx], bytes);
1342 rx_frags[i].page = rx->info[idx].page;
1343 rx_frags[i].page_offset = rx->info[idx].page_offset;
1344 if (remainder < MYRI10GE_ALLOC_SIZE)
1345 rx_frags[i].size = remainder;
1346 else
1347 rx_frags[i].size = MYRI10GE_ALLOC_SIZE;
1348 rx->cnt++;
1349 idx = rx->cnt & rx->mask;
1350 remainder -= MYRI10GE_ALLOC_SIZE;
1351 }
1352
1353 if (lro_enabled) {
1354 rx_frags[0].page_offset += MXGEFW_PAD;
1355 rx_frags[0].size -= MXGEFW_PAD;
1356 len -= MXGEFW_PAD;
1357 lro_receive_frags(&ss->rx_done.lro_mgr, rx_frags,
1358 /* opaque, will come back in get_frag_header */
1359 len, len,
1360 (void *)(__force unsigned long)csum, csum);
1361
1362 return 1;
1363 }
1364
1365 hlen = MYRI10GE_HLEN > len ? len : MYRI10GE_HLEN;
1366
1367 /* allocate an skb to attach the page(s) to. This is done
1368 * after trying LRO, so as to avoid skb allocation overheads */
1369
1370 skb = netdev_alloc_skb(dev, MYRI10GE_HLEN + 16);
1371 if (unlikely(skb == NULL)) {
1372 ss->stats.rx_dropped++;
1373 do {
1374 i--;
1375 put_page(rx_frags[i].page);
1376 } while (i != 0);
1377 return 0;
1378 }
1379
1380 /* Attach the pages to the skb, and trim off any padding */
1381 myri10ge_rx_skb_build(skb, va, rx_frags, len, hlen);
1382 if (skb_shinfo(skb)->frags[0].size <= 0) {
1383 put_page(skb_shinfo(skb)->frags[0].page);
1384 skb_shinfo(skb)->nr_frags = 0;
1385 }
1386 skb->protocol = eth_type_trans(skb, dev);
1387 skb_record_rx_queue(skb, ss - &mgp->ss[0]);
1388
1389 if (dev->features & NETIF_F_RXCSUM) {
1390 if ((skb->protocol == htons(ETH_P_IP)) ||
1391 (skb->protocol == htons(ETH_P_IPV6))) {
1392 skb->csum = csum;
1393 skb->ip_summed = CHECKSUM_COMPLETE;
1394 } else
1395 myri10ge_vlan_ip_csum(skb, csum);
1396 }
1397 netif_receive_skb(skb);
1398 return 1;
1399 }
1400
1401 static inline void
1402 myri10ge_tx_done(struct myri10ge_slice_state *ss, int mcp_index)
1403 {
1404 struct pci_dev *pdev = ss->mgp->pdev;
1405 struct myri10ge_tx_buf *tx = &ss->tx;
1406 struct netdev_queue *dev_queue;
1407 struct sk_buff *skb;
1408 int idx, len;
1409
1410 while (tx->pkt_done != mcp_index) {
1411 idx = tx->done & tx->mask;
1412 skb = tx->info[idx].skb;
1413
1414 /* Mark as free */
1415 tx->info[idx].skb = NULL;
1416 if (tx->info[idx].last) {
1417 tx->pkt_done++;
1418 tx->info[idx].last = 0;
1419 }
1420 tx->done++;
1421 len = dma_unmap_len(&tx->info[idx], len);
1422 dma_unmap_len_set(&tx->info[idx], len, 0);
1423 if (skb) {
1424 ss->stats.tx_bytes += skb->len;
1425 ss->stats.tx_packets++;
1426 dev_kfree_skb_irq(skb);
1427 if (len)
1428 pci_unmap_single(pdev,
1429 dma_unmap_addr(&tx->info[idx],
1430 bus), len,
1431 PCI_DMA_TODEVICE);
1432 } else {
1433 if (len)
1434 pci_unmap_page(pdev,
1435 dma_unmap_addr(&tx->info[idx],
1436 bus), len,
1437 PCI_DMA_TODEVICE);
1438 }
1439 }
1440
1441 dev_queue = netdev_get_tx_queue(ss->dev, ss - ss->mgp->ss);
1442 /*
1443 * Make a minimal effort to prevent the NIC from polling an
1444 * idle tx queue. If we can't get the lock we leave the queue
1445 * active. In this case, either a thread was about to start
1446 * using the queue anyway, or we lost a race and the NIC will
1447 * waste some of its resources polling an inactive queue for a
1448 * while.
1449 */
1450
1451 if ((ss->mgp->dev->real_num_tx_queues > 1) &&
1452 __netif_tx_trylock(dev_queue)) {
1453 if (tx->req == tx->done) {
1454 tx->queue_active = 0;
1455 put_be32(htonl(1), tx->send_stop);
1456 mb();
1457 mmiowb();
1458 }
1459 __netif_tx_unlock(dev_queue);
1460 }
1461
1462 /* start the queue if we've stopped it */
1463 if (netif_tx_queue_stopped(dev_queue) &&
1464 tx->req - tx->done < (tx->mask >> 1)) {
1465 tx->wake_queue++;
1466 netif_tx_wake_queue(dev_queue);
1467 }
1468 }
1469
1470 static inline int
1471 myri10ge_clean_rx_done(struct myri10ge_slice_state *ss, int budget)
1472 {
1473 struct myri10ge_rx_done *rx_done = &ss->rx_done;
1474 struct myri10ge_priv *mgp = ss->mgp;
1475
1476 unsigned long rx_bytes = 0;
1477 unsigned long rx_packets = 0;
1478 unsigned long rx_ok;
1479
1480 int idx = rx_done->idx;
1481 int cnt = rx_done->cnt;
1482 int work_done = 0;
1483 u16 length;
1484 __wsum checksum;
1485
1486 /*
1487 * Prevent compiler from generating more than one ->features memory
1488 * access to avoid theoretical race condition with functions that
1489 * change NETIF_F_LRO flag at runtime.
1490 */
1491 bool lro_enabled = ACCESS_ONCE(mgp->dev->features) & NETIF_F_LRO;
1492
1493 while (rx_done->entry[idx].length != 0 && work_done < budget) {
1494 length = ntohs(rx_done->entry[idx].length);
1495 rx_done->entry[idx].length = 0;
1496 checksum = csum_unfold(rx_done->entry[idx].checksum);
1497 rx_ok = myri10ge_rx_done(ss, length, checksum, lro_enabled);
1498 rx_packets += rx_ok;
1499 rx_bytes += rx_ok * (unsigned long)length;
1500 cnt++;
1501 idx = cnt & (mgp->max_intr_slots - 1);
1502 work_done++;
1503 }
1504 rx_done->idx = idx;
1505 rx_done->cnt = cnt;
1506 ss->stats.rx_packets += rx_packets;
1507 ss->stats.rx_bytes += rx_bytes;
1508
1509 if (lro_enabled)
1510 lro_flush_all(&rx_done->lro_mgr);
1511
1512 /* restock receive rings if needed */
1513 if (ss->rx_small.fill_cnt - ss->rx_small.cnt < myri10ge_fill_thresh)
1514 myri10ge_alloc_rx_pages(mgp, &ss->rx_small,
1515 mgp->small_bytes + MXGEFW_PAD, 0);
1516 if (ss->rx_big.fill_cnt - ss->rx_big.cnt < myri10ge_fill_thresh)
1517 myri10ge_alloc_rx_pages(mgp, &ss->rx_big, mgp->big_bytes, 0);
1518
1519 return work_done;
1520 }
1521
1522 static inline void myri10ge_check_statblock(struct myri10ge_priv *mgp)
1523 {
1524 struct mcp_irq_data *stats = mgp->ss[0].fw_stats;
1525
1526 if (unlikely(stats->stats_updated)) {
1527 unsigned link_up = ntohl(stats->link_up);
1528 if (mgp->link_state != link_up) {
1529 mgp->link_state = link_up;
1530
1531 if (mgp->link_state == MXGEFW_LINK_UP) {
1532 if (netif_msg_link(mgp))
1533 netdev_info(mgp->dev, "link up\n");
1534 netif_carrier_on(mgp->dev);
1535 mgp->link_changes++;
1536 } else {
1537 if (netif_msg_link(mgp))
1538 netdev_info(mgp->dev, "link %s\n",
1539 link_up == MXGEFW_LINK_MYRINET ?
1540 "mismatch (Myrinet detected)" :
1541 "down");
1542 netif_carrier_off(mgp->dev);
1543 mgp->link_changes++;
1544 }
1545 }
1546 if (mgp->rdma_tags_available !=
1547 ntohl(stats->rdma_tags_available)) {
1548 mgp->rdma_tags_available =
1549 ntohl(stats->rdma_tags_available);
1550 netdev_warn(mgp->dev, "RDMA timed out! %d tags left\n",
1551 mgp->rdma_tags_available);
1552 }
1553 mgp->down_cnt += stats->link_down;
1554 if (stats->link_down)
1555 wake_up(&mgp->down_wq);
1556 }
1557 }
1558
1559 static int myri10ge_poll(struct napi_struct *napi, int budget)
1560 {
1561 struct myri10ge_slice_state *ss =
1562 container_of(napi, struct myri10ge_slice_state, napi);
1563 int work_done;
1564
1565 #ifdef CONFIG_MYRI10GE_DCA
1566 if (ss->mgp->dca_enabled)
1567 myri10ge_update_dca(ss);
1568 #endif
1569
1570 /* process as many rx events as NAPI will allow */
1571 work_done = myri10ge_clean_rx_done(ss, budget);
1572
1573 if (work_done < budget) {
1574 napi_complete(napi);
1575 put_be32(htonl(3), ss->irq_claim);
1576 }
1577 return work_done;
1578 }
1579
1580 static irqreturn_t myri10ge_intr(int irq, void *arg)
1581 {
1582 struct myri10ge_slice_state *ss = arg;
1583 struct myri10ge_priv *mgp = ss->mgp;
1584 struct mcp_irq_data *stats = ss->fw_stats;
1585 struct myri10ge_tx_buf *tx = &ss->tx;
1586 u32 send_done_count;
1587 int i;
1588
1589 /* an interrupt on a non-zero receive-only slice is implicitly
1590 * valid since MSI-X irqs are not shared */
1591 if ((mgp->dev->real_num_tx_queues == 1) && (ss != mgp->ss)) {
1592 napi_schedule(&ss->napi);
1593 return IRQ_HANDLED;
1594 }
1595
1596 /* make sure it is our IRQ, and that the DMA has finished */
1597 if (unlikely(!stats->valid))
1598 return IRQ_NONE;
1599
1600 /* low bit indicates receives are present, so schedule
1601 * napi poll handler */
1602 if (stats->valid & 1)
1603 napi_schedule(&ss->napi);
1604
1605 if (!mgp->msi_enabled && !mgp->msix_enabled) {
1606 put_be32(0, mgp->irq_deassert);
1607 if (!myri10ge_deassert_wait)
1608 stats->valid = 0;
1609 mb();
1610 } else
1611 stats->valid = 0;
1612
1613 /* Wait for IRQ line to go low, if using INTx */
1614 i = 0;
1615 while (1) {
1616 i++;
1617 /* check for transmit completes and receives */
1618 send_done_count = ntohl(stats->send_done_count);
1619 if (send_done_count != tx->pkt_done)
1620 myri10ge_tx_done(ss, (int)send_done_count);
1621 if (unlikely(i > myri10ge_max_irq_loops)) {
1622 netdev_err(mgp->dev, "irq stuck?\n");
1623 stats->valid = 0;
1624 schedule_work(&mgp->watchdog_work);
1625 }
1626 if (likely(stats->valid == 0))
1627 break;
1628 cpu_relax();
1629 barrier();
1630 }
1631
1632 /* Only slice 0 updates stats */
1633 if (ss == mgp->ss)
1634 myri10ge_check_statblock(mgp);
1635
1636 put_be32(htonl(3), ss->irq_claim + 1);
1637 return IRQ_HANDLED;
1638 }
1639
1640 static int
1641 myri10ge_get_settings(struct net_device *netdev, struct ethtool_cmd *cmd)
1642 {
1643 struct myri10ge_priv *mgp = netdev_priv(netdev);
1644 char *ptr;
1645 int i;
1646
1647 cmd->autoneg = AUTONEG_DISABLE;
1648 ethtool_cmd_speed_set(cmd, SPEED_10000);
1649 cmd->duplex = DUPLEX_FULL;
1650
1651 /*
1652 * parse the product code to deterimine the interface type
1653 * (CX4, XFP, Quad Ribbon Fiber) by looking at the character
1654 * after the 3rd dash in the driver's cached copy of the
1655 * EEPROM's product code string.
