1 /*- 2 * Copyright (c) 1997, 1998-2003 3 * Bill Paul <wpaul@windriver.com>. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by Bill Paul. 16 * 4. Neither the name of the author nor the names of any co-contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD 24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF 30 * THE POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33 #include <sys/cdefs.h> 34 __FBSDID("$FreeBSD$"); 35 36 /* 37 * RealTek 8139C+/8169/8169S/8110S/8168/8111/8101E PCI NIC driver 38 * 39 * Written by Bill Paul <wpaul@windriver.com> 40 * Senior Networking Software Engineer 41 * Wind River Systems 42 */ 43 44 /* 45 * This driver is designed to support RealTek's next generation of 46 * 10/100 and 10/100/1000 PCI ethernet controllers. There are currently 47 * seven devices in this family: the RTL8139C+, the RTL8169, the RTL8169S, 48 * RTL8110S, the RTL8168, the RTL8111 and the RTL8101E. 49 * 50 * The 8139C+ is a 10/100 ethernet chip. It is backwards compatible 51 * with the older 8139 family, however it also supports a special 52 * C+ mode of operation that provides several new performance enhancing 53 * features. These include: 54 * 55 * o Descriptor based DMA mechanism. Each descriptor represents 56 * a single packet fragment. Data buffers may be aligned on 57 * any byte boundary. 58 * 59 * o 64-bit DMA 60 * 61 * o TCP/IP checksum offload for both RX and TX 62 * 63 * o High and normal priority transmit DMA rings 64 * 65 * o VLAN tag insertion and extraction 66 * 67 * o TCP large send (segmentation offload) 68 * 69 * Like the 8139, the 8139C+ also has a built-in 10/100 PHY. The C+ 70 * programming API is fairly straightforward. The RX filtering, EEPROM 71 * access and PHY access is the same as it is on the older 8139 series 72 * chips. 73 * 74 * The 8169 is a 64-bit 10/100/1000 gigabit ethernet MAC. It has almost the 75 * same programming API and feature set as the 8139C+ with the following 76 * differences and additions: 77 * 78 * o 1000Mbps mode 79 * 80 * o Jumbo frames 81 * 82 * o GMII and TBI ports/registers for interfacing with copper 83 * or fiber PHYs 84 * 85 * o RX and TX DMA rings can have up to 1024 descriptors 86 * (the 8139C+ allows a maximum of 64) 87 * 88 * o Slight differences in register layout from the 8139C+ 89 * 90 * The TX start and timer interrupt registers are at different locations 91 * on the 8169 than they are on the 8139C+. Also, the status word in the 92 * RX descriptor has a slightly different bit layout. The 8169 does not 93 * have a built-in PHY. Most reference boards use a Marvell 88E1000 'Alaska' 94 * copper gigE PHY. 95 * 96 * The 8169S/8110S 10/100/1000 devices have built-in copper gigE PHYs 97 * (the 'S' stands for 'single-chip'). These devices have the same 98 * programming API as the older 8169, but also have some vendor-specific 99 * registers for the on-board PHY. The 8110S is a LAN-on-motherboard 100 * part designed to be pin-compatible with the RealTek 8100 10/100 chip. 101 * 102 * This driver takes advantage of the RX and TX checksum offload and 103 * VLAN tag insertion/extraction features. It also implements TX 104 * interrupt moderation using the timer interrupt registers, which 105 * significantly reduces TX interrupt load. There is also support 106 * for jumbo frames, however the 8169/8169S/8110S can not transmit 107 * jumbo frames larger than 7440, so the max MTU possible with this 108 * driver is 7422 bytes. 109 */ 110 111 #ifdef HAVE_KERNEL_OPTION_HEADERS 112 #include "opt_device_polling.h" 113 #endif 114 115 #include <sys/param.h> 116 #include <sys/endian.h> 117 #include <sys/systm.h> 118 #include <sys/sockio.h> 119 #include <sys/mbuf.h> 120 #include <sys/malloc.h> 121 #include <sys/module.h> 122 #include <sys/kernel.h> 123 #include <sys/socket.h> 124 #include <sys/lock.h> 125 #include <sys/mutex.h> 126 #include <sys/taskqueue.h> 127 128 #include <net/if.h> 129 #include <net/if_arp.h> 130 #include <net/ethernet.h> 131 #include <net/if_dl.h> 132 #include <net/if_media.h> 133 #include <net/if_types.h> 134 #include <net/if_vlan_var.h> 135 136 #include <net/bpf.h> 137 138 #include <machine/bus.h> 139 #include <machine/resource.h> 140 #include <sys/bus.h> 141 #include <sys/rman.h> 142 143 #include <dev/mii/mii.h> 144 #include <dev/mii/miivar.h> 145 146 #include <dev/pci/pcireg.h> 147 #include <dev/pci/pcivar.h> 148 149 #include <pci/if_rlreg.h> 150 151 MODULE_DEPEND(re, pci, 1, 1, 1); 152 MODULE_DEPEND(re, ether, 1, 1, 1); 153 MODULE_DEPEND(re, miibus, 1, 1, 1); 154 155 /* "device miibus" required. See GENERIC if you get errors here. */ 156 #include "miibus_if.h" 157 158 /* Tunables. */ 159 static int msi_disable = 0; 160 TUNABLE_INT("hw.re.msi_disable", &msi_disable); 161 static int prefer_iomap = 0; 162 TUNABLE_INT("hw.re.prefer_iomap", &prefer_iomap); 163 164 #define RE_CSUM_FEATURES (CSUM_IP | CSUM_TCP | CSUM_UDP) 165 166 /* 167 * Various supported device vendors/types and their names. 168 */ 169 static struct rl_type re_devs[] = { 170 { DLINK_VENDORID, DLINK_DEVICEID_528T, 0, 171 "D-Link DGE-528(T) Gigabit Ethernet Adapter" }, 172 { RT_VENDORID, RT_DEVICEID_8139, 0, 173 "RealTek 8139C+ 10/100BaseTX" }, 174 { RT_VENDORID, RT_DEVICEID_8101E, 0, 175 "RealTek 8101E/8102E/8102EL PCIe 10/100baseTX" }, 176 { RT_VENDORID, RT_DEVICEID_8168, 0, 177 "RealTek 8168/8168B/8168C/8168CP/8168D/8111B/8111C/8111CP PCIe " 178 "Gigabit Ethernet" }, 179 { RT_VENDORID, RT_DEVICEID_8169, 0, 180 "RealTek 8169/8169S/8169SB(L)/8110S/8110SB(L) Gigabit Ethernet" }, 181 { RT_VENDORID, RT_DEVICEID_8169SC, 0, 182 "RealTek 8169SC/8110SC Single-chip Gigabit Ethernet" }, 183 { COREGA_VENDORID, COREGA_DEVICEID_CGLAPCIGT, 0, 184 "Corega CG-LAPCIGT (RTL8169S) Gigabit Ethernet" }, 185 { LINKSYS_VENDORID, LINKSYS_DEVICEID_EG1032, 0, 186 "Linksys EG1032 (RTL8169S) Gigabit Ethernet" }, 187 { USR_VENDORID, USR_DEVICEID_997902, 0, 188 "US Robotics 997902 (RTL8169S) Gigabit Ethernet" } 189 }; 190 191 static struct rl_hwrev re_hwrevs[] = { 192 { RL_HWREV_8139, RL_8139, "" }, 193 { RL_HWREV_8139A, RL_8139, "A" }, 194 { RL_HWREV_8139AG, RL_8139, "A-G" }, 195 { RL_HWREV_8139B, RL_8139, "B" }, 196 { RL_HWREV_8130, RL_8139, "8130" }, 197 { RL_HWREV_8139C, RL_8139, "C" }, 198 { RL_HWREV_8139D, RL_8139, "8139D/8100B/8100C" }, 199 { RL_HWREV_8139CPLUS, RL_8139CPLUS, "C+"}, 200 { RL_HWREV_8168_SPIN1, RL_8169, "8168"}, 201 { RL_HWREV_8169, RL_8169, "8169"}, 202 { RL_HWREV_8169S, RL_8169, "8169S"}, 203 { RL_HWREV_8110S, RL_8169, "8110S"}, 204 { RL_HWREV_8169_8110SB, RL_8169, "8169SB/8110SB"}, 205 { RL_HWREV_8169_8110SC, RL_8169, "8169SC/8110SC"}, 206 { RL_HWREV_8169_8110SBL, RL_8169, "8169SBL/8110SBL"}, 207 { RL_HWREV_8169_8110SCE, RL_8169, "8169SC/8110SC"}, 208 { RL_HWREV_8100, RL_8139, "8100"}, 209 { RL_HWREV_8101, RL_8139, "8101"}, 210 { RL_HWREV_8100E, RL_8169, "8100E"}, 211 { RL_HWREV_8101E, RL_8169, "8101E"}, 212 { RL_HWREV_8102E, RL_8169, "8102E"}, 213 { RL_HWREV_8102EL, RL_8169, "8102EL"}, 214 { RL_HWREV_8102EL_SPIN1, RL_8169, "8102EL"}, 215 { RL_HWREV_8168_SPIN2, RL_8169, "8168"}, 216 { RL_HWREV_8168_SPIN3, RL_8169, "8168"}, 217 { RL_HWREV_8168C, RL_8169, "8168C/8111C"}, 218 { RL_HWREV_8168C_SPIN2, RL_8169, "8168C/8111C"}, 219 { RL_HWREV_8168CP, RL_8169, "8168CP/8111CP"}, 220 { RL_HWREV_8168D, RL_8169, "8168D"}, 221 { 0, 0, NULL } 222 }; 223 224 static int re_probe (device_t); 225 static int re_attach (device_t); 226 static int re_detach (device_t); 227 228 static int re_encap (struct rl_softc *, struct mbuf **); 229 230 static void re_dma_map_addr (void *, bus_dma_segment_t *, int, int); 231 static int re_allocmem (device_t, struct rl_softc *); 232 static __inline void re_discard_rxbuf 233 (struct rl_softc *, int); 234 static int re_newbuf (struct rl_softc *, int); 235 static int re_rx_list_init (struct rl_softc *); 236 static int re_tx_list_init (struct rl_softc *); 237 #ifdef RE_FIXUP_RX 238 static __inline void re_fixup_rx 239 (struct mbuf *); 240 #endif 241 static int re_rxeof (struct rl_softc *, int *); 242 static void re_txeof (struct rl_softc *); 243 #ifdef DEVICE_POLLING 244 static int re_poll (struct ifnet *, enum poll_cmd, int); 245 static int re_poll_locked (struct ifnet *, enum poll_cmd, int); 246 #endif 247 static int re_intr (void *); 248 static void re_tick (void *); 249 static void re_tx_task (void *, int); 250 static void re_int_task (void *, int); 251 static void re_start (struct ifnet *); 252 static int re_ioctl (struct ifnet *, u_long, caddr_t); 253 static void re_init (void *); 254 static void re_init_locked (struct rl_softc *); 255 static void re_stop (struct rl_softc *); 256 static void re_watchdog (struct rl_softc *); 257 static int re_suspend (device_t); 258 static int re_resume (device_t); 259 static int re_shutdown (device_t); 260 static int re_ifmedia_upd (struct ifnet *); 261 static void re_ifmedia_sts (struct ifnet *, struct ifmediareq *); 262 263 static void re_eeprom_putbyte (struct rl_softc *, int); 264 static void re_eeprom_getword (struct rl_softc *, int, u_int16_t *); 265 static void re_read_eeprom (struct rl_softc *, caddr_t, int, int); 266 static int re_gmii_readreg (device_t, int, int); 267 static int re_gmii_writereg (device_t, int, int, int); 268 269 static int re_miibus_readreg (device_t, int, int); 270 static int re_miibus_writereg (device_t, int, int, int); 271 static void re_miibus_statchg (device_t); 272 273 static void re_set_rxmode (struct rl_softc *); 274 static void re_reset (struct rl_softc *); 275 static void re_setwol (struct rl_softc *); 276 static void re_clrwol (struct rl_softc *); 277 278 #ifdef RE_DIAG 279 static int re_diag (struct rl_softc *); 280 #endif 281 282 static device_method_t re_methods[] = { 283 /* Device interface */ 284 DEVMETHOD(device_probe, re_probe), 285 DEVMETHOD(device_attach, re_attach), 286 DEVMETHOD(device_detach, re_detach), 287 DEVMETHOD(device_suspend, re_suspend), 288 DEVMETHOD(device_resume, re_resume), 289 DEVMETHOD(device_shutdown, re_shutdown), 290 291 /* bus interface */ 292 DEVMETHOD(bus_print_child, bus_generic_print_child), 293 DEVMETHOD(bus_driver_added, bus_generic_driver_added), 294 295 /* MII interface */ 296 DEVMETHOD(miibus_readreg, re_miibus_readreg), 297 DEVMETHOD(miibus_writereg, re_miibus_writereg), 298 DEVMETHOD(miibus_statchg, re_miibus_statchg), 299 300 { 0, 0 } 301 }; 302 303 static driver_t re_driver = { 304 "re", 305 re_methods, 306 sizeof(struct rl_softc) 307 }; 308 309 static devclass_t re_devclass; 310 311 DRIVER_MODULE(re, pci, re_driver, re_devclass, 0, 0); 312 DRIVER_MODULE(miibus, re, miibus_driver, miibus_devclass, 0, 0); 313 314 #define EE_SET(x) \ 315 CSR_WRITE_1(sc, RL_EECMD, \ 316 CSR_READ_1(sc, RL_EECMD) | x) 317 318 #define EE_CLR(x) \ 319 CSR_WRITE_1(sc, RL_EECMD, \ 320 CSR_READ_1(sc, RL_EECMD) & ~x) 321 322 /* 323 * Send a read command and address to the EEPROM, check for ACK. 324 */ 325 static void 326 re_eeprom_putbyte(struct rl_softc *sc, int addr) 327 { 328 int d, i; 329 330 d = addr | (RL_9346_READ << sc->rl_eewidth); 331 332 /* 333 * Feed in each bit and strobe the clock. 