Committed the original NetBSD version of this driver to get the changes

made for FreeBSD as cvs diffs. This is the 3C509 ethernet driver.
This commit is contained in:
ats 1994-01-10 18:56:26 +00:00
parent 24d7941166
commit b5e6b11d1f
4 changed files with 2288 additions and 0 deletions

849
sys/dev/ep/if_ep.c Normal file
View File

@ -0,0 +1,849 @@
/*
* Copyright (c) 1993 Herb Peyerl (hpeyerl@novatel.ca)
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software withough specific prior written permission
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* $Id: if_ep.c,v 1.5 1993/12/20 09:05:55 mycroft Exp $
*/
/*
* TODO:
* Multi-509 configs.
* don't pass unit into epstop.
* epintr returns an int for magnum. 0=not for me. 1=for me. -1=whoknows?
*/
#include "ep.h"
#if NEP > 0
#include "bpfilter.h"
#include <sys/param.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/syslog.h>
#include <sys/select.h>
#include <net/if.h>
#include <net/netisr.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/netisr.h>
#ifdef INET
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/if_ether.h>
#endif
#ifdef NS
#include <netns/ns.h>
#include <netns/ns_if.h>
#endif
#if NBPFILTER > 0
#include <net/bpf.h>
#include <net/bpfdesc.h>
#endif
#include <machine/pio.h>
#include <i386/isa/isa.h>
#include <i386/isa/isa_device.h>
#include <i386/isa/icu.h>
#include <i386/isa/if_epreg.h>
#define ETHER_MIN_LEN 64
#define ETHER_MAX_LEN 1518
#define ETHER_ADDR_LEN 6
/*
* Ethernet software status per interface.
*/
struct ep_softc {
struct arpcom ep_ac; /* Ethernet common part */
#define ep_if ep_ac.ac_if /* network-visible interface */
#define ep_addr ep_ac.ac_enaddr /* hardware Ethernet address */
short ep_io_addr; /* i/o bus address */
char ep_connectors; /* Connectors on this card. */
#define MAX_MBS 8 /* # of mbufs we keep around */
struct mbuf *mb[MAX_MBS]; /* spare mbuf storage. */
int next_mb; /* Which mbuf to use next. */
int last_mb; /* Last mbuf. */
int tx_start_thresh; /* Current TX_start_thresh. */
caddr_t bpf; /* BPF "magic cookie" */
} ep_softc[NEP];
int ether_output(),
epprobe(),
epattach(),
epintr(),
epinit(),
epioctl(),
epreset(),
epwatchdog(),
epstart(),
fill_mbuf_queue();
struct isa_driver epdriver = {
epprobe,
epattach,
"ep"
};
extern u_short get_eeprom_data(int id_port, int offset);
extern int is_eeprom_busy(struct isa_device *is);
/*
* Rudimentary support for multiple cards is here but is not
* currently handled. In the future we will have to add code
* for tagging the cards for later activation. We wanna do something
* about the id_port. We're limited due to current config procedure.
* Magnum config holds promise of a fix but we'll have to wait a bit.
*/
epprobe(is)
struct isa_device *is;
{
struct ep_softc *sc = &ep_softc[is->id_unit];
u_short k;
char buf[8];
int id_port = 0x100; /* XXX */
outw(BASE+EP_COMMAND, GLOBAL_RESET);
DELAY(1000);
outb(id_port, 0xc0); /* Global reset to id_port. */
DELAY(1000);
send_ID_sequence(id_port);
DELAY(1000);
/*
* MFG_ID should have 0x6d50.
* PROD_ID should be 0x9[0-f]50
*/
k = get_eeprom_data(id_port, EEPROM_MFG_ID);
if (k != MFG_ID)
return(0);
k = get_eeprom_data(id_port, EEPROM_PROD_ID);
if ((k & 0xf0ff) != (PROD_ID & 0xf0ff))
return(0);
k = get_eeprom_data(id_port, EEPROM_ADDR_CFG); /* get addr cfg */
k = (k & 0x1f)*0x10+0x200; /* decode base addr. */
if (k != is->id_iobase)
return(0);
k = get_eeprom_data(id_port, EEPROM_RESOURCE_CFG);
k >>= 12;
if (is->id_irq != (1<<((k==2) ? 9 : k)))
return(0);
outb(id_port, ACTIVATE_ADAPTER_TO_CONFIG);
return(0x10); /* 16 bytes of I/O space used. */
}
epattach(is)
struct isa_device *is;
{
struct ep_softc *sc = &ep_softc[is->id_unit];
struct ifnet *ifp = &sc->ep_if;
u_short i;
struct ifaddr *ifa;
struct sockaddr_dl *sdl;
sc->ep_io_addr = is->id_iobase;
sc->ep_connectors = 0;
i=inw(is->id_iobase+EP_W0_CONFIG_CTRL);
printf("ep%d: ", is->id_unit);
if (i & IS_AUI) {
if (sc->ep_connectors)
printf("/");
printf("aui");
sc->ep_connectors |= AUI;
}
if (i & IS_BNC) {
if (sc->ep_connectors)
printf("/");
printf("bnc");
sc->ep_connectors |= BNC;
}
if (i & IS_UTP) {
if (sc->ep_connectors)
printf("/");
printf("utp");
sc->ep_connectors |= UTP;
}
if (!sc->ep_connectors)
printf("!no connectors!");
/*
* Read the station address from the eeprom
*/
for (i=0; i<3; i++) {
u_short *p;
GO_WINDOW(0);
if (is_eeprom_busy(is))
return;
outw(BASE+EP_W0_EEPROM_COMMAND, READ_EEPROM | i);
if (is_eeprom_busy(is))
return;
p =(u_short *)&sc->ep_addr[i*2];
*p=htons(inw(BASE+EP_W0_EEPROM_DATA));
GO_WINDOW(2);
outw(BASE+EP_W2_ADDR_0+(i*2), ntohs(*p));
}
printf(" address %s\n", ether_sprintf(sc->ep_addr));
ifp->if_unit = is->id_unit;
ifp->if_name = "ep" ;
ifp->if_mtu = ETHERMTU;
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_NOTRAILERS;
ifp->if_init = epinit;
ifp->if_output = ether_output;
ifp->if_start = epstart;
ifp->if_ioctl = epioctl;
ifp->if_watchdog = epwatchdog;
if_attach(ifp);
/*
* Fill the hardware address into ifa_addr if we find an AF_LINK entry.
* We need to do this so bpf's can get the hardware addr of this card.
* netstat likes this too!
*/
ifa = ifp->if_addrlist;
while ((ifa != 0) && (ifa->ifa_addr != 0) &&
(ifa->ifa_addr->sa_family != AF_LINK))
ifa = ifa->ifa_next;
if ((ifa != 0) && (ifa->ifa_addr != 0)) {
sdl = (struct sockaddr_dl *)ifa->ifa_addr;
sdl->sdl_type = IFT_ETHER;
sdl->sdl_alen = ETHER_ADDR_LEN;
sdl->sdl_slen = 0;
bcopy(sc->ep_addr, LLADDR(sdl), ETHER_ADDR_LEN);
}
#if NBPFILTER > 0
bpfattach(&sc->bpf, ifp, DLT_EN10MB, sizeof(struct ether_header));
#endif
}
/* The order in here seems important. Otherwise we may not receive interrupts. ?! */
epinit(unit)
int unit;
{
register struct ep_softc *sc = &ep_softc[unit];
register struct ifnet *ifp = &sc->ep_if;
int s,i;
s=splnet();
if (ifp->if_addrlist == (struct ifaddr *) 0) {
printf("ep: address not known...\n");
splx(s);
return;
}
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
GO_WINDOW(0);
outw(BASE+EP_W0_CONFIG_CTRL, 0); /* Disable the card */
outw(BASE+EP_W0_CONFIG_CTRL, ENABLE_DRQ_IRQ); /* Enable the card */
GO_WINDOW(2);
for(i=0;i<6;i++) /* Reload the ether_addr. */
outb(BASE+EP_W2_ADDR_0+i, sc->ep_addr[i]);
outw(BASE+EP_COMMAND, RX_RESET);
outw(BASE+EP_COMMAND, TX_RESET);
GO_WINDOW(1); /* Window 1 is operating window */
for(i=0;i<31;i++)
inb(BASE+EP_W1_TX_STATUS);
outw(BASE+EP_COMMAND, ACK_INTR | 0xff); /* get rid of stray intr's */
outw(BASE+EP_COMMAND, SET_RD_0_MASK | S_CARD_FAILURE | S_RX_COMPLETE |
S_TX_COMPLETE | S_TX_AVAIL);
outw(BASE+EP_COMMAND, SET_INTR_MASK | S_CARD_FAILURE | S_RX_COMPLETE |
S_TX_COMPLETE | S_TX_AVAIL);
outw(BASE+EP_COMMAND, SET_RX_FILTER | FIL_INDIVIDUAL |
FIL_GROUP | FIL_BRDCST);
if (!(ifp->if_flags & IFF_LINK0) && (sc->ep_connectors & BNC)) { /* Want BNC? */
outw(BASE+EP_COMMAND, START_TRANSCEIVER);
DELAY(1000);
}
if ((sc->ep_connectors & UTP) & !(ifp->if_flags & IFF_LINK0)) { /* Want UTP? */
GO_WINDOW(4);
outw(BASE+EP_W4_MEDIA_TYPE, ENABLE_UTP);
GO_WINDOW(1);
}
outw(BASE+EP_COMMAND, RX_ENABLE);
outw(BASE+EP_COMMAND, TX_ENABLE);
ifp->if_flags |= IFF_RUNNING;
ifp->if_flags &= ~IFF_OACTIVE; /* just in case */
sc->tx_start_thresh = 20; /* probably a good starting point. */
/*
* Store up a bunch of mbuf's for use later. (MAX_MBS). First we
* free up any that we had in case we're being called from intr or
* somewhere else.