1656 */
1657 ptr = mgp->product_code_string;
1658 if (ptr == NULL) {
1659 netdev_err(netdev, "Missing product code\n");
1660 return 0;
1661 }
1662 for (i = 0; i < 3; i++, ptr++) {
1663 ptr = strchr(ptr, '-');
1664 if (ptr == NULL) {
1665 netdev_err(netdev, "Invalid product code %s\n",
1666 mgp->product_code_string);
1667 return 0;
1668 }
1669 }
1670 if (*ptr == '2')
1671 ptr++;
1672 if (*ptr == 'R' || *ptr == 'Q' || *ptr == 'S') {
1673 /* We've found either an XFP, quad ribbon fiber, or SFP+ */
1674 cmd->port = PORT_FIBRE;
1675 cmd->supported |= SUPPORTED_FIBRE;
1676 cmd->advertising |= ADVERTISED_FIBRE;
1677 } else {
1678 cmd->port = PORT_OTHER;
1679 }
1680 if (*ptr == 'R' || *ptr == 'S')
1681 cmd->transceiver = XCVR_EXTERNAL;
1682 else
1683 cmd->transceiver = XCVR_INTERNAL;
1684
1685 return 0;
1686 }
1687
1688 static void
1689 myri10ge_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *info)
1690 {
1691 struct myri10ge_priv *mgp = netdev_priv(netdev);
1692
1693 strlcpy(info->driver, "myri10ge", sizeof(info->driver));
1694 strlcpy(info->version, MYRI10GE_VERSION_STR, sizeof(info->version));
1695 strlcpy(info->fw_version, mgp->fw_version, sizeof(info->fw_version));
1696 strlcpy(info->bus_info, pci_name(mgp->pdev), sizeof(info->bus_info));
1697 }
1698
1699 static int
1700 myri10ge_get_coalesce(struct net_device *netdev, struct ethtool_coalesce *coal)
1701 {
1702 struct myri10ge_priv *mgp = netdev_priv(netdev);
1703
1704 coal->rx_coalesce_usecs = mgp->intr_coal_delay;
1705 return 0;
1706 }
1707
1708 static int
1709 myri10ge_set_coalesce(struct net_device *netdev, struct ethtool_coalesce *coal)
1710 {
1711 struct myri10ge_priv *mgp = netdev_priv(netdev);
1712
1713 mgp->intr_coal_delay = coal->rx_coalesce_usecs;
1714 put_be32(htonl(mgp->intr_coal_delay), mgp->intr_coal_delay_ptr);
1715 return 0;
1716 }
1717
1718 static void
1719 myri10ge_get_pauseparam(struct net_device *netdev,
1720 struct ethtool_pauseparam *pause)
1721 {
1722 struct myri10ge_priv *mgp = netdev_priv(netdev);
1723
1724 pause->autoneg = 0;
1725 pause->rx_pause = mgp->pause;
1726 pause->tx_pause = mgp->pause;
1727 }
1728
1729 static int
1730 myri10ge_set_pauseparam(struct net_device *netdev,
1731 struct ethtool_pauseparam *pause)
1732 {
1733 struct myri10ge_priv *mgp = netdev_priv(netdev);
1734
1735 if (pause->tx_pause != mgp->pause)
1736 return myri10ge_change_pause(mgp, pause->tx_pause);
1737 if (pause->rx_pause != mgp->pause)
1738 return myri10ge_change_pause(mgp, pause->rx_pause);
1739 if (pause->autoneg != 0)
1740 return -EINVAL;
1741 return 0;
1742 }
1743
1744 static void
1745 myri10ge_get_ringparam(struct net_device *netdev,
1746 struct ethtool_ringparam *ring)
1747 {
1748 struct myri10ge_priv *mgp = netdev_priv(netdev);
1749
1750 ring->rx_mini_max_pending = mgp->ss[0].rx_small.mask + 1;
1751 ring->rx_max_pending = mgp->ss[0].rx_big.mask + 1;
1752 ring->rx_jumbo_max_pending = 0;
1753 ring->tx_max_pending = mgp->ss[0].tx.mask + 1;
1754 ring->rx_mini_pending = ring->rx_mini_max_pending;
1755 ring->rx_pending = ring->rx_max_pending;
1756 ring->rx_jumbo_pending = ring->rx_jumbo_max_pending;
1757 ring->tx_pending = ring->tx_max_pending;
1758 }
1759
1760 static const char myri10ge_gstrings_main_stats[][ETH_GSTRING_LEN] = {
1761 "rx_packets", "tx_packets", "rx_bytes", "tx_bytes", "rx_errors",
1762 "tx_errors", "rx_dropped", "tx_dropped", "multicast", "collisions",
1763 "rx_length_errors", "rx_over_errors", "rx_crc_errors",
1764 "rx_frame_errors", "rx_fifo_errors", "rx_missed_errors",
1765 "tx_aborted_errors", "tx_carrier_errors", "tx_fifo_errors",
1766 "tx_heartbeat_errors", "tx_window_errors",
1767 /* device-specific stats */
1768 "tx_boundary", "WC", "irq", "MSI", "MSIX",
1769 "read_dma_bw_MBs", "write_dma_bw_MBs", "read_write_dma_bw_MBs",
1770 "serial_number", "watchdog_resets",
1771 #ifdef CONFIG_MYRI10GE_DCA
1772 "dca_capable_firmware", "dca_device_present",
1773 #endif
1774 "link_changes", "link_up", "dropped_link_overflow",
1775 "dropped_link_error_or_filtered",
1776 "dropped_pause", "dropped_bad_phy", "dropped_bad_crc32",
1777 "dropped_unicast_filtered", "dropped_multicast_filtered",
1778 "dropped_runt", "dropped_overrun", "dropped_no_small_buffer",
1779 "dropped_no_big_buffer"
1780 };
1781
1782 static const char myri10ge_gstrings_slice_stats[][ETH_GSTRING_LEN] = {
1783 "----------- slice ---------",
1784 "tx_pkt_start", "tx_pkt_done", "tx_req", "tx_done",
1785 "rx_small_cnt", "rx_big_cnt",
1786 "wake_queue", "stop_queue", "tx_linearized", "LRO aggregated",
1787 "LRO flushed",
1788 "LRO avg aggr", "LRO no_desc"
1789 };
1790
1791 #define MYRI10GE_NET_STATS_LEN 21
1792 #define MYRI10GE_MAIN_STATS_LEN ARRAY_SIZE(myri10ge_gstrings_main_stats)
1793 #define MYRI10GE_SLICE_STATS_LEN ARRAY_SIZE(myri10ge_gstrings_slice_stats)
1794
1795 static void
1796 myri10ge_get_strings(struct net_device *netdev, u32 stringset, u8 * data)
1797 {
1798 struct myri10ge_priv *mgp = netdev_priv(netdev);
1799 int i;
1800
1801 switch (stringset) {
1802 case ETH_SS_STATS:
1803 memcpy(data, *myri10ge_gstrings_main_stats,
1804 sizeof(myri10ge_gstrings_main_stats));
1805 data += sizeof(myri10ge_gstrings_main_stats);
1806 for (i = 0; i < mgp->num_slices; i++) {
1807 memcpy(data, *myri10ge_gstrings_slice_stats,
1808 sizeof(myri10ge_gstrings_slice_stats));
1809 data += sizeof(myri10ge_gstrings_slice_stats);
1810 }
1811 break;
1812 }
1813 }
1814
1815 static int myri10ge_get_sset_count(struct net_device *netdev, int sset)
1816 {
1817 struct myri10ge_priv *mgp = netdev_priv(netdev);
1818
1819 switch (sset) {
1820 case ETH_SS_STATS:
1821 return MYRI10GE_MAIN_STATS_LEN +
1822 mgp->num_slices * MYRI10GE_SLICE_STATS_LEN;
1823 default:
1824 return -EOPNOTSUPP;
1825 }
1826 }
1827
1828 static void
1829 myri10ge_get_ethtool_stats(struct net_device *netdev,
1830 struct ethtool_stats *stats, u64 * data)
1831 {
1832 struct myri10ge_priv *mgp = netdev_priv(netdev);
1833 struct myri10ge_slice_state *ss;
1834 int slice;
1835 int i;
1836
1837 /* force stats update */
1838 (void)myri10ge_get_stats(netdev);
1839 for (i = 0; i < MYRI10GE_NET_STATS_LEN; i++)
1840 data[i] = ((unsigned long *)&netdev->stats)[i];
1841
1842 data[i++] = (unsigned int)mgp->tx_boundary;
1843 data[i++] = (unsigned int)mgp->wc_enabled;
1844 data[i++] = (unsigned int)mgp->pdev->irq;
1845 data[i++] = (unsigned int)mgp->msi_enabled;
1846 data[i++] = (unsigned int)mgp->msix_enabled;
1847 data[i++] = (unsigned int)mgp->read_dma;
1848 data[i++] = (unsigned int)mgp->write_dma;
1849 data[i++] = (unsigned int)mgp->read_write_dma;
1850 data[i++] = (unsigned int)mgp->serial_number;
1851 data[i++] = (unsigned int)mgp->watchdog_resets;
1852 #ifdef CONFIG_MYRI10GE_DCA
1853 data[i++] = (unsigned int)(mgp->ss[0].dca_tag != NULL);
1854 data[i++] = (unsigned int)(mgp->dca_enabled);
1855 #endif
1856 data[i++] = (unsigned int)mgp->link_changes;
1857
1858 /* firmware stats are useful only in the first slice */
1859 ss = &mgp->ss[0];
1860 data[i++] = (unsigned int)ntohl(ss->fw_stats->link_up);
1861 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_link_overflow);
1862 data[i++] =
1863 (unsigned int)ntohl(ss->fw_stats->dropped_link_error_or_filtered);
1864 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_pause);
1865 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_bad_phy);
1866 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_bad_crc32);
1867 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_unicast_filtered);
1868 data[i++] =
1869 (unsigned int)ntohl(ss->fw_stats->dropped_multicast_filtered);
1870 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_runt);
1871 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_overrun);
1872 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_no_small_buffer);
1873 data[i++] = (unsigned int)ntohl(ss->fw_stats->dropped_no_big_buffer);
1874
1875 for (slice = 0; slice < mgp->num_slices; slice++) {
1876 ss = &mgp->ss[slice];
1877 data[i++] = slice;
1878 data[i++] = (unsigned int)ss->tx.pkt_start;
1879 data[i++] = (unsigned int)ss->tx.pkt_done;
1880 data[i++] = (unsigned int)ss->tx.req;
1881 data[i++] = (unsigned int)ss->tx.done;
1882 data[i++] = (unsigned int)ss->rx_small.cnt;
1883 data[i++] = (unsigned int)ss->rx_big.cnt;
1884 data[i++] = (unsigned int)ss->tx.wake_queue;
1885 data[i++] = (unsigned int)ss->tx.stop_queue;
1886 data[i++] = (unsigned int)ss->tx.linearized;
1887 data[i++] = ss->rx_done.lro_mgr.stats.aggregated;
1888 data[i++] = ss->rx_done.lro_mgr.stats.flushed;
1889 if (ss->rx_done.lro_mgr.stats.flushed)
1890 data[i++] = ss->rx_done.lro_mgr.stats.aggregated /
1891 ss->rx_done.lro_mgr.stats.flushed;
1892 else
1893 data[i++] = 0;
1894 data[i++] = ss->rx_done.lro_mgr.stats.no_desc;
1895 }
1896 }
1897
1898 static void myri10ge_set_msglevel(struct net_device *netdev, u32 value)
1899 {
1900 struct myri10ge_priv *mgp = netdev_priv(netdev);
1901 mgp->msg_enable = value;
1902 }
1903
1904 static u32 myri10ge_get_msglevel(struct net_device *netdev)
1905 {
1906 struct myri10ge_priv *mgp = netdev_priv(netdev);
1907 return mgp->msg_enable;
1908 }
1909
1910 static const struct ethtool_ops myri10ge_ethtool_ops = {
1911 .get_settings = myri10ge_get_settings,
1912 .get_drvinfo = myri10ge_get_drvinfo,
1913 .get_coalesce = myri10ge_get_coalesce,
1914 .set_coalesce = myri10ge_set_coalesce,
1915 .get_pauseparam = myri10ge_get_pauseparam,
1916 .set_pauseparam = myri10ge_set_pauseparam,
1917 .get_ringparam = myri10ge_get_ringparam,
1918 .get_link = ethtool_op_get_link,
1919 .get_strings = myri10ge_get_strings,
1920 .get_sset_count = myri10ge_get_sset_count,
1921 .get_ethtool_stats = myri10ge_get_ethtool_stats,
1922 .set_msglevel = myri10ge_set_msglevel,
1923 .get_msglevel = myri10ge_get_msglevel,
1924 };
1925
1926 static int myri10ge_allocate_rings(struct myri10ge_slice_state *ss)
1927 {
1928 struct myri10ge_priv *mgp = ss->mgp;
1929 struct myri10ge_cmd cmd;
1930 struct net_device *dev = mgp->dev;
1931 int tx_ring_size, rx_ring_size;
1932 int tx_ring_entries, rx_ring_entries;
1933 int i, slice, status;
1934 size_t bytes;
1935
1936 /* get ring sizes */
1937 slice = ss - mgp->ss;
1938 cmd.data0 = slice;
1939 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_SEND_RING_SIZE, &cmd, 0);
1940 tx_ring_size = cmd.data0;
1941 cmd.data0 = slice;
1942 status |= myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_RX_RING_SIZE, &cmd, 0);
1943 if (status != 0)
1944 return status;
1945 rx_ring_size = cmd.data0;
1946
1947 tx_ring_entries = tx_ring_size / sizeof(struct mcp_kreq_ether_send);
1948 rx_ring_entries = rx_ring_size / sizeof(struct mcp_dma_addr);
1949 ss->tx.mask = tx_ring_entries - 1;
1950 ss->rx_small.mask = ss->rx_big.mask = rx_ring_entries - 1;
1951
1952 status = -ENOMEM;
1953
1954 /* allocate the host shadow rings */
1955
1956 bytes = 8 + (MYRI10GE_MAX_SEND_DESC_TSO + 4)
1957 * sizeof(*ss->tx.req_list);
1958 ss->tx.req_bytes = kzalloc(bytes, GFP_KERNEL);
1959 if (ss->tx.req_bytes == NULL)
1960 goto abort_with_nothing;
1961
1962 /* ensure req_list entries are aligned to 8 bytes */
1963 ss->tx.req_list = (struct mcp_kreq_ether_send *)
1964 ALIGN((unsigned long)ss->tx.req_bytes, 8);
1965 ss->tx.queue_active = 0;
1966
1967 bytes = rx_ring_entries * sizeof(*ss->rx_small.shadow);
1968 ss->rx_small.shadow = kzalloc(bytes, GFP_KERNEL);
1969 if (ss->rx_small.shadow == NULL)
1970 goto abort_with_tx_req_bytes;
1971
1972 bytes = rx_ring_entries * sizeof(*ss->rx_big.shadow);
1973 ss->rx_big.shadow = kzalloc(bytes, GFP_KERNEL);
1974 if (ss->rx_big.shadow == NULL)
1975 goto abort_with_rx_small_shadow;
1976
1977 /* allocate the host info rings */
1978
1979 bytes = tx_ring_entries * sizeof(*ss->tx.info);
1980 ss->tx.info = kzalloc(bytes, GFP_KERNEL);
1981 if (ss->tx.info == NULL)
1982 goto abort_with_rx_big_shadow;
1983
1984 bytes = rx_ring_entries * sizeof(*ss->rx_small.info);
1985 ss->rx_small.info = kzalloc(bytes, GFP_KERNEL);
1986 if (ss->rx_small.info == NULL)
1987 goto abort_with_tx_info;
1988
1989 bytes = rx_ring_entries * sizeof(*ss->rx_big.info);
1990 ss->rx_big.info = kzalloc(bytes, GFP_KERNEL);
1991 if (ss->rx_big.info == NULL)