334 */ 335 336 for (i = 1 << (sc->rl_eewidth + 3); i; i >>= 1) { 337 if (d & i) { 338 EE_SET(RL_EE_DATAIN); 339 } else { 340 EE_CLR(RL_EE_DATAIN); 341 } 342 DELAY(100); 343 EE_SET(RL_EE_CLK); 344 DELAY(150); 345 EE_CLR(RL_EE_CLK); 346 DELAY(100); 347 } 348 } 349 350 /* 351 * Read a word of data stored in the EEPROM at address 'addr.' 352 */ 353 static void 354 re_eeprom_getword(struct rl_softc *sc, int addr, u_int16_t *dest) 355 { 356 int i; 357 u_int16_t word = 0; 358 359 /* 360 * Send address of word we want to read. 361 */ 362 re_eeprom_putbyte(sc, addr); 363 364 /* 365 * Start reading bits from EEPROM. 366 */ 367 for (i = 0x8000; i; i >>= 1) { 368 EE_SET(RL_EE_CLK); 369 DELAY(100); 370 if (CSR_READ_1(sc, RL_EECMD) & RL_EE_DATAOUT) 371 word |= i; 372 EE_CLR(RL_EE_CLK); 373 DELAY(100); 374 } 375 376 *dest = word; 377 } 378 379 /* 380 * Read a sequence of words from the EEPROM. 381 */ 382 static void 383 re_read_eeprom(struct rl_softc *sc, caddr_t dest, int off, int cnt) 384 { 385 int i; 386 u_int16_t word = 0, *ptr; 387 388 CSR_SETBIT_1(sc, RL_EECMD, RL_EEMODE_PROGRAM); 389 390 DELAY(100); 391 392 for (i = 0; i < cnt; i++) { 393 CSR_SETBIT_1(sc, RL_EECMD, RL_EE_SEL); 394 re_eeprom_getword(sc, off + i, &word); 395 CSR_CLRBIT_1(sc, RL_EECMD, RL_EE_SEL); 396 ptr = (u_int16_t *)(dest + (i * 2)); 397 *ptr = word; 398 } 399 400 CSR_CLRBIT_1(sc, RL_EECMD, RL_EEMODE_PROGRAM); 401 } 402 403 static int 404 re_gmii_readreg(device_t dev, int phy, int reg) 405 { 406 struct rl_softc *sc; 407 u_int32_t rval; 408 int i; 409 410 if (phy != 1) 411 return (0); 412 413 sc = device_get_softc(dev); 414 415 /* Let the rgephy driver read the GMEDIASTAT register */ 416 417 if (reg == RL_GMEDIASTAT) { 418 rval = CSR_READ_1(sc, RL_GMEDIASTAT); 419 return (rval); 420 } 421 422 CSR_WRITE_4(sc, RL_PHYAR, reg << 16); 423 DELAY(1000); 424 425 for (i = 0; i < RL_PHY_TIMEOUT; i++) { 426 rval = CSR_READ_4(sc, RL_PHYAR); 427 if (rval & RL_PHYAR_BUSY) 428 break; 429 DELAY(100); 430 } 431 432 if (i == RL_PHY_TIMEOUT) { 433 device_printf(sc->rl_dev, "PHY read failed\n"); 434 return (0); 435 } 436 437 return (rval & RL_PHYAR_PHYDATA); 438 } 439 440 static int 441 re_gmii_writereg(device_t dev, int phy, int reg, int data) 442 { 443 struct rl_softc *sc; 444 u_int32_t rval; 445 int i; 446 447 sc = device_get_softc(dev); 448 449 CSR_WRITE_4(sc, RL_PHYAR, (reg << 16) | 450 (data & RL_PHYAR_PHYDATA) | RL_PHYAR_BUSY); 451 DELAY(1000); 452 453 for (i = 0; i < RL_PHY_TIMEOUT; i++) { 454 rval = CSR_READ_4(sc, RL_PHYAR); 455 if (!(rval & RL_PHYAR_BUSY)) 456 break; 457 DELAY(100); 458 } 459 460 if (i == RL_PHY_TIMEOUT) { 461 device_printf(sc->rl_dev, "PHY write failed\n"); 462 return (0); 463 } 464 465 return (0); 466 } 467 468 static int 469 re_miibus_readreg(device_t dev, int phy, int reg) 470 { 471 struct rl_softc *sc; 472 u_int16_t rval = 0; 473 u_int16_t re8139_reg = 0; 474 475 sc = device_get_softc(dev); 476 477 if (sc->rl_type == RL_8169) { 478 rval = re_gmii_readreg(dev, phy, reg); 479 return (rval); 480 } 481 482 /* Pretend the internal PHY is only at address 0 */ 483 if (phy) { 484 return (0); 485 } 486 switch (reg) { 487 case MII_BMCR: 488 re8139_reg = RL_BMCR; 489 break; 490 case MII_BMSR: 491 re8139_reg = RL_BMSR; 492 break; 493 case MII_ANAR: 494 re8139_reg = RL_ANAR; 495 break; 496 case MII_ANER: 497 re8139_reg = RL_ANER; 498 break; 499 case MII_ANLPAR: 500 re8139_reg = RL_LPAR; 501 break; 502 case MII_PHYIDR1: 503 case MII_PHYIDR2: 504 return (0); 505 /* 506 * Allow the rlphy driver to read the media status 507 * register. If we have a link partner which does not 508 * support NWAY, this is the register which will tell 509 * us the results of parallel detection. 510 */ 511 case RL_MEDIASTAT: 512 rval = CSR_READ_1(sc, RL_MEDIASTAT); 513 return (rval); 514 default: 515 device_printf(sc->rl_dev, "bad phy register\n"); 516 return (0); 517 } 518 rval = CSR_READ_2(sc, re8139_reg); 519 if (sc->rl_type == RL_8139CPLUS && re8139_reg == RL_BMCR) { 520 /* 8139C+ has different bit layout. */ 521 rval &= ~(BMCR_LOOP | BMCR_ISO); 522 } 523 return (rval); 524 } 525 526 static int 527 re_miibus_writereg(device_t dev, int phy, int reg, int data) 528 { 529 struct rl_softc *sc; 530 u_int16_t re8139_reg = 0; 531 int rval = 0; 532 533 sc = device_get_softc(dev); 534 535 if (sc->rl_type == RL_8169) { 536 rval = re_gmii_writereg(dev, phy, reg, data); 537 return (rval); 538 } 539 540 /* Pretend the internal PHY is only at address 0 */ 541 if (phy) 542 return (0); 543 544 switch (reg) { 545 case MII_BMCR: 546 re8139_reg = RL_BMCR; 547 if (sc->rl_type == RL_8139CPLUS) { 548 /* 8139C+ has different bit layout. */ 549 data &= ~(BMCR_LOOP | BMCR_ISO); 550 } 551 break; 552 case MII_BMSR: 553 re8139_reg = RL_BMSR; 554 break; 555 case MII_ANAR: 556 re8139_reg = RL_ANAR; 557 break; 558 case MII_ANER: 559 re8139_reg = RL_ANER; 560 break; 561 case MII_ANLPAR: 562 re8139_reg = RL_LPAR; 563 break; 564 case MII_PHYIDR1: 565 case MII_PHYIDR2: 566 return (0); 567 break; 568 default: 569 device_printf(sc->rl_dev, "bad phy register\n"); 570 return (0); 571 } 572 CSR_WRITE_2(sc, re8139_reg, data); 573 return (0); 574 } 575 576 static void 577 re_miibus_statchg(device_t dev) 578 { 579 struct rl_softc *sc; 580 struct ifnet *ifp; 581 struct mii_data *mii; 582 583 sc = device_get_softc(dev); 584 mii = device_get_softc(sc->rl_miibus); 585 ifp = sc->rl_ifp; 586 if (mii == NULL || ifp == NULL || 587 (ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) 588 return; 589 590 sc->rl_flags &= ~RL_FLAG_LINK; 591 if ((mii->mii_media_status & (IFM_ACTIVE | IFM_AVALID)) == 592 (IFM_ACTIVE | IFM_AVALID)) { 593 switch (IFM_SUBTYPE(mii->mii_media_active)) { 594 case IFM_10_T: 595 case IFM_100_TX: 596 sc->rl_flags |= RL_FLAG_LINK; 597 break; 598 case IFM_1000_T: 599 if ((sc->rl_flags & RL_FLAG_FASTETHER) != 0) 600 break; 601 sc->rl_flags |= RL_FLAG_LINK; 602 break; 603 default: 604 break; 605 } 606 } 607 /* 608 * RealTek controllers does not provide any interface to 609 * Tx/Rx MACs for resolved speed, duplex and flow-control 610 * parameters. 611 */ 612 } 613 614 /* 615 * Set the RX configuration and 64-bit multicast hash filter. 616 */ 617 static void 618 re_set_rxmode(struct rl_softc *sc) 619 { 620 struct ifnet *ifp; 621 struct ifmultiaddr *ifma; 622 uint32_t hashes[2] = { 0, 0 }; 623 uint32_t h, rxfilt; 624 625 RL_LOCK_ASSERT(sc); 626 627 ifp = sc->rl_ifp; 628 629 rxfilt = RL_RXCFG_CONFIG | RL_RXCFG_RX_INDIV | RL_RXCFG_RX_BROAD; 630 631 if (ifp->if_flags & (IFF_ALLMULTI | IFF_PROMISC)) { 632 if (ifp->if_flags & IFF_PROMISC) 633 rxfilt |= RL_RXCFG_RX_ALLPHYS; 634 /* 635 * Unlike other hardwares, we have to explicitly set 636 * RL_RXCFG_RX_MULTI to receive multicast frames in 637 * promiscuous mode. 638 */ 639 rxfilt |= RL_RXCFG_RX_MULTI; 640 hashes[0] = hashes[1] = 0xffffffff; 641 goto done; 642 } 643 644 if_maddr_rlock(ifp); 645 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { 646 if (ifma->ifma_addr->sa_family != AF_LINK) 647 continue; 648 h = ether_crc32_be(LLADDR((struct sockaddr_dl *) 649 ifma->ifma_addr), ETHER_ADDR_LEN) >> 26; 650 if (h < 32) 651 hashes[0] |= (1 << h); 652 else 653 hashes[1] |= (1 << (h - 32)); 654 } 655 if_maddr_runlock(ifp); 656 657 if (hashes[0] != 0 || hashes[1] != 0) { 658 /* 659 * For some unfathomable reason, RealTek decided to 660 * reverse the order of the multicast hash registers 661 * in the PCI Express parts. This means we have to 662 * write the hash pattern in reverse order for those 663 * devices. 664 */ 665 if ((sc->rl_flags & RL_FLAG_PCIE) != 0) { 666 h = bswap32(hashes[0]); 667 hashes[0] = bswap32(hashes[1]); 668 hashes[1] = h; 669 } 670 rxfilt |= RL_RXCFG_RX_MULTI; 671 } 672 673 done: 674 CSR_WRITE_4(sc, RL_MAR0, hashes[0]); 675 CSR_WRITE_4(sc, RL_MAR4, hashes[1]); 676 CSR_WRITE_4(sc, RL_RXCFG, rxfilt); 677 } 678 679 static void 680 re_reset(struct rl_softc *sc) 681 { 682 int i; 683 684 RL_LOCK_ASSERT(sc); 685 686 CSR_WRITE_1(sc, RL_COMMAND, RL_CMD_RESET); 687 688 for (i = 0; i < RL_TIMEOUT; i++) { 689 DELAY(10); 690 if (!(CSR_READ_1(sc, RL_COMMAND) & RL_CMD_RESET)) 691 break; 692 } 693 if (i == RL_TIMEOUT) 694 device_printf(sc->rl_dev, "reset never completed!\n"); 695 696 if ((sc->rl_flags & RL_FLAG_MACRESET) != 0) 697 CSR_WRITE_1(sc, 0x82, 1); 698 if (sc->rl_hwrev == RL_HWREV_8169S) 699 re_gmii_writereg(sc->rl_dev, 1, 0x0b, 0); 700 } 701 702 #ifdef RE_DIAG 703 704 /* 705 * The following routine is designed to test for a defect on some 706 * 32-bit 8169 cards. Some of these NICs have the REQ64# and ACK64# 707 * lines connected to the bus, however for a 32-bit only card, they 708 * should be pulled high. The result of this defect is that the 709 * NIC will not work right if you plug it into a 64-bit slot: DMA 710 * operations will be done with 64-bit transfers, which will fail 711 * because the 64-bit data lines aren't connected. 712 * 713 * There's no way to work around this (short of talking a soldering 714 * iron to the board), however we can detect it. The method we use 715 * here is to put the NIC into digital loopback mode, set the receiver 716 * to promiscuous mode, and then try to send a frame. We then compare 717 * the frame data we sent to what was received. If the data matches, 718 * then the NIC is working correctly, otherwise we know the user has 719 * a defective NIC which has been mistakenly plugged into a 64-bit PCI 720 * slot. In the latter case, there's no way the NIC can work correctly, 721 * so we print out a message on the console and abort the device attach. 722 */ 723 724 static int 725 re_diag(struct rl_softc *sc) 726 { 727 struct ifnet *ifp = sc->rl_ifp; 728 struct mbuf *m0; 729 struct ether_header *eh; 730 struct rl_desc *cur_rx; 731 u_int16_t status; 732 u_int32_t rxstat; 733 int total_len, i, error = 0, phyaddr; 734 u_int8_t dst[] = { 0x00, 'h', 'e', 'l', 'l', 'o' }; 735 u_int8_t src[] = { 0x00, 'w', 'o', 'r', 'l', 'd' }; 736 737 /* Allocate a single mbuf */ 738 MGETHDR(m0, M_DONTWAIT, MT_DATA); 739 if (m0 == NULL) 740 return (ENOBUFS); 741 742 RL_LOCK(sc); 743 744 /* 745 * Initialize the NIC in test mode. This sets the chip up 746 * so that it can send and receive frames, but performs the 747 * following special functions: 748 * - Puts receiver in promiscuous mode 749 * - Enables digital loopback mode 750 * - Leaves interrupts turned off 751 */ 752 753 ifp->if_flags |= IFF_PROMISC; 754 sc->rl_testmode = 1; 755 re_init_locked(sc); 756 sc->rl_flags |= RL_FLAG_LINK; 757 if (sc->rl_type == RL_8169) 758 phyaddr = 1; 759 else 760 phyaddr = 0; 761 762 re_miibus_writereg(sc->rl_dev, phyaddr, MII_BMCR, BMCR_RESET); 763 for (i = 0; i < RL_TIMEOUT; i++) { 764 status = re_miibus_readreg(sc->rl_dev, phyaddr, MII_BMCR); 765 if (!(status & BMCR_RESET)) 766 break; 767 } 768 769 re_miibus_writereg(sc->rl_dev, phyaddr, MII_BMCR, BMCR_LOOP); 770 CSR_WRITE_2(sc, RL_ISR, RL_INTRS); 771 772 DELAY(100000); 773 774 /* Put some data in the mbuf */ 775 776 eh = mtod(m0, struct ether_header *); 777 bcopy ((char *)&dst, eh->ether_dhost, ETHER_ADDR_LEN); 778 bcopy ((char *)&src, eh->ether_shost, ETHER_ADDR_LEN); 779 eh->ether_type = htons(ETHERTYPE_IP); 780 m0->m_pkthdr.len = m0->m_len = ETHER_MIN_LEN - ETHER_CRC_LEN; 781 782 /* 783 * Queue the packet, start transmission. 