*/
sc->last_mb = 0;
sc->next_mb = 0;
fill_mbuf_queue(sc);
epstart(ifp);
splx(s);
}
epstart(ifp)
struct ifnet *ifp;
{
register struct ep_softc *sc = &ep_softc[ifp->if_unit];
struct mbuf *m, *top;
int s, len, pad;
s=splnet();
if (sc->ep_if.if_flags & IFF_OACTIVE) {
splx(s);
return;
}
startagain:
m = sc->ep_if.if_snd.ifq_head; /* Sneak a peek at the next packet */
if (m == 0) {
splx(s);
return;
}
pad = (4-(m->m_pkthdr.len%4)+4)%4; /* icky pooh!! */
if ((inw(BASE+EP_W1_FREE_TX)) < (m->m_pkthdr.len)+pad+4) { /* no room in FIFO */
outw(BASE+EP_COMMAND, SET_TX_AVAIL_THRESH | (m->m_pkthdr.len)+pad+4);
sc->ep_if.if_flags |= IFF_OACTIVE;
splx(s);
return;
}
IF_DEQUEUE(&sc->ep_if.if_snd, m);
if (m == 0) { /* Could make this go away. */
splx(s);
return;
}
outw(BASE+EP_COMMAND, SET_TX_START_THRESH |
(m->m_pkthdr.len/4 + sc->tx_start_thresh));
outw(BASE+EP_W1_TX_PIO_WR_1, m->m_pkthdr.len);
outw(BASE+EP_W1_TX_PIO_WR_1, 0xffff); /* Second dword meaningless */
for(top = m; m != 0; m = m->m_next) {
outsw(BASE+EP_W1_TX_PIO_WR_1, mtod(m, caddr_t), m->m_len/2);
if(m->m_len & 1)
outb(BASE+EP_W1_TX_PIO_WR_1,
*(mtod(m, caddr_t)+m->m_len-1));
}
while (pad--)
outb(BASE+EP_W1_TX_PIO_WR_1, 0); /* Padding */
#if NBPFILTER > 0
if (sc->bpf) {
u_short etype;
int off, datasize, resid;
struct ether_header *eh;
struct trailer_header {
u_short ether_type;
u_short ether_residual;
} trailer_header;
char ether_packet[ETHER_MAX_LEN];
char *ep;
ep = ether_packet;
/*
* We handle trailers below:
* Copy ether header first, then residual data,
* then data. Put all this in a temporary buffer
* 'ether_packet' and send off to bpf. Since the
* system has generated this packet, we assume
* that all of the offsets in the packet are
* correct; if they're not, the system will almost
* certainly crash in m_copydata.
* We make no assumptions about how the data is
* arranged in the mbuf chain (i.e. how much
* data is in each mbuf, if mbuf clusters are
* used, etc.), which is why we use m_copydata
* to get the ether header rather than assume
* that this is located in the first mbuf.
*/
/* copy ether header */
m_copydata(top, 0, sizeof(struct ether_header), ep);
eh = (struct ether_header *) ep;
ep += sizeof(struct ether_header);
etype = ntohs(eh->ether_type);
if (etype >= ETHERTYPE_TRAIL &&
etype < ETHERTYPE_TRAIL+ETHERTYPE_NTRAILER) {
datasize = ((etype - ETHERTYPE_TRAIL) << 9);
off = datasize + sizeof(struct ether_header);
/* copy trailer_header into a data structure */
m_copydata(top, off, sizeof(struct trailer_header),
&trailer_header.ether_type);
/* copy residual data */
resid = trailer_header.ether_residual -
sizeof(struct trailer_header);
resid = ntohs(resid);
m_copydata(top, off+sizeof(struct trailer_header),
resid, ep);
ep += resid;
/* copy data */
m_copydata(top, sizeof(struct ether_header),
datasize, ep);
ep += datasize;
/* restore original ether packet type */
eh->ether_type = trailer_header.ether_type;
bpf_tap(sc->bpf, ether_packet, ep - ether_packet);
} else
bpf_mtap(sc->bpf, top);
}
#endif
m_freem(top);
++sc->ep_if.if_opackets;
/*
* Is another packet coming in? We don't want to overflow the
* tiny RX fifo.
*/
if (inw(BASE+EP_W1_RX_STATUS) & RX_BYTES_MASK) {
splx(s);
return;
}
goto startagain;
}
epintr(unit)
int unit;
{
int status, i;
register struct ep_softc *sc = &ep_softc[unit];
struct ifnet *ifp = &sc->ep_if;
struct mbuf *m;
status=0;
checkintr:
status = inw(BASE + EP_STATUS) & (S_TX_COMPLETE|S_TX_AVAIL|S_RX_COMPLETE|S_CARD_FAILURE);
if (status == 0) { /* No interrupts. */
outw(BASE+EP_COMMAND, C_INTR_LATCH);
return;
}
outw(BASE+EP_COMMAND, ACK_INTR | status); /* important that we do this first. */
if (status & S_TX_AVAIL) {
status &= ~S_TX_AVAIL;
inw(BASE+EP_W1_FREE_TX);
sc->ep_if.if_flags &= ~IFF_OACTIVE;
epstart(&sc->ep_if);
}
if (status & S_RX_COMPLETE) {
status &= ~S_RX_COMPLETE;
epread(sc);
}
if (status & S_CARD_FAILURE) {
printf("ep%d: reset (status: %x)\n", unit, status);
outw(BASE+EP_COMMAND, C_INTR_LATCH);
epinit(unit);
return;
}
if (status & S_TX_COMPLETE) {
status &= ~S_TX_COMPLETE;
/*
* We need to read TX_STATUS until we get a 0 status in
* order to turn off the interrupt flag.
*/
while ((i=inb(BASE+EP_W1_TX_STATUS)) & TXS_COMPLETE) {
outw(BASE+EP_W1_TX_STATUS, 0x0);
if (i & (TXS_MAX_COLLISION|TXS_JABBER|TXS_UNDERRUN)) {
if (i & TXS_MAX_COLLISION)
++sc->ep_if.if_collisions;
if (i & (TXS_JABBER|TXS_UNDERRUN)) {
outw(BASE+EP_COMMAND, TX_RESET);
if(i & TXS_UNDERRUN) {
if (sc->tx_start_thresh < ETHER_MAX_LEN) {
sc->tx_start_thresh += 20;
outw(BASE+EP_COMMAND,
SET_TX_START_THRESH |
sc->tx_start_thresh);
}
}
}
outw(BASE+EP_COMMAND, TX_ENABLE);
++sc->ep_if.if_oerrors;
}
}
epstart(ifp);
}
goto checkintr;
}
epread(sc)
register struct ep_softc *sc;
{
struct ether_header *eh;
struct mbuf *mcur, *m, *m0, *top;
int totlen, lenthisone;
int save_totlen;
u_short etype;
int off, resid;
int count, spinwait;
int i;
totlen = inw(BASE + EP_W1_RX_STATUS);
off = 0;
top = 0;
if (totlen & ERR_RX) {
++sc->ep_if.if_ierrors;
goto out;
}
save_totlen = totlen &= RX_BYTES_MASK; /* Lower 10 bits = RX bytes. */
m = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (m == 0) {
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == 0)
goto out;
} else { /* Convert one of our saved mbuf's */
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
m->m_data = m->m_pktdat;
m->m_flags = M_PKTHDR;
}
top = m0 = m; /* We assign top so we can "goto out" */
# define EROUND ((sizeof(struct ether_header) + 3) & ~3)
# define EOFF (EROUND - sizeof(struct ether_header))
m0->m_data += EOFF;
/* Read what should be the header. */
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m0, caddr_t), sizeof(struct ether_header)/2);
m->m_len = sizeof(struct ether_header);
totlen -= sizeof(struct ether_header);
/*
* mostly deal with trailer here. (untested)
* We do this in a couple of parts. First we check for a trailer, if
* we have one we convert the mbuf back to a regular mbuf and set the offset and
* subtract sizeof(struct ether_header) from the pktlen.
* After we've read the packet off the interface (all except for the trailer
* header, we then get a header mbuf, read the trailer into it, and fix up
* the mbuf pointer chain.