1992 goto abort_with_rx_small_info;
1993
1994 /* Fill the receive rings */
1995 ss->rx_big.cnt = 0;
1996 ss->rx_small.cnt = 0;
1997 ss->rx_big.fill_cnt = 0;
1998 ss->rx_small.fill_cnt = 0;
1999 ss->rx_small.page_offset = MYRI10GE_ALLOC_SIZE;
2000 ss->rx_big.page_offset = MYRI10GE_ALLOC_SIZE;
2001 ss->rx_small.watchdog_needed = 0;
2002 ss->rx_big.watchdog_needed = 0;
2003 myri10ge_alloc_rx_pages(mgp, &ss->rx_small,
2004 mgp->small_bytes + MXGEFW_PAD, 0);
2005
2006 if (ss->rx_small.fill_cnt < ss->rx_small.mask + 1) {
2007 netdev_err(dev, "slice-%d: alloced only %d small bufs\n",
2008 slice, ss->rx_small.fill_cnt);
2009 goto abort_with_rx_small_ring;
2010 }
2011
2012 myri10ge_alloc_rx_pages(mgp, &ss->rx_big, mgp->big_bytes, 0);
2013 if (ss->rx_big.fill_cnt < ss->rx_big.mask + 1) {
2014 netdev_err(dev, "slice-%d: alloced only %d big bufs\n",
2015 slice, ss->rx_big.fill_cnt);
2016 goto abort_with_rx_big_ring;
2017 }
2018
2019 return 0;
2020
2021 abort_with_rx_big_ring:
2022 for (i = ss->rx_big.cnt; i < ss->rx_big.fill_cnt; i++) {
2023 int idx = i & ss->rx_big.mask;
2024 myri10ge_unmap_rx_page(mgp->pdev, &ss->rx_big.info[idx],
2025 mgp->big_bytes);
2026 put_page(ss->rx_big.info[idx].page);
2027 }
2028
2029 abort_with_rx_small_ring:
2030 for (i = ss->rx_small.cnt; i < ss->rx_small.fill_cnt; i++) {
2031 int idx = i & ss->rx_small.mask;
2032 myri10ge_unmap_rx_page(mgp->pdev, &ss->rx_small.info[idx],
2033 mgp->small_bytes + MXGEFW_PAD);
2034 put_page(ss->rx_small.info[idx].page);
2035 }
2036
2037 kfree(ss->rx_big.info);
2038
2039 abort_with_rx_small_info:
2040 kfree(ss->rx_small.info);
2041
2042 abort_with_tx_info:
2043 kfree(ss->tx.info);
2044
2045 abort_with_rx_big_shadow:
2046 kfree(ss->rx_big.shadow);
2047
2048 abort_with_rx_small_shadow:
2049 kfree(ss->rx_small.shadow);
2050
2051 abort_with_tx_req_bytes:
2052 kfree(ss->tx.req_bytes);
2053 ss->tx.req_bytes = NULL;
2054 ss->tx.req_list = NULL;
2055
2056 abort_with_nothing:
2057 return status;
2058 }
2059
2060 static void myri10ge_free_rings(struct myri10ge_slice_state *ss)
2061 {
2062 struct myri10ge_priv *mgp = ss->mgp;
2063 struct sk_buff *skb;
2064 struct myri10ge_tx_buf *tx;
2065 int i, len, idx;
2066
2067 /* If not allocated, skip it */
2068 if (ss->tx.req_list == NULL)
2069 return;
2070
2071 for (i = ss->rx_big.cnt; i < ss->rx_big.fill_cnt; i++) {
2072 idx = i & ss->rx_big.mask;
2073 if (i == ss->rx_big.fill_cnt - 1)
2074 ss->rx_big.info[idx].page_offset = MYRI10GE_ALLOC_SIZE;
2075 myri10ge_unmap_rx_page(mgp->pdev, &ss->rx_big.info[idx],
2076 mgp->big_bytes);
2077 put_page(ss->rx_big.info[idx].page);
2078 }
2079
2080 for (i = ss->rx_small.cnt; i < ss->rx_small.fill_cnt; i++) {
2081 idx = i & ss->rx_small.mask;
2082 if (i == ss->rx_small.fill_cnt - 1)
2083 ss->rx_small.info[idx].page_offset =
2084 MYRI10GE_ALLOC_SIZE;
2085 myri10ge_unmap_rx_page(mgp->pdev, &ss->rx_small.info[idx],
2086 mgp->small_bytes + MXGEFW_PAD);
2087 put_page(ss->rx_small.info[idx].page);
2088 }
2089 tx = &ss->tx;
2090 while (tx->done != tx->req) {
2091 idx = tx->done & tx->mask;
2092 skb = tx->info[idx].skb;
2093
2094 /* Mark as free */
2095 tx->info[idx].skb = NULL;
2096 tx->done++;
2097 len = dma_unmap_len(&tx->info[idx], len);
2098 dma_unmap_len_set(&tx->info[idx], len, 0);
2099 if (skb) {
2100 ss->stats.tx_dropped++;
2101 dev_kfree_skb_any(skb);
2102 if (len)
2103 pci_unmap_single(mgp->pdev,
2104 dma_unmap_addr(&tx->info[idx],
2105 bus), len,
2106 PCI_DMA_TODEVICE);
2107 } else {
2108 if (len)
2109 pci_unmap_page(mgp->pdev,
2110 dma_unmap_addr(&tx->info[idx],
2111 bus), len,
2112 PCI_DMA_TODEVICE);
2113 }
2114 }
2115 kfree(ss->rx_big.info);
2116
2117 kfree(ss->rx_small.info);
2118
2119 kfree(ss->tx.info);
2120
2121 kfree(ss->rx_big.shadow);
2122
2123 kfree(ss->rx_small.shadow);
2124
2125 kfree(ss->tx.req_bytes);
2126 ss->tx.req_bytes = NULL;
2127 ss->tx.req_list = NULL;
2128 }
2129
2130 static int myri10ge_request_irq(struct myri10ge_priv *mgp)
2131 {
2132 struct pci_dev *pdev = mgp->pdev;
2133 struct myri10ge_slice_state *ss;
2134 struct net_device *netdev = mgp->dev;
2135 int i;
2136 int status;
2137
2138 mgp->msi_enabled = 0;
2139 mgp->msix_enabled = 0;
2140 status = 0;
2141 if (myri10ge_msi) {
2142 if (mgp->num_slices > 1) {
2143 status =
2144 pci_enable_msix(pdev, mgp->msix_vectors,
2145 mgp->num_slices);
2146 if (status == 0) {
2147 mgp->msix_enabled = 1;
2148 } else {
2149 dev_err(&pdev->dev,
2150 "Error %d setting up MSI-X\n", status);
2151 return status;
2152 }
2153 }
2154 if (mgp->msix_enabled == 0) {
2155 status = pci_enable_msi(pdev);
2156 if (status != 0) {
2157 dev_err(&pdev->dev,
2158 "Error %d setting up MSI; falling back to xPIC\n",
2159 status);
2160 } else {
2161 mgp->msi_enabled = 1;
2162 }
2163 }
2164 }
2165 if (mgp->msix_enabled) {
2166 for (i = 0; i < mgp->num_slices; i++) {
2167 ss = &mgp->ss[i];
2168 snprintf(ss->irq_desc, sizeof(ss->irq_desc),
2169 "%s:slice-%d", netdev->name, i);
2170 status = request_irq(mgp->msix_vectors[i].vector,
2171 myri10ge_intr, 0, ss->irq_desc,
2172 ss);
2173 if (status != 0) {
2174 dev_err(&pdev->dev,
2175 "slice %d failed to allocate IRQ\n", i);
2176 i--;
2177 while (i >= 0) {
2178 free_irq(mgp->msix_vectors[i].vector,
2179 &mgp->ss[i]);
2180 i--;
2181 }
2182 pci_disable_msix(pdev);
2183 return status;
2184 }
2185 }
2186 } else {
2187 status = request_irq(pdev->irq, myri10ge_intr, IRQF_SHARED,
2188 mgp->dev->name, &mgp->ss[0]);
2189 if (status != 0) {
2190 dev_err(&pdev->dev, "failed to allocate IRQ\n");
2191 if (mgp->msi_enabled)
2192 pci_disable_msi(pdev);
2193 }
2194 }
2195 return status;
2196 }
2197
2198 static void myri10ge_free_irq(struct myri10ge_priv *mgp)
2199 {
2200 struct pci_dev *pdev = mgp->pdev;
2201 int i;
2202
2203 if (mgp->msix_enabled) {
2204 for (i = 0; i < mgp->num_slices; i++)
2205 free_irq(mgp->msix_vectors[i].vector, &mgp->ss[i]);
2206 } else {
2207 free_irq(pdev->irq, &mgp->ss[0]);
2208 }
2209 if (mgp->msi_enabled)
2210 pci_disable_msi(pdev);
2211 if (mgp->msix_enabled)
2212 pci_disable_msix(pdev);
2213 }
2214
2215 static int
2216 myri10ge_get_frag_header(struct skb_frag_struct *frag, void **mac_hdr,
2217 void **ip_hdr, void **tcpudp_hdr,
2218 u64 * hdr_flags, void *priv)
2219 {
2220 struct ethhdr *eh;
2221 struct vlan_ethhdr *veh;
2222 struct iphdr *iph;
2223 u8 *va = page_address(frag->page) + frag->page_offset;
2224 unsigned long ll_hlen;
2225 /* passed opaque through lro_receive_frags() */
2226 __wsum csum = (__force __wsum) (unsigned long)priv;
2227
2228 /* find the mac header, aborting if not IPv4 */
2229
2230 eh = (struct ethhdr *)va;
2231 *mac_hdr = eh;
2232 ll_hlen = ETH_HLEN;
2233 if (eh->h_proto != htons(ETH_P_IP)) {
2234 if (eh->h_proto == htons(ETH_P_8021Q)) {
2235 veh = (struct vlan_ethhdr *)va;
2236 if (veh->h_vlan_encapsulated_proto != htons(ETH_P_IP))
2237 return -1;
2238
2239 ll_hlen += VLAN_HLEN;
2240
2241 /*
2242 * HW checksum starts ETH_HLEN bytes into
2243 * frame, so we must subtract off the VLAN
2244 * header's checksum before csum can be used
2245 */
2246 csum = csum_sub(csum, csum_partial(va + ETH_HLEN,
2247 VLAN_HLEN, 0));
2248 } else {
2249 return -1;
2250 }
2251 }
2252 *hdr_flags = LRO_IPV4;
2253
2254 iph = (struct iphdr *)(va + ll_hlen);
2255 *ip_hdr = iph;
2256 if (iph->protocol != IPPROTO_TCP)
2257 return -1;
2258 if (iph->frag_off & htons(IP_MF | IP_OFFSET))
2259 return -1;
2260 *hdr_flags |= LRO_TCP;
2261 *tcpudp_hdr = (u8 *) (*ip_hdr) + (iph->ihl << 2);
2262
2263 /* verify the IP checksum */
2264 if (unlikely(ip_fast_csum((u8 *) iph, iph->ihl)))
2265 return -1;
2266
2267 /* verify the checksum */
2268 if (unlikely(csum_tcpudp_magic(iph->saddr, iph->daddr,
2269 ntohs(iph->tot_len) - (iph->ihl << 2),
2270 IPPROTO_TCP, csum)))
2271 return -1;
2272
2273 return 0;
2274 }
2275
2276 static int myri10ge_get_txrx(struct myri10ge_priv *mgp, int slice)
2277 {
2278 struct myri10ge_cmd cmd;
2279 struct myri10ge_slice_state *ss;
2280 int status;
2281
2282 ss = &mgp->ss[slice];
2283 status = 0;
2284 if (slice == 0 || (mgp->dev->real_num_tx_queues > 1)) {
2285 cmd.data0 = slice;
2286 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_SEND_OFFSET,
2287 &cmd, 0);
2288 ss->tx.lanai = (struct mcp_kreq_ether_send __iomem *)
2289 (mgp->sram + cmd.data0);
2290 }
2291 cmd.data0 = slice;
2292 status |= myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_SMALL_RX_OFFSET,
2293 &cmd, 0);
2294 ss->rx_small.lanai = (struct mcp_kreq_ether_recv __iomem *)
2295 (mgp->sram + cmd.data0);
2296
2297 cmd.data0 = slice;
2298 status |= myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_BIG_RX_OFFSET, &cmd, 0);
2299 ss->rx_big.lanai = (struct mcp_kreq_ether_recv __iomem *)
2300 (mgp->sram + cmd.data0);
2301
2302 ss->tx.send_go = (__iomem __be32 *)
2303 (mgp->sram + MXGEFW_ETH_SEND_GO + 64 * slice);
2304 ss->tx.send_stop = (__iomem __be32 *)
2305 (mgp->sram + MXGEFW_ETH_SEND_STOP + 64 * slice);
2306 return status;
2307
2308 }
2309
2310 static int myri10ge_set_stats(struct myri10ge_priv *mgp, int slice)
2311 {
2312 struct myri10ge_cmd cmd;
2313 struct myri10ge_slice_state *ss;
2314 int status;
2315
2316 ss = &mgp->ss[slice];
2317 cmd.data0 = MYRI10GE_LOWPART_TO_U32(ss->fw_stats_bus);
2318 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(ss->fw_stats_bus);
2319 cmd.data2 = sizeof(struct mcp_irq_data) | (slice << 16);
2320 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_STATS_DMA_V2, &cmd, 0);
2321 if (status == -ENOSYS) {
2322 dma_addr_t bus = ss->fw_stats_bus;
2323 if (slice != 0)
2324 return -EINVAL;
2325 bus += offsetof(struct mcp_irq_data, send_done_count);
2326 cmd.data0 = MYRI10GE_LOWPART_TO_U32(bus);
2327 cmd.data1 = MYRI10GE_HIGHPART_TO_U32(bus);
2328 status = myri10ge_send_cmd(mgp,
2329 MXGEFW_CMD_SET_STATS_DMA_OBSOLETE,
2330 &cmd, 0);
2331 /* Firmware cannot support multicast without STATS_DMA_V2 */
2332 mgp->fw_multicast_support = 0;
2333 } else {
2334 mgp->fw_multicast_support = 1;
2335 }
2336 return 0;
2337 }
2338
2339 static int myri10ge_open(struct net_device *dev)
2340 {
2341 struct myri10ge_slice_state *ss;
2342 struct myri10ge_priv *mgp = netdev_priv(dev);
2343 struct myri10ge_cmd cmd;
2344 int i, status, big_pow2, slice;
2345 u8 *itable;
2346 struct net_lro_mgr *lro_mgr;
2347
2348 if (mgp->running != MYRI10GE_ETH_STOPPED)
2349 return -EBUSY;
2350
2351 mgp->running = MYRI10GE_ETH_STARTING;
2352 status = myri10ge_reset(mgp);
2353 if (status != 0) {
2354 netdev_err(dev, "failed reset\n");
2355 goto abort_with_nothing;
2356 }
2357
2358 if (mgp->num_slices > 1) {
2359 cmd.data0 = mgp->num_slices;
2360 cmd.data1 = MXGEFW_SLICE_INTR_MODE_ONE_PER_SLICE;
2361 if (mgp->dev->real_num_tx_queues > 1)
2362 cmd.data1 |= MXGEFW_SLICE_ENABLE_MULTIPLE_TX_QUEUES;
2363 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_ENABLE_RSS_QUEUES,
2364 &cmd, 0);
2365 if (status != 0) {
2366 netdev_err(dev, "failed to set number of slices\n");
2367 goto abort_with_nothing;
2368 }
2369 /* setup the indirection table */
2370 cmd.data0 = mgp->num_slices;
2371 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_RSS_TABLE_SIZE,
2372 &cmd, 0);
2373
2374 status |= myri10ge_send_cmd(mgp,
2375 MXGEFW_CMD_GET_RSS_TABLE_OFFSET,
2376 &cmd, 0);
2377 if (status != 0) {
2378 netdev_err(dev, "failed to setup rss tables\n");
2379 goto abort_with_nothing;
2380 }
2381
2382 /* just enable an identity mapping */
2383 itable = mgp->sram + cmd.data0;
2384 for (i = 0; i < mgp->num_slices; i++)
2385 __raw_writeb(i, &itable[i]);
2386
2387 cmd.data0 = 1;
2388 cmd.data1 = myri10ge_rss_hash;
2389 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_RSS_ENABLE,
2390 &cmd, 0);
2391 if (status != 0) {
2392 netdev_err(dev, "failed to enable slices\n");
2393 goto abort_with_nothing;
2394 }
2395 }
2396
2397 status = myri10ge_request_irq(mgp);
2398 if (status != 0)
2399 goto abort_with_nothing;
2400
2401 /* decide what small buffer size to use. For good TCP rx
2402 * performance, it is important to not receive 1514 byte
2403 * frames into jumbo buffers, as it confuses the socket buffer
2404 * accounting code, leading to drops and erratic performance.