784 * Note: IF_HANDOFF() ultimately calls re_start() for us. 785 */ 786 787 CSR_WRITE_2(sc, RL_ISR, 0xFFFF); 788 RL_UNLOCK(sc); 789 /* XXX: re_diag must not be called when in ALTQ mode */ 790 IF_HANDOFF(&ifp->if_snd, m0, ifp); 791 RL_LOCK(sc); 792 m0 = NULL; 793 794 /* Wait for it to propagate through the chip */ 795 796 DELAY(100000); 797 for (i = 0; i < RL_TIMEOUT; i++) { 798 status = CSR_READ_2(sc, RL_ISR); 799 CSR_WRITE_2(sc, RL_ISR, status); 800 if ((status & (RL_ISR_TIMEOUT_EXPIRED|RL_ISR_RX_OK)) == 801 (RL_ISR_TIMEOUT_EXPIRED|RL_ISR_RX_OK)) 802 break; 803 DELAY(10); 804 } 805 806 if (i == RL_TIMEOUT) { 807 device_printf(sc->rl_dev, 808 "diagnostic failed, failed to receive packet in" 809 " loopback mode\n"); 810 error = EIO; 811 goto done; 812 } 813 814 /* 815 * The packet should have been dumped into the first 816 * entry in the RX DMA ring. Grab it from there. 817 */ 818 819 bus_dmamap_sync(sc->rl_ldata.rl_rx_list_tag, 820 sc->rl_ldata.rl_rx_list_map, 821 BUS_DMASYNC_POSTREAD); 822 bus_dmamap_sync(sc->rl_ldata.rl_rx_mtag, 823 sc->rl_ldata.rl_rx_desc[0].rx_dmamap, 824 BUS_DMASYNC_POSTREAD); 825 bus_dmamap_unload(sc->rl_ldata.rl_rx_mtag, 826 sc->rl_ldata.rl_rx_desc[0].rx_dmamap); 827 828 m0 = sc->rl_ldata.rl_rx_desc[0].rx_m; 829 sc->rl_ldata.rl_rx_desc[0].rx_m = NULL; 830 eh = mtod(m0, struct ether_header *); 831 832 cur_rx = &sc->rl_ldata.rl_rx_list[0]; 833 total_len = RL_RXBYTES(cur_rx); 834 rxstat = le32toh(cur_rx->rl_cmdstat); 835 836 if (total_len != ETHER_MIN_LEN) { 837 device_printf(sc->rl_dev, 838 "diagnostic failed, received short packet\n"); 839 error = EIO; 840 goto done; 841 } 842 843 /* Test that the received packet data matches what we sent. */ 844 845 if (bcmp((char *)&eh->ether_dhost, (char *)&dst, ETHER_ADDR_LEN) || 846 bcmp((char *)&eh->ether_shost, (char *)&src, ETHER_ADDR_LEN) || 847 ntohs(eh->ether_type) != ETHERTYPE_IP) { 848 device_printf(sc->rl_dev, "WARNING, DMA FAILURE!\n"); 849 device_printf(sc->rl_dev, "expected TX data: %6D/%6D/0x%x\n", 850 dst, ":", src, ":", ETHERTYPE_IP); 851 device_printf(sc->rl_dev, "received RX data: %6D/%6D/0x%x\n", 852 eh->ether_dhost, ":", eh->ether_shost, ":", 853 ntohs(eh->ether_type)); 854 device_printf(sc->rl_dev, "You may have a defective 32-bit " 855 "NIC plugged into a 64-bit PCI slot.\n"); 856 device_printf(sc->rl_dev, "Please re-install the NIC in a " 857 "32-bit slot for proper operation.\n"); 858 device_printf(sc->rl_dev, "Read the re(4) man page for more " 859 "details.\n"); 860 error = EIO; 861 } 862 863 done: 864 /* Turn interface off, release resources */ 865 866 sc->rl_testmode = 0; 867 sc->rl_flags &= ~RL_FLAG_LINK; 868 ifp->if_flags &= ~IFF_PROMISC; 869 re_stop(sc); 870 if (m0 != NULL) 871 m_freem(m0); 872 873 RL_UNLOCK(sc); 874 875 return (error); 876 } 877 878 #endif 879 880 /* 881 * Probe for a RealTek 8139C+/8169/8110 chip. Check the PCI vendor and device 882 * IDs against our list and return a device name if we find a match. 883 */ 884 static int 885 re_probe(device_t dev) 886 { 887 struct rl_type *t; 888 uint16_t devid, vendor; 889 uint16_t revid, sdevid; 890 int i; 891 892 vendor = pci_get_vendor(dev); 893 devid = pci_get_device(dev); 894 revid = pci_get_revid(dev); 895 sdevid = pci_get_subdevice(dev); 896 897 if (vendor == LINKSYS_VENDORID && devid == LINKSYS_DEVICEID_EG1032) { 898 if (sdevid != LINKSYS_SUBDEVICE_EG1032_REV3) { 899 /* 900 * Only attach to rev. 3 of the Linksys EG1032 adapter. 901 * Rev. 2 is supported by sk(4). 902 */ 903 return (ENXIO); 904 } 905 } 906 907 if (vendor == RT_VENDORID && devid == RT_DEVICEID_8139) { 908 if (revid != 0x20) { 909 /* 8139, let rl(4) take care of this device. */ 910 return (ENXIO); 911 } 912 } 913 914 t = re_devs; 915 for (i = 0; i < sizeof(re_devs) / sizeof(re_devs[0]); i++, t++) { 916 if (vendor == t->rl_vid && devid == t->rl_did) { 917 device_set_desc(dev, t->rl_name); 918 return (BUS_PROBE_DEFAULT); 919 } 920 } 921 922 return (ENXIO); 923 } 924 925 /* 926 * Map a single buffer address. 927 */ 928 929 static void 930 re_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nseg, int error) 931 { 932 bus_addr_t *addr; 933 934 if (error) 935 return; 936 937 KASSERT(nseg == 1, ("too many DMA segments, %d should be 1", nseg)); 938 addr = arg; 939 *addr = segs->ds_addr; 940 } 941 942 static int 943 re_allocmem(device_t dev, struct rl_softc *sc) 944 { 945 bus_size_t rx_list_size, tx_list_size; 946 int error; 947 int i; 948 949 rx_list_size = sc->rl_ldata.rl_rx_desc_cnt * sizeof(struct rl_desc); 950 tx_list_size = sc->rl_ldata.rl_tx_desc_cnt * sizeof(struct rl_desc); 951 952 /* 953 * Allocate the parent bus DMA tag appropriate for PCI. 954 * In order to use DAC, RL_CPLUSCMD_PCI_DAC bit of RL_CPLUS_CMD 955 * register should be set. However some RealTek chips are known 956 * to be buggy on DAC handling, therefore disable DAC by limiting 957 * DMA address space to 32bit. PCIe variants of RealTek chips 958 * may not have the limitation but I took safer path. 959 */ 960 error = bus_dma_tag_create(bus_get_dma_tag(dev), 1, 0, 961 BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL, 962 BUS_SPACE_MAXSIZE_32BIT, 0, BUS_SPACE_MAXSIZE_32BIT, 0, 963 NULL, NULL, &sc->rl_parent_tag); 964 if (error) { 965 device_printf(dev, "could not allocate parent DMA tag\n"); 966 return (error); 967 } 968 969 /* 970 * Allocate map for TX mbufs. 971 */ 972 error = bus_dma_tag_create(sc->rl_parent_tag, 1, 0, 973 BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, 974 NULL, MCLBYTES * RL_NTXSEGS, RL_NTXSEGS, 4096, 0, 975 NULL, NULL, &sc->rl_ldata.rl_tx_mtag); 976 if (error) { 977 device_printf(dev, "could not allocate TX DMA tag\n"); 978 return (error); 979 } 980 981 /* 982 * Allocate map for RX mbufs. 983 */ 984 985 error = bus_dma_tag_create(sc->rl_parent_tag, sizeof(uint64_t), 0, 986 BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, 987 MCLBYTES, 1, MCLBYTES, 0, NULL, NULL, &sc->rl_ldata.rl_rx_mtag); 988 if (error) { 989 device_printf(dev, "could not allocate RX DMA tag\n"); 990 return (error); 991 } 992 993 /* 994 * Allocate map for TX descriptor list. 995 */ 996 error = bus_dma_tag_create(sc->rl_parent_tag, RL_RING_ALIGN, 997 0, BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, 998 NULL, tx_list_size, 1, tx_list_size, 0, 999 NULL, NULL, &sc->rl_ldata.rl_tx_list_tag); 1000 if (error) { 1001 device_printf(dev, "could not allocate TX DMA ring tag\n"); 1002 return (error); 1003 } 1004 1005 /* Allocate DMA'able memory for the TX ring */ 1006 1007 error = bus_dmamem_alloc(sc->rl_ldata.rl_tx_list_tag, 1008 (void **)&sc->rl_ldata.rl_tx_list, 1009 BUS_DMA_WAITOK | BUS_DMA_COHERENT | BUS_DMA_ZERO, 1010 &sc->rl_ldata.rl_tx_list_map); 1011 if (error) { 1012 device_printf(dev, "could not allocate TX DMA ring\n"); 1013 return (error); 1014 } 1015 1016 /* Load the map for the TX ring. */ 1017 1018 sc->rl_ldata.rl_tx_list_addr = 0; 1019 error = bus_dmamap_load(sc->rl_ldata.rl_tx_list_tag, 1020 sc->rl_ldata.rl_tx_list_map, sc->rl_ldata.rl_tx_list, 1021 tx_list_size, re_dma_map_addr, 1022 &sc->rl_ldata.rl_tx_list_addr, BUS_DMA_NOWAIT); 1023 if (error != 0 || sc->rl_ldata.rl_tx_list_addr == 0) { 1024 device_printf(dev, "could not load TX DMA ring\n"); 1025 return (ENOMEM); 1026 } 1027 1028 /* Create DMA maps for TX buffers */ 1029 1030 for (i = 0; i < sc->rl_ldata.rl_tx_desc_cnt; i++) { 1031 error = bus_dmamap_create(sc->rl_ldata.rl_tx_mtag, 0, 1032 &sc->rl_ldata.rl_tx_desc[i].tx_dmamap); 1033 if (error) { 1034 device_printf(dev, "could not create DMA map for TX\n"); 1035 return (error); 1036 } 1037 } 1038 1039 /* 1040 * Allocate map for RX descriptor list. 1041 */ 1042 error = bus_dma_tag_create(sc->rl_parent_tag, RL_RING_ALIGN, 1043 0, BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, 1044 NULL, rx_list_size, 1, rx_list_size, 0, 1045 NULL, NULL, &sc->rl_ldata.rl_rx_list_tag); 1046 if (error) { 1047 device_printf(dev, "could not create RX DMA ring tag\n"); 1048 return (error); 1049 } 1050 1051 /* Allocate DMA'able memory for the RX ring */ 1052 1053 error = bus_dmamem_alloc(sc->rl_ldata.rl_rx_list_tag, 1054 (void **)&sc->rl_ldata.rl_rx_list, 1055 BUS_DMA_WAITOK | BUS_DMA_COHERENT | BUS_DMA_ZERO, 1056 &sc->rl_ldata.rl_rx_list_map); 1057 if (error) { 1058 device_printf(dev, "could not allocate RX DMA ring\n"); 1059 return (error); 1060 } 1061 1062 /* Load the map for the RX ring. */ 1063 1064 sc->rl_ldata.rl_rx_list_addr = 0; 1065 error = bus_dmamap_load(sc->rl_ldata.rl_rx_list_tag, 1066 sc->rl_ldata.rl_rx_list_map, sc->rl_ldata.rl_rx_list, 1067 rx_list_size, re_dma_map_addr, 1068 &sc->rl_ldata.rl_rx_list_addr, BUS_DMA_NOWAIT); 1069 if (error != 0 || sc->rl_ldata.rl_rx_list_addr == 0) { 1070 device_printf(dev, "could not load RX DMA ring\n"); 1071 return (ENOMEM); 1072 } 1073 1074 /* Create DMA maps for RX buffers */ 1075 1076 error = bus_dmamap_create(sc->rl_ldata.rl_rx_mtag, 0, 1077 &sc->rl_ldata.rl_rx_sparemap); 1078 if (error) { 1079 device_printf(dev, "could not create spare DMA map for RX\n"); 1080 return (error); 1081 } 1082 for (i = 0; i < sc->rl_ldata.rl_rx_desc_cnt; i++) { 1083 error = bus_dmamap_create(sc->rl_ldata.rl_rx_mtag, 0, 1084 &sc->rl_ldata.rl_rx_desc[i].rx_dmamap); 1085 if (error) { 1086 device_printf(dev, "could not create DMA map for RX\n"); 1087 return (error); 1088 } 1089 } 1090 1091 return (0); 1092 } 1093 1094 /* 1095 * Attach the interface. Allocate softc structures, do ifmedia 1096 * setup and ethernet/BPF attach. 1097 */ 1098 static int 1099 re_attach(device_t dev) 1100 { 1101 u_char eaddr[ETHER_ADDR_LEN]; 1102 u_int16_t as[ETHER_ADDR_LEN / 2]; 1103 struct rl_softc *sc; 1104 struct ifnet *ifp; 1105 struct rl_hwrev *hw_rev; 1106 int hwrev; 1107 u_int16_t devid, re_did = 0; 1108 int error = 0, rid, i; 1109 int msic, reg; 1110 uint8_t cfg; 1111 1112 sc = device_get_softc(dev); 1113 sc->rl_dev = dev; 1114 1115 mtx_init(&sc->rl_mtx, device_get_nameunit(dev), MTX_NETWORK_LOCK, 1116 MTX_DEF); 1117 callout_init_mtx(&sc->rl_stat_callout, &sc->rl_mtx, 0); 1118 1119 /* 1120 * Map control/status registers. 1121 */ 1122 pci_enable_busmaster(dev); 1123 1124 devid = pci_get_device(dev); 1125 /* 1126 * Prefer memory space register mapping over IO space. 1127 * Because RTL8169SC does not seem to work when memory mapping 1128 * is used always activate io mapping. 1129 */ 1130 if (devid == RT_DEVICEID_8169SC) 1131 prefer_iomap = 1; 1132 if (prefer_iomap == 0) { 1133 sc->rl_res_id = PCIR_BAR(1); 1134 sc->rl_res_type = SYS_RES_MEMORY; 1135 /* RTL8168/8101E seems to use different BARs. */ 1136 if (devid == RT_DEVICEID_8168 || devid == RT_DEVICEID_8101E) 1137 sc->rl_res_id = PCIR_BAR(2); 1138 } else { 1139 sc->rl_res_id = PCIR_BAR(0); 1140 sc->rl_res_type = SYS_RES_IOPORT; 1141 } 1142 sc->rl_res = bus_alloc_resource_any(dev, sc->rl_res_type, 1143 &sc->rl_res_id, RF_ACTIVE); 1144 if (sc->rl_res == NULL && prefer_iomap == 0) { 1145 sc->rl_res_id = PCIR_BAR(0); 1146 sc->rl_res_type = SYS_RES_IOPORT; 1147 sc->rl_res = bus_alloc_resource_any(dev, sc->rl_res_type, 1148 &sc->rl_res_id, RF_ACTIVE); 1149 } 1150 if (sc->rl_res == NULL) { 1151 device_printf(dev, "couldn't map ports/memory\n"); 1152 error = ENXIO; 1153 goto fail; 1154 } 1155 1156 sc->rl_btag = rman_get_bustag(sc->rl_res); 1157 sc->rl_bhandle = rman_get_bushandle(sc->rl_res); 1158 1159 msic = 0; 1160 if (pci_find_extcap(dev, PCIY_EXPRESS, ®) == 0) { 1161 sc->rl_flags |= RL_FLAG_PCIE; 1162 msic = pci_msi_count(dev); 1163 if (bootverbose) 1164 device_printf(dev, "MSI count : %d\n", msic); 1165 } 1166 if (msic > 0 && msi_disable == 0) { 1167 msic = 1; 1168 if (pci_alloc_msi(dev, &msic) == 0) { 1169 if (msic == RL_MSI_MESSAGES) { 1170 device_printf(dev, "Using %d MSI messages\n", 1171 msic); 1172 sc->rl_flags |= RL_FLAG_MSI; 1173 /* Explicitly set MSI enable bit. */ 1174 CSR_WRITE_1(sc, RL_EECMD, RL_EE_MODE); 1175 cfg = CSR_READ_1(sc, RL_CFG2); 1176 cfg |= RL_CFG2_MSI; 1177 CSR_WRITE_1(sc, RL_CFG2, cfg); 1178 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 1179 } else 1180 pci_release_msi(dev); 1181 } 1182 } 1183 1184 /* Allocate interrupt */ 1185 if ((sc->rl_flags & RL_FLAG_MSI) == 0) { 1186 rid = 0; 1187 sc->rl_irq[0] = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, 1188 RF_SHAREABLE | RF_ACTIVE); 1189 if (sc->rl_irq[0] == NULL) { 1190 device_printf(dev, "couldn't allocate IRQ resources\n"); 1191 error = ENXIO; 1192 goto fail; 1193 } 1194 } else { 1195 for (i = 0, rid = 1; i < RL_MSI_MESSAGES; i++, rid++) { 1196 sc->rl_irq[i] = bus_alloc_resource_any(dev, 1197 SYS_RES_IRQ, &rid, RF_ACTIVE); 1198 if (sc->rl_irq[i] == NULL) { 1199 device_printf(dev, 1200 "couldn't llocate IRQ resources for " 1201 "message %d\n", rid); 1202 error = ENXIO; 1203 goto fail; 1204 } 1205 } 1206 } 1207 1208 if ((sc->rl_flags & RL_FLAG_MSI) == 0) { 1209 CSR_WRITE_1(sc, RL_EECMD, RL_EE_MODE); 1210 cfg = CSR_READ_1(sc, RL_CFG2); 1211 if ((cfg & RL_CFG2_MSI) != 0) { 1212 device_printf(dev, "turning off MSI enable bit.