*/
eh=mtod(m, struct ether_header *);
eh->ether_type = ntohs((u_short)eh->ether_type);
if (eh->ether_type >= ETHERTYPE_TRAIL &&
eh->ether_type < ETHERTYPE_TRAIL+ETHERTYPE_NTRAILER) {
m->m_data = m->m_dat; /* Convert back to regular mbuf. */
m->m_flags = 0; /* This sucks but non-trailers are the norm */
off = (eh->ether_type - ETHERTYPE_TRAIL) * 512;
if (off >= ETHERMTU) {
m_freem(m);
return; /* sanity */
}
totlen -= sizeof(struct ether_header); /* We don't read the trailer */
m->m_data += 2 * sizeof(u_short); /* Get rid of type & len*/
}
while (totlen>0) {
lenthisone=min(totlen, M_TRAILINGSPACE(m));
if (lenthisone == 0) { /* no room in this one */
mcur = m;
m = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (!m) {
MGET(m, M_DONTWAIT, MT_DATA);
if (m==0)
goto out;
} else {
timeout(fill_mbuf_queue, sc, 1);
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
}
if (totlen >= MINCLSIZE)
MCLGET(m, M_DONTWAIT);
m->m_len = 0;
mcur->m_next = m;
lenthisone = min(totlen, M_TRAILINGSPACE(m));
}
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m, caddr_t)+m->m_len, lenthisone/2);
m->m_len += lenthisone;
if (lenthisone & 1)
*(mtod(m, caddr_t)+m->m_len-1) = inb(BASE+EP_W1_RX_PIO_RD_1);
totlen -= lenthisone;
}
if (off) {
top = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (top == 0) {
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (top == 0)
goto out;
} else { /* Convert one of our saved mbuf's */
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
top->m_data = top->m_pktdat;
top->m_flags = M_PKTHDR;
}
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m, caddr_t)+m->m_len,
sizeof(struct ether_header));
eh->ether_type = ntohs(eh->ether_type);
top->m_next = m0;
top->m_len = sizeof(struct ether_header);
/* XXX Accomodate for type and len from beginning of trailer data */
top->m_pkthdr.len = save_totlen - (2 * sizeof(u_short));
} else {
top = m0;
top->m_pkthdr.len = save_totlen;
}
top->m_pkthdr.rcvif = &sc->ep_if;
outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
++sc->ep_if.if_ipackets;
m_adj(top, sizeof(struct ether_header));
ether_input(&sc->ep_if, eh, top);
return;
out: outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
if (top)
m_freem(top);
}
/*
* Look familiar?
*/
epioctl(ifp, cmd, data)
register struct ifnet *ifp;
int cmd;
caddr_t data;
{
register struct ifaddr *ifa = (struct ifaddr *)data;
struct ep_softc *sc = &ep_softc[ifp->if_unit];
struct ifreq *ifr = (struct ifreq *)data;
int s, error=0;
switch(cmd){
case SIOCSIFADDR:
ifp->if_flags |= IFF_UP;
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
epinit(ifp->if_unit); /* before arpwhohas */
((struct arpcom *)ifp)->ac_ipaddr = IA_SIN(ifa)->sin_addr;
arpwhohas((struct arpcom *)ifp, &IA_SIN(ifa)->sin_addr);
break;
#endif
#ifdef NS
case AF_NS:
{
register struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
if (ns_nullhost(*ina))
ina->x_host = *(union ns_host *)(sc->ep_ac.ac_enaddr);
else {
ifp->if_flags &= ~IFF_RUNNING;
bcopy((caddr_t)ina->x_host.c_host,
(caddr_t)sc->ns_addr,
sizeof(sc->ep_ac.ac_enaddr));
}
epinit(ifp->if_unit);
break;
}
#endif
default:
epinit(ifp->if_unit);
break;
}
break;
case SIOCSIFFLAGS:
if ((ifp->if_flags & IFF_UP) == 0 && ifp->if_flags & IFF_RUNNING) {
ifp->if_flags &= ~IFF_RUNNING;
epstop(ifp->if_unit);
} else if (ifp->if_flags & IFF_UP && (ifp->if_flags & IFF_RUNNING) == 0)
epinit(ifp->if_unit);
break;
#ifdef notdef
case SIOCGHWADDR:
bcopy((caddr_t)sc->sc_addr, (caddr_t) &ifr->ifr_data, sizeof(sc->sc_addr));
break;
#endif
default:
error = EINVAL;
}
return (error);
}
epreset(unit)
int unit;
{
int s;
epstop(unit);
epinit(unit);
return;
}
epwatchdog(unit)
int unit;
{
return;
}
epstop(unit)
int unit;
{
register struct ep_softc *sc = &ep_softc[unit];
outw(BASE+EP_COMMAND, RX_DISABLE);
outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
outw(BASE+EP_COMMAND, TX_DISABLE);
outw(BASE+EP_COMMAND, STOP_TRANSCEIVER);
outw(BASE+EP_COMMAND, RX_RESET);
outw(BASE+EP_COMMAND, TX_RESET);
outw(BASE+EP_COMMAND, C_INTR_LATCH);
outw(BASE+EP_COMMAND, SET_RD_0_MASK);
outw(BASE+EP_COMMAND, SET_INTR_MASK);
outw(BASE+EP_COMMAND, SET_RX_FILTER);
return;
}
/*
* This is adapted straight from the book. There's probably a better way.
*/
send_ID_sequence(port)
u_short port;
{
char cx, al;
cx=0x0ff;
al=0x0ff;
outb(port, 0x0);
DELAY(1000);
outb(port, 0x0);
DELAY(1000);
loop1: cx--;
outb(port, al);
if (!(al & 0x80)) {
al=al<<1;
goto loop1;
}
al=al<<1;
al^=0xcf;
if (cx)
goto loop1;
}
/*
* We get eeprom data from the id_port given an offset into the
* eeprom. Basically; after the ID_sequence is sent to all of
* the cards; they enter the ID_CMD state where they will accept
* command requests. 0x80-0xbf loads the eeprom data. We then
* read the port 16 times and with every read; the cards check
* for contention (ie: if one card writes a 0 bit and another
* writes a 1 bit then the host sees a 0. At the end of the cycle;
* each card compares the data on the bus; if there is a difference
* then that card goes into ID_WAIT state again). In the meantime;
* one bit of data is returned in the AX register which is conveniently
* returned to us by inb(). Hence; we read 16 times getting one
* bit of data with each read.
*/
u_short get_eeprom_data(id_port, offset)
int id_port;
int offset;
{
int i, data=0;
outb(id_port, 0x80+offset);
DELAY(1000);
for (i=0; i<16; i++)
data = (data<<1) | (inw(id_port) & 1);
return(data);
}
int
is_eeprom_busy(is)
struct isa_device *is;
{
int i=0, j;
register struct ep_softc *sc = &ep_softc[is->id_unit];
while (i++<100) {
j=inw(BASE+EP_W0_EEPROM_COMMAND);
if (j & EEPROM_BUSY)
DELAY(100);
else
break;
}
if (i>=100) {
printf("\nep%d: eeprom failed to come ready.\n", is->id_unit);
return(1);
}
if (j & EEPROM_TST_MODE) {
printf("\nep%d: 3c509 in test mode. Erase pencil mark!\n", is->id_unit);
return(1);
}
return(0);
}
int
fill_mbuf_queue(sc)
struct ep_softc *sc;
{
int i=0;
if (sc->mb[sc->last_mb])
return;
i=sc->last_mb;
do {
MGET(sc->mb[i], M_DONTWAIT, MT_DATA);
if (!sc->mb[i])
break;
i = (i+1) % MAX_MBS;
} while(i != sc->next_mb);
sc->last_mb = i;
}
#endif /* NEP > 0 */

295
sys/dev/ep/if_epreg.h Normal file
View File

@ -0,0 +1,295 @@
/*
* Copyright (c) 1993 Herb Peyerl (hpeyerl@novatel.ca)
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software withough specific prior written permission
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* $Id: if_epreg.h,v 1.1 1993/12/14 04:26:47 hpeyerl Exp $
*/
/**************************************************************************
* *
* These define the EEPROM data structure. They are used in the probe
* function to verify the existance of the adapter after having sent
* the ID_Sequence.
*
* There are others but only the ones we use are defined here.
*
**************************************************************************/
#define EEPROM_NODE_ADDR_0 0x0 /* Word */
#define EEPROM_NODE_ADDR_1 0x1 /* Word */
#define EEPROM_NODE_ADDR_2 0x2 /* Word */
#define EEPROM_PROD_ID 0x3 /* 0x9[0-f]50 */
#define EEPROM_MFG_ID 0x7 /* 0x6d50 */
#define EEPROM_ADDR_CFG 0x8 /* Base addr */
#define EEPROM_RESOURCE_CFG 0x9 /* IRQ. Bits 12-15 */
/**************************************************************************
* *
* These are the registers for the 3Com 3c509 and their bit patterns when *
* applicable. They have been taken out the the "EtherLink III Parallel *
* Tasking EISA and ISA Technical Reference" "Beta Draft 10/30/92" manual *
* from 3com. *
* *
**************************************************************************/
#define EP_COMMAND 0x0e /* Write. BASE+0x0e is always a command reg. */
#define EP_STATUS 0x0e /* Read. BASE+0x0e is always status reg. */
#define EP_WINDOW 0x0f /* Read. BASE+0x0f is always window reg. */
/*
* Window 0 registers. Setup.