2405 */
2406
2407 if (dev->mtu <= ETH_DATA_LEN)
2408 /* enough for a TCP header */
2409 mgp->small_bytes = (128 > SMP_CACHE_BYTES)
2410 ? (128 - MXGEFW_PAD)
2411 : (SMP_CACHE_BYTES - MXGEFW_PAD);
2412 else
2413 /* enough for a vlan encapsulated ETH_DATA_LEN frame */
2414 mgp->small_bytes = VLAN_ETH_FRAME_LEN;
2415
2416 /* Override the small buffer size? */
2417 if (myri10ge_small_bytes > 0)
2418 mgp->small_bytes = myri10ge_small_bytes;
2419
2420 /* Firmware needs the big buff size as a power of 2. Lie and
2421 * tell him the buffer is larger, because we only use 1
2422 * buffer/pkt, and the mtu will prevent overruns.
2423 */
2424 big_pow2 = dev->mtu + ETH_HLEN + VLAN_HLEN + MXGEFW_PAD;
2425 if (big_pow2 < MYRI10GE_ALLOC_SIZE / 2) {
2426 while (!is_power_of_2(big_pow2))
2427 big_pow2++;
2428 mgp->big_bytes = dev->mtu + ETH_HLEN + VLAN_HLEN + MXGEFW_PAD;
2429 } else {
2430 big_pow2 = MYRI10GE_ALLOC_SIZE;
2431 mgp->big_bytes = big_pow2;
2432 }
2433
2434 /* setup the per-slice data structures */
2435 for (slice = 0; slice < mgp->num_slices; slice++) {
2436 ss = &mgp->ss[slice];
2437
2438 status = myri10ge_get_txrx(mgp, slice);
2439 if (status != 0) {
2440 netdev_err(dev, "failed to get ring sizes or locations\n");
2441 goto abort_with_rings;
2442 }
2443 status = myri10ge_allocate_rings(ss);
2444 if (status != 0)
2445 goto abort_with_rings;
2446
2447 /* only firmware which supports multiple TX queues
2448 * supports setting up the tx stats on non-zero
2449 * slices */
2450 if (slice == 0 || mgp->dev->real_num_tx_queues > 1)
2451 status = myri10ge_set_stats(mgp, slice);
2452 if (status) {
2453 netdev_err(dev, "Couldn't set stats DMA\n");
2454 goto abort_with_rings;
2455 }
2456
2457 lro_mgr = &ss->rx_done.lro_mgr;
2458 lro_mgr->dev = dev;
2459 lro_mgr->features = LRO_F_NAPI;
2460 lro_mgr->ip_summed = CHECKSUM_COMPLETE;
2461 lro_mgr->ip_summed_aggr = CHECKSUM_UNNECESSARY;
2462 lro_mgr->max_desc = MYRI10GE_MAX_LRO_DESCRIPTORS;
2463 lro_mgr->lro_arr = ss->rx_done.lro_desc;
2464 lro_mgr->get_frag_header = myri10ge_get_frag_header;
2465 lro_mgr->max_aggr = myri10ge_lro_max_pkts;
2466 lro_mgr->frag_align_pad = 2;
2467 if (lro_mgr->max_aggr > MAX_SKB_FRAGS)
2468 lro_mgr->max_aggr = MAX_SKB_FRAGS;
2469
2470 /* must happen prior to any irq */
2471 napi_enable(&(ss)->napi);
2472 }
2473
2474 /* now give firmware buffers sizes, and MTU */
2475 cmd.data0 = dev->mtu + ETH_HLEN + VLAN_HLEN;
2476 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_MTU, &cmd, 0);
2477 cmd.data0 = mgp->small_bytes;
2478 status |=
2479 myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_SMALL_BUFFER_SIZE, &cmd, 0);
2480 cmd.data0 = big_pow2;
2481 status |=
2482 myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_BIG_BUFFER_SIZE, &cmd, 0);
2483 if (status) {
2484 netdev_err(dev, "Couldn't set buffer sizes\n");
2485 goto abort_with_rings;
2486 }
2487
2488 /*
2489 * Set Linux style TSO mode; this is needed only on newer
2490 * firmware versions. Older versions default to Linux
2491 * style TSO
2492 */
2493 cmd.data0 = 0;
2494 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_TSO_MODE, &cmd, 0);
2495 if (status && status != -ENOSYS) {
2496 netdev_err(dev, "Couldn't set TSO mode\n");
2497 goto abort_with_rings;
2498 }
2499
2500 mgp->link_state = ~0U;
2501 mgp->rdma_tags_available = 15;
2502
2503 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_ETHERNET_UP, &cmd, 0);
2504 if (status) {
2505 netdev_err(dev, "Couldn't bring up link\n");
2506 goto abort_with_rings;
2507 }
2508
2509 mgp->running = MYRI10GE_ETH_RUNNING;
2510 mgp->watchdog_timer.expires = jiffies + myri10ge_watchdog_timeout * HZ;
2511 add_timer(&mgp->watchdog_timer);
2512 netif_tx_wake_all_queues(dev);
2513
2514 return 0;
2515
2516 abort_with_rings:
2517 while (slice) {
2518 slice--;
2519 napi_disable(&mgp->ss[slice].napi);
2520 }
2521 for (i = 0; i < mgp->num_slices; i++)
2522 myri10ge_free_rings(&mgp->ss[i]);
2523
2524 myri10ge_free_irq(mgp);
2525
2526 abort_with_nothing:
2527 mgp->running = MYRI10GE_ETH_STOPPED;
2528 return -ENOMEM;
2529 }
2530
2531 static int myri10ge_close(struct net_device *dev)
2532 {
2533 struct myri10ge_priv *mgp = netdev_priv(dev);
2534 struct myri10ge_cmd cmd;
2535 int status, old_down_cnt;
2536 int i;
2537
2538 if (mgp->running != MYRI10GE_ETH_RUNNING)
2539 return 0;
2540
2541 if (mgp->ss[0].tx.req_bytes == NULL)
2542 return 0;
2543
2544 del_timer_sync(&mgp->watchdog_timer);
2545 mgp->running = MYRI10GE_ETH_STOPPING;
2546 for (i = 0; i < mgp->num_slices; i++) {
2547 napi_disable(&mgp->ss[i].napi);
2548 }
2549 netif_carrier_off(dev);
2550
2551 netif_tx_stop_all_queues(dev);
2552 if (mgp->rebooted == 0) {
2553 old_down_cnt = mgp->down_cnt;
2554 mb();
2555 status =
2556 myri10ge_send_cmd(mgp, MXGEFW_CMD_ETHERNET_DOWN, &cmd, 0);
2557 if (status)
2558 netdev_err(dev, "Couldn't bring down link\n");
2559
2560 wait_event_timeout(mgp->down_wq, old_down_cnt != mgp->down_cnt,
2561 HZ);
2562 if (old_down_cnt == mgp->down_cnt)
2563 netdev_err(dev, "never got down irq\n");
2564 }
2565 netif_tx_disable(dev);
2566 myri10ge_free_irq(mgp);
2567 for (i = 0; i < mgp->num_slices; i++)
2568 myri10ge_free_rings(&mgp->ss[i]);
2569
2570 mgp->running = MYRI10GE_ETH_STOPPED;
2571 return 0;
2572 }
2573
2574 /* copy an array of struct mcp_kreq_ether_send's to the mcp. Copy
2575 * backwards one at a time and handle ring wraps */
2576
2577 static inline void
2578 myri10ge_submit_req_backwards(struct myri10ge_tx_buf *tx,
2579 struct mcp_kreq_ether_send *src, int cnt)
2580 {
2581 int idx, starting_slot;
2582 starting_slot = tx->req;
2583 while (cnt > 1) {
2584 cnt--;
2585 idx = (starting_slot + cnt) & tx->mask;
2586 myri10ge_pio_copy(&tx->lanai[idx], &src[cnt], sizeof(*src));
2587 mb();
2588 }
2589 }
2590
2591 /*
2592 * copy an array of struct mcp_kreq_ether_send's to the mcp. Copy
2593 * at most 32 bytes at a time, so as to avoid involving the software
2594 * pio handler in the nic. We re-write the first segment's flags
2595 * to mark them valid only after writing the entire chain.
2596 */
2597
2598 static inline void
2599 myri10ge_submit_req(struct myri10ge_tx_buf *tx, struct mcp_kreq_ether_send *src,
2600 int cnt)
2601 {
2602 int idx, i;
2603 struct mcp_kreq_ether_send __iomem *dstp, *dst;
2604 struct mcp_kreq_ether_send *srcp;
2605 u8 last_flags;
2606
2607 idx = tx->req & tx->mask;
2608
2609 last_flags = src->flags;
2610 src->flags = 0;
2611 mb();
2612 dst = dstp = &tx->lanai[idx];
2613 srcp = src;
2614
2615 if ((idx + cnt) < tx->mask) {
2616 for (i = 0; i < (cnt - 1); i += 2) {
2617 myri10ge_pio_copy(dstp, srcp, 2 * sizeof(*src));
2618 mb(); /* force write every 32 bytes */
2619 srcp += 2;
2620 dstp += 2;
2621 }
2622 } else {
2623 /* submit all but the first request, and ensure
2624 * that it is submitted below */
2625 myri10ge_submit_req_backwards(tx, src, cnt);
2626 i = 0;
2627 }
2628 if (i < cnt) {
2629 /* submit the first request */
2630 myri10ge_pio_copy(dstp, srcp, sizeof(*src));
2631 mb(); /* barrier before setting valid flag */
2632 }
2633
2634 /* re-write the last 32-bits with the valid flags */
2635 src->flags = last_flags;
2636 put_be32(*((__be32 *) src + 3), (__be32 __iomem *) dst + 3);
2637 tx->req += cnt;
2638 mb();
2639 }
2640
2641 /*
2642 * Transmit a packet. We need to split the packet so that a single
2643 * segment does not cross myri10ge->tx_boundary, so this makes segment
2644 * counting tricky. So rather than try to count segments up front, we
2645 * just give up if there are too few segments to hold a reasonably
2646 * fragmented packet currently available. If we run
2647 * out of segments while preparing a packet for DMA, we just linearize
2648 * it and try again.
2649 */
2650
2651 static netdev_tx_t myri10ge_xmit(struct sk_buff *skb,
2652 struct net_device *dev)
2653 {
2654 struct myri10ge_priv *mgp = netdev_priv(dev);
2655 struct myri10ge_slice_state *ss;
2656 struct mcp_kreq_ether_send *req;
2657 struct myri10ge_tx_buf *tx;
2658 struct skb_frag_struct *frag;
2659 struct netdev_queue *netdev_queue;
2660 dma_addr_t bus;
2661 u32 low;
2662 __be32 high_swapped;
2663 unsigned int len;
2664 int idx, last_idx, avail, frag_cnt, frag_idx, count, mss, max_segments;
2665 u16 pseudo_hdr_offset, cksum_offset, queue;
2666 int cum_len, seglen, boundary, rdma_count;
2667 u8 flags, odd_flag;
2668
2669 queue = skb_get_queue_mapping(skb);
2670 ss = &mgp->ss[queue];
2671 netdev_queue = netdev_get_tx_queue(mgp->dev, queue);
2672 tx = &ss->tx;
2673
2674 again:
2675 req = tx->req_list;
2676 avail = tx->mask - 1 - (tx->req - tx->done);
2677
2678 mss = 0;
2679 max_segments = MXGEFW_MAX_SEND_DESC;
2680
2681 if (skb_is_gso(skb)) {
2682 mss = skb_shinfo(skb)->gso_size;
2683 max_segments = MYRI10GE_MAX_SEND_DESC_TSO;
2684 }
2685
2686 if ((unlikely(avail < max_segments))) {
2687 /* we are out of transmit resources */
2688 tx->stop_queue++;
2689 netif_tx_stop_queue(netdev_queue);
2690 return NETDEV_TX_BUSY;
2691 }
2692
2693 /* Setup checksum offloading, if needed */
2694 cksum_offset = 0;
2695 pseudo_hdr_offset = 0;
2696 odd_flag = 0;
2697 flags = (MXGEFW_FLAGS_NO_TSO | MXGEFW_FLAGS_FIRST);
2698 if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
2699 cksum_offset = skb_checksum_start_offset(skb);
2700 pseudo_hdr_offset = cksum_offset + skb->csum_offset;
2701 /* If the headers are excessively large, then we must
2702 * fall back to a software checksum */
2703 if (unlikely(!mss && (cksum_offset > 255 ||
2704 pseudo_hdr_offset > 127))) {
2705 if (skb_checksum_help(skb))
2706 goto drop;
2707 cksum_offset = 0;
2708 pseudo_hdr_offset = 0;
2709 } else {
2710 odd_flag = MXGEFW_FLAGS_ALIGN_ODD;
2711 flags |= MXGEFW_FLAGS_CKSUM;
2712 }
2713 }
2714
2715 cum_len = 0;
2716
2717 if (mss) { /* TSO */
2718 /* this removes any CKSUM flag from before */
2719 flags = (MXGEFW_FLAGS_TSO_HDR | MXGEFW_FLAGS_FIRST);
2720
2721 /* negative cum_len signifies to the
2722 * send loop that we are still in the
2723 * header portion of the TSO packet.