\n"); 1213 cfg &= ~RL_CFG2_MSI; 1214 CSR_WRITE_1(sc, RL_CFG2, cfg); 1215 } 1216 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 1217 } 1218 1219 /* Reset the adapter. */ 1220 RL_LOCK(sc); 1221 re_reset(sc); 1222 RL_UNLOCK(sc); 1223 1224 hw_rev = re_hwrevs; 1225 hwrev = CSR_READ_4(sc, RL_TXCFG); 1226 switch (hwrev & 0x70000000) { 1227 case 0x00000000: 1228 case 0x10000000: 1229 device_printf(dev, "Chip rev. 0x%08x\n", hwrev & 0xfc800000); 1230 hwrev &= (RL_TXCFG_HWREV | 0x80000000); 1231 break; 1232 default: 1233 device_printf(dev, "Chip rev. 0x%08x\n", hwrev & 0x7c800000); 1234 hwrev &= RL_TXCFG_HWREV; 1235 break; 1236 } 1237 device_printf(dev, "MAC rev. 0x%08x\n", hwrev & 0x00700000); 1238 while (hw_rev->rl_desc != NULL) { 1239 if (hw_rev->rl_rev == hwrev) { 1240 sc->rl_type = hw_rev->rl_type; 1241 sc->rl_hwrev = hw_rev->rl_rev; 1242 break; 1243 } 1244 hw_rev++; 1245 } 1246 if (hw_rev->rl_desc == NULL) { 1247 device_printf(dev, "Unknown H/W revision: 0x%08x\n", hwrev); 1248 error = ENXIO; 1249 goto fail; 1250 } 1251 1252 switch (hw_rev->rl_rev) { 1253 case RL_HWREV_8139CPLUS: 1254 sc->rl_flags |= RL_FLAG_NOJUMBO | RL_FLAG_FASTETHER | 1255 RL_FLAG_AUTOPAD; 1256 break; 1257 case RL_HWREV_8100E: 1258 case RL_HWREV_8101E: 1259 sc->rl_flags |= RL_FLAG_NOJUMBO | RL_FLAG_PHYWAKE | 1260 RL_FLAG_FASTETHER; 1261 break; 1262 case RL_HWREV_8102E: 1263 case RL_HWREV_8102EL: 1264 case RL_HWREV_8102EL_SPIN1: 1265 sc->rl_flags |= RL_FLAG_NOJUMBO | RL_FLAG_PHYWAKE | 1266 RL_FLAG_PAR | RL_FLAG_DESCV2 | RL_FLAG_MACSTAT | 1267 RL_FLAG_FASTETHER | RL_FLAG_CMDSTOP | RL_FLAG_AUTOPAD; 1268 break; 1269 case RL_HWREV_8168_SPIN1: 1270 case RL_HWREV_8168_SPIN2: 1271 sc->rl_flags |= RL_FLAG_WOLRXENB; 1272 /* FALLTHROUGH */ 1273 case RL_HWREV_8168_SPIN3: 1274 sc->rl_flags |= RL_FLAG_PHYWAKE | RL_FLAG_MACSTAT; 1275 break; 1276 case RL_HWREV_8168C_SPIN2: 1277 sc->rl_flags |= RL_FLAG_MACSLEEP; 1278 /* FALLTHROUGH */ 1279 case RL_HWREV_8168C: 1280 if ((hwrev & 0x00700000) == 0x00200000) 1281 sc->rl_flags |= RL_FLAG_MACSLEEP; 1282 /* FALLTHROUGH */ 1283 case RL_HWREV_8168CP: 1284 case RL_HWREV_8168D: 1285 sc->rl_flags |= RL_FLAG_PHYWAKE | RL_FLAG_PAR | 1286 RL_FLAG_DESCV2 | RL_FLAG_MACSTAT | RL_FLAG_CMDSTOP | 1287 RL_FLAG_AUTOPAD; 1288 /* 1289 * These controllers support jumbo frame but it seems 1290 * that enabling it requires touching additional magic 1291 * registers. Depending on MAC revisions some 1292 * controllers need to disable checksum offload. So 1293 * disable jumbo frame until I have better idea what 1294 * it really requires to make it support. 1295 * RTL8168C/CP : supports up to 6KB jumbo frame. 1296 * RTL8111C/CP : supports up to 9KB jumbo frame. 1297 */ 1298 sc->rl_flags |= RL_FLAG_NOJUMBO; 1299 break; 1300 case RL_HWREV_8169_8110SB: 1301 case RL_HWREV_8169_8110SBL: 1302 case RL_HWREV_8169_8110SC: 1303 case RL_HWREV_8169_8110SCE: 1304 sc->rl_flags |= RL_FLAG_PHYWAKE; 1305 /* FALLTHROUGH */ 1306 case RL_HWREV_8169: 1307 case RL_HWREV_8169S: 1308 case RL_HWREV_8110S: 1309 sc->rl_flags |= RL_FLAG_MACRESET; 1310 break; 1311 default: 1312 break; 1313 } 1314 1315 /* Enable PME. */ 1316 CSR_WRITE_1(sc, RL_EECMD, RL_EE_MODE); 1317 cfg = CSR_READ_1(sc, RL_CFG1); 1318 cfg |= RL_CFG1_PME; 1319 CSR_WRITE_1(sc, RL_CFG1, cfg); 1320 cfg = CSR_READ_1(sc, RL_CFG5); 1321 cfg &= RL_CFG5_PME_STS; 1322 CSR_WRITE_1(sc, RL_CFG5, cfg); 1323 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 1324 1325 if ((sc->rl_flags & RL_FLAG_PAR) != 0) { 1326 /* 1327 * XXX Should have a better way to extract station 1328 * address from EEPROM. 1329 */ 1330 for (i = 0; i < ETHER_ADDR_LEN; i++) 1331 eaddr[i] = CSR_READ_1(sc, RL_IDR0 + i); 1332 } else { 1333 sc->rl_eewidth = RL_9356_ADDR_LEN; 1334 re_read_eeprom(sc, (caddr_t)&re_did, 0, 1); 1335 if (re_did != 0x8129) 1336 sc->rl_eewidth = RL_9346_ADDR_LEN; 1337 1338 /* 1339 * Get station address from the EEPROM. 1340 */ 1341 re_read_eeprom(sc, (caddr_t)as, RL_EE_EADDR, 3); 1342 for (i = 0; i < ETHER_ADDR_LEN / 2; i++) 1343 as[i] = le16toh(as[i]); 1344 bcopy(as, eaddr, sizeof(eaddr)); 1345 } 1346 1347 if (sc->rl_type == RL_8169) { 1348 /* Set RX length mask and number of descriptors. */ 1349 sc->rl_rxlenmask = RL_RDESC_STAT_GFRAGLEN; 1350 sc->rl_txstart = RL_GTXSTART; 1351 sc->rl_ldata.rl_tx_desc_cnt = RL_8169_TX_DESC_CNT; 1352 sc->rl_ldata.rl_rx_desc_cnt = RL_8169_RX_DESC_CNT; 1353 } else { 1354 /* Set RX length mask and number of descriptors. */ 1355 sc->rl_rxlenmask = RL_RDESC_STAT_FRAGLEN; 1356 sc->rl_txstart = RL_TXSTART; 1357 sc->rl_ldata.rl_tx_desc_cnt = RL_8139_TX_DESC_CNT; 1358 sc->rl_ldata.rl_rx_desc_cnt = RL_8139_RX_DESC_CNT; 1359 } 1360 1361 error = re_allocmem(dev, sc); 1362 if (error) 1363 goto fail; 1364 1365 ifp = sc->rl_ifp = if_alloc(IFT_ETHER); 1366 if (ifp == NULL) { 1367 device_printf(dev, "can not if_alloc()\n"); 1368 error = ENOSPC; 1369 goto fail; 1370 } 1371 1372 /* Take controller out of deep sleep mode. */ 1373 if ((sc->rl_flags & RL_FLAG_MACSLEEP) != 0) { 1374 if ((CSR_READ_1(sc, RL_MACDBG) & 0x80) == 0x80) 1375 CSR_WRITE_1(sc, RL_GPIO, 1376 CSR_READ_1(sc, RL_GPIO) | 0x01); 1377 else 1378 CSR_WRITE_1(sc, RL_GPIO, 1379 CSR_READ_1(sc, RL_GPIO) & ~0x01); 1380 } 1381 1382 /* Take PHY out of power down mode. */ 1383 if ((sc->rl_flags & RL_FLAG_PHYWAKE) != 0) { 1384 re_gmii_writereg(dev, 1, 0x1f, 0); 1385 re_gmii_writereg(dev, 1, 0x0e, 0); 1386 } 1387 1388 /* Do MII setup */ 1389 if (mii_phy_probe(dev, &sc->rl_miibus, 1390 re_ifmedia_upd, re_ifmedia_sts)) { 1391 device_printf(dev, "MII without any phy!\n"); 1392 error = ENXIO; 1393 goto fail; 1394 } 1395 1396 ifp->if_softc = sc; 1397 if_initname(ifp, device_get_name(dev), device_get_unit(dev)); 1398 ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; 1399 ifp->if_ioctl = re_ioctl; 1400 ifp->if_start = re_start; 1401 ifp->if_hwassist = RE_CSUM_FEATURES; 1402 ifp->if_capabilities = IFCAP_HWCSUM; 1403 ifp->if_capenable = ifp->if_capabilities; 1404 ifp->if_init = re_init; 1405 IFQ_SET_MAXLEN(&ifp->if_snd, RL_IFQ_MAXLEN); 1406 ifp->if_snd.ifq_drv_maxlen = RL_IFQ_MAXLEN; 1407 IFQ_SET_READY(&ifp->if_snd); 1408 1409 TASK_INIT(&sc->rl_txtask, 1, re_tx_task, ifp); 1410 TASK_INIT(&sc->rl_inttask, 0, re_int_task, sc); 1411 1412 /* 1413 * XXX 1414 * Still have no idea how to make TSO work on 8168C, 8168CP, 1415 * 8111C and 8111CP. 1416 */ 1417 if ((sc->rl_flags & RL_FLAG_DESCV2) == 0) { 1418 ifp->if_hwassist |= CSUM_TSO; 1419 ifp->if_capabilities |= IFCAP_TSO4; 1420 } 1421 1422 /* 1423 * Call MI attach routine. 1424 */ 1425 ether_ifattach(ifp, eaddr); 1426 1427 /* VLAN capability setup */ 1428 ifp->if_capabilities |= IFCAP_VLAN_MTU | IFCAP_VLAN_HWTAGGING; 1429 if (ifp->if_capabilities & IFCAP_HWCSUM) 1430 ifp->if_capabilities |= IFCAP_VLAN_HWCSUM; 1431 /* Enable WOL if PM is supported. */ 1432 if (pci_find_extcap(sc->rl_dev, PCIY_PMG, ®) == 0) 1433 ifp->if_capabilities |= IFCAP_WOL; 1434 ifp->if_capenable = ifp->if_capabilities; 1435 /* 1436 * Don't enable TSO by default. Under certain 1437 * circumtances the controller generated corrupted 1438 * packets in TSO size. 1439 */ 1440 ifp->if_hwassist &= ~CSUM_TSO; 1441 ifp->if_capenable &= ~IFCAP_TSO4; 1442 #ifdef DEVICE_POLLING 1443 ifp->if_capabilities |= IFCAP_POLLING; 1444 #endif 1445 /* 1446 * Tell the upper layer(s) we support long frames. 1447 * Must appear after the call to ether_ifattach() because 1448 * ether_ifattach() sets ifi_hdrlen to the default value. 1449 */ 1450 ifp->if_data.ifi_hdrlen = sizeof(struct ether_vlan_header); 1451 1452 #ifdef RE_DIAG 1453 /* 1454 * Perform hardware diagnostic on the original RTL8169. 1455 * Some 32-bit cards were incorrectly wired and would 1456 * malfunction if plugged into a 64-bit slot. 1457 */ 1458 1459 if (hwrev == RL_HWREV_8169) { 1460 error = re_diag(sc); 1461 if (error) { 1462 device_printf(dev, 1463 "attach aborted due to hardware diag failure\n"); 1464 ether_ifdetach(ifp); 1465 goto fail; 1466 } 1467 } 1468 #endif 1469 1470 /* Hook interrupt last to avoid having to lock softc */ 1471 if ((sc->rl_flags & RL_FLAG_MSI) == 0) 1472 error = bus_setup_intr(dev, sc->rl_irq[0], 1473 INTR_TYPE_NET | INTR_MPSAFE, re_intr, NULL, sc, 1474 &sc->rl_intrhand[0]); 1475 else { 1476 for (i = 0; i < RL_MSI_MESSAGES; i++) { 1477 error = bus_setup_intr(dev, sc->rl_irq[i], 1478 INTR_TYPE_NET | INTR_MPSAFE, re_intr, NULL, sc, 1479 &sc->rl_intrhand[i]); 1480 if (error != 0) 1481 break; 1482 } 1483 } 1484 if (error) { 1485 device_printf(dev, "couldn't set up irq\n"); 1486 ether_ifdetach(ifp); 1487 } 1488 1489 fail: 1490 1491 if (error) 1492 re_detach(dev); 1493 1494 return (error); 1495 } 1496 1497 /* 1498 * Shutdown hardware and free up resources. This can be called any 1499 * time after the mutex has been initialized. It is called in both 1500 * the error case in attach and the normal detach case so it needs 1501 * to be careful about only freeing resources that have actually been 1502 * allocated. 1503 */ 1504 static int 1505 re_detach(device_t dev) 1506 { 1507 struct rl_softc *sc; 1508 struct ifnet *ifp; 1509 int i, rid; 1510 1511 sc = device_get_softc(dev); 1512 ifp = sc->rl_ifp; 1513 KASSERT(mtx_initialized(&sc->rl_mtx), ("re mutex not initialized")); 1514 1515 /* These should only be active if attach succeeded */ 1516 if (device_is_attached(dev)) { 1517 #ifdef DEVICE_POLLING 1518 if (ifp->if_capenable & IFCAP_POLLING) 1519 ether_poll_deregister(ifp); 1520 #endif 1521 RL_LOCK(sc); 1522 #if 0 1523 sc->suspended = 1; 1524 #endif 1525 re_stop(sc); 1526 RL_UNLOCK(sc); 1527 callout_drain(&sc->rl_stat_callout); 1528 taskqueue_drain(taskqueue_fast, &sc->rl_inttask); 1529 taskqueue_drain(taskqueue_fast, &sc->rl_txtask); 1530 /* 1531 * Force off the IFF_UP flag here, in case someone 1532 * still had a BPF descriptor attached to this 1533 * interface. If they do, ether_ifdetach() will cause 1534 * the BPF code to try and clear the promisc mode 1535 * flag, which will bubble down to re_ioctl(), 1536 * which will try to call re_init() again. This will 1537 * turn the NIC back on and restart the MII ticker, 1538 * which will panic the system when the kernel tries 1539 * to invoke the re_tick() function that isn't there 1540 * anymore. 