*/
/* Write */
#define EP_W0_EEPROM_DATA 0x0c
#define EP_W0_EEPROM_COMMAND 0x0a
#define EP_W0_RESOURCE_CFG 0x08
#define EP_W0_ADDRESS_CFG 0x06
#define EP_W0_CONFIG_CTRL 0x04
/* Read */
#define EP_W0_PRODUCT_ID 0x02
#define EP_W0_MFG_ID 0x00
/*
* Window 1 registers. Operating Set.
*/
/* Write */
#define EP_W1_TX_PIO_WR_2 0x02
#define EP_W1_TX_PIO_WR_1 0x00
/* Read */
#define EP_W1_FREE_TX 0x0c
#define EP_W1_TX_STATUS 0x0b /* byte */
#define EP_W1_TIMER 0x0a /* byte */
#define EP_W1_RX_STATUS 0x08
#define EP_W1_RX_PIO_RD_2 0x02
#define EP_W1_RX_PIO_RD_1 0x00
/*
* Window 2 registers. Station Address Setup/Read
*/
/* Read/Write */
#define EP_W2_ADDR_5 0x05
#define EP_W2_ADDR_4 0x04
#define EP_W2_ADDR_3 0x03
#define EP_W2_ADDR_2 0x02
#define EP_W2_ADDR_1 0x01
#define EP_W2_ADDR_0 0x00
/*
* Window 3 registers. FIFO Management.
*/
/* Read */
#define EP_W3_FREE_TX 0x0c
#define EP_W3_FREE_RX 0x0a
/*
* Window 4 registers. Diagnostics.
*/
/* Read/Write */
#define EP_W4_MEDIA_TYPE 0x0a
#define EP_W4_CTRLR_STATUS 0x08
#define EP_W4_NET_DIAG 0x06
#define EP_W4_FIFO_DIAG 0x04
#define EP_W4_HOST_DIAG 0x02
#define EP_W4_TX_DIAG 0x00
/*
* Window 5 Registers. Results and Internal status.
*/
/* Read */
#define EP_W5_READ_0_MASK 0x0c
#define EP_W5_INTR_MASK 0x0a
#define EP_W5_RX_FILTER 0x08
#define EP_W5_RX_EARLY_THRESH 0x06
#define EP_W5_TX_AVAIL_THRESH 0x02
#define EP_W5_TX_START_THRESH 0x00
/*
* Window 6 registers. Statistics.
*/
/* Read/Write */
#define TX_TOTAL_OK 0x0c
#define RX_TOTAL_OK 0x0a
#define TX_DEFERRALS 0x08
#define RX_FRAMES_OK 0x07
#define TX_FRAMES_OK 0x06
#define RX_OVERRUNS 0x05
#define TX_COLLISIONS 0x04
#define TX_AFTER_1_COLLISION 0x03
#define TX_AFTER_X_COLLISIONS 0x02
#define TX_NO_SQE 0x01
#define TX_CD_LOST 0x00
/****************************************
*
* Register definitions.
*
****************************************/
/*
* Command register. All windows.
*
* 16 bit register.
* 15-11: 5-bit code for command to be executed.
* 10-0: 11-bit arg if any. For commands with no args;
* this can be set to anything.
*/
#define GLOBAL_RESET (u_short) 0x0000 /* Wait at least 1ms after issuing */
#define WINDOW_SELECT (u_short) (0x1<<11)
#define START_TRANSCEIVER (u_short) (0x2<<11) /* Read ADDR_CFG reg to determine
whether this is needed. If so;
wait 800 uSec before using trans-
ceiver. */
#define RX_DISABLE (u_short) (0x3<<11) /* state disabled on power-up */
#define RX_ENABLE (u_short) (0x4<<11)
#define RX_RESET (u_short) (0x5<<11)
#define RX_DISCARD_TOP_PACK (u_short) (0x8<<11)
#define TX_ENABLE (u_short) (0x9<<11)
#define TX_DISABLE (u_short) (0xa<<11)
#define TX_RESET (u_short) (0xb<<11)
#define REQ_INTR (u_short) (0xc<<11)
/*
* The following C_* acknowledge the various interrupts.
* Some of them don't do anything. See the manual.
*/
#define ACK_INTR (u_short) (0x6800)
# define C_INTR_LATCH (u_short) (ACK_INTR|0x1)
# define C_CARD_FAILURE (u_short) (ACK_INTR|0x2)
# define C_TX_COMPLETE (u_short) (ACK_INTR|0x4)
# define C_TX_AVAIL (u_short) (ACK_INTR|0x8)
# define C_RX_COMPLETE (u_short) (ACK_INTR|0x10)
# define C_RX_EARLY (u_short) (ACK_INTR|0x20)
# define C_INT_RQD (u_short) (ACK_INTR|0x40)
# define C_UPD_STATS (u_short) (ACK_INTR|0x80)
#define SET_INTR_MASK (u_short) (0xe<<11)
#define SET_RD_0_MASK (u_short) (0xf<<11)
#define SET_RX_FILTER (u_short) (0x10<<11)
# define FIL_INDIVIDUAL (u_short) (0x1)
# define FIL_GROUP (u_short) (0x2)
# define FIL_BRDCST (u_short) (0x4)
# define FIL_ALL (u_short) (0x8)
#define SET_RX_EARLY_THRESH (u_short) (0x11<<11)
#define SET_TX_AVAIL_THRESH (u_short) (0x12<<11)
#define SET_TX_START_THRESH (u_short) (0x13<<11)
#define STATS_ENABLE (u_short) (0x15<<11)
#define STATS_DISABLE (u_short) (0x16<<11)
#define STOP_TRANSCEIVER (u_short) (0x17<<11)
/*
* Status register. All windows.
*
* 15-13: Window number(0-7).
* 12: Command_in_progress.
* 11: reserved.
* 10: reserved.
* 9: reserved.
* 8: reserved.
* 7: Update Statistics.
* 6: Interrupt Requested.
* 5: RX Early.
* 4: RX Complete.
* 3: TX Available.
* 2: TX Complete.
* 1: Adapter Failure.
* 0: Interrupt Latch.
*/
#define S_INTR_LATCH (u_short) (0x1)
#define S_CARD_FAILURE (u_short) (0x2)
#define S_TX_COMPLETE (u_short) (0x4)
#define S_TX_AVAIL (u_short) (0x8)
#define S_RX_COMPLETE (u_short) (0x10)
#define S_RX_EARLY (u_short) (0x20)
#define S_INT_RQD (u_short) (0x40)
#define S_UPD_STATS (u_short) (0x80)
#define S_COMMAND_IN_PROGRESS (u_short) (0x1000)
/*
* FIFO Registers. RX Status.
*
* 15: Incomplete or FIFO empty.
* 14: 1: Error in RX Packet 0: Incomplete or no error.
* 13-11: Type of error.
* 1000 = Overrun.
* 1011 = Run Packet Error.
* 1100 = Alignment Error.
* 1101 = CRC Error.
* 1001 = Oversize Packet Error (>1514 bytes)
* 0010 = Dribble Bits.
* (all other error codes, no errors.)
*
* 10-0: RX Bytes (0-1514)
*/
#define ERR_INCOMPLETE (u_short) (0x8000)
#define ERR_RX (u_short) (0x4000)
#define ERR_RX_PACKET (u_short) (0x2000)
#define ERR_OVERRUN (u_short) (0x1000)
#define ERR_RUNT (u_short) (0x1300)
#define ERR_ALIGNMENT (u_short) (0x1400)
#define ERR_CRC (u_short) (0x1500)
#define ERR_OVERSIZE (u_short) (0x1100)
#define ERR_DRIBBLE (u_short) (0x200)
/*
* TX Status
*
* Reports the transmit status of a completed transmission. Writing this
* register pops the transmit completion stack.
*
* Window 1/Port 0x0b.
*
* 7: Complete
* 6: Interrupt on successful transmission requested.
* 5: Jabber Error (TP Only, TX Reset required. )
* 4: Underrun (TX Reset required. )
* 3: Maximum Collisions.
* 2: TX Status Overflow.
* 1-0: Undefined.
*
*/
#define TXS_COMPLETE 0x80
#define TXS_INTR_REQ 0x40
#define TXS_JABBER 0x20
#define TXS_UNDERRUN 0x10
#define TXS_MAX_COLLISION 0x8
#define TXS_STATUS_OVERFLOW 0x4
/*
* Misc defines for various things.
*/
#define TAG_ADAPTER_0 0xd0
#define ACTIVATE_ADAPTER_TO_CONFIG 0xff
#define ENABLE_DRQ_IRQ 0x0001
#define MFG_ID 0x6d50
#define PROD_ID 0x9150
#define BASE sc->ep_io_addr
#define GO_WINDOW(x) outw(BASE+EP_COMMAND, WINDOW_SELECT|x)
#define AUI 0x1
#define BNC 0x2
#define UTP 0x4
#define IS_AUI (1<<13)
#define IS_BNC (1<<12)
#define IS_UTP (1<<9)
#define EEPROM_BUSY (1<<15)
#define EEPROM_TST_MODE (1<<14)
#define READ_EEPROM (1<<7)
#define ETHER_ADDR_LEN 6
#define ETHER_MAX 1536
#define ENABLE_UTP 0xc0
#define DISABLE_UTP 0x0
#define RX_BYTES_MASK (u_short) (0x07ff)

849
sys/i386/isa/if_ep.c Normal file
View File

@ -0,0 +1,849 @@
/*
* Copyright (c) 1993 Herb Peyerl (hpeyerl@novatel.ca)
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software withough specific prior written permission
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* $Id: if_ep.c,v 1.5 1993/12/20 09:05:55 mycroft Exp $
*/
/*
* TODO:
* Multi-509 configs.