2724 * TSO header can be at most 1KB long */
2725 cum_len = -(skb_transport_offset(skb) + tcp_hdrlen(skb));
2726
2727 /* for IPv6 TSO, the checksum offset stores the
2728 * TCP header length, to save the firmware from
2729 * the need to parse the headers */
2730 if (skb_is_gso_v6(skb)) {
2731 cksum_offset = tcp_hdrlen(skb);
2732 /* Can only handle headers <= max_tso6 long */
2733 if (unlikely(-cum_len > mgp->max_tso6))
2734 return myri10ge_sw_tso(skb, dev);
2735 }
2736 /* for TSO, pseudo_hdr_offset holds mss.
2737 * The firmware figures out where to put
2738 * the checksum by parsing the header. */
2739 pseudo_hdr_offset = mss;
2740 } else
2741 /* Mark small packets, and pad out tiny packets */
2742 if (skb->len <= MXGEFW_SEND_SMALL_SIZE) {
2743 flags |= MXGEFW_FLAGS_SMALL;
2744
2745 /* pad frames to at least ETH_ZLEN bytes */
2746 if (unlikely(skb->len < ETH_ZLEN)) {
2747 if (skb_padto(skb, ETH_ZLEN)) {
2748 /* The packet is gone, so we must
2749 * return 0 */
2750 ss->stats.tx_dropped += 1;
2751 return NETDEV_TX_OK;
2752 }
2753 /* adjust the len to account for the zero pad
2754 * so that the nic can know how long it is */
2755 skb->len = ETH_ZLEN;
2756 }
2757 }
2758
2759 /* map the skb for DMA */
2760 len = skb_headlen(skb);
2761 idx = tx->req & tx->mask;
2762 tx->info[idx].skb = skb;
2763 bus = pci_map_single(mgp->pdev, skb->data, len, PCI_DMA_TODEVICE);
2764 dma_unmap_addr_set(&tx->info[idx], bus, bus);
2765 dma_unmap_len_set(&tx->info[idx], len, len);
2766
2767 frag_cnt = skb_shinfo(skb)->nr_frags;
2768 frag_idx = 0;
2769 count = 0;
2770 rdma_count = 0;
2771
2772 /* "rdma_count" is the number of RDMAs belonging to the
2773 * current packet BEFORE the current send request. For
2774 * non-TSO packets, this is equal to "count".
2775 * For TSO packets, rdma_count needs to be reset
2776 * to 0 after a segment cut.
2777 *
2778 * The rdma_count field of the send request is
2779 * the number of RDMAs of the packet starting at
2780 * that request. For TSO send requests with one ore more cuts
2781 * in the middle, this is the number of RDMAs starting
2782 * after the last cut in the request. All previous
2783 * segments before the last cut implicitly have 1 RDMA.
2784 *
2785 * Since the number of RDMAs is not known beforehand,
2786 * it must be filled-in retroactively - after each
2787 * segmentation cut or at the end of the entire packet.
2788 */
2789
2790 while (1) {
2791 /* Break the SKB or Fragment up into pieces which
2792 * do not cross mgp->tx_boundary */
2793 low = MYRI10GE_LOWPART_TO_U32(bus);
2794 high_swapped = htonl(MYRI10GE_HIGHPART_TO_U32(bus));
2795 while (len) {
2796 u8 flags_next;
2797 int cum_len_next;
2798
2799 if (unlikely(count == max_segments))
2800 goto abort_linearize;
2801
2802 boundary =
2803 (low + mgp->tx_boundary) & ~(mgp->tx_boundary - 1);
2804 seglen = boundary - low;
2805 if (seglen > len)
2806 seglen = len;
2807 flags_next = flags & ~MXGEFW_FLAGS_FIRST;
2808 cum_len_next = cum_len + seglen;
2809 if (mss) { /* TSO */
2810 (req - rdma_count)->rdma_count = rdma_count + 1;
2811
2812 if (likely(cum_len >= 0)) { /* payload */
2813 int next_is_first, chop;
2814
2815 chop = (cum_len_next > mss);
2816 cum_len_next = cum_len_next % mss;
2817 next_is_first = (cum_len_next == 0);
2818 flags |= chop * MXGEFW_FLAGS_TSO_CHOP;
2819 flags_next |= next_is_first *
2820 MXGEFW_FLAGS_FIRST;
2821 rdma_count |= -(chop | next_is_first);
2822 rdma_count += chop & !next_is_first;
2823 } else if (likely(cum_len_next >= 0)) { /* header ends */
2824 int small;
2825
2826 rdma_count = -1;
2827 cum_len_next = 0;
2828 seglen = -cum_len;
2829 small = (mss <= MXGEFW_SEND_SMALL_SIZE);
2830 flags_next = MXGEFW_FLAGS_TSO_PLD |
2831 MXGEFW_FLAGS_FIRST |
2832 (small * MXGEFW_FLAGS_SMALL);
2833 }
2834 }
2835 req->addr_high = high_swapped;
2836 req->addr_low = htonl(low);
2837 req->pseudo_hdr_offset = htons(pseudo_hdr_offset);
2838 req->pad = 0; /* complete solid 16-byte block; does this matter? */
2839 req->rdma_count = 1;
2840 req->length = htons(seglen);
2841 req->cksum_offset = cksum_offset;
2842 req->flags = flags | ((cum_len & 1) * odd_flag);
2843
2844 low += seglen;
2845 len -= seglen;
2846 cum_len = cum_len_next;
2847 flags = flags_next;
2848 req++;
2849 count++;
2850 rdma_count++;
2851 if (cksum_offset != 0 && !(mss && skb_is_gso_v6(skb))) {
2852 if (unlikely(cksum_offset > seglen))
2853 cksum_offset -= seglen;
2854 else
2855 cksum_offset = 0;
2856 }
2857 }
2858 if (frag_idx == frag_cnt)
2859 break;
2860
2861 /* map next fragment for DMA */
2862 idx = (count + tx->req) & tx->mask;
2863 frag = &skb_shinfo(skb)->frags[frag_idx];
2864 frag_idx++;
2865 len = frag->size;
2866 bus = pci_map_page(mgp->pdev, frag->page, frag->page_offset,
2867 len, PCI_DMA_TODEVICE);
2868 dma_unmap_addr_set(&tx->info[idx], bus, bus);
2869 dma_unmap_len_set(&tx->info[idx], len, len);
2870 }
2871
2872 (req - rdma_count)->rdma_count = rdma_count;
2873 if (mss)
2874 do {
2875 req--;
2876 req->flags |= MXGEFW_FLAGS_TSO_LAST;
2877 } while (!(req->flags & (MXGEFW_FLAGS_TSO_CHOP |
2878 MXGEFW_FLAGS_FIRST)));
2879 idx = ((count - 1) + tx->req) & tx->mask;
2880 tx->info[idx].last = 1;
2881 myri10ge_submit_req(tx, tx->req_list, count);
2882 /* if using multiple tx queues, make sure NIC polls the
2883 * current slice */
2884 if ((mgp->dev->real_num_tx_queues > 1) && tx->queue_active == 0) {
2885 tx->queue_active = 1;
2886 put_be32(htonl(1), tx->send_go);
2887 mb();
2888 mmiowb();
2889 }
2890 tx->pkt_start++;
2891 if ((avail - count) < MXGEFW_MAX_SEND_DESC) {
2892 tx->stop_queue++;
2893 netif_tx_stop_queue(netdev_queue);
2894 }
2895 return NETDEV_TX_OK;
2896
2897 abort_linearize:
2898 /* Free any DMA resources we've alloced and clear out the skb
2899 * slot so as to not trip up assertions, and to avoid a
2900 * double-free if linearizing fails */
2901
2902 last_idx = (idx + 1) & tx->mask;
2903 idx = tx->req & tx->mask;
2904 tx->info[idx].skb = NULL;
2905 do {
2906 len = dma_unmap_len(&tx->info[idx], len);
2907 if (len) {
2908 if (tx->info[idx].skb != NULL)
2909 pci_unmap_single(mgp->pdev,
2910 dma_unmap_addr(&tx->info[idx],
2911 bus), len,
2912 PCI_DMA_TODEVICE);
2913 else
2914 pci_unmap_page(mgp->pdev,
2915 dma_unmap_addr(&tx->info[idx],
2916 bus), len,
2917 PCI_DMA_TODEVICE);
2918 dma_unmap_len_set(&tx->info[idx], len, 0);
2919 tx->info[idx].skb = NULL;
2920 }
2921 idx = (idx + 1) & tx->mask;
2922 } while (idx != last_idx);
2923 if (skb_is_gso(skb)) {
2924 netdev_err(mgp->dev, "TSO but wanted to linearize?!?!?\n");
2925 goto drop;
2926 }
2927
2928 if (skb_linearize(skb))
2929 goto drop;
2930
2931 tx->linearized++;
2932 goto again;
2933
2934 drop:
2935 dev_kfree_skb_any(skb);
2936 ss->stats.tx_dropped += 1;
2937 return NETDEV_TX_OK;
2938
2939 }
2940
2941 static netdev_tx_t myri10ge_sw_tso(struct sk_buff *skb,
2942 struct net_device *dev)
2943 {
2944 struct sk_buff *segs, *curr;
2945 struct myri10ge_priv *mgp = netdev_priv(dev);
2946 struct myri10ge_slice_state *ss;
2947 netdev_tx_t status;
2948
2949 segs = skb_gso_segment(skb, dev->features & ~NETIF_F_TSO6);
2950 if (IS_ERR(segs))
2951 goto drop;
2952
2953 while (segs) {
2954 curr = segs;
2955 segs = segs->next;
2956 curr->next = NULL;
2957 status = myri10ge_xmit(curr, dev);
2958 if (status != 0) {
2959 dev_kfree_skb_any(curr);
2960 if (segs != NULL) {
2961 curr = segs;
2962 segs = segs->next;
2963 curr->next = NULL;
2964 dev_kfree_skb_any(segs);
2965 }
2966 goto drop;
2967 }
2968 }
2969 dev_kfree_skb_any(skb);
2970 return NETDEV_TX_OK;
2971
2972 drop:
2973 ss = &mgp->ss[skb_get_queue_mapping(skb)];
2974 dev_kfree_skb_any(skb);
2975 ss->stats.tx_dropped += 1;
2976 return NETDEV_TX_OK;
2977 }
2978
2979 static struct net_device_stats *myri10ge_get_stats(struct net_device *dev)
2980 {
2981 struct myri10ge_priv *mgp = netdev_priv(dev);
2982 struct myri10ge_slice_netstats *slice_stats;
2983 struct net_device_stats *stats = &dev->stats;
2984 int i;
2985
2986 spin_lock(&mgp->stats_lock);
2987 memset(stats, 0, sizeof(*stats));
2988 for (i = 0; i < mgp->num_slices; i++) {
2989 slice_stats = &mgp->ss[i].stats;
2990 stats->rx_packets += slice_stats->rx_packets;
2991 stats->tx_packets += slice_stats->tx_packets;
2992 stats->rx_bytes += slice_stats->rx_bytes;
2993 stats->tx_bytes += slice_stats->tx_bytes;
2994 stats->rx_dropped += slice_stats->rx_dropped;
2995 stats->tx_dropped += slice_stats->tx_dropped;
2996 }
2997 spin_unlock(&mgp->stats_lock);
2998 return stats;
2999 }
3000
3001 static void myri10ge_set_multicast_list(struct net_device *dev)
3002 {
3003 struct myri10ge_priv *mgp = netdev_priv(dev);
3004 struct myri10ge_cmd cmd;
3005 struct netdev_hw_addr *ha;
3006 __be32 data[2] = { 0, 0 };
3007 int err;
3008
3009 /* can be called from atomic contexts,
3010 * pass 1 to force atomicity in myri10ge_send_cmd() */
3011 myri10ge_change_promisc(mgp, dev->flags & IFF_PROMISC, 1);
3012
3013 /* This firmware is known to not support multicast */
3014 if (!mgp->fw_multicast_support)
3015 return;
3016
3017 /* Disable multicast filtering */
3018
3019 err = myri10ge_send_cmd(mgp, MXGEFW_ENABLE_ALLMULTI, &cmd, 1);
3020 if (err != 0) {
3021 netdev_err(dev, "Failed MXGEFW_ENABLE_ALLMULTI, error status: %d\n",
3022 err);
3023 goto abort;
3024 }
3025
3026 if ((dev->flags & IFF_ALLMULTI) || mgp->adopted_rx_filter_bug) {
3027 /* request to disable multicast filtering, so quit here */
3028 return;
3029 }
3030
3031 /* Flush the filters */
3032
3033 err = myri10ge_send_cmd(mgp, MXGEFW_LEAVE_ALL_MULTICAST_GROUPS,
3034 &cmd, 1);
3035 if (err != 0) {
3036 netdev_err(dev, "Failed MXGEFW_LEAVE_ALL_MULTICAST_GROUPS, error status: %d\n",
3037 err);
3038 goto abort;
3039 }
3040
3041 /* Walk the multicast list, and add each address */
3042 netdev_for_each_mc_addr(ha, dev) {
3043 memcpy(data, &ha->addr, 6);
3044 cmd.data0 = ntohl(data[0]);
3045 cmd.data1 = ntohl(data[1]);
3046 err = myri10ge_send_cmd(mgp, MXGEFW_JOIN_MULTICAST_GROUP,
3047 &cmd, 1);
3048
3049 if (err != 0) {
3050 netdev_err(dev, "Failed MXGEFW_JOIN_MULTICAST_GROUP, error status:%d %pM\n",
3051 err, ha->addr);
3052 goto abort;
3053 }
3054 }
3055 /* Enable multicast filtering */
3056 err = myri10ge_send_cmd(mgp, MXGEFW_DISABLE_ALLMULTI, &cmd, 1);
3057 if (err != 0) {
3058 netdev_err(dev, "Failed MXGEFW_DISABLE_ALLMULTI, error status: %d\n",
3059 err);
3060 goto abort;
3061 }
3062
3063 return;
3064
3065 abort:
3066 return;
3067 }
3068
3069 static int myri10ge_set_mac_address(struct net_device *dev, void *addr)
3070 {
3071 struct sockaddr *sa = addr;
3072 struct myri10ge_priv *mgp = netdev_priv(dev);
3073 int status;
3074
3075 if (!is_valid_ether_addr(sa->sa_data))
3076 return -EADDRNOTAVAIL;
3077
3078 status = myri10ge_update_mac_address(mgp, sa->sa_data);
3079 if (status != 0) {
3080 netdev_err(dev, "changing mac address failed with %d\n",
3081 status);
3082 return status;
3083 }
3084
3085 /* change the dev structure */
3086 memcpy(dev->dev_addr, sa->sa_data, 6);
3087 return 0;
3088 }
3089
3090 static u32 myri10ge_fix_features(struct net_device *dev, u32 features)
3091 {
3092 if (!(features & NETIF_F_RXCSUM))
3093 features &= ~NETIF_F_LRO;
3094
3095 return features;
3096 }
3097
3098 static int myri10ge_change_mtu(struct net_device *dev, int new_mtu)
3099 {
3100 struct myri10ge_priv *mgp = netdev_priv(dev);
3101 int error = 0;
3102
3103 if ((new_mtu < 68) || (ETH_HLEN + new_mtu > MYRI10GE_MAX_ETHER_MTU)) {
3104 netdev_err(dev, "new mtu (%d) is not valid\n", new_mtu);
3105 return -EINVAL;
3106 }
3107 netdev_info(dev, "changing mtu from %d to %d\n", dev->mtu, new_mtu);
3108 if (mgp->running) {
3109 /* if we change the mtu on an active device, we must
3110 * reset the device so the firmware sees the change */
3111 myri10ge_close(dev);
3112 dev->mtu = new_mtu;
3113 myri10ge_open(dev);
3114 } else
3115 dev->mtu = new_mtu;
3116
3117 return error;
3118 }
3119
3120 /*
3121 * Enable ECRC to align PCI-E Completion packets on an 8-byte boundary.