1541 */ 1542 ifp->if_flags &= ~IFF_UP; 1543 ether_ifdetach(ifp); 1544 } 1545 if (sc->rl_miibus) 1546 device_delete_child(dev, sc->rl_miibus); 1547 bus_generic_detach(dev); 1548 1549 /* 1550 * The rest is resource deallocation, so we should already be 1551 * stopped here. 1552 */ 1553 1554 for (i = 0; i < RL_MSI_MESSAGES; i++) { 1555 if (sc->rl_intrhand[i] != NULL) { 1556 bus_teardown_intr(dev, sc->rl_irq[i], 1557 sc->rl_intrhand[i]); 1558 sc->rl_intrhand[i] = NULL; 1559 } 1560 } 1561 if (ifp != NULL) 1562 if_free(ifp); 1563 if ((sc->rl_flags & RL_FLAG_MSI) == 0) { 1564 if (sc->rl_irq[0] != NULL) { 1565 bus_release_resource(dev, SYS_RES_IRQ, 0, 1566 sc->rl_irq[0]); 1567 sc->rl_irq[0] = NULL; 1568 } 1569 } else { 1570 for (i = 0, rid = 1; i < RL_MSI_MESSAGES; i++, rid++) { 1571 if (sc->rl_irq[i] != NULL) { 1572 bus_release_resource(dev, SYS_RES_IRQ, rid, 1573 sc->rl_irq[i]); 1574 sc->rl_irq[i] = NULL; 1575 } 1576 } 1577 pci_release_msi(dev); 1578 } 1579 if (sc->rl_res) 1580 bus_release_resource(dev, sc->rl_res_type, sc->rl_res_id, 1581 sc->rl_res); 1582 1583 /* Unload and free the RX DMA ring memory and map */ 1584 1585 if (sc->rl_ldata.rl_rx_list_tag) { 1586 bus_dmamap_unload(sc->rl_ldata.rl_rx_list_tag, 1587 sc->rl_ldata.rl_rx_list_map); 1588 bus_dmamem_free(sc->rl_ldata.rl_rx_list_tag, 1589 sc->rl_ldata.rl_rx_list, 1590 sc->rl_ldata.rl_rx_list_map); 1591 bus_dma_tag_destroy(sc->rl_ldata.rl_rx_list_tag); 1592 } 1593 1594 /* Unload and free the TX DMA ring memory and map */ 1595 1596 if (sc->rl_ldata.rl_tx_list_tag) { 1597 bus_dmamap_unload(sc->rl_ldata.rl_tx_list_tag, 1598 sc->rl_ldata.rl_tx_list_map); 1599 bus_dmamem_free(sc->rl_ldata.rl_tx_list_tag, 1600 sc->rl_ldata.rl_tx_list, 1601 sc->rl_ldata.rl_tx_list_map); 1602 bus_dma_tag_destroy(sc->rl_ldata.rl_tx_list_tag); 1603 } 1604 1605 /* Destroy all the RX and TX buffer maps */ 1606 1607 if (sc->rl_ldata.rl_tx_mtag) { 1608 for (i = 0; i < sc->rl_ldata.rl_tx_desc_cnt; i++) 1609 bus_dmamap_destroy(sc->rl_ldata.rl_tx_mtag, 1610 sc->rl_ldata.rl_tx_desc[i].tx_dmamap); 1611 bus_dma_tag_destroy(sc->rl_ldata.rl_tx_mtag); 1612 } 1613 if (sc->rl_ldata.rl_rx_mtag) { 1614 for (i = 0; i < sc->rl_ldata.rl_rx_desc_cnt; i++) 1615 bus_dmamap_destroy(sc->rl_ldata.rl_rx_mtag, 1616 sc->rl_ldata.rl_rx_desc[i].rx_dmamap); 1617 if (sc->rl_ldata.rl_rx_sparemap) 1618 bus_dmamap_destroy(sc->rl_ldata.rl_rx_mtag, 1619 sc->rl_ldata.rl_rx_sparemap); 1620 bus_dma_tag_destroy(sc->rl_ldata.rl_rx_mtag); 1621 } 1622 1623 /* Unload and free the stats buffer and map */ 1624 1625 if (sc->rl_ldata.rl_stag) { 1626 bus_dmamap_unload(sc->rl_ldata.rl_stag, 1627 sc->rl_ldata.rl_rx_list_map); 1628 bus_dmamem_free(sc->rl_ldata.rl_stag, 1629 sc->rl_ldata.rl_stats, 1630 sc->rl_ldata.rl_smap); 1631 bus_dma_tag_destroy(sc->rl_ldata.rl_stag); 1632 } 1633 1634 if (sc->rl_parent_tag) 1635 bus_dma_tag_destroy(sc->rl_parent_tag); 1636 1637 mtx_destroy(&sc->rl_mtx); 1638 1639 return (0); 1640 } 1641 1642 static __inline void 1643 re_discard_rxbuf(struct rl_softc *sc, int idx) 1644 { 1645 struct rl_desc *desc; 1646 struct rl_rxdesc *rxd; 1647 uint32_t cmdstat; 1648 1649 rxd = &sc->rl_ldata.rl_rx_desc[idx]; 1650 desc = &sc->rl_ldata.rl_rx_list[idx]; 1651 desc->rl_vlanctl = 0; 1652 cmdstat = rxd->rx_size; 1653 if (idx == sc->rl_ldata.rl_rx_desc_cnt - 1) 1654 cmdstat |= RL_RDESC_CMD_EOR; 1655 desc->rl_cmdstat = htole32(cmdstat | RL_RDESC_CMD_OWN); 1656 } 1657 1658 static int 1659 re_newbuf(struct rl_softc *sc, int idx) 1660 { 1661 struct mbuf *m; 1662 struct rl_rxdesc *rxd; 1663 bus_dma_segment_t segs[1]; 1664 bus_dmamap_t map; 1665 struct rl_desc *desc; 1666 uint32_t cmdstat; 1667 int error, nsegs; 1668 1669 m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR); 1670 if (m == NULL) 1671 return (ENOBUFS); 1672 1673 m->m_len = m->m_pkthdr.len = MCLBYTES; 1674 #ifdef RE_FIXUP_RX 1675 /* 1676 * This is part of an evil trick to deal with non-x86 platforms. 1677 * The RealTek chip requires RX buffers to be aligned on 64-bit 1678 * boundaries, but that will hose non-x86 machines. To get around 1679 * this, we leave some empty space at the start of each buffer 1680 * and for non-x86 hosts, we copy the buffer back six bytes 1681 * to achieve word alignment. This is slightly more efficient 1682 * than allocating a new buffer, copying the contents, and 1683 * discarding the old buffer. 1684 */ 1685 m_adj(m, RE_ETHER_ALIGN); 1686 #endif 1687 error = bus_dmamap_load_mbuf_sg(sc->rl_ldata.rl_rx_mtag, 1688 sc->rl_ldata.rl_rx_sparemap, m, segs, &nsegs, BUS_DMA_NOWAIT); 1689 if (error != 0) { 1690 m_freem(m); 1691 return (ENOBUFS); 1692 } 1693 KASSERT(nsegs == 1, ("%s: %d segment returned!", __func__, nsegs)); 1694 1695 rxd = &sc->rl_ldata.rl_rx_desc[idx]; 1696 if (rxd->rx_m != NULL) { 1697 bus_dmamap_sync(sc->rl_ldata.rl_rx_mtag, rxd->rx_dmamap, 1698 BUS_DMASYNC_POSTREAD); 1699 bus_dmamap_unload(sc->rl_ldata.rl_rx_mtag, rxd->rx_dmamap); 1700 } 1701 1702 rxd->rx_m = m; 1703 map = rxd->rx_dmamap; 1704 rxd->rx_dmamap = sc->rl_ldata.rl_rx_sparemap; 1705 rxd->rx_size = segs[0].ds_len; 1706 sc->rl_ldata.rl_rx_sparemap = map; 1707 bus_dmamap_sync(sc->rl_ldata.rl_rx_mtag, rxd->rx_dmamap, 1708 BUS_DMASYNC_PREREAD); 1709 1710 desc = &sc->rl_ldata.rl_rx_list[idx]; 1711 desc->rl_vlanctl = 0; 1712 desc->rl_bufaddr_lo = htole32(RL_ADDR_LO(segs[0].ds_addr)); 1713 desc->rl_bufaddr_hi = htole32(RL_ADDR_HI(segs[0].ds_addr)); 1714 cmdstat = segs[0].ds_len; 1715 if (idx == sc->rl_ldata.rl_rx_desc_cnt - 1) 1716 cmdstat |= RL_RDESC_CMD_EOR; 1717 desc->rl_cmdstat = htole32(cmdstat | RL_RDESC_CMD_OWN); 1718 1719 return (0); 1720 } 1721 1722 #ifdef RE_FIXUP_RX 1723 static __inline void 1724 re_fixup_rx(struct mbuf *m) 1725 { 1726 int i; 1727 uint16_t *src, *dst; 1728 1729 src = mtod(m, uint16_t *); 1730 dst = src - (RE_ETHER_ALIGN - ETHER_ALIGN) / sizeof *src; 1731 1732 for (i = 0; i < (m->m_len / sizeof(uint16_t) + 1); i++) 1733 *dst++ = *src++; 1734 1735 m->m_data -= RE_ETHER_ALIGN - ETHER_ALIGN; 1736 } 1737 #endif 1738 1739 static int 1740 re_tx_list_init(struct rl_softc *sc) 1741 { 1742 struct rl_desc *desc; 1743 int i; 1744 1745 RL_LOCK_ASSERT(sc); 1746 1747 bzero(sc->rl_ldata.rl_tx_list, 1748 sc->rl_ldata.rl_tx_desc_cnt * sizeof(struct rl_desc)); 1749 for (i = 0; i < sc->rl_ldata.rl_tx_desc_cnt; i++) 1750 sc->rl_ldata.rl_tx_desc[i].tx_m = NULL; 1751 /* Set EOR. */ 1752 desc = &sc->rl_ldata.rl_tx_list[sc->rl_ldata.rl_tx_desc_cnt - 1]; 1753 desc->rl_cmdstat |= htole32(RL_TDESC_CMD_EOR); 1754 1755 bus_dmamap_sync(sc->rl_ldata.rl_tx_list_tag, 1756 sc->rl_ldata.rl_tx_list_map, 1757 BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); 1758 1759 sc->rl_ldata.rl_tx_prodidx = 0; 1760 sc->rl_ldata.rl_tx_considx = 0; 1761 sc->rl_ldata.rl_tx_free = sc->rl_ldata.rl_tx_desc_cnt; 1762 1763 return (0); 1764 } 1765 1766 static int 1767 re_rx_list_init(struct rl_softc *sc) 1768 { 1769 int error, i; 1770 1771 bzero(sc->rl_ldata.rl_rx_list, 1772 sc->rl_ldata.rl_rx_desc_cnt * sizeof(struct rl_desc)); 1773 for (i = 0; i < sc->rl_ldata.rl_rx_desc_cnt; i++) { 1774 sc->rl_ldata.rl_rx_desc[i].rx_m = NULL; 1775 if ((error = re_newbuf(sc, i)) != 0) 1776 return (error); 1777 } 1778 1779 /* Flush the RX descriptors */ 1780 1781 bus_dmamap_sync(sc->rl_ldata.rl_rx_list_tag, 1782 sc->rl_ldata.rl_rx_list_map, 1783 BUS_DMASYNC_PREWRITE|BUS_DMASYNC_PREREAD); 1784 1785 sc->rl_ldata.rl_rx_prodidx = 0; 1786 sc->rl_head = sc->rl_tail = NULL; 1787 1788 return (0); 1789 } 1790 1791 /* 1792 * RX handler for C+ and 8169. For the gigE chips, we support 1793 * the reception of jumbo frames that have been fragmented 1794 * across multiple 2K mbuf cluster buffers. 1795 */ 1796 static int 1797 re_rxeof(struct rl_softc *sc, int *rx_npktsp) 1798 { 1799 struct mbuf *m; 1800 struct ifnet *ifp; 1801 int i, total_len; 1802 struct rl_desc *cur_rx; 1803 u_int32_t rxstat, rxvlan; 1804 int maxpkt = 16, rx_npkts = 0; 1805 1806 RL_LOCK_ASSERT(sc); 1807 1808 ifp = sc->rl_ifp; 1809 1810 /* Invalidate the descriptor memory */ 1811 1812 bus_dmamap_sync(sc->rl_ldata.rl_rx_list_tag, 1813 sc->rl_ldata.rl_rx_list_map, 1814 BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); 1815 1816 for (i = sc->rl_ldata.rl_rx_prodidx; maxpkt > 0; 1817 i = RL_RX_DESC_NXT(sc, i)) { 1818 cur_rx = &sc->rl_ldata.rl_rx_list[i]; 1819 rxstat = le32toh(cur_rx->rl_cmdstat); 1820 if ((rxstat & RL_RDESC_STAT_OWN) != 0) 1821 break; 1822 total_len = rxstat & sc->rl_rxlenmask; 1823 rxvlan = le32toh(cur_rx->rl_vlanctl); 1824 m = sc->rl_ldata.rl_rx_desc[i].rx_m; 1825 1826 if (!(rxstat & RL_RDESC_STAT_EOF)) { 1827 if (re_newbuf(sc, i) != 0) { 1828 /* 1829 * If this is part of a multi-fragment packet, 1830 * discard all the pieces. 1831 */ 1832 if (sc->rl_head != NULL) { 1833 m_freem(sc->rl_head); 1834 sc->rl_head = sc->rl_tail = NULL; 1835 } 1836 re_discard_rxbuf(sc, i); 1837 continue; 1838 } 1839 m->m_len = RE_RX_DESC_BUFLEN; 1840 if (sc->rl_head == NULL) 1841 sc->rl_head = sc->rl_tail = m; 1842 else { 1843 m->m_flags &= ~M_PKTHDR; 1844 sc->rl_tail->m_next = m; 1845 sc->rl_tail = m; 1846 } 1847 continue; 1848 } 1849 1850 /* 1851 * NOTE: for the 8139C+, the frame length field 1852 * is always 12 bits in size, but for the gigE chips, 1853 * it is 13 bits (since the max RX frame length is 16K). 1854 * Unfortunately, all 32 bits in the status word 1855 * were already used, so to make room for the extra 1856 * length bit, RealTek took out the 'frame alignment 1857 * error' bit and shifted the other status bits 1858 * over one slot. The OWN, EOR, FS and LS bits are 1859 * still in the same places. We have already extracted 1860 * the frame length and checked the OWN bit, so rather 1861 * than using an alternate bit mapping, we shift the 1862 * status bits one space to the right so we can evaluate 1863 * them using the 8169 status as though it was in the 1864 * same format as that of the 8139C+. 1865 */ 1866 if (sc->rl_type == RL_8169) 1867 rxstat >>= 1; 1868 1869 /* 1870 * if total_len > 2^13-1, both _RXERRSUM and _GIANT will be 1871 * set, but if CRC is clear, it will still be a valid frame. 1872 */ 1873 if (rxstat & RL_RDESC_STAT_RXERRSUM && !(total_len > 8191 && 1874 (rxstat & RL_RDESC_STAT_ERRS) == RL_RDESC_STAT_GIANT)) { 1875 ifp->if_ierrors++; 1876 /* 1877 * If this is part of a multi-fragment packet, 1878 * discard all the pieces. 1879 */ 1880 if (sc->rl_head != NULL) { 1881 m_freem(sc->rl_head); 1882 sc->rl_head = sc->rl_tail = NULL; 1883 } 1884 re_discard_rxbuf(sc, i); 1885 continue; 1886 } 1887 1888 /* 1889 * If allocating a replacement mbuf fails, 1890 * reload the current one. 1891 */ 1892 1893 if (re_newbuf(sc, i) != 0) { 1894 ifp->if_iqdrops++; 1895 if (sc->rl_head != NULL) { 1896 m_freem(sc->rl_head); 1897 sc->rl_head = sc->rl_tail = NULL; 1898 } 1899 re_discard_rxbuf(sc, i); 1900 continue; 1901 } 1902 1903 if (sc->rl_head != NULL) { 1904 m->m_len = total_len % RE_RX_DESC_BUFLEN; 1905 if (m->m_len == 0) 1906 m->m_len = RE_RX_DESC_BUFLEN; 1907 /* 1908 * Special case: if there's 4 bytes or less 1909 * in this buffer, the mbuf can be discarded: 1910 * the last 4 bytes is the CRC, which we don't 1911 * care about anyway. 