* don't pass unit into epstop.
* epintr returns an int for magnum. 0=not for me. 1=for me. -1=whoknows?
*/
#include "ep.h"
#if NEP > 0
#include "bpfilter.h"
#include <sys/param.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/syslog.h>
#include <sys/select.h>
#include <net/if.h>
#include <net/netisr.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/netisr.h>
#ifdef INET
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/if_ether.h>
#endif
#ifdef NS
#include <netns/ns.h>
#include <netns/ns_if.h>
#endif
#if NBPFILTER > 0
#include <net/bpf.h>
#include <net/bpfdesc.h>
#endif
#include <machine/pio.h>
#include <i386/isa/isa.h>
#include <i386/isa/isa_device.h>
#include <i386/isa/icu.h>
#include <i386/isa/if_epreg.h>
#define ETHER_MIN_LEN 64
#define ETHER_MAX_LEN 1518
#define ETHER_ADDR_LEN 6
/*
* Ethernet software status per interface.
*/
struct ep_softc {
struct arpcom ep_ac; /* Ethernet common part */
#define ep_if ep_ac.ac_if /* network-visible interface */
#define ep_addr ep_ac.ac_enaddr /* hardware Ethernet address */
short ep_io_addr; /* i/o bus address */
char ep_connectors; /* Connectors on this card. */
#define MAX_MBS 8 /* # of mbufs we keep around */
struct mbuf *mb[MAX_MBS]; /* spare mbuf storage. */
int next_mb; /* Which mbuf to use next. */
int last_mb; /* Last mbuf. */
int tx_start_thresh; /* Current TX_start_thresh. */
caddr_t bpf; /* BPF "magic cookie" */
} ep_softc[NEP];
int ether_output(),
epprobe(),
epattach(),
epintr(),
epinit(),
epioctl(),
epreset(),
epwatchdog(),
epstart(),
fill_mbuf_queue();
struct isa_driver epdriver = {
epprobe,
epattach,
"ep"
};
extern u_short get_eeprom_data(int id_port, int offset);
extern int is_eeprom_busy(struct isa_device *is);
/*
* Rudimentary support for multiple cards is here but is not
* currently handled. In the future we will have to add code
* for tagging the cards for later activation. We wanna do something
* about the id_port. We're limited due to current config procedure.
* Magnum config holds promise of a fix but we'll have to wait a bit.
*/
epprobe(is)
struct isa_device *is;
{
struct ep_softc *sc = &ep_softc[is->id_unit];
u_short k;
char buf[8];
int id_port = 0x100; /* XXX */
outw(BASE+EP_COMMAND, GLOBAL_RESET);
DELAY(1000);
outb(id_port, 0xc0); /* Global reset to id_port. */
DELAY(1000);
send_ID_sequence(id_port);
DELAY(1000);
/*
* MFG_ID should have 0x6d50.
* PROD_ID should be 0x9[0-f]50
*/
k = get_eeprom_data(id_port, EEPROM_MFG_ID);
if (k != MFG_ID)
return(0);
k = get_eeprom_data(id_port, EEPROM_PROD_ID);
if ((k & 0xf0ff) != (PROD_ID & 0xf0ff))
return(0);
k = get_eeprom_data(id_port, EEPROM_ADDR_CFG); /* get addr cfg */
k = (k & 0x1f)*0x10+0x200; /* decode base addr. */
if (k != is->id_iobase)
return(0);
k = get_eeprom_data(id_port, EEPROM_RESOURCE_CFG);
k >>= 12;
if (is->id_irq != (1<<((k==2) ? 9 : k)))
return(0);
outb(id_port, ACTIVATE_ADAPTER_TO_CONFIG);
return(0x10); /* 16 bytes of I/O space used. */
}
epattach(is)
struct isa_device *is;
{
struct ep_softc *sc = &ep_softc[is->id_unit];
struct ifnet *ifp = &sc->ep_if;
u_short i;
struct ifaddr *ifa;
struct sockaddr_dl *sdl;
sc->ep_io_addr = is->id_iobase;
sc->ep_connectors = 0;
i=inw(is->id_iobase+EP_W0_CONFIG_CTRL);
printf("ep%d: ", is->id_unit);
if (i & IS_AUI) {
if (sc->ep_connectors)
printf("/");
printf("aui");
sc->ep_connectors |= AUI;
}
if (i & IS_BNC) {
if (sc->ep_connectors)
printf("/");
printf("bnc");
sc->ep_connectors |= BNC;
}
if (i & IS_UTP) {
if (sc->ep_connectors)
printf("/");
printf("utp");
sc->ep_connectors |= UTP;
}
if (!sc->ep_connectors)
printf("!no connectors!");
/*
* Read the station address from the eeprom
*/
for (i=0; i<3; i++) {
u_short *p;
GO_WINDOW(0);
if (is_eeprom_busy(is))
return;
outw(BASE+EP_W0_EEPROM_COMMAND, READ_EEPROM | i);
if (is_eeprom_busy(is))
return;
p =(u_short *)&sc->ep_addr[i*2];
*p=htons(inw(BASE+EP_W0_EEPROM_DATA));
GO_WINDOW(2);
outw(BASE+EP_W2_ADDR_0+(i*2), ntohs(*p));
}
printf(" address %s\n", ether_sprintf(sc->ep_addr));
ifp->if_unit = is->id_unit;
ifp->if_name = "ep" ;
ifp->if_mtu = ETHERMTU;
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_NOTRAILERS;
ifp->if_init = epinit;
ifp->if_output = ether_output;
ifp->if_start = epstart;
ifp->if_ioctl = epioctl;
ifp->if_watchdog = epwatchdog;
if_attach(ifp);
/*
* Fill the hardware address into ifa_addr if we find an AF_LINK entry.
* We need to do this so bpf's can get the hardware addr of this card.
* netstat likes this too!
*/
ifa = ifp->if_addrlist;
while ((ifa != 0) && (ifa->ifa_addr != 0) &&
(ifa->ifa_addr->sa_family != AF_LINK))
ifa = ifa->ifa_next;
if ((ifa != 0) && (ifa->ifa_addr != 0)) {
sdl = (struct sockaddr_dl *)ifa->ifa_addr;
sdl->sdl_type = IFT_ETHER;
sdl->sdl_alen = ETHER_ADDR_LEN;
sdl->sdl_slen = 0;
bcopy(sc->ep_addr, LLADDR(sdl), ETHER_ADDR_LEN);
}
#if NBPFILTER > 0
bpfattach(&sc->bpf, ifp, DLT_EN10MB, sizeof(struct ether_header));
#endif
}
/* The order in here seems important. Otherwise we may not receive interrupts. ?! */
epinit(unit)
int unit;
{
register struct ep_softc *sc = &ep_softc[unit];
register struct ifnet *ifp = &sc->ep_if;
int s,i;
s=splnet();
if (ifp->if_addrlist == (struct ifaddr *) 0) {
printf("ep: address not known...\n");
splx(s);
return;
}
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
GO_WINDOW(0);
outw(BASE+EP_W0_CONFIG_CTRL, 0); /* Disable the card */
outw(BASE+EP_W0_CONFIG_CTRL, ENABLE_DRQ_IRQ); /* Enable the card */
GO_WINDOW(2);
for(i=0;i<6;i++) /* Reload the ether_addr. */
outb(BASE+EP_W2_ADDR_0+i, sc->ep_addr[i]);
outw(BASE+EP_COMMAND, RX_RESET);
outw(BASE+EP_COMMAND, TX_RESET);
GO_WINDOW(1); /* Window 1 is operating window */
for(i=0;i<31;i++)
inb(BASE+EP_W1_TX_STATUS);
outw(BASE+EP_COMMAND, ACK_INTR | 0xff); /* get rid of stray intr's */
outw(BASE+EP_COMMAND, SET_RD_0_MASK | S_CARD_FAILURE | S_RX_COMPLETE |
S_TX_COMPLETE | S_TX_AVAIL);
outw(BASE+EP_COMMAND, SET_INTR_MASK | S_CARD_FAILURE | S_RX_COMPLETE |
S_TX_COMPLETE | S_TX_AVAIL);
outw(BASE+EP_COMMAND, SET_RX_FILTER | FIL_INDIVIDUAL |
FIL_GROUP | FIL_BRDCST);
if (!(ifp->if_flags & IFF_LINK0) && (sc->ep_connectors & BNC)) { /* Want BNC? */
outw(BASE+EP_COMMAND, START_TRANSCEIVER);
DELAY(1000);
}
if ((sc->ep_connectors & UTP) & !(ifp->if_flags & IFF_LINK0)) { /* Want UTP? */
GO_WINDOW(4);
outw(BASE+EP_W4_MEDIA_TYPE, ENABLE_UTP);
GO_WINDOW(1);
}
outw(BASE+EP_COMMAND, RX_ENABLE);
outw(BASE+EP_COMMAND, TX_ENABLE);
ifp->if_flags |= IFF_RUNNING;
ifp->if_flags &= ~IFF_OACTIVE; /* just in case */
sc->tx_start_thresh = 20; /* probably a good starting point. */
/*
* Store up a bunch of mbuf's for use later. (MAX_MBS). First we
* free up any that we had in case we're being called from intr or
* somewhere else.