3122 * Only do it if the bridge is a root port since we don't want to disturb
3123 * any other device, except if forced with myri10ge_ecrc_enable > 1.
3124 */
3125
3126 static void myri10ge_enable_ecrc(struct myri10ge_priv *mgp)
3127 {
3128 struct pci_dev *bridge = mgp->pdev->bus->self;
3129 struct device *dev = &mgp->pdev->dev;
3130 unsigned cap;
3131 unsigned err_cap;
3132 u16 val;
3133 u8 ext_type;
3134 int ret;
3135
3136 if (!myri10ge_ecrc_enable || !bridge)
3137 return;
3138
3139 /* check that the bridge is a root port */
3140 cap = pci_find_capability(bridge, PCI_CAP_ID_EXP);
3141 pci_read_config_word(bridge, cap + PCI_CAP_FLAGS, &val);
3142 ext_type = (val & PCI_EXP_FLAGS_TYPE) >> 4;
3143 if (ext_type != PCI_EXP_TYPE_ROOT_PORT) {
3144 if (myri10ge_ecrc_enable > 1) {
3145 struct pci_dev *prev_bridge, *old_bridge = bridge;
3146
3147 /* Walk the hierarchy up to the root port
3148 * where ECRC has to be enabled */
3149 do {
3150 prev_bridge = bridge;
3151 bridge = bridge->bus->self;
3152 if (!bridge || prev_bridge == bridge) {
3153 dev_err(dev,
3154 "Failed to find root port"
3155 " to force ECRC\n");
3156 return;
3157 }
3158 cap =
3159 pci_find_capability(bridge, PCI_CAP_ID_EXP);
3160 pci_read_config_word(bridge,
3161 cap + PCI_CAP_FLAGS, &val);
3162 ext_type = (val & PCI_EXP_FLAGS_TYPE) >> 4;
3163 } while (ext_type != PCI_EXP_TYPE_ROOT_PORT);
3164
3165 dev_info(dev,
3166 "Forcing ECRC on non-root port %s"
3167 " (enabling on root port %s)\n",
3168 pci_name(old_bridge), pci_name(bridge));
3169 } else {
3170 dev_err(dev,
3171 "Not enabling ECRC on non-root port %s\n",
3172 pci_name(bridge));
3173 return;
3174 }
3175 }
3176
3177 cap = pci_find_ext_capability(bridge, PCI_EXT_CAP_ID_ERR);
3178 if (!cap)
3179 return;
3180
3181 ret = pci_read_config_dword(bridge, cap + PCI_ERR_CAP, &err_cap);
3182 if (ret) {
3183 dev_err(dev, "failed reading ext-conf-space of %s\n",
3184 pci_name(bridge));
3185 dev_err(dev, "\t pci=nommconf in use? "
3186 "or buggy/incomplete/absent ACPI MCFG attr?\n");
3187 return;
3188 }
3189 if (!(err_cap & PCI_ERR_CAP_ECRC_GENC))
3190 return;
3191
3192 err_cap |= PCI_ERR_CAP_ECRC_GENE;
3193 pci_write_config_dword(bridge, cap + PCI_ERR_CAP, err_cap);
3194 dev_info(dev, "Enabled ECRC on upstream bridge %s\n", pci_name(bridge));
3195 }
3196
3197 /*
3198 * The Lanai Z8E PCI-E interface achieves higher Read-DMA throughput
3199 * when the PCI-E Completion packets are aligned on an 8-byte
3200 * boundary. Some PCI-E chip sets always align Completion packets; on
3201 * the ones that do not, the alignment can be enforced by enabling
3202 * ECRC generation (if supported).
3203 *
3204 * When PCI-E Completion packets are not aligned, it is actually more
3205 * efficient to limit Read-DMA transactions to 2KB, rather than 4KB.
3206 *
3207 * If the driver can neither enable ECRC nor verify that it has
3208 * already been enabled, then it must use a firmware image which works
3209 * around unaligned completion packets (myri10ge_rss_ethp_z8e.dat), and it
3210 * should also ensure that it never gives the device a Read-DMA which is
3211 * larger than 2KB by setting the tx_boundary to 2KB. If ECRC is
3212 * enabled, then the driver should use the aligned (myri10ge_rss_eth_z8e.dat)
3213 * firmware image, and set tx_boundary to 4KB.
3214 */
3215
3216 static void myri10ge_firmware_probe(struct myri10ge_priv *mgp)
3217 {
3218 struct pci_dev *pdev = mgp->pdev;
3219 struct device *dev = &pdev->dev;
3220 int status;
3221
3222 mgp->tx_boundary = 4096;
3223 /*
3224 * Verify the max read request size was set to 4KB
3225 * before trying the test with 4KB.
3226 */
3227 status = pcie_get_readrq(pdev);
3228 if (status < 0) {
3229 dev_err(dev, "Couldn't read max read req size: %d\n", status);
3230 goto abort;
3231 }
3232 if (status != 4096) {
3233 dev_warn(dev, "Max Read Request size != 4096 (%d)\n", status);
3234 mgp->tx_boundary = 2048;
3235 }
3236 /*
3237 * load the optimized firmware (which assumes aligned PCIe
3238 * completions) in order to see if it works on this host.
3239 */
3240 set_fw_name(mgp, myri10ge_fw_aligned, false);
3241 status = myri10ge_load_firmware(mgp, 1);
3242 if (status != 0) {
3243 goto abort;
3244 }
3245
3246 /*
3247 * Enable ECRC if possible
3248 */
3249 myri10ge_enable_ecrc(mgp);
3250
3251 /*
3252 * Run a DMA test which watches for unaligned completions and
3253 * aborts on the first one seen.
3254 */
3255
3256 status = myri10ge_dma_test(mgp, MXGEFW_CMD_UNALIGNED_TEST);
3257 if (status == 0)
3258 return; /* keep the aligned firmware */
3259
3260 if (status != -E2BIG)
3261 dev_warn(dev, "DMA test failed: %d\n", status);
3262 if (status == -ENOSYS)
3263 dev_warn(dev, "Falling back to ethp! "
3264 "Please install up to date fw\n");
3265 abort:
3266 /* fall back to using the unaligned firmware */
3267 mgp->tx_boundary = 2048;
3268 set_fw_name(mgp, myri10ge_fw_unaligned, false);
3269
3270 }
3271
3272 static void myri10ge_select_firmware(struct myri10ge_priv *mgp)
3273 {
3274 int overridden = 0;
3275
3276 if (myri10ge_force_firmware == 0) {
3277 int link_width, exp_cap;
3278 u16 lnk;
3279
3280 exp_cap = pci_find_capability(mgp->pdev, PCI_CAP_ID_EXP);
3281 pci_read_config_word(mgp->pdev, exp_cap + PCI_EXP_LNKSTA, &lnk);
3282 link_width = (lnk >> 4) & 0x3f;
3283
3284 /* Check to see if Link is less than 8 or if the
3285 * upstream bridge is known to provide aligned
3286 * completions */
3287 if (link_width < 8) {
3288 dev_info(&mgp->pdev->dev, "PCIE x%d Link\n",
3289 link_width);
3290 mgp->tx_boundary = 4096;
3291 set_fw_name(mgp, myri10ge_fw_aligned, false);
3292 } else {
3293 myri10ge_firmware_probe(mgp);
3294 }
3295 } else {
3296 if (myri10ge_force_firmware == 1) {
3297 dev_info(&mgp->pdev->dev,
3298 "Assuming aligned completions (forced)\n");
3299 mgp->tx_boundary = 4096;
3300 set_fw_name(mgp, myri10ge_fw_aligned, false);
3301 } else {
3302 dev_info(&mgp->pdev->dev,
3303 "Assuming unaligned completions (forced)\n");
3304 mgp->tx_boundary = 2048;
3305 set_fw_name(mgp, myri10ge_fw_unaligned, false);
3306 }
3307 }
3308
3309 kparam_block_sysfs_write(myri10ge_fw_name);
3310 if (myri10ge_fw_name != NULL) {
3311 char *fw_name = kstrdup(myri10ge_fw_name, GFP_KERNEL);
3312 if (fw_name) {
3313 overridden = 1;
3314 set_fw_name(mgp, fw_name, true);
3315 }
3316 }
3317 kparam_unblock_sysfs_write(myri10ge_fw_name);
3318
3319 if (mgp->board_number < MYRI10GE_MAX_BOARDS &&
3320 myri10ge_fw_names[mgp->board_number] != NULL &&
3321 strlen(myri10ge_fw_names[mgp->board_number])) {
3322 set_fw_name(mgp, myri10ge_fw_names[mgp->board_number], false);
3323 overridden = 1;
3324 }
3325 if (overridden)
3326 dev_info(&mgp->pdev->dev, "overriding firmware to %s\n",
3327 mgp->fw_name);
3328 }
3329
3330 #ifdef CONFIG_PM
3331 static int myri10ge_suspend(struct pci_dev *pdev, pm_message_t state)
3332 {
3333 struct myri10ge_priv *mgp;
3334 struct net_device *netdev;
3335
3336 mgp = pci_get_drvdata(pdev);
3337 if (mgp == NULL)
3338 return -EINVAL;
3339 netdev = mgp->dev;
3340
3341 netif_device_detach(netdev);
3342 if (netif_running(netdev)) {
3343 netdev_info(netdev, "closing\n");
3344 rtnl_lock();
3345 myri10ge_close(netdev);
3346 rtnl_unlock();
3347 }
3348 myri10ge_dummy_rdma(mgp, 0);
3349 pci_save_state(pdev);
3350 pci_disable_device(pdev);
3351
3352 return pci_set_power_state(pdev, pci_choose_state(pdev, state));
3353 }
3354
3355 static int myri10ge_resume(struct pci_dev *pdev)
3356 {
3357 struct myri10ge_priv *mgp;
3358 struct net_device *netdev;
3359 int status;
3360 u16 vendor;
3361
3362 mgp = pci_get_drvdata(pdev);
3363 if (mgp == NULL)
3364 return -EINVAL;
3365 netdev = mgp->dev;
3366 pci_set_power_state(pdev, 0); /* zeros conf space as a side effect */
3367 msleep(5); /* give card time to respond */
3368 pci_read_config_word(mgp->pdev, PCI_VENDOR_ID, &vendor);
3369 if (vendor == 0xffff) {
3370 netdev_err(mgp->dev, "device disappeared!\n");
3371 return -EIO;
3372 }
3373
3374 pci_restore_state(pdev);
3375
3376 status = pci_enable_device(pdev);
3377 if (status) {
3378 dev_err(&pdev->dev, "failed to enable device\n");
3379 return status;
3380 }
3381
3382 pci_set_master(pdev);
3383
3384 myri10ge_reset(mgp);
3385 myri10ge_dummy_rdma(mgp, 1);
3386
3387 /* Save configuration space to be restored if the
3388 * nic resets due to a parity error */
3389 pci_save_state(pdev);
3390
3391 if (netif_running(netdev)) {
3392 rtnl_lock();
3393 status = myri10ge_open(netdev);
3394 rtnl_unlock();
3395 if (status != 0)
3396 goto abort_with_enabled;
3397
3398 }
3399 netif_device_attach(netdev);
3400
3401 return 0;
3402
3403 abort_with_enabled:
3404 pci_disable_device(pdev);
3405 return -EIO;
3406
3407 }
3408 #endif /* CONFIG_PM */
3409
3410 static u32 myri10ge_read_reboot(struct myri10ge_priv *mgp)
3411 {
3412 struct pci_dev *pdev = mgp->pdev;
3413 int vs = mgp->vendor_specific_offset;
3414 u32 reboot;
3415
3416 /*enter read32 mode */
3417 pci_write_config_byte(pdev, vs + 0x10, 0x3);
3418
3419 /*read REBOOT_STATUS (0xfffffff0) */
3420 pci_write_config_dword(pdev, vs + 0x18, 0xfffffff0);
3421 pci_read_config_dword(pdev, vs + 0x14, &reboot);
3422 return reboot;
3423 }
3424
3425 /*
3426 * This watchdog is used to check whether the board has suffered
3427 * from a parity error and needs to be recovered.
3428 */
3429 static void myri10ge_watchdog(struct work_struct *work)
3430 {
3431 struct myri10ge_priv *mgp =
3432 container_of(work, struct myri10ge_priv, watchdog_work);
3433 struct myri10ge_tx_buf *tx;
3434 u32 reboot;
3435 int status, rebooted;
3436 int i;
3437 u16 cmd, vendor;
3438
3439 mgp->watchdog_resets++;
3440 pci_read_config_word(mgp->pdev, PCI_COMMAND, &cmd);
3441 rebooted = 0;
3442 if ((cmd & PCI_COMMAND_MASTER) == 0) {
3443 /* Bus master DMA disabled? Check to see
3444 * if the card rebooted due to a parity error
3445 * For now, just report it */
3446 reboot = myri10ge_read_reboot(mgp);
3447 netdev_err(mgp->dev, "NIC rebooted (0x%x),%s resetting\n",
3448 reboot,
3449 myri10ge_reset_recover ? "" : " not");
3450 if (myri10ge_reset_recover == 0)
3451 return;
3452 rtnl_lock();
3453 mgp->rebooted = 1;
3454 rebooted = 1;
3455 myri10ge_close(mgp->dev);
3456 myri10ge_reset_recover--;
3457 mgp->rebooted = 0;
3458 /*
3459 * A rebooted nic will come back with config space as
3460 * it was after power was applied to PCIe bus.