1912 */ 1913 if (m->m_len <= ETHER_CRC_LEN) { 1914 sc->rl_tail->m_len -= 1915 (ETHER_CRC_LEN - m->m_len); 1916 m_freem(m); 1917 } else { 1918 m->m_len -= ETHER_CRC_LEN; 1919 m->m_flags &= ~M_PKTHDR; 1920 sc->rl_tail->m_next = m; 1921 } 1922 m = sc->rl_head; 1923 sc->rl_head = sc->rl_tail = NULL; 1924 m->m_pkthdr.len = total_len - ETHER_CRC_LEN; 1925 } else 1926 m->m_pkthdr.len = m->m_len = 1927 (total_len - ETHER_CRC_LEN); 1928 1929 #ifdef RE_FIXUP_RX 1930 re_fixup_rx(m); 1931 #endif 1932 ifp->if_ipackets++; 1933 m->m_pkthdr.rcvif = ifp; 1934 1935 /* Do RX checksumming if enabled */ 1936 1937 if (ifp->if_capenable & IFCAP_RXCSUM) { 1938 if ((sc->rl_flags & RL_FLAG_DESCV2) == 0) { 1939 /* Check IP header checksum */ 1940 if (rxstat & RL_RDESC_STAT_PROTOID) 1941 m->m_pkthdr.csum_flags |= 1942 CSUM_IP_CHECKED; 1943 if (!(rxstat & RL_RDESC_STAT_IPSUMBAD)) 1944 m->m_pkthdr.csum_flags |= 1945 CSUM_IP_VALID; 1946 1947 /* Check TCP/UDP checksum */ 1948 if ((RL_TCPPKT(rxstat) && 1949 !(rxstat & RL_RDESC_STAT_TCPSUMBAD)) || 1950 (RL_UDPPKT(rxstat) && 1951 !(rxstat & RL_RDESC_STAT_UDPSUMBAD))) { 1952 m->m_pkthdr.csum_flags |= 1953 CSUM_DATA_VALID|CSUM_PSEUDO_HDR; 1954 m->m_pkthdr.csum_data = 0xffff; 1955 } 1956 } else { 1957 /* 1958 * RTL8168C/RTL816CP/RTL8111C/RTL8111CP 1959 */ 1960 if ((rxstat & RL_RDESC_STAT_PROTOID) && 1961 (rxvlan & RL_RDESC_IPV4)) 1962 m->m_pkthdr.csum_flags |= 1963 CSUM_IP_CHECKED; 1964 if (!(rxstat & RL_RDESC_STAT_IPSUMBAD) && 1965 (rxvlan & RL_RDESC_IPV4)) 1966 m->m_pkthdr.csum_flags |= 1967 CSUM_IP_VALID; 1968 if (((rxstat & RL_RDESC_STAT_TCP) && 1969 !(rxstat & RL_RDESC_STAT_TCPSUMBAD)) || 1970 ((rxstat & RL_RDESC_STAT_UDP) && 1971 !(rxstat & RL_RDESC_STAT_UDPSUMBAD))) { 1972 m->m_pkthdr.csum_flags |= 1973 CSUM_DATA_VALID|CSUM_PSEUDO_HDR; 1974 m->m_pkthdr.csum_data = 0xffff; 1975 } 1976 } 1977 } 1978 maxpkt--; 1979 if (rxvlan & RL_RDESC_VLANCTL_TAG) { 1980 m->m_pkthdr.ether_vtag = 1981 bswap16((rxvlan & RL_RDESC_VLANCTL_DATA)); 1982 m->m_flags |= M_VLANTAG; 1983 } 1984 RL_UNLOCK(sc); 1985 (*ifp->if_input)(ifp, m); 1986 RL_LOCK(sc); 1987 rx_npkts++; 1988 } 1989 1990 /* Flush the RX DMA ring */ 1991 1992 bus_dmamap_sync(sc->rl_ldata.rl_rx_list_tag, 1993 sc->rl_ldata.rl_rx_list_map, 1994 BUS_DMASYNC_PREWRITE|BUS_DMASYNC_PREREAD); 1995 1996 sc->rl_ldata.rl_rx_prodidx = i; 1997 1998 if (rx_npktsp != NULL) 1999 *rx_npktsp = rx_npkts; 2000 if (maxpkt) 2001 return(EAGAIN); 2002 2003 return(0); 2004 } 2005 2006 static void 2007 re_txeof(struct rl_softc *sc) 2008 { 2009 struct ifnet *ifp; 2010 struct rl_txdesc *txd; 2011 u_int32_t txstat; 2012 int cons; 2013 2014 cons = sc->rl_ldata.rl_tx_considx; 2015 if (cons == sc->rl_ldata.rl_tx_prodidx) 2016 return; 2017 2018 ifp = sc->rl_ifp; 2019 /* Invalidate the TX descriptor list */ 2020 bus_dmamap_sync(sc->rl_ldata.rl_tx_list_tag, 2021 sc->rl_ldata.rl_tx_list_map, 2022 BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); 2023 2024 for (; cons != sc->rl_ldata.rl_tx_prodidx; 2025 cons = RL_TX_DESC_NXT(sc, cons)) { 2026 txstat = le32toh(sc->rl_ldata.rl_tx_list[cons].rl_cmdstat); 2027 if (txstat & RL_TDESC_STAT_OWN) 2028 break; 2029 /* 2030 * We only stash mbufs in the last descriptor 2031 * in a fragment chain, which also happens to 2032 * be the only place where the TX status bits 2033 * are valid. 2034 */ 2035 if (txstat & RL_TDESC_CMD_EOF) { 2036 txd = &sc->rl_ldata.rl_tx_desc[cons]; 2037 bus_dmamap_sync(sc->rl_ldata.rl_tx_mtag, 2038 txd->tx_dmamap, BUS_DMASYNC_POSTWRITE); 2039 bus_dmamap_unload(sc->rl_ldata.rl_tx_mtag, 2040 txd->tx_dmamap); 2041 KASSERT(txd->tx_m != NULL, 2042 ("%s: freeing NULL mbufs!", __func__)); 2043 m_freem(txd->tx_m); 2044 txd->tx_m = NULL; 2045 if (txstat & (RL_TDESC_STAT_EXCESSCOL| 2046 RL_TDESC_STAT_COLCNT)) 2047 ifp->if_collisions++; 2048 if (txstat & RL_TDESC_STAT_TXERRSUM) 2049 ifp->if_oerrors++; 2050 else 2051 ifp->if_opackets++; 2052 } 2053 sc->rl_ldata.rl_tx_free++; 2054 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; 2055 } 2056 sc->rl_ldata.rl_tx_considx = cons; 2057 2058 /* No changes made to the TX ring, so no flush needed */ 2059 2060 if (sc->rl_ldata.rl_tx_free != sc->rl_ldata.rl_tx_desc_cnt) { 2061 #ifdef RE_TX_MODERATION 2062 /* 2063 * If not all descriptors have been reaped yet, reload 2064 * the timer so that we will eventually get another 2065 * interrupt that will cause us to re-enter this routine. 2066 * This is done in case the transmitter has gone idle. 2067 */ 2068 CSR_WRITE_4(sc, RL_TIMERCNT, 1); 2069 #endif 2070 } else 2071 sc->rl_watchdog_timer = 0; 2072 } 2073 2074 static void 2075 re_tick(void *xsc) 2076 { 2077 struct rl_softc *sc; 2078 struct mii_data *mii; 2079 2080 sc = xsc; 2081 2082 RL_LOCK_ASSERT(sc); 2083 2084 mii = device_get_softc(sc->rl_miibus); 2085 mii_tick(mii); 2086 if ((sc->rl_flags & RL_FLAG_LINK) == 0) 2087 re_miibus_statchg(sc->rl_dev); 2088 /* 2089 * Reclaim transmitted frames here. Technically it is not 2090 * necessary to do here but it ensures periodic reclamation 2091 * regardless of Tx completion interrupt which seems to be 2092 * lost on PCIe based controllers under certain situations. 2093 */ 2094 re_txeof(sc); 2095 re_watchdog(sc); 2096 callout_reset(&sc->rl_stat_callout, hz, re_tick, sc); 2097 } 2098 2099 #ifdef DEVICE_POLLING 2100 static int 2101 re_poll(struct ifnet *ifp, enum poll_cmd cmd, int count) 2102 { 2103 struct rl_softc *sc = ifp->if_softc; 2104 int rx_npkts = 0; 2105 2106 RL_LOCK(sc); 2107 if (ifp->if_drv_flags & IFF_DRV_RUNNING) 2108 rx_npkts = re_poll_locked(ifp, cmd, count); 2109 RL_UNLOCK(sc); 2110 return (rx_npkts); 2111 } 2112 2113 static int 2114 re_poll_locked(struct ifnet *ifp, enum poll_cmd cmd, int count) 2115 { 2116 struct rl_softc *sc = ifp->if_softc; 2117 int rx_npkts; 2118 2119 RL_LOCK_ASSERT(sc); 2120 2121 sc->rxcycles = count; 2122 re_rxeof(sc, &rx_npkts); 2123 re_txeof(sc); 2124 2125 if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) 2126 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_txtask); 2127 2128 if (cmd == POLL_AND_CHECK_STATUS) { /* also check status register */ 2129 u_int16_t status; 2130 2131 status = CSR_READ_2(sc, RL_ISR); 2132 if (status == 0xffff) 2133 return (rx_npkts); 2134 if (status) 2135 CSR_WRITE_2(sc, RL_ISR, status); 2136 if ((status & (RL_ISR_TX_OK | RL_ISR_TX_DESC_UNAVAIL)) && 2137 (sc->rl_flags & RL_FLAG_PCIE)) 2138 CSR_WRITE_1(sc, sc->rl_txstart, RL_TXSTART_START); 2139 2140 /* 2141 * XXX check behaviour on receiver stalls. 2142 */ 2143 2144 if (status & RL_ISR_SYSTEM_ERR) 2145 re_init_locked(sc); 2146 } 2147 return (rx_npkts); 2148 } 2149 #endif /* DEVICE_POLLING */ 2150 2151 static int 2152 re_intr(void *arg) 2153 { 2154 struct rl_softc *sc; 2155 uint16_t status; 2156 2157 sc = arg; 2158 2159 status = CSR_READ_2(sc, RL_ISR); 2160 if (status == 0xFFFF || (status & RL_INTRS_CPLUS) == 0) 2161 return (FILTER_STRAY); 2162 CSR_WRITE_2(sc, RL_IMR, 0); 2163 2164 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_inttask); 2165 2166 return (FILTER_HANDLED); 2167 } 2168 2169 static void 2170 re_int_task(void *arg, int npending) 2171 { 2172 struct rl_softc *sc; 2173 struct ifnet *ifp; 2174 u_int16_t status; 2175 int rval = 0; 2176 2177 sc = arg; 2178 ifp = sc->rl_ifp; 2179 2180 RL_LOCK(sc); 2181 2182 status = CSR_READ_2(sc, RL_ISR); 2183 CSR_WRITE_2(sc, RL_ISR, status); 2184 2185 if (sc->suspended || 2186 (ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { 2187 RL_UNLOCK(sc); 2188 return; 2189 } 2190 2191 #ifdef DEVICE_POLLING 2192 if (ifp->if_capenable & IFCAP_POLLING) { 2193 RL_UNLOCK(sc); 2194 return; 2195 } 2196 #endif 2197 2198 if (status & (RL_ISR_RX_OK|RL_ISR_RX_ERR|RL_ISR_FIFO_OFLOW)) 2199 rval = re_rxeof(sc, NULL); 2200 2201 /* 2202 * Some chips will ignore a second TX request issued 2203 * while an existing transmission is in progress. If 2204 * the transmitter goes idle but there are still 2205 * packets waiting to be sent, we need to restart the 2206 * channel here to flush them out. This only seems to 2207 * be required with the PCIe devices. 2208 */ 2209 if ((status & (RL_ISR_TX_OK | RL_ISR_TX_DESC_UNAVAIL)) && 2210 (sc->rl_flags & RL_FLAG_PCIE)) 2211 CSR_WRITE_1(sc, sc->rl_txstart, RL_TXSTART_START); 2212 if (status & ( 2213 #ifdef RE_TX_MODERATION 2214 RL_ISR_TIMEOUT_EXPIRED| 2215 #else 2216 RL_ISR_TX_OK| 2217 #endif 2218 RL_ISR_TX_ERR|RL_ISR_TX_DESC_UNAVAIL)) 2219 re_txeof(sc); 2220 2221 if (status & RL_ISR_SYSTEM_ERR) 2222 re_init_locked(sc); 2223 2224 if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) 2225 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_txtask); 2226 2227 RL_UNLOCK(sc); 2228 2229 if ((CSR_READ_2(sc, RL_ISR) & RL_INTRS_CPLUS) || rval) { 2230 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_inttask); 2231 return; 2232 } 2233 2234 CSR_WRITE_2(sc, RL_IMR, RL_INTRS_CPLUS); 2235 } 2236 2237 static int 2238 re_encap(struct rl_softc *sc, struct mbuf **m_head) 2239 { 2240 struct rl_txdesc *txd, *txd_last; 2241 bus_dma_segment_t segs[RL_NTXSEGS]; 2242 bus_dmamap_t map; 2243 struct mbuf *m_new; 2244 struct rl_desc *desc; 2245 int nsegs, prod; 2246 int i, error, ei, si; 2247 int padlen; 2248 uint32_t cmdstat, csum_flags, vlanctl; 2249 2250 RL_LOCK_ASSERT(sc); 2251 M_ASSERTPKTHDR((*m_head)); 2252 2253 /* 2254 * With some of the RealTek chips, using the checksum offload 2255 * support in conjunction with the autopadding feature results 2256 * in the transmission of corrupt frames. For example, if we 2257 * need to send a really small IP fragment that's less than 60 2258 * bytes in size, and IP header checksumming is enabled, the 2259 * resulting ethernet frame that appears on the wire will 2260 * have garbled payload. To work around this, if TX IP checksum 2261 * offload is enabled, we always manually pad short frames out 2262 * to the minimum ethernet frame size. 2263 */ 2264 if ((sc->rl_flags & RL_FLAG_AUTOPAD) == 0 && 2265 (*m_head)->m_pkthdr.len < RL_IP4CSUMTX_PADLEN && 2266 ((*m_head)->m_pkthdr.csum_flags & CSUM_IP) != 0) { 2267 padlen = RL_MIN_FRAMELEN - (*m_head)->m_pkthdr.len; 2268 if (M_WRITABLE(*m_head) == 0) { 2269 /* Get a writable copy. */ 2270 m_new = m_dup(*m_head, M_DONTWAIT); 2271 m_freem(*m_head); 2272 if (m_new == NULL) { 2273 *m_head = NULL; 2274 return (ENOBUFS); 2275 } 2276 *m_head = m_new; 2277 } 2278 if ((*m_head)->m_next != NULL || 2279 M_TRAILINGSPACE(*m_head) < padlen) { 2280 m_new = m_defrag(*m_head, M_DONTWAIT); 2281 if (m_new == NULL) { 2282 m_freem(*m_head); 2283 *m_head = NULL; 2284 return (ENOBUFS); 2285 } 2286 } else 2287 m_new = *m_head; 2288 2289 /* 2290 * Manually pad short frames, and zero the pad space 2291 * to avoid leaking data. 2292 */ 2293 bzero(mtod(m_new, char *) + m_new->m_pkthdr.len, padlen); 2294 m_new->m_pkthdr.len += padlen; 2295 m_new->m_len = m_new->m_pkthdr.len; 2296 *m_head = m_new; 2297 } 2298 2299 prod = sc->rl_ldata.rl_tx_prodidx; 2300 txd = &sc->rl_ldata.rl_tx_desc[prod]; 2301 error = bus_dmamap_load_mbuf_sg(sc->rl_ldata.rl_tx_mtag, txd->tx_dmamap, 2302 *m_head, segs, &nsegs, BUS_DMA_NOWAIT); 2303 if (error == EFBIG) { 2304 m_new = m_collapse(*m_head, M_DONTWAIT, RL_NTXSEGS); 2305 if (m_new == NULL) { 2306 m_freem(*m_head); 2307 *m_head = NULL; 2308 return (ENOBUFS); 2309 } 2310 *m_head = m_new; 2311 error = bus_dmamap_load_mbuf_sg(sc->rl_ldata.rl_tx_mtag, 2312 txd->tx_dmamap, *m_head, segs, &nsegs, BUS_DMA_NOWAIT); 2313 if (error != 0) { 2314 m_freem(*m_head); 2315 *m_head = NULL; 2316 return (error); 2317 } 2318 } else if (error != 0) 2319 return (error); 2320 if (nsegs == 0) { 2321 m_freem(*m_head); 2322 *m_head = NULL; 2323 return (EIO); 2324 } 2325 2326 /* Check for number of available descriptors. */ 2327 if (sc->rl_ldata.rl_tx_free - nsegs <= 1) { 2328 bus_dmamap_unload(sc->rl_ldata.rl_tx_mtag, txd->tx_dmamap); 2329 return (ENOBUFS); 2330 } 2331 2332 bus_dmamap_sync(sc->rl_ldata.rl_tx_mtag, txd->tx_dmamap, 2333 BUS_DMASYNC_PREWRITE); 2334 2335 /* 2336 * Set up checksum offload. Note: checksum offload bits must 2337 * appear in all descriptors of a multi-descriptor transmit 2338 * attempt. This is according to testing done with an 8169 2339 * chip. This is a requirement. 