*/
sc->last_mb = 0;
sc->next_mb = 0;
fill_mbuf_queue(sc);
epstart(ifp);
splx(s);
}
epstart(ifp)
struct ifnet *ifp;
{
register struct ep_softc *sc = &ep_softc[ifp->if_unit];
struct mbuf *m, *top;
int s, len, pad;
s=splnet();
if (sc->ep_if.if_flags & IFF_OACTIVE) {
splx(s);
return;
}
startagain:
m = sc->ep_if.if_snd.ifq_head; /* Sneak a peek at the next packet */
if (m == 0) {
splx(s);
return;
}
pad = (4-(m->m_pkthdr.len%4)+4)%4; /* icky pooh!! */
if ((inw(BASE+EP_W1_FREE_TX)) < (m->m_pkthdr.len)+pad+4) { /* no room in FIFO */
outw(BASE+EP_COMMAND, SET_TX_AVAIL_THRESH | (m->m_pkthdr.len)+pad+4);
sc->ep_if.if_flags |= IFF_OACTIVE;
splx(s);
return;
}
IF_DEQUEUE(&sc->ep_if.if_snd, m);
if (m == 0) { /* Could make this go away. */
splx(s);
return;
}
outw(BASE+EP_COMMAND, SET_TX_START_THRESH |
(m->m_pkthdr.len/4 + sc->tx_start_thresh));
outw(BASE+EP_W1_TX_PIO_WR_1, m->m_pkthdr.len);
outw(BASE+EP_W1_TX_PIO_WR_1, 0xffff); /* Second dword meaningless */
for(top = m; m != 0; m = m->m_next) {
outsw(BASE+EP_W1_TX_PIO_WR_1, mtod(m, caddr_t), m->m_len/2);
if(m->m_len & 1)
outb(BASE+EP_W1_TX_PIO_WR_1,
*(mtod(m, caddr_t)+m->m_len-1));
}
while (pad--)
outb(BASE+EP_W1_TX_PIO_WR_1, 0); /* Padding */
#if NBPFILTER > 0
if (sc->bpf) {
u_short etype;
int off, datasize, resid;
struct ether_header *eh;
struct trailer_header {
u_short ether_type;
u_short ether_residual;
} trailer_header;
char ether_packet[ETHER_MAX_LEN];
char *ep;
ep = ether_packet;
/*
* We handle trailers below:
* Copy ether header first, then residual data,
* then data. Put all this in a temporary buffer
* 'ether_packet' and send off to bpf. Since the
* system has generated this packet, we assume
* that all of the offsets in the packet are
* correct; if they're not, the system will almost
* certainly crash in m_copydata.
* We make no assumptions about how the data is
* arranged in the mbuf chain (i.e. how much
* data is in each mbuf, if mbuf clusters are
* used, etc.), which is why we use m_copydata
* to get the ether header rather than assume
* that this is located in the first mbuf.
*/
/* copy ether header */
m_copydata(top, 0, sizeof(struct ether_header), ep);
eh = (struct ether_header *) ep;
ep += sizeof(struct ether_header);
etype = ntohs(eh->ether_type);
if (etype >= ETHERTYPE_TRAIL &&
etype < ETHERTYPE_TRAIL+ETHERTYPE_NTRAILER) {
datasize = ((etype - ETHERTYPE_TRAIL) << 9);
off = datasize + sizeof(struct ether_header);
/* copy trailer_header into a data structure */
m_copydata(top, off, sizeof(struct trailer_header),
&trailer_header.ether_type);
/* copy residual data */
resid = trailer_header.ether_residual -
sizeof(struct trailer_header);
resid = ntohs(resid);
m_copydata(top, off+sizeof(struct trailer_header),
resid, ep);
ep += resid;
/* copy data */
m_copydata(top, sizeof(struct ether_header),
datasize, ep);
ep += datasize;
/* restore original ether packet type */
eh->ether_type = trailer_header.ether_type;
bpf_tap(sc->bpf, ether_packet, ep - ether_packet);
} else
bpf_mtap(sc->bpf, top);
}
#endif
m_freem(top);
++sc->ep_if.if_opackets;
/*
* Is another packet coming in? We don't want to overflow the
* tiny RX fifo.
*/
if (inw(BASE+EP_W1_RX_STATUS) & RX_BYTES_MASK) {
splx(s);
return;
}
goto startagain;
}
epintr(unit)
int unit;
{
int status, i;
register struct ep_softc *sc = &ep_softc[unit];
struct ifnet *ifp = &sc->ep_if;
struct mbuf *m;
status=0;
checkintr:
status = inw(BASE + EP_STATUS) & (S_TX_COMPLETE|S_TX_AVAIL|S_RX_COMPLETE|S_CARD_FAILURE);
if (status == 0) { /* No interrupts. */
outw(BASE+EP_COMMAND, C_INTR_LATCH);
return;
}
outw(BASE+EP_COMMAND, ACK_INTR | status); /* important that we do this first. */
if (status & S_TX_AVAIL) {
status &= ~S_TX_AVAIL;
inw(BASE+EP_W1_FREE_TX);
sc->ep_if.if_flags &= ~IFF_OACTIVE;
epstart(&sc->ep_if);
}
if (status & S_RX_COMPLETE) {
status &= ~S_RX_COMPLETE;
epread(sc);
}
if (status & S_CARD_FAILURE) {
printf("ep%d: reset (status: %x)\n", unit, status);
outw(BASE+EP_COMMAND, C_INTR_LATCH);
epinit(unit);
return;
}
if (status & S_TX_COMPLETE) {
status &= ~S_TX_COMPLETE;
/*
* We need to read TX_STATUS until we get a 0 status in
* order to turn off the interrupt flag.
*/
while ((i=inb(BASE+EP_W1_TX_STATUS)) & TXS_COMPLETE) {
outw(BASE+EP_W1_TX_STATUS, 0x0);
if (i & (TXS_MAX_COLLISION|TXS_JABBER|TXS_UNDERRUN)) {
if (i & TXS_MAX_COLLISION)
++sc->ep_if.if_collisions;
if (i & (TXS_JABBER|TXS_UNDERRUN)) {
outw(BASE+EP_COMMAND, TX_RESET);
if(i & TXS_UNDERRUN) {
if (sc->tx_start_thresh < ETHER_MAX_LEN) {
sc->tx_start_thresh += 20;
outw(BASE+EP_COMMAND,
SET_TX_START_THRESH |
sc->tx_start_thresh);
}
}
}
outw(BASE+EP_COMMAND, TX_ENABLE);
++sc->ep_if.if_oerrors;
}
}
epstart(ifp);
}
goto checkintr;
}
epread(sc)
register struct ep_softc *sc;
{
struct ether_header *eh;
struct mbuf *mcur, *m, *m0, *top;
int totlen, lenthisone;
int save_totlen;
u_short etype;
int off, resid;
int count, spinwait;
int i;
totlen = inw(BASE + EP_W1_RX_STATUS);
off = 0;
top = 0;
if (totlen & ERR_RX) {
++sc->ep_if.if_ierrors;
goto out;
}
save_totlen = totlen &= RX_BYTES_MASK; /* Lower 10 bits = RX bytes. */
m = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (m == 0) {
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == 0)
goto out;
} else { /* Convert one of our saved mbuf's */
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
m->m_data = m->m_pktdat;
m->m_flags = M_PKTHDR;
}
top = m0 = m; /* We assign top so we can "goto out" */
# define EROUND ((sizeof(struct ether_header) + 3) & ~3)
# define EOFF (EROUND - sizeof(struct ether_header))
m0->m_data += EOFF;
/* Read what should be the header. */
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m0, caddr_t), sizeof(struct ether_header)/2);
m->m_len = sizeof(struct ether_header);
totlen -= sizeof(struct ether_header);
/*
* mostly deal with trailer here. (untested)
* We do this in a couple of parts. First we check for a trailer, if
* we have one we convert the mbuf back to a regular mbuf and set the offset and
* subtract sizeof(struct ether_header) from the pktlen.
* After we've read the packet off the interface (all except for the trailer
* header, we then get a header mbuf, read the trailer into it, and fix up
* the mbuf pointer chain.