3461 * Attempt to restore config space which was saved
3462 * when the driver was loaded, or the last time the
3463 * nic was resumed from power saving mode.
3464 */
3465 pci_restore_state(mgp->pdev);
3466
3467 /* save state again for accounting reasons */
3468 pci_save_state(mgp->pdev);
3469
3470 } else {
3471 /* if we get back -1's from our slot, perhaps somebody
3472 * powered off our card. Don't try to reset it in
3473 * this case */
3474 if (cmd == 0xffff) {
3475 pci_read_config_word(mgp->pdev, PCI_VENDOR_ID, &vendor);
3476 if (vendor == 0xffff) {
3477 netdev_err(mgp->dev, "device disappeared!\n");
3478 return;
3479 }
3480 }
3481 /* Perhaps it is a software error. Try to reset */
3482
3483 netdev_err(mgp->dev, "device timeout, resetting\n");
3484 for (i = 0; i < mgp->num_slices; i++) {
3485 tx = &mgp->ss[i].tx;
3486 netdev_err(mgp->dev, "(%d): %d %d %d %d %d %d\n",
3487 i, tx->queue_active, tx->req,
3488 tx->done, tx->pkt_start, tx->pkt_done,
3489 (int)ntohl(mgp->ss[i].fw_stats->
3490 send_done_count));
3491 msleep(2000);
3492 netdev_info(mgp->dev, "(%d): %d %d %d %d %d %d\n",
3493 i, tx->queue_active, tx->req,
3494 tx->done, tx->pkt_start, tx->pkt_done,
3495 (int)ntohl(mgp->ss[i].fw_stats->
3496 send_done_count));
3497 }
3498 }
3499
3500 if (!rebooted) {
3501 rtnl_lock();
3502 myri10ge_close(mgp->dev);
3503 }
3504 status = myri10ge_load_firmware(mgp, 1);
3505 if (status != 0)
3506 netdev_err(mgp->dev, "failed to load firmware\n");
3507 else
3508 myri10ge_open(mgp->dev);
3509 rtnl_unlock();
3510 }
3511
3512 /*
3513 * We use our own timer routine rather than relying upon
3514 * netdev->tx_timeout because we have a very large hardware transmit
3515 * queue. Due to the large queue, the netdev->tx_timeout function
3516 * cannot detect a NIC with a parity error in a timely fashion if the
3517 * NIC is lightly loaded.
3518 */
3519 static void myri10ge_watchdog_timer(unsigned long arg)
3520 {
3521 struct myri10ge_priv *mgp;
3522 struct myri10ge_slice_state *ss;
3523 int i, reset_needed, busy_slice_cnt;
3524 u32 rx_pause_cnt;
3525 u16 cmd;
3526
3527 mgp = (struct myri10ge_priv *)arg;
3528
3529 rx_pause_cnt = ntohl(mgp->ss[0].fw_stats->dropped_pause);
3530 busy_slice_cnt = 0;
3531 for (i = 0, reset_needed = 0;
3532 i < mgp->num_slices && reset_needed == 0; ++i) {
3533
3534 ss = &mgp->ss[i];
3535 if (ss->rx_small.watchdog_needed) {
3536 myri10ge_alloc_rx_pages(mgp, &ss->rx_small,
3537 mgp->small_bytes + MXGEFW_PAD,
3538 1);
3539 if (ss->rx_small.fill_cnt - ss->rx_small.cnt >=
3540 myri10ge_fill_thresh)
3541 ss->rx_small.watchdog_needed = 0;
3542 }
3543 if (ss->rx_big.watchdog_needed) {
3544 myri10ge_alloc_rx_pages(mgp, &ss->rx_big,
3545 mgp->big_bytes, 1);
3546 if (ss->rx_big.fill_cnt - ss->rx_big.cnt >=
3547 myri10ge_fill_thresh)
3548 ss->rx_big.watchdog_needed = 0;
3549 }
3550
3551 if (ss->tx.req != ss->tx.done &&
3552 ss->tx.done == ss->watchdog_tx_done &&
3553 ss->watchdog_tx_req != ss->watchdog_tx_done) {
3554 /* nic seems like it might be stuck.. */
3555 if (rx_pause_cnt != mgp->watchdog_pause) {
3556 if (net_ratelimit())
3557 netdev_err(mgp->dev, "slice %d: TX paused, check link partner\n",
3558 i);
3559 } else {
3560 netdev_warn(mgp->dev, "slice %d stuck:", i);
3561 reset_needed = 1;
3562 }
3563 }
3564 if (ss->watchdog_tx_done != ss->tx.done ||
3565 ss->watchdog_rx_done != ss->rx_done.cnt) {
3566 busy_slice_cnt++;
3567 }
3568 ss->watchdog_tx_done = ss->tx.done;
3569 ss->watchdog_tx_req = ss->tx.req;
3570 ss->watchdog_rx_done = ss->rx_done.cnt;
3571 }
3572 /* if we've sent or received no traffic, poll the NIC to
3573 * ensure it is still there. Otherwise, we risk not noticing
3574 * an error in a timely fashion */
3575 if (busy_slice_cnt == 0) {
3576 pci_read_config_word(mgp->pdev, PCI_COMMAND, &cmd);
3577 if ((cmd & PCI_COMMAND_MASTER) == 0) {
3578 reset_needed = 1;
3579 }
3580 }
3581 mgp->watchdog_pause = rx_pause_cnt;
3582
3583 if (reset_needed) {
3584 schedule_work(&mgp->watchdog_work);
3585 } else {
3586 /* rearm timer */
3587 mod_timer(&mgp->watchdog_timer,
3588 jiffies + myri10ge_watchdog_timeout * HZ);
3589 }
3590 }
3591
3592 static void myri10ge_free_slices(struct myri10ge_priv *mgp)
3593 {
3594 struct myri10ge_slice_state *ss;
3595 struct pci_dev *pdev = mgp->pdev;
3596 size_t bytes;
3597 int i;
3598
3599 if (mgp->ss == NULL)
3600 return;
3601
3602 for (i = 0; i < mgp->num_slices; i++) {
3603 ss = &mgp->ss[i];
3604 if (ss->rx_done.entry != NULL) {
3605 bytes = mgp->max_intr_slots *
3606 sizeof(*ss->rx_done.entry);
3607 dma_free_coherent(&pdev->dev, bytes,
3608 ss->rx_done.entry, ss->rx_done.bus);
3609 ss->rx_done.entry = NULL;
3610 }
3611 if (ss->fw_stats != NULL) {
3612 bytes = sizeof(*ss->fw_stats);
3613 dma_free_coherent(&pdev->dev, bytes,
3614 ss->fw_stats, ss->fw_stats_bus);
3615 ss->fw_stats = NULL;
3616 netif_napi_del(&ss->napi);
3617 }
3618 }
3619 kfree(mgp->ss);
3620 mgp->ss = NULL;
3621 }
3622
3623 static int myri10ge_alloc_slices(struct myri10ge_priv *mgp)
3624 {
3625 struct myri10ge_slice_state *ss;
3626 struct pci_dev *pdev = mgp->pdev;
3627 size_t bytes;
3628 int i;
3629
3630 bytes = sizeof(*mgp->ss) * mgp->num_slices;
3631 mgp->ss = kzalloc(bytes, GFP_KERNEL);
3632 if (mgp->ss == NULL) {
3633 return -ENOMEM;
3634 }
3635
3636 for (i = 0; i < mgp->num_slices; i++) {
3637 ss = &mgp->ss[i];
3638 bytes = mgp->max_intr_slots * sizeof(*ss->rx_done.entry);
3639 ss->rx_done.entry = dma_alloc_coherent(&pdev->dev, bytes,
3640 &ss->rx_done.bus,
3641 GFP_KERNEL);
3642 if (ss->rx_done.entry == NULL)
3643 goto abort;
3644 memset(ss->rx_done.entry, 0, bytes);
3645 bytes = sizeof(*ss->fw_stats);
3646 ss->fw_stats = dma_alloc_coherent(&pdev->dev, bytes,
3647 &ss->fw_stats_bus,
3648 GFP_KERNEL);
3649 if (ss->fw_stats == NULL)
3650 goto abort;
3651 ss->mgp = mgp;
3652 ss->dev = mgp->dev;
3653 netif_napi_add(ss->dev, &ss->napi, myri10ge_poll,
3654 myri10ge_napi_weight);
3655 }
3656 return 0;
3657 abort:
3658 myri10ge_free_slices(mgp);
3659 return -ENOMEM;
3660 }
3661
3662 /*
3663 * This function determines the number of slices supported.
3664 * The number slices is the minimum of the number of CPUS,
3665 * the number of MSI-X irqs supported, the number of slices
3666 * supported by the firmware
3667 */
3668 static void myri10ge_probe_slices(struct myri10ge_priv *mgp)
3669 {
3670 struct myri10ge_cmd cmd;
3671 struct pci_dev *pdev = mgp->pdev;
3672 char *old_fw;
3673 bool old_allocated;
3674 int i, status, ncpus, msix_cap;
3675
3676 mgp->num_slices = 1;
3677 msix_cap = pci_find_capability(pdev, PCI_CAP_ID_MSIX);
3678 ncpus = num_online_cpus();
3679
3680 if (myri10ge_max_slices == 1 || msix_cap == 0 ||
3681 (myri10ge_max_slices == -1 && ncpus < 2))
3682 return;
3683
3684 /* try to load the slice aware rss firmware */
3685 old_fw = mgp->fw_name;
3686 old_allocated = mgp->fw_name_allocated;
3687 /* don't free old_fw if we override it. */
3688 mgp->fw_name_allocated = false;
3689
3690 if (myri10ge_fw_name != NULL) {
3691 dev_info(&mgp->pdev->dev, "overriding rss firmware to %s\n",
3692 myri10ge_fw_name);
3693 set_fw_name(mgp, myri10ge_fw_name, false);
3694 } else if (old_fw == myri10ge_fw_aligned)
3695 set_fw_name(mgp, myri10ge_fw_rss_aligned, false);
3696 else
3697 set_fw_name(mgp, myri10ge_fw_rss_unaligned, false);
3698 status = myri10ge_load_firmware(mgp, 0);
3699 if (status != 0) {
3700 dev_info(&pdev->dev, "Rss firmware not found\n");
3701 if (old_allocated)
3702 kfree(old_fw);
3703 return;
3704 }
3705
3706 /* hit the board with a reset to ensure it is alive */
3707 memset(&cmd, 0, sizeof(cmd));
3708 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_RESET, &cmd, 0);
3709 if (status != 0) {
3710 dev_err(&mgp->pdev->dev, "failed reset\n");
3711 goto abort_with_fw;
3712 }
3713
3714 mgp->max_intr_slots = cmd.data0 / sizeof(struct mcp_slot);
3715
3716 /* tell it the size of the interrupt queues */
3717 cmd.data0 = mgp->max_intr_slots * sizeof(struct mcp_slot);
3718 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_SET_INTRQ_SIZE, &cmd, 0);
3719 if (status != 0) {
3720 dev_err(&mgp->pdev->dev, "failed MXGEFW_CMD_SET_INTRQ_SIZE\n");
3721 goto abort_with_fw;
3722 }
3723
3724 /* ask the maximum number of slices it supports */
3725 status = myri10ge_send_cmd(mgp, MXGEFW_CMD_GET_MAX_RSS_QUEUES, &cmd, 0);
3726 if (status != 0)
3727 goto abort_with_fw;
3728 else
3729 mgp->num_slices = cmd.data0;
3730
3731 /* Only allow multiple slices if MSI-X is usable */
3732 if (!myri10ge_msi) {
3733 goto abort_with_fw;
3734 }
3735
3736 /* if the admin did not specify a limit to how many
3737 * slices we should use, cap it automatically to the
3738 * number of CPUs currently online */
3739 if (myri10ge_max_slices == -1)
3740 myri10ge_max_slices = ncpus;
3741
3742 if (mgp->num_slices > myri10ge_max_slices)
3743 mgp->num_slices = myri10ge_max_slices;
3744
3745 /* Now try to allocate as many MSI-X vectors as we have
3746 * slices. We give up on MSI-X if we can only get a single
3747 * vector. */
3748
3749 mgp->msix_vectors = kcalloc(mgp->num_slices, sizeof(*mgp->msix_vectors),
3750 GFP_KERNEL);
3751 if (mgp->msix_vectors == NULL)
3752 goto disable_msix;
3753 for (i = 0; i < mgp->num_slices; i++) {
3754 mgp->msix_vectors[i].entry = i;
3755 }
3756
3757 while (mgp->num_slices > 1) {
3758 /* make sure it is a power of two */
3759 while (!is_power_of_2(mgp->num_slices))
3760 mgp->num_slices--;
3761 if (mgp->num_slices == 1)
3762 goto disable_msix;
3763 status = pci_enable_msix(pdev, mgp->msix_vectors,
3764 mgp->num_slices);
3765 if (status == 0) {
3766 pci_disable_msix(pdev);
3767 if (old_allocated)
3768 kfree(old_fw);
3769 return;
3770 }
3771 if (status > 0)
3772 mgp->num_slices = status;
3773 else
3774 goto disable_msix;
3775 }
3776
3777 disable_msix:
3778 if (mgp->msix_vectors != NULL) {
3779 kfree(mgp->msix_vectors);
3780 mgp->msix_vectors = NULL;
3781 }
3782
3783 abort_with_fw:
3784 mgp->num_slices = 1;
3785 set_fw_name(mgp, old_fw, old_allocated);
3786 myri10ge_load_firmware(mgp, 0);
3787 }
3788
3789 static const struct net_device_ops myri10ge_netdev_ops = {
3790 .ndo_open = myri10ge_open,
3791 .ndo_stop = myri10ge_close,
3792 .ndo_start_xmit = myri10ge_xmit,
3793 .ndo_get_stats = myri10ge_get_stats,
3794 .ndo_validate_addr = eth_validate_addr,
3795 .ndo_change_mtu = myri10ge_change_mtu,
3796 .ndo_fix_features = myri10ge_fix_features,
3797 .ndo_set_multicast_list = myri10ge_set_multicast_list,
3798 .ndo_set_mac_address = myri10ge_set_mac_address,
3799 };
3800
3801 static int myri10ge_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