2340 */ 2341 vlanctl = 0; 2342 csum_flags = 0; 2343 if (((*m_head)->m_pkthdr.csum_flags & CSUM_TSO) != 0) 2344 csum_flags = RL_TDESC_CMD_LGSEND | 2345 ((uint32_t)(*m_head)->m_pkthdr.tso_segsz << 2346 RL_TDESC_CMD_MSSVAL_SHIFT); 2347 else { 2348 /* 2349 * Unconditionally enable IP checksum if TCP or UDP 2350 * checksum is required. Otherwise, TCP/UDP checksum 2351 * does't make effects. 2352 */ 2353 if (((*m_head)->m_pkthdr.csum_flags & RE_CSUM_FEATURES) != 0) { 2354 if ((sc->rl_flags & RL_FLAG_DESCV2) == 0) { 2355 csum_flags |= RL_TDESC_CMD_IPCSUM; 2356 if (((*m_head)->m_pkthdr.csum_flags & 2357 CSUM_TCP) != 0) 2358 csum_flags |= RL_TDESC_CMD_TCPCSUM; 2359 if (((*m_head)->m_pkthdr.csum_flags & 2360 CSUM_UDP) != 0) 2361 csum_flags |= RL_TDESC_CMD_UDPCSUM; 2362 } else { 2363 vlanctl |= RL_TDESC_CMD_IPCSUMV2; 2364 if (((*m_head)->m_pkthdr.csum_flags & 2365 CSUM_TCP) != 0) 2366 vlanctl |= RL_TDESC_CMD_TCPCSUMV2; 2367 if (((*m_head)->m_pkthdr.csum_flags & 2368 CSUM_UDP) != 0) 2369 vlanctl |= RL_TDESC_CMD_UDPCSUMV2; 2370 } 2371 } 2372 } 2373 2374 /* 2375 * Set up hardware VLAN tagging. Note: vlan tag info must 2376 * appear in all descriptors of a multi-descriptor 2377 * transmission attempt. 2378 */ 2379 if ((*m_head)->m_flags & M_VLANTAG) 2380 vlanctl |= bswap16((*m_head)->m_pkthdr.ether_vtag) | 2381 RL_TDESC_VLANCTL_TAG; 2382 2383 si = prod; 2384 for (i = 0; i < nsegs; i++, prod = RL_TX_DESC_NXT(sc, prod)) { 2385 desc = &sc->rl_ldata.rl_tx_list[prod]; 2386 desc->rl_vlanctl = htole32(vlanctl); 2387 desc->rl_bufaddr_lo = htole32(RL_ADDR_LO(segs[i].ds_addr)); 2388 desc->rl_bufaddr_hi = htole32(RL_ADDR_HI(segs[i].ds_addr)); 2389 cmdstat = segs[i].ds_len; 2390 if (i != 0) 2391 cmdstat |= RL_TDESC_CMD_OWN; 2392 if (prod == sc->rl_ldata.rl_tx_desc_cnt - 1) 2393 cmdstat |= RL_TDESC_CMD_EOR; 2394 desc->rl_cmdstat = htole32(cmdstat | csum_flags); 2395 sc->rl_ldata.rl_tx_free--; 2396 } 2397 /* Update producer index. */ 2398 sc->rl_ldata.rl_tx_prodidx = prod; 2399 2400 /* Set EOF on the last descriptor. */ 2401 ei = RL_TX_DESC_PRV(sc, prod); 2402 desc = &sc->rl_ldata.rl_tx_list[ei]; 2403 desc->rl_cmdstat |= htole32(RL_TDESC_CMD_EOF); 2404 2405 desc = &sc->rl_ldata.rl_tx_list[si]; 2406 /* Set SOF and transfer ownership of packet to the chip. */ 2407 desc->rl_cmdstat |= htole32(RL_TDESC_CMD_OWN | RL_TDESC_CMD_SOF); 2408 2409 /* 2410 * Insure that the map for this transmission 2411 * is placed at the array index of the last descriptor 2412 * in this chain. (Swap last and first dmamaps.) 2413 */ 2414 txd_last = &sc->rl_ldata.rl_tx_desc[ei]; 2415 map = txd->tx_dmamap; 2416 txd->tx_dmamap = txd_last->tx_dmamap; 2417 txd_last->tx_dmamap = map; 2418 txd_last->tx_m = *m_head; 2419 2420 return (0); 2421 } 2422 2423 static void 2424 re_tx_task(void *arg, int npending) 2425 { 2426 struct ifnet *ifp; 2427 2428 ifp = arg; 2429 re_start(ifp); 2430 } 2431 2432 /* 2433 * Main transmit routine for C+ and gigE NICs. 2434 */ 2435 static void 2436 re_start(struct ifnet *ifp) 2437 { 2438 struct rl_softc *sc; 2439 struct mbuf *m_head; 2440 int queued; 2441 2442 sc = ifp->if_softc; 2443 2444 RL_LOCK(sc); 2445 2446 if ((ifp->if_drv_flags & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) != 2447 IFF_DRV_RUNNING || (sc->rl_flags & RL_FLAG_LINK) == 0) { 2448 RL_UNLOCK(sc); 2449 return; 2450 } 2451 2452 for (queued = 0; !IFQ_DRV_IS_EMPTY(&ifp->if_snd) && 2453 sc->rl_ldata.rl_tx_free > 1;) { 2454 IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head); 2455 if (m_head == NULL) 2456 break; 2457 2458 if (re_encap(sc, &m_head) != 0) { 2459 if (m_head == NULL) 2460 break; 2461 IFQ_DRV_PREPEND(&ifp->if_snd, m_head); 2462 ifp->if_drv_flags |= IFF_DRV_OACTIVE; 2463 break; 2464 } 2465 2466 /* 2467 * If there's a BPF listener, bounce a copy of this frame 2468 * to him. 2469 */ 2470 ETHER_BPF_MTAP(ifp, m_head); 2471 2472 queued++; 2473 } 2474 2475 if (queued == 0) { 2476 #ifdef RE_TX_MODERATION 2477 if (sc->rl_ldata.rl_tx_free != sc->rl_ldata.rl_tx_desc_cnt) 2478 CSR_WRITE_4(sc, RL_TIMERCNT, 1); 2479 #endif 2480 RL_UNLOCK(sc); 2481 return; 2482 } 2483 2484 /* Flush the TX descriptors */ 2485 2486 bus_dmamap_sync(sc->rl_ldata.rl_tx_list_tag, 2487 sc->rl_ldata.rl_tx_list_map, 2488 BUS_DMASYNC_PREWRITE|BUS_DMASYNC_PREREAD); 2489 2490 CSR_WRITE_1(sc, sc->rl_txstart, RL_TXSTART_START); 2491 2492 #ifdef RE_TX_MODERATION 2493 /* 2494 * Use the countdown timer for interrupt moderation. 2495 * 'TX done' interrupts are disabled. Instead, we reset the 2496 * countdown timer, which will begin counting until it hits 2497 * the value in the TIMERINT register, and then trigger an 2498 * interrupt. Each time we write to the TIMERCNT register, 2499 * the timer count is reset to 0. 2500 */ 2501 CSR_WRITE_4(sc, RL_TIMERCNT, 1); 2502 #endif 2503 2504 /* 2505 * Set a timeout in case the chip goes out to lunch. 2506 */ 2507 sc->rl_watchdog_timer = 5; 2508 2509 RL_UNLOCK(sc); 2510 } 2511 2512 static void 2513 re_init(void *xsc) 2514 { 2515 struct rl_softc *sc = xsc; 2516 2517 RL_LOCK(sc); 2518 re_init_locked(sc); 2519 RL_UNLOCK(sc); 2520 } 2521 2522 static void 2523 re_init_locked(struct rl_softc *sc) 2524 { 2525 struct ifnet *ifp = sc->rl_ifp; 2526 struct mii_data *mii; 2527 uint32_t reg; 2528 uint16_t cfg; 2529 union { 2530 uint32_t align_dummy; 2531 u_char eaddr[ETHER_ADDR_LEN]; 2532 } eaddr; 2533 2534 RL_LOCK_ASSERT(sc); 2535 2536 mii = device_get_softc(sc->rl_miibus); 2537 2538 /* 2539 * Cancel pending I/O and free all RX/TX buffers. 2540 */ 2541 re_stop(sc); 2542 2543 /* Put controller into known state. */ 2544 re_reset(sc); 2545 2546 /* 2547 * Enable C+ RX and TX mode, as well as VLAN stripping and 2548 * RX checksum offload. We must configure the C+ register 2549 * before all others. 2550 */ 2551 cfg = RL_CPLUSCMD_PCI_MRW; 2552 if ((ifp->if_capenable & IFCAP_RXCSUM) != 0) 2553 cfg |= RL_CPLUSCMD_RXCSUM_ENB; 2554 if ((ifp->if_capenable & IFCAP_VLAN_HWTAGGING) != 0) 2555 cfg |= RL_CPLUSCMD_VLANSTRIP; 2556 if ((sc->rl_flags & RL_FLAG_MACSTAT) != 0) { 2557 cfg |= RL_CPLUSCMD_MACSTAT_DIS; 2558 /* XXX magic. */ 2559 cfg |= 0x0001; 2560 } else 2561 cfg |= RL_CPLUSCMD_RXENB | RL_CPLUSCMD_TXENB; 2562 CSR_WRITE_2(sc, RL_CPLUS_CMD, cfg); 2563 if (sc->rl_hwrev == RL_HWREV_8169_8110SC || 2564 sc->rl_hwrev == RL_HWREV_8169_8110SCE) { 2565 reg = 0x000fff00; 2566 if ((CSR_READ_1(sc, RL_CFG2) & RL_CFG2_PCI66MHZ) != 0) 2567 reg |= 0x000000ff; 2568 if (sc->rl_hwrev == RL_HWREV_8169_8110SCE) 2569 reg |= 0x00f00000; 2570 CSR_WRITE_4(sc, 0x7c, reg); 2571 /* Disable interrupt mitigation. */ 2572 CSR_WRITE_2(sc, 0xe2, 0); 2573 } 2574 /* 2575 * Disable TSO if interface MTU size is greater than MSS 2576 * allowed in controller. 2577 */ 2578 if (ifp->if_mtu > RL_TSO_MTU && (ifp->if_capenable & IFCAP_TSO4) != 0) { 2579 ifp->if_capenable &= ~IFCAP_TSO4; 2580 ifp->if_hwassist &= ~CSUM_TSO; 2581 } 2582 2583 /* 2584 * Init our MAC address. Even though the chipset 2585 * documentation doesn't mention it, we need to enter "Config 2586 * register write enable" mode to modify the ID registers. 2587 */ 2588 /* Copy MAC address on stack to align. */ 2589 bcopy(IF_LLADDR(ifp), eaddr.eaddr, ETHER_ADDR_LEN); 2590 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_WRITECFG); 2591 CSR_WRITE_4(sc, RL_IDR0, 2592 htole32(*(u_int32_t *)(&eaddr.eaddr[0]))); 2593 CSR_WRITE_4(sc, RL_IDR4, 2594 htole32(*(u_int32_t *)(&eaddr.eaddr[4]))); 2595 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 2596 2597 /* 2598 * For C+ mode, initialize the RX descriptors and mbufs. 2599 */ 2600 re_rx_list_init(sc); 2601 re_tx_list_init(sc); 2602 2603 /* 2604 * Load the addresses of the RX and TX lists into the chip. 2605 */ 2606 2607 CSR_WRITE_4(sc, RL_RXLIST_ADDR_HI, 2608 RL_ADDR_HI(sc->rl_ldata.rl_rx_list_addr)); 2609 CSR_WRITE_4(sc, RL_RXLIST_ADDR_LO, 2610 RL_ADDR_LO(sc->rl_ldata.rl_rx_list_addr)); 2611 2612 CSR_WRITE_4(sc, RL_TXLIST_ADDR_HI, 2613 RL_ADDR_HI(sc->rl_ldata.rl_tx_list_addr)); 2614 CSR_WRITE_4(sc, RL_TXLIST_ADDR_LO, 2615 RL_ADDR_LO(sc->rl_ldata.rl_tx_list_addr)); 2616 2617 /* 2618 * Enable transmit and receive. 2619 */ 2620 CSR_WRITE_1(sc, RL_COMMAND, RL_CMD_TX_ENB|RL_CMD_RX_ENB); 2621 2622 /* 2623 * Set the initial TX configuration. 2624 */ 2625 if (sc->rl_testmode) { 2626 if (sc->rl_type == RL_8169) 2627 CSR_WRITE_4(sc, RL_TXCFG, 2628 RL_TXCFG_CONFIG|RL_LOOPTEST_ON); 2629 else 2630 CSR_WRITE_4(sc, RL_TXCFG, 2631 RL_TXCFG_CONFIG|RL_LOOPTEST_ON_CPLUS); 2632 } else 2633 CSR_WRITE_4(sc, RL_TXCFG, RL_TXCFG_CONFIG); 2634 2635 CSR_WRITE_1(sc, RL_EARLY_TX_THRESH, 16); 2636 2637 /* 2638 * Set the initial RX configuration. 2639 */ 2640 re_set_rxmode(sc); 2641 2642 #ifdef DEVICE_POLLING 2643 /* 2644 * Disable interrupts if we are polling. 2645 */ 2646 if (ifp->if_capenable & IFCAP_POLLING) 2647 CSR_WRITE_2(sc, RL_IMR, 0); 2648 else /* otherwise ... */ 2649 #endif 2650 2651 /* 2652 * Enable interrupts. 2653 */ 2654 if (sc->rl_testmode) 2655 CSR_WRITE_2(sc, RL_IMR, 0); 2656 else 2657 CSR_WRITE_2(sc, RL_IMR, RL_INTRS_CPLUS); 2658 CSR_WRITE_2(sc, RL_ISR, RL_INTRS_CPLUS); 2659 2660 /* Set initial TX threshold */ 2661 sc->rl_txthresh = RL_TX_THRESH_INIT; 2662 2663 /* Start RX/TX process. */ 2664 CSR_WRITE_4(sc, RL_MISSEDPKT, 0); 2665 #ifdef notdef 2666 /* Enable receiver and transmitter. */ 2667 CSR_WRITE_1(sc, RL_COMMAND, RL_CMD_TX_ENB|RL_CMD_RX_ENB); 2668 #endif 2669 2670 #ifdef RE_TX_MODERATION 2671 /* 2672 * Initialize the timer interrupt register so that 2673 * a timer interrupt will be generated once the timer 2674 * reaches a certain number of ticks. The timer is 2675 * reloaded on each transmit. This gives us TX interrupt 2676 * moderation, which dramatically improves TX frame rate. 2677 */ 2678 if (sc->rl_type == RL_8169) 2679 CSR_WRITE_4(sc, RL_TIMERINT_8169, 0x800); 2680 else 2681 CSR_WRITE_4(sc, RL_TIMERINT, 0x400); 2682 #endif 2683 2684 /* 2685 * For 8169 gigE NICs, set the max allowed RX packet 2686 * size so we can receive jumbo frames. 2687 */ 2688 if (sc->rl_type == RL_8169) 2689 CSR_WRITE_2(sc, RL_MAXRXPKTLEN, 16383); 2690 2691 if (sc->rl_testmode) 2692 return; 2693 2694 mii_mediachg(mii); 2695 2696 CSR_WRITE_1(sc, RL_CFG1, CSR_READ_1(sc, RL_CFG1) | RL_CFG1_DRVLOAD); 2697 2698 ifp->if_drv_flags |= IFF_DRV_RUNNING; 2699 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; 2700 2701 sc->rl_flags &= ~RL_FLAG_LINK; 2702 sc->rl_watchdog_timer = 0; 2703 callout_reset(&sc->rl_stat_callout, hz, re_tick, sc); 2704 } 2705 2706 /* 2707 * Set media options. 2708 */ 2709 static int 2710 re_ifmedia_upd(struct ifnet *ifp) 2711 { 2712 struct rl_softc *sc; 2713 struct mii_data *mii; 2714 int error; 2715 2716 sc = ifp->if_softc; 2717 mii = device_get_softc(sc->rl_miibus); 2718 RL_LOCK(sc); 2719 error = mii_mediachg(mii); 2720 RL_UNLOCK(sc); 2721 2722 return (error); 2723 } 2724 2725 /* 2726 * Report current media status. 2727 */ 2728 static void 2729 re_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr) 2730 { 2731 struct rl_softc *sc; 2732 struct mii_data *mii; 2733 2734 sc = ifp->if_softc; 2735 mii = device_get_softc(sc->rl_miibus); 2736 2737 RL_LOCK(sc); 2738 mii_pollstat(mii); 2739 RL_UNLOCK(sc); 2740 ifmr->ifm_active = mii->mii_media_active; 2741 ifmr->ifm_status = mii->mii_media_status; 2742 } 2743 2744 static int 2745 re_ioctl(struct ifnet *ifp, u_long command, caddr_t data) 2746 { 2747 struct rl_softc *sc = ifp->if_softc; 2748 struct ifreq *ifr = (struct ifreq *) data; 2749 struct mii_data *mii; 2750 int error = 0; 2751 2752 switch (command) { 2753 case SIOCSIFMTU: 2754 if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > RL_JUMBO_MTU) { 2755 error = EINVAL; 2756 break; 2757 } 2758 if ((sc->rl_flags & RL_FLAG_NOJUMBO) != 0 && 2759 ifr->ifr_mtu > RL_MAX_FRAMELEN) { 2760 error = EINVAL; 2761 break; 2762 } 2763 RL_LOCK(sc); 2764 if (ifp->if_mtu != ifr->ifr_mtu) 2765 ifp->if_mtu = ifr->ifr_mtu; 2766 if (ifp->if_mtu > RL_TSO_MTU && 2767 (ifp->if_capenable & IFCAP_TSO4) != 0) { 2768 ifp->if_capenable &= ~IFCAP_TSO4; 2769 ifp->if_hwassist &= ~CSUM_TSO; 2770 } 2771 RL_UNLOCK(sc); 2772 break; 2773 case SIOCSIFFLAGS: 2774 RL_LOCK(sc); 2775 if ((ifp->if_flags & IFF_UP) != 0) { 2776 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) != 0) { 2777 if (((ifp->if_flags ^ sc->rl_if_flags) 2778 & (IFF_PROMISC | IFF_ALLMULTI)) != 0) 2779 re_set_rxmode(sc); 2780 } else 2781 re_init_locked(sc); 2782 } else { 2783 if ((ifp->if_drv_flags & IFF_DRV_RUNNING) != 0) 2784 re_stop(sc); 2785 } 2786 sc->rl_if_flags = ifp->if_flags; 2787 RL_UNLOCK(sc); 2788 break; 2789 case SIOCADDMULTI: 2790 case SIOCDELMULTI: 2791 RL_LOCK(sc); 2792 re_set_rxmode(sc); 2793 RL_UNLOCK(sc); 2794 break; 2795 case SIOCGIFMEDIA: 2796 case SIOCSIFMEDIA: 2797 mii = device_get_softc(sc->rl_miibus); 2798 error = ifmedia_ioctl(ifp, ifr, &mii->mii_media, command); 2799 break; 2800 case SIOCSIFCAP: 2801 { 2802 int mask, reinit; 2803 2804 mask = ifr->ifr_reqcap ^ ifp->if_capenable; 2805 reinit = 0; 2806 #ifdef DEVICE_POLLING 2807 if (mask & IFCAP_POLLING) { 2808 if (ifr->ifr_reqcap & IFCAP_POLLING) { 2809 error = ether_poll_register(re_poll, ifp); 2810 if (error) 2811 return(error); 2812 RL_LOCK(sc); 2813 /* Disable interrupts */ 2814 CSR_WRITE_2(sc, RL_IMR, 0x0000); 2815 ifp->if_capenable |= IFCAP_POLLING; 2816 RL_UNLOCK(sc); 2817 } else { 2818 error = ether_poll_deregister(ifp); 2819 /* Enable interrupts. */ 2820 RL_LOCK(sc); 2821 CSR_WRITE_2(sc, RL_IMR, RL_INTRS_CPLUS); 2822 ifp->if_capenable &= ~IFCAP_POLLING; 2823 RL_UNLOCK(sc); 2824 } 2825 } 2826 #endif /* DEVICE_POLLING */ 2827 if (mask & IFCAP_HWCSUM) { 2828 ifp->if_capenable ^= IFCAP_HWCSUM; 2829 if (ifp->if_capenable & IFCAP_TXCSUM) 2830 ifp->if_hwassist |= RE_CSUM_FEATURES; 2831 else 2832 ifp->if_hwassist &= ~RE_CSUM_FEATURES; 2833 reinit = 1; 2834 } 2835 if (mask & IFCAP_VLAN_HWTAGGING) { 2836 ifp->if_capenable ^= IFCAP_VLAN_HWTAGGING; 2837 reinit = 1; 2838 } 2839 if (mask & IFCAP_TSO4) { 2840 ifp->if_capenable ^= IFCAP_TSO4; 2841 if ((IFCAP_TSO4 & ifp->if_capenable) && 2842 (IFCAP_TSO4 & ifp->if_capabilities)) 2843 ifp->if_hwassist |= CSUM_TSO; 2844 else 2845 ifp->if_hwassist &= ~CSUM_TSO; 2846 if (ifp->if_mtu > RL_TSO_MTU && 2847 (ifp->if_capenable & IFCAP_TSO4) != 0) { 2848 ifp->if_capenable &= ~IFCAP_TSO4; 2849 ifp->if_hwassist &= ~CSUM_TSO; 2850 } 2851 } 2852 if ((mask & IFCAP_WOL) != 0 && 2853 (ifp->if_capabilities & IFCAP_WOL) != 0) { 2854 if ((mask & IFCAP_WOL_UCAST) != 0) 2855 ifp->if_capenable ^= IFCAP_WOL_UCAST; 2856 if ((mask & IFCAP_WOL_MCAST) != 0) 2857 ifp->if_capenable ^= IFCAP_WOL_MCAST; 2858 if ((mask & IFCAP_WOL_MAGIC) != 0) 2859 ifp->if_capenable ^= IFCAP_WOL_MAGIC; 2860 } 2861 if (reinit && ifp->if_drv_flags & IFF_DRV_RUNNING) 2862 re_init(sc); 2863 VLAN_CAPABILITIES(ifp); 2864 } 2865 break; 2866 default: 2867 error = ether_ioctl(ifp, command, data); 2868 break; 2869 } 2870 2871 return (error); 2872 } 2873 2874 static void 2875 re_watchdog(struct rl_softc *sc) 2876 { 2877 struct ifnet *ifp; 2878 2879 RL_LOCK_ASSERT(sc); 2880 2881 if (sc->rl_watchdog_timer == 0 || --sc->rl_watchdog_timer != 0) 2882 return; 2883 2884 ifp = sc->rl_ifp; 2885 re_txeof(sc); 2886 if (sc->rl_ldata.rl_tx_free == sc->rl_ldata.rl_tx_desc_cnt) { 2887 if_printf(ifp, "watchdog timeout (missed Tx interrupts) " 2888 "-- recovering\n"); 2889 if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) 2890 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_txtask); 2891 return; 2892 } 2893 2894 if_printf(ifp, "watchdog timeout\n"); 2895 ifp->if_oerrors++; 2896 2897 re_rxeof(sc, NULL); 2898 re_init_locked(sc); 2899 if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) 2900 taskqueue_enqueue_fast(taskqueue_fast, &sc->rl_txtask); 2901 } 2902 2903 /* 2904 * Stop the adapter and free any mbufs allocated to the 2905 * RX and TX lists. 2906 */ 2907 static void 2908 re_stop(struct rl_softc *sc) 2909 { 2910 int i; 2911 struct ifnet *ifp; 2912 struct rl_txdesc *txd; 2913 struct rl_rxdesc *rxd; 2914 2915 RL_LOCK_ASSERT(sc); 2916 2917 ifp = sc->rl_ifp; 2918 2919 sc->rl_watchdog_timer = 0; 2920 callout_stop(&sc->rl_stat_callout); 2921 ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); 2922 2923 if ((sc->rl_flags & RL_FLAG_CMDSTOP) != 0) 2924 CSR_WRITE_1(sc, RL_COMMAND, RL_CMD_STOPREQ | RL_CMD_TX_ENB | 2925 RL_CMD_RX_ENB); 2926 else 2927 CSR_WRITE_1(sc, RL_COMMAND, 0x00); 2928 DELAY(1000); 2929 CSR_WRITE_2(sc, RL_IMR, 0x0000); 2930 CSR_WRITE_2(sc, RL_ISR, 0xFFFF); 2931 2932 if (sc->rl_head != NULL) { 2933 m_freem(sc->rl_head); 2934 sc->rl_head = sc->rl_tail = NULL; 2935 } 2936 2937 /* Free the TX list buffers. */ 2938 2939 for (i = 0; i < sc->rl_ldata.rl_tx_desc_cnt; i++) { 2940 txd = &sc->rl_ldata.rl_tx_desc[i]; 2941 if (txd->tx_m != NULL) { 2942 bus_dmamap_sync(sc->rl_ldata.rl_tx_mtag, 2943 txd->tx_dmamap, BUS_DMASYNC_POSTWRITE); 2944 bus_dmamap_unload(sc->rl_ldata.rl_tx_mtag, 2945 txd->tx_dmamap); 2946 m_freem(txd->tx_m); 2947 txd->tx_m = NULL; 2948 } 2949 } 2950 2951 /* Free the RX list buffers. */ 2952 2953 for (i = 0; i < sc->rl_ldata.rl_rx_desc_cnt; i++) { 2954 rxd = &sc->rl_ldata.rl_rx_desc[i]; 2955 if (rxd->rx_m != NULL) { 2956 bus_dmamap_sync(sc->rl_ldata.rl_tx_mtag, 2957 rxd->rx_dmamap, BUS_DMASYNC_POSTREAD); 2958 bus_dmamap_unload(sc->rl_ldata.rl_rx_mtag, 2959 rxd->rx_dmamap); 2960 m_freem(rxd->rx_m); 2961 rxd->rx_m = NULL; 2962 } 2963 } 2964 } 2965 2966 /* 2967 * Device suspend routine. Stop the interface and save some PCI 2968 * settings in case the BIOS doesn't restore them properly on 2969 * resume. 2970 */ 2971 static int 2972 re_suspend(device_t dev) 2973 { 2974 struct rl_softc *sc; 2975 2976 sc = device_get_softc(dev); 2977 2978 RL_LOCK(sc); 2979 re_stop(sc); 2980 re_setwol(sc); 2981 sc->suspended = 1; 2982 RL_UNLOCK(sc); 2983 2984 return (0); 2985 } 2986 2987 /* 2988 * Device resume routine. Restore some PCI settings in case the BIOS 2989 * doesn't, re-enable busmastering, and restart the interface if 2990 * appropriate. 2991 */ 2992 static int 2993 re_resume(device_t dev) 2994 { 2995 struct rl_softc *sc; 2996 struct ifnet *ifp; 2997 2998 sc = device_get_softc(dev); 2999 3000 RL_LOCK(sc); 3001 3002 ifp = sc->rl_ifp; 3003 /* Take controller out of sleep mode. */ 3004 if ((sc->rl_flags & RL_FLAG_MACSLEEP) != 0) { 3005 if ((CSR_READ_1(sc, RL_MACDBG) & 0x80) == 0x80) 3006 CSR_WRITE_1(sc, RL_GPIO, 3007 CSR_READ_1(sc, RL_GPIO) | 0x01); 3008 } 3009 3010 /* reinitialize interface if necessary */ 3011 if (ifp->if_flags & IFF_UP) 3012 re_init_locked(sc); 3013 3014 /* 3015 * Clear WOL matching such that normal Rx filtering 3016 * wouldn't interfere with WOL patterns. 3017 */ 3018 re_clrwol(sc); 3019 sc->suspended = 0; 3020 RL_UNLOCK(sc); 3021 3022 return (0); 3023 } 3024 3025 /* 3026 * Stop all chip I/O so that the kernel's probe routines don't 3027 * get confused by errant DMAs when rebooting. 3028 */ 3029 static int 3030 re_shutdown(device_t dev) 3031 { 3032 struct rl_softc *sc; 3033 3034 sc = device_get_softc(dev); 3035 3036 RL_LOCK(sc); 3037 re_stop(sc); 3038 /* 3039 * Mark interface as down since otherwise we will panic if 3040 * interrupt comes in later on, which can happen in some 3041 * cases. 3042 */ 3043 sc->rl_ifp->if_flags &= ~IFF_UP; 3044 re_setwol(sc); 3045 RL_UNLOCK(sc); 3046 3047 return (0); 3048 } 3049 3050 static void 3051 re_setwol(struct rl_softc *sc) 3052 { 3053 struct ifnet *ifp; 3054 int pmc; 3055 uint16_t pmstat; 3056 uint8_t v; 3057 3058 RL_LOCK_ASSERT(sc); 3059 3060 if (pci_find_extcap(sc->rl_dev, PCIY_PMG, &pmc) != 0) 3061 return; 3062 3063 ifp = sc->rl_ifp; 3064 /* Put controller into sleep mode. */ 3065 if ((sc->rl_flags & RL_FLAG_MACSLEEP) != 0) { 3066 if ((CSR_READ_1(sc, RL_MACDBG) & 0x80) == 0x80) 3067 CSR_WRITE_1(sc, RL_GPIO, 3068 CSR_READ_1(sc, RL_GPIO) & ~0x01); 3069 } 3070 if ((ifp->if_capenable & IFCAP_WOL) != 0 && 3071 (sc->rl_flags & RL_FLAG_WOLRXENB) != 0) 3072 CSR_WRITE_1(sc, RL_COMMAND, RL_CMD_RX_ENB); 3073 /* Enable config register write. */ 3074 CSR_WRITE_1(sc, RL_EECMD, RL_EE_MODE); 3075 3076 /* Enable PME. */ 3077 v = CSR_READ_1(sc, RL_CFG1); 3078 v &= ~RL_CFG1_PME; 3079 if ((ifp->if_capenable & IFCAP_WOL) != 0) 3080 v |= RL_CFG1_PME; 3081 CSR_WRITE_1(sc, RL_CFG1, v); 3082 3083 v = CSR_READ_1(sc, RL_CFG3); 3084 v &= ~(RL_CFG3_WOL_LINK | RL_CFG3_WOL_MAGIC); 3085 if ((ifp->if_capenable & IFCAP_WOL_MAGIC) != 0) 3086 v |= RL_CFG3_WOL_MAGIC; 3087 CSR_WRITE_1(sc, RL_CFG3, v); 3088 3089 /* Config register write done. */ 3090 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 3091 3092 v = CSR_READ_1(sc, RL_CFG5); 3093 v &= ~(RL_CFG5_WOL_BCAST | RL_CFG5_WOL_MCAST | RL_CFG5_WOL_UCAST); 3094 v &= ~RL_CFG5_WOL_LANWAKE; 3095 if ((ifp->if_capenable & IFCAP_WOL_UCAST) != 0) 3096 v |= RL_CFG5_WOL_UCAST; 3097 if ((ifp->if_capenable & IFCAP_WOL_MCAST) != 0) 3098 v |= RL_CFG5_WOL_MCAST | RL_CFG5_WOL_BCAST; 3099 if ((ifp->if_capenable & IFCAP_WOL) != 0) 3100 v |= RL_CFG5_WOL_LANWAKE; 3101 CSR_WRITE_1(sc, RL_CFG5, v); 3102 3103 /* 3104 * It seems that hardware resets its link speed to 100Mbps in 3105 * power down mode so switching to 100Mbps in driver is not 3106 * needed. 3107 */ 3108 3109 /* Request PME if WOL is requested. */ 3110 pmstat = pci_read_config(sc->rl_dev, pmc + PCIR_POWER_STATUS, 2); 3111 pmstat &= ~(PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE); 3112 if ((ifp->if_capenable & IFCAP_WOL) != 0) 3113 pmstat |= PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE; 3114 pci_write_config(sc->rl_dev, pmc + PCIR_POWER_STATUS, pmstat, 2); 3115 } 3116 3117 static void 3118 re_clrwol(struct rl_softc *sc) 3119 { 3120 int pmc; 3121 uint8_t v; 3122 3123 RL_LOCK_ASSERT(sc); 3124 3125 if (pci_find_extcap(sc->rl_dev, PCIY_PMG, &pmc) != 0) 3126 return; 3127 3128 /* Enable config register write. */ 3129 CSR_WRITE_1(sc, RL_EECMD, RL_EE_MODE); 3130 3131 v = CSR_READ_1(sc, RL_CFG3); 3132 v &= ~(RL_CFG3_WOL_LINK | RL_CFG3_WOL_MAGIC); 3133 CSR_WRITE_1(sc, RL_CFG3, v); 3134 3135 /* Config register write done. */ 3136 CSR_WRITE_1(sc, RL_EECMD, RL_EEMODE_OFF); 3137 3138 v = CSR_READ_1(sc, RL_CFG5); 3139 v &= ~(RL_CFG5_WOL_BCAST | RL_CFG5_WOL_MCAST | RL_CFG5_WOL_UCAST); 3140 v &= ~RL_CFG5_WOL_LANWAKE; 3141 CSR_WRITE_1(sc, RL_CFG5, v); 3142 } 3143