*/
eh=mtod(m, struct ether_header *);
eh->ether_type = ntohs((u_short)eh->ether_type);
if (eh->ether_type >= ETHERTYPE_TRAIL &&
eh->ether_type < ETHERTYPE_TRAIL+ETHERTYPE_NTRAILER) {
m->m_data = m->m_dat; /* Convert back to regular mbuf. */
m->m_flags = 0; /* This sucks but non-trailers are the norm */
off = (eh->ether_type - ETHERTYPE_TRAIL) * 512;
if (off >= ETHERMTU) {
m_freem(m);
return; /* sanity */
}
totlen -= sizeof(struct ether_header); /* We don't read the trailer */
m->m_data += 2 * sizeof(u_short); /* Get rid of type & len*/
}
while (totlen>0) {
lenthisone=min(totlen, M_TRAILINGSPACE(m));
if (lenthisone == 0) { /* no room in this one */
mcur = m;
m = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (!m) {
MGET(m, M_DONTWAIT, MT_DATA);
if (m==0)
goto out;
} else {
timeout(fill_mbuf_queue, sc, 1);
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
}
if (totlen >= MINCLSIZE)
MCLGET(m, M_DONTWAIT);
m->m_len = 0;
mcur->m_next = m;
lenthisone = min(totlen, M_TRAILINGSPACE(m));
}
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m, caddr_t)+m->m_len, lenthisone/2);
m->m_len += lenthisone;
if (lenthisone & 1)
*(mtod(m, caddr_t)+m->m_len-1) = inb(BASE+EP_W1_RX_PIO_RD_1);
totlen -= lenthisone;
}
if (off) {
top = sc->mb[sc->next_mb];
sc->mb[sc->next_mb] = 0;
if (top == 0) {
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (top == 0)
goto out;
} else { /* Convert one of our saved mbuf's */
sc->next_mb = (sc->next_mb + 1) % MAX_MBS;
top->m_data = top->m_pktdat;
top->m_flags = M_PKTHDR;
}
insw(BASE+EP_W1_RX_PIO_RD_1, mtod(m, caddr_t)+m->m_len,
sizeof(struct ether_header));
eh->ether_type = ntohs(eh->ether_type);
top->m_next = m0;
top->m_len = sizeof(struct ether_header);
/* XXX Accomodate for type and len from beginning of trailer data */
top->m_pkthdr.len = save_totlen - (2 * sizeof(u_short));
} else {
top = m0;
top->m_pkthdr.len = save_totlen;
}
top->m_pkthdr.rcvif = &sc->ep_if;
outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
++sc->ep_if.if_ipackets;
m_adj(top, sizeof(struct ether_header));
ether_input(&sc->ep_if, eh, top);
return;
out: outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
if (top)
m_freem(top);
}
/*
* Look familiar?
*/
epioctl(ifp, cmd, data)
register struct ifnet *ifp;
int cmd;
caddr_t data;
{
register struct ifaddr *ifa = (struct ifaddr *)data;
struct ep_softc *sc = &ep_softc[ifp->if_unit];
struct ifreq *ifr = (struct ifreq *)data;
int s, error=0;
switch(cmd){
case SIOCSIFADDR:
ifp->if_flags |= IFF_UP;
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
epinit(ifp->if_unit); /* before arpwhohas */
((struct arpcom *)ifp)->ac_ipaddr = IA_SIN(ifa)->sin_addr;
arpwhohas((struct arpcom *)ifp, &IA_SIN(ifa)->sin_addr);
break;
#endif
#ifdef NS
case AF_NS:
{
register struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr);
if (ns_nullhost(*ina))
ina->x_host = *(union ns_host *)(sc->ep_ac.ac_enaddr);
else {
ifp->if_flags &= ~IFF_RUNNING;
bcopy((caddr_t)ina->x_host.c_host,
(caddr_t)sc->ns_addr,
sizeof(sc->ep_ac.ac_enaddr));
}
epinit(ifp->if_unit);
break;
}
#endif
default:
epinit(ifp->if_unit);
break;
}
break;
case SIOCSIFFLAGS:
if ((ifp->if_flags & IFF_UP) == 0 && ifp->if_flags & IFF_RUNNING) {
ifp->if_flags &= ~IFF_RUNNING;
epstop(ifp->if_unit);
} else if (ifp->if_flags & IFF_UP && (ifp->if_flags & IFF_RUNNING) == 0)
epinit(ifp->if_unit);
break;
#ifdef notdef
case SIOCGHWADDR:
bcopy((caddr_t)sc->sc_addr, (caddr_t) &ifr->ifr_data, sizeof(sc->sc_addr));
break;
#endif
default:
error = EINVAL;
}
return (error);
}
epreset(unit)
int unit;
{
int s;
epstop(unit);
epinit(unit);
return;
}
epwatchdog(unit)
int unit;
{
return;
}
epstop(unit)
int unit;
{
register struct ep_softc *sc = &ep_softc[unit];
outw(BASE+EP_COMMAND, RX_DISABLE);
outw(BASE+EP_COMMAND, RX_DISCARD_TOP_PACK);
while (inb(BASE+EP_STATUS) & S_COMMAND_IN_PROGRESS)
;
outw(BASE+EP_COMMAND, TX_DISABLE);
outw(BASE+EP_COMMAND, STOP_TRANSCEIVER);
outw(BASE+EP_COMMAND, RX_RESET);
outw(BASE+EP_COMMAND, TX_RESET);
outw(BASE+EP_COMMAND, C_INTR_LATCH);
outw(BASE+EP_COMMAND, SET_RD_0_MASK);
outw(BASE+EP_COMMAND, SET_INTR_MASK);
outw(BASE+EP_COMMAND, SET_RX_FILTER);
return;
}
/*
* This is adapted straight from the book. There's probably a better way.
*/
send_ID_sequence(port)
u_short port;
{
char cx, al;
cx=0x0ff;
al=0x0ff;
outb(port, 0x0);
DELAY(1000);
outb(port, 0x0);
DELAY(1000);
loop1: cx--;
outb(port, al);
if (!(al & 0x80)) {
al=al<<1;
goto loop1;
}
al=al<<1;
al^=0xcf;
if (cx)
goto loop1;
}
/*
* We get eeprom data from the id_port given an offset into the
* eeprom. Basically; after the ID_sequence is sent to all of
* the cards; they enter the ID_CMD state where they will accept
* command requests. 0x80-0xbf loads the eeprom data. We then
* read the port 16 times and with every read; the cards check
* for contention (ie: if one card writes a 0 bit and another
* writes a 1 bit then the host sees a 0. At the end of the cycle;
* each card compares the data on the bus; if there is a difference
* then that card goes into ID_WAIT state again). In the meantime;
* one bit of data is returned in the AX register which is conveniently
* returned to us by inb(). Hence; we read 16 times getting one
* bit of data with each read.
*/
u_short get_eeprom_data(id_port, offset)
int id_port;
int offset;
{
int i, data=0;
outb(id_port, 0x80+offset);
DELAY(1000);
for (i=0; i<16; i++)
data = (data<<1) | (inw(id_port) & 1);
return(data);
}
int
is_eeprom_busy(is)
struct isa_device *is;
{
int i=0, j;
register struct ep_softc *sc = &ep_softc[is->id_unit];
while (i++<100) {
j=inw(BASE+EP_W0_EEPROM_COMMAND);
if (j & EEPROM_BUSY)
DELAY(100);
else
break;
}
if (i>=100) {
printf("\nep%d: eeprom failed to come ready.\n", is->id_unit);
return(1);
}
if (j & EEPROM_TST_MODE) {
printf("\nep%d: 3c509 in test mode. Erase pencil mark!\n", is->id_unit);
return(1);
}
return(0);
}
int
fill_mbuf_queue(sc)
struct ep_softc *sc;
{
int i=0;
if (sc->mb[sc->last_mb])
return;
i=sc->last_mb;
do {
MGET(sc->mb[i], M_DONTWAIT, MT_DATA);
if (!sc->mb[i])
break;
i = (i+1) % MAX_MBS;
} while(i != sc->next_mb);
sc->last_mb = i;
}
#endif /* NEP > 0 */

295
sys/i386/isa/if_epreg.h Normal file
View File

@ -0,0 +1,295 @@
/*
* Copyright (c) 1993 Herb Peyerl (hpeyerl@novatel.ca)
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software withough specific prior written permission
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* $Id: if_epreg.h,v 1.1 1993/12/14 04:26:47 hpeyerl Exp $
*/
/**************************************************************************
* *
* These define the EEPROM data structure. They are used in the probe
* function to verify the existance of the adapter after having sent
* the ID_Sequence.
*
* There are others but only the ones we use are defined here.
*
**************************************************************************/
#define EEPROM_NODE_ADDR_0 0x0 /* Word */
#define EEPROM_NODE_ADDR_1 0x1 /* Word */
#define EEPROM_NODE_ADDR_2 0x2 /* Word */
#define EEPROM_PROD_ID 0x3 /* 0x9[0-f]50 */
#define EEPROM_MFG_ID 0x7 /* 0x6d50 */
#define EEPROM_ADDR_CFG 0x8 /* Base addr */
#define EEPROM_RESOURCE_CFG 0x9 /* IRQ. Bits 12-15 */
/**************************************************************************
* *
* These are the registers for the 3Com 3c509 and their bit patterns when *
* applicable. They have been taken out the the "EtherLink III Parallel *
* Tasking EISA and ISA Technical Reference" "Beta Draft 10/30/92" manual *
* from 3com. *
* *
**************************************************************************/
#define EP_COMMAND 0x0e /* Write. BASE+0x0e is always a command reg. */
#define EP_STATUS 0x0e /* Read. BASE+0x0e is always status reg. */
#define EP_WINDOW 0x0f /* Read. BASE+0x0f is always window reg. */
/*
* Window 0 registers. Setup.
*/
/* Write */
#define EP_W0_EEPROM_DATA 0x0c
#define EP_W0_EEPROM_COMMAND 0x0a
#define EP_W0_RESOURCE_CFG 0x08
#define EP_W0_ADDRESS_CFG 0x06
#define EP_W0_CONFIG_CTRL 0x04
/* Read */
#define EP_W0_PRODUCT_ID 0x02
#define EP_W0_MFG_ID 0x00
/*
* Window 1 registers. Operating Set.
*/
/* Write */
#define EP_W1_TX_PIO_WR_2 0x02
#define EP_W1_TX_PIO_WR_1 0x00
/* Read */
#define EP_W1_FREE_TX 0x0c
#define EP_W1_TX_STATUS 0x0b /* byte */
#define EP_W1_TIMER 0x0a /* byte */
#define EP_W1_RX_STATUS 0x08
#define EP_W1_RX_PIO_RD_2 0x02
#define EP_W1_RX_PIO_RD_1 0x00
/*
* Window 2 registers. Station Address Setup/Read
*/
/* Read/Write */
#define EP_W2_ADDR_5 0x05
#define EP_W2_ADDR_4 0x04
#define EP_W2_ADDR_3 0x03
#define EP_W2_ADDR_2 0x02
#define EP_W2_ADDR_1 0x01
#define EP_W2_ADDR_0 0x00
/*
* Window 3 registers. FIFO Management.
*/
/* Read */
#define EP_W3_FREE_TX 0x0c
#define EP_W3_FREE_RX 0x0a
/*
* Window 4 registers. Diagnostics.
*/
/* Read/Write */
#define EP_W4_MEDIA_TYPE 0x0a
#define EP_W4_CTRLR_STATUS 0x08
#define EP_W4_NET_DIAG 0x06
#define EP_W4_FIFO_DIAG 0x04
#define EP_W4_HOST_DIAG 0x02
#define EP_W4_TX_DIAG 0x00
/*
* Window 5 Registers. Results and Internal status.
*/
/* Read */
#define EP_W5_READ_0_MASK 0x0c
#define EP_W5_INTR_MASK 0x0a
#define EP_W5_RX_FILTER 0x08
#define EP_W5_RX_EARLY_THRESH 0x06
#define EP_W5_TX_AVAIL_THRESH 0x02
#define EP_W5_TX_START_THRESH 0x00
/*
* Window 6 registers. Statistics.
*/
/* Read/Write */
#define TX_TOTAL_OK 0x0c
#define RX_TOTAL_OK 0x0a
#define TX_DEFERRALS 0x08
#define RX_FRAMES_OK 0x07
#define TX_FRAMES_OK 0x06
#define RX_OVERRUNS 0x05
#define TX_COLLISIONS 0x04
#define TX_AFTER_1_COLLISION 0x03
#define TX_AFTER_X_COLLISIONS 0x02
#define TX_NO_SQE 0x01
#define TX_CD_LOST 0x00
/****************************************
*
* Register definitions.
*
****************************************/
/*
* Command register. All windows.
*
* 16 bit register.
* 15-11: 5-bit code for command to be executed.
* 10-0: 11-bit arg if any. For commands with no args;
* this can be set to anything.
*/
#define GLOBAL_RESET (u_short) 0x0000 /* Wait at least 1ms after issuing */
#define WINDOW_SELECT (u_short) (0x1<<11)
#define START_TRANSCEIVER (u_short) (0x2<<11) /* Read ADDR_CFG reg to determine
whether this is needed. If so;
wait 800 uSec before using trans-
ceiver. */
#define RX_DISABLE (u_short) (0x3<<11) /* state disabled on power-up */
#define RX_ENABLE (u_short) (0x4<<11)
#define RX_RESET (u_short) (0x5<<11)
#define RX_DISCARD_TOP_PACK (u_short) (0x8<<11)
#define TX_ENABLE (u_short) (0x9<<11)
#define TX_DISABLE (u_short) (0xa<<11)
#define TX_RESET (u_short) (0xb<<11)
#define REQ_INTR (u_short) (0xc<<11)
/*
* The following C_* acknowledge the various interrupts.
* Some of them don't do anything. See the manual.
*/
#define ACK_INTR (u_short) (0x6800)
# define C_INTR_LATCH (u_short) (ACK_INTR|0x1)
# define C_CARD_FAILURE (u_short) (ACK_INTR|0x2)
# define C_TX_COMPLETE (u_short) (ACK_INTR|0x4)
# define C_TX_AVAIL (u_short) (ACK_INTR|0x8)
# define C_RX_COMPLETE (u_short) (ACK_INTR|0x10)
# define C_RX_EARLY (u_short) (ACK_INTR|0x20)
# define C_INT_RQD (u_short) (ACK_INTR|0x40)
# define C_UPD_STATS (u_short) (ACK_INTR|0x80)
#define SET_INTR_MASK (u_short) (0xe<<11)
#define SET_RD_0_MASK (u_short) (0xf<<11)
#define SET_RX_FILTER (u_short) (0x10<<11)
# define FIL_INDIVIDUAL (u_short) (0x1)
# define FIL_GROUP (u_short) (0x2)
# define FIL_BRDCST (u_short) (0x4)
# define FIL_ALL (u_short) (0x8)
#define SET_RX_EARLY_THRESH (u_short) (0x11<<11)
#define SET_TX_AVAIL_THRESH (u_short) (0x12<<11)
#define SET_TX_START_THRESH (u_short) (0x13<<11)
#define STATS_ENABLE (u_short) (0x15<<11)
#define STATS_DISABLE (u_short) (0x16<<11)
#define STOP_TRANSCEIVER (u_short) (0x17<<11)
/*
* Status register. All windows.
*
* 15-13: Window number(0-7).
* 12: Command_in_progress.
* 11: reserved.
* 10: reserved.
* 9: reserved.
* 8: reserved.
* 7: Update Statistics.
* 6: Interrupt Requested.
* 5: RX Early.
* 4: RX Complete.
* 3: TX Available.
* 2: TX Complete.
* 1: Adapter Failure.
* 0: Interrupt Latch.
*/
#define S_INTR_LATCH (u_short) (0x1)
#define S_CARD_FAILURE (u_short) (0x2)
#define S_TX_COMPLETE (u_short) (0x4)
#define S_TX_AVAIL (u_short) (0x8)
#define S_RX_COMPLETE (u_short) (0x10)
#define S_RX_EARLY (u_short) (0x20)
#define S_INT_RQD (u_short) (0x40)
#define S_UPD_STATS (u_short) (0x80)
#define S_COMMAND_IN_PROGRESS (u_short) (0x1000)
/*
* FIFO Registers. RX Status.
*
* 15: Incomplete or FIFO empty.
* 14: 1: Error in RX Packet 0: Incomplete or no error.
* 13-11: Type of error.
* 1000 = Overrun.
* 1011 = Run Packet Error.
* 1100 = Alignment Error.
* 1101 = CRC Error.
* 1001 = Oversize Packet Error (>1514 bytes)
* 0010 = Dribble Bits.
* (all other error codes, no errors.)
*
* 10-0: RX Bytes (0-1514)
*/
#define ERR_INCOMPLETE (u_short) (0x8000)
#define ERR_RX (u_short) (0x4000)
#define ERR_RX_PACKET (u_short) (0x2000)
#define ERR_OVERRUN (u_short) (0x1000)
#define ERR_RUNT (u_short) (0x1300)
#define ERR_ALIGNMENT (u_short) (0x1400)
#define ERR_CRC (u_short) (0x1500)
#define ERR_OVERSIZE (u_short) (0x1100)
#define ERR_DRIBBLE (u_short) (0x200)
/*
* TX Status
*
* Reports the transmit status of a completed transmission. Writing this
* register pops the transmit completion stack.
*
* Window 1/Port 0x0b.
*
* 7: Complete
* 6: Interrupt on successful transmission requested.
* 5: Jabber Error (TP Only, TX Reset required. )
* 4: Underrun (TX Reset required. )
* 3: Maximum Collisions.
* 2: TX Status Overflow.
* 1-0: Undefined.
*
*/
#define TXS_COMPLETE 0x80
#define TXS_INTR_REQ 0x40
#define TXS_JABBER 0x20
#define TXS_UNDERRUN 0x10
#define TXS_MAX_COLLISION 0x8
#define TXS_STATUS_OVERFLOW 0x4
/*
* Misc defines for various things.
*/
#define TAG_ADAPTER_0 0xd0
#define ACTIVATE_ADAPTER_TO_CONFIG 0xff
#define ENABLE_DRQ_IRQ 0x0001
#define MFG_ID 0x6d50
#define PROD_ID 0x9150
#define BASE sc->ep_io_addr
#define GO_WINDOW(x) outw(BASE+EP_COMMAND, WINDOW_SELECT|x)
#define AUI 0x1
#define BNC 0x2
#define UTP 0x4
#define IS_AUI (1<<13)
#define IS_BNC (1<<12)
#define IS_UTP (1<<9)
#define EEPROM_BUSY (1<<15)
#define EEPROM_TST_MODE (1<<14)
#define READ_EEPROM (1<<7)
#define ETHER_ADDR_LEN 6
#define ETHER_MAX 1536
#define ENABLE_UTP 0xc0
#define DISABLE_UTP 0x0
#define RX_BYTES_MASK (u_short) (0x07ff)