3802 {
3803 struct net_device *netdev;
3804 struct myri10ge_priv *mgp;
3805 struct device *dev = &pdev->dev;
3806 int i;
3807 int status = -ENXIO;
3808 int dac_enabled;
3809 unsigned hdr_offset, ss_offset;
3810 static int board_number;
3811
3812 netdev = alloc_etherdev_mq(sizeof(*mgp), MYRI10GE_MAX_SLICES);
3813 if (netdev == NULL) {
3814 dev_err(dev, "Could not allocate ethernet device\n");
3815 return -ENOMEM;
3816 }
3817
3818 SET_NETDEV_DEV(netdev, &pdev->dev);
3819
3820 mgp = netdev_priv(netdev);
3821 mgp->dev = netdev;
3822 mgp->pdev = pdev;
3823 mgp->pause = myri10ge_flow_control;
3824 mgp->intr_coal_delay = myri10ge_intr_coal_delay;
3825 mgp->msg_enable = netif_msg_init(myri10ge_debug, MYRI10GE_MSG_DEFAULT);
3826 mgp->board_number = board_number;
3827 init_waitqueue_head(&mgp->down_wq);
3828
3829 if (pci_enable_device(pdev)) {
3830 dev_err(&pdev->dev, "pci_enable_device call failed\n");
3831 status = -ENODEV;
3832 goto abort_with_netdev;
3833 }
3834
3835 /* Find the vendor-specific cap so we can check
3836 * the reboot register later on */
3837 mgp->vendor_specific_offset
3838 = pci_find_capability(pdev, PCI_CAP_ID_VNDR);
3839
3840 /* Set our max read request to 4KB */
3841 status = pcie_set_readrq(pdev, 4096);
3842 if (status != 0) {
3843 dev_err(&pdev->dev, "Error %d writing PCI_EXP_DEVCTL\n",
3844 status);
3845 goto abort_with_enabled;
3846 }
3847
3848 pci_set_master(pdev);
3849 dac_enabled = 1;
3850 status = pci_set_dma_mask(pdev, DMA_BIT_MASK(64));
3851 if (status != 0) {
3852 dac_enabled = 0;
3853 dev_err(&pdev->dev,
3854 "64-bit pci address mask was refused, "
3855 "trying 32-bit\n");
3856 status = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
3857 }
3858 if (status != 0) {
3859 dev_err(&pdev->dev, "Error %d setting DMA mask\n", status);
3860 goto abort_with_enabled;
3861 }
3862 (void)pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64));
3863 mgp->cmd = dma_alloc_coherent(&pdev->dev, sizeof(*mgp->cmd),
3864 &mgp->cmd_bus, GFP_KERNEL);
3865 if (mgp->cmd == NULL)
3866 goto abort_with_enabled;
3867
3868 mgp->board_span = pci_resource_len(pdev, 0);
3869 mgp->iomem_base = pci_resource_start(pdev, 0);
3870 mgp->mtrr = -1;
3871 mgp->wc_enabled = 0;
3872 #ifdef CONFIG_MTRR
3873 mgp->mtrr = mtrr_add(mgp->iomem_base, mgp->board_span,
3874 MTRR_TYPE_WRCOMB, 1);
3875 if (mgp->mtrr >= 0)
3876 mgp->wc_enabled = 1;
3877 #endif
3878 mgp->sram = ioremap_wc(mgp->iomem_base, mgp->board_span);
3879 if (mgp->sram == NULL) {
3880 dev_err(&pdev->dev, "ioremap failed for %ld bytes at 0x%lx\n",
3881 mgp->board_span, mgp->iomem_base);
3882 status = -ENXIO;
3883 goto abort_with_mtrr;
3884 }
3885 hdr_offset =
3886 ntohl(__raw_readl(mgp->sram + MCP_HEADER_PTR_OFFSET)) & 0xffffc;
3887 ss_offset = hdr_offset + offsetof(struct mcp_gen_header, string_specs);
3888 mgp->sram_size = ntohl(__raw_readl(mgp->sram + ss_offset));
3889 if (mgp->sram_size > mgp->board_span ||
3890 mgp->sram_size <= MYRI10GE_FW_OFFSET) {
3891 dev_err(&pdev->dev,
3892 "invalid sram_size %dB or board span %ldB\n",
3893 mgp->sram_size, mgp->board_span);
3894 goto abort_with_ioremap;
3895 }
3896 memcpy_fromio(mgp->eeprom_strings,
3897 mgp->sram + mgp->sram_size, MYRI10GE_EEPROM_STRINGS_SIZE);
3898 memset(mgp->eeprom_strings + MYRI10GE_EEPROM_STRINGS_SIZE - 2, 0, 2);
3899 status = myri10ge_read_mac_addr(mgp);
3900 if (status)
3901 goto abort_with_ioremap;
3902
3903 for (i = 0; i < ETH_ALEN; i++)
3904 netdev->dev_addr[i] = mgp->mac_addr[i];
3905
3906 myri10ge_select_firmware(mgp);
3907
3908 status = myri10ge_load_firmware(mgp, 1);
3909 if (status != 0) {
3910 dev_err(&pdev->dev, "failed to load firmware\n");
3911 goto abort_with_ioremap;
3912 }
3913 myri10ge_probe_slices(mgp);
3914 status = myri10ge_alloc_slices(mgp);
3915 if (status != 0) {
3916 dev_err(&pdev->dev, "failed to alloc slice state\n");
3917 goto abort_with_firmware;
3918 }
3919 netif_set_real_num_tx_queues(netdev, mgp->num_slices);
3920 netif_set_real_num_rx_queues(netdev, mgp->num_slices);
3921 status = myri10ge_reset(mgp);
3922 if (status != 0) {
3923 dev_err(&pdev->dev, "failed reset\n");
3924 goto abort_with_slices;
3925 }
3926 #ifdef CONFIG_MYRI10GE_DCA
3927 myri10ge_setup_dca(mgp);
3928 #endif
3929 pci_set_drvdata(pdev, mgp);
3930 if ((myri10ge_initial_mtu + ETH_HLEN) > MYRI10GE_MAX_ETHER_MTU)
3931 myri10ge_initial_mtu = MYRI10GE_MAX_ETHER_MTU - ETH_HLEN;
3932 if ((myri10ge_initial_mtu + ETH_HLEN) < 68)
3933 myri10ge_initial_mtu = 68;
3934
3935 netdev->netdev_ops = &myri10ge_netdev_ops;
3936 netdev->mtu = myri10ge_initial_mtu;
3937 netdev->base_addr = mgp->iomem_base;
3938 netdev->hw_features = mgp->features | NETIF_F_LRO | NETIF_F_RXCSUM;
3939 netdev->features = netdev->hw_features;
3940
3941 if (dac_enabled)
3942 netdev->features |= NETIF_F_HIGHDMA;
3943
3944 netdev->vlan_features |= mgp->features;
3945 if (mgp->fw_ver_tiny < 37)
3946 netdev->vlan_features &= ~NETIF_F_TSO6;
3947 if (mgp->fw_ver_tiny < 32)
3948 netdev->vlan_features &= ~NETIF_F_TSO;
3949
3950 /* make sure we can get an irq, and that MSI can be
3951 * setup (if available). Also ensure netdev->irq
3952 * is set to correct value if MSI is enabled */
3953 status = myri10ge_request_irq(mgp);
3954 if (status != 0)
3955 goto abort_with_firmware;
3956 netdev->irq = pdev->irq;
3957 myri10ge_free_irq(mgp);
3958
3959 /* Save configuration space to be restored if the
3960 * nic resets due to a parity error */
3961 pci_save_state(pdev);
3962
3963 /* Setup the watchdog timer */
3964 setup_timer(&mgp->watchdog_timer, myri10ge_watchdog_timer,
3965 (unsigned long)mgp);
3966
3967 spin_lock_init(&mgp->stats_lock);
3968 SET_ETHTOOL_OPS(netdev, &myri10ge_ethtool_ops);
3969 INIT_WORK(&mgp->watchdog_work, myri10ge_watchdog);
3970 status = register_netdev(netdev);
3971 if (status != 0) {
3972 dev_err(&pdev->dev, "register_netdev failed: %d\n", status);
3973 goto abort_with_state;
3974 }
3975 if (mgp->msix_enabled)
3976 dev_info(dev, "%d MSI-X IRQs, tx bndry %d, fw %s, WC %s\n",
3977 mgp->num_slices, mgp->tx_boundary, mgp->fw_name,
3978 (mgp->wc_enabled ? "Enabled" : "Disabled"));
3979 else
3980 dev_info(dev, "%s IRQ %d, tx bndry %d, fw %s, WC %s\n",
3981 mgp->msi_enabled ? "MSI" : "xPIC",
3982 netdev->irq, mgp->tx_boundary, mgp->fw_name,
3983 (mgp->wc_enabled ? "Enabled" : "Disabled"));
3984
3985 board_number++;
3986 return 0;
3987
3988 abort_with_state:
3989 pci_restore_state(pdev);
3990
3991 abort_with_slices:
3992 myri10ge_free_slices(mgp);
3993
3994 abort_with_firmware:
3995 myri10ge_dummy_rdma(mgp, 0);
3996
3997 abort_with_ioremap:
3998 if (mgp->mac_addr_string != NULL)
3999 dev_err(&pdev->dev,
4000 "myri10ge_probe() failed: MAC=%s, SN=%ld\n",
4001 mgp->mac_addr_string, mgp->serial_number);
4002 iounmap(mgp->sram);
4003
4004 abort_with_mtrr:
4005 #ifdef CONFIG_MTRR
4006 if (mgp->mtrr >= 0)
4007 mtrr_del(mgp->mtrr, mgp->iomem_base, mgp->board_span);
4008 #endif
4009 dma_free_coherent(&pdev->dev, sizeof(*mgp->cmd),
4010 mgp->cmd, mgp->cmd_bus);
4011
4012 abort_with_enabled:
4013 pci_disable_device(pdev);
4014
4015 abort_with_netdev:
4016 set_fw_name(mgp, NULL, false);
4017 free_netdev(netdev);
4018 return status;
4019 }
4020
4021 /*
4022 * myri10ge_remove
4023 *
4024 * Does what is necessary to shutdown one Myrinet device. Called
4025 * once for each Myrinet card by the kernel when a module is
4026 * unloaded.
4027 */
4028 static void myri10ge_remove(struct pci_dev *pdev)
4029 {
4030 struct myri10ge_priv *mgp;
4031 struct net_device *netdev;
4032
4033 mgp = pci_get_drvdata(pdev);
4034 if (mgp == NULL)
4035 return;
4036
4037 cancel_work_sync(&mgp->watchdog_work);
4038 netdev = mgp->dev;
4039 unregister_netdev(netdev);
4040
4041 #ifdef CONFIG_MYRI10GE_DCA
4042 myri10ge_teardown_dca(mgp);
4043 #endif
4044 myri10ge_dummy_rdma(mgp, 0);
4045
4046 /* avoid a memory leak */
4047 pci_restore_state(pdev);
4048
4049 iounmap(mgp->sram);
4050
4051 #ifdef CONFIG_MTRR
4052 if (mgp->mtrr >= 0)
4053 mtrr_del(mgp->mtrr, mgp->iomem_base, mgp->board_span);
4054 #endif
4055 myri10ge_free_slices(mgp);
4056 if (mgp->msix_vectors != NULL)
4057 kfree(mgp->msix_vectors);
4058 dma_free_coherent(&pdev->dev, sizeof(*mgp->cmd),
4059 mgp->cmd, mgp->cmd_bus);
4060
4061 set_fw_name(mgp, NULL, false);
4062 free_netdev(netdev);
4063 pci_disable_device(pdev);
4064 pci_set_drvdata(pdev, NULL);
4065 }
4066
4067 #define PCI_DEVICE_ID_MYRICOM_MYRI10GE_Z8E 0x0008
4068 #define PCI_DEVICE_ID_MYRICOM_MYRI10GE_Z8E_9 0x0009
4069
4070 static DEFINE_PCI_DEVICE_TABLE(myri10ge_pci_tbl) = {
4071 {PCI_DEVICE(PCI_VENDOR_ID_MYRICOM, PCI_DEVICE_ID_MYRICOM_MYRI10GE_Z8E)},
4072 {PCI_DEVICE
4073 (PCI_VENDOR_ID_MYRICOM, PCI_DEVICE_ID_MYRICOM_MYRI10GE_Z8E_9)},
4074 {0},
4075 };
4076
4077 MODULE_DEVICE_TABLE(pci, myri10ge_pci_tbl);
4078
4079 static struct pci_driver myri10ge_driver = {
4080 .name = "myri10ge",
4081 .probe = myri10ge_probe,
4082 .remove = myri10ge_remove,
4083 .id_table = myri10ge_pci_tbl,
4084 #ifdef CONFIG_PM
4085 .suspend = myri10ge_suspend,
4086 .resume = myri10ge_resume,
4087 #endif
4088 };
4089
4090 #ifdef CONFIG_MYRI10GE_DCA
4091 static int
4092 myri10ge_notify_dca(struct notifier_block *nb, unsigned long event, void *p)
4093 {
4094 int err = driver_for_each_device(&myri10ge_driver.driver,
4095 NULL, &event,
4096 myri10ge_notify_dca_device);
4097
4098 if (err)
4099 return NOTIFY_BAD;
4100 return NOTIFY_DONE;
4101 }
4102
4103 static struct notifier_block myri10ge_dca_notifier = {
4104 .notifier_call = myri10ge_notify_dca,
4105 .next = NULL,
4106 .priority = 0,
4107 };
4108 #endif /* CONFIG_MYRI10GE_DCA */
4109
4110 static __init int myri10ge_init_module(void)
4111 {
4112 pr_info("Version %s\n", MYRI10GE_VERSION_STR);
4113
4114 if (myri10ge_rss_hash > MXGEFW_RSS_HASH_TYPE_MAX) {
4115 pr_err("Illegal rssh hash type %d, defaulting to source port\n",
4116 myri10ge_rss_hash);
4117 myri10ge_rss_hash = MXGEFW_RSS_HASH_TYPE_SRC_PORT;
4118 }
4119 #ifdef CONFIG_MYRI10GE_DCA
4120 dca_register_notify(&myri10ge_dca_notifier);
4121 #endif
4122 if (myri10ge_max_slices > MYRI10GE_MAX_SLICES)
4123 myri10ge_max_slices = MYRI10GE_MAX_SLICES;
4124
4125 return pci_register_driver(&myri10ge_driver);
4126 }
4127
4128 module_init(myri10ge_init_module);
4129
4130 static __exit void myri10ge_cleanup_module(void)
4131 {
4132 #ifdef CONFIG_MYRI10GE_DCA
4133 dca_unregister_notify(&myri10ge_dca_notifier);
4134 #endif
4135 pci_unregister_driver(&myri10ge_driver);
4136 }
4137
4138 module_exit(myri10ge_cleanup_module);
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree