NetBSD/sys/dev/ic/aic6915.c

1429 lines
35 KiB
C

/* $NetBSD: aic6915.c,v 1.44 2020/03/15 22:19:00 thorpej Exp $ */
/*-
* Copyright (c) 2001 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Jason R. Thorpe.
*
* 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. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``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 FOUNDATION OR CONTRIBUTORS
* 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.
*/
/*
* Device driver for the Adaptec AIC-6915 (``Starfire'')
* 10/100 Ethernet controller.
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: aic6915.c,v 1.44 2020/03/15 22:19:00 thorpej Exp $");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/callout.h>
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/device.h>
#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_media.h>
#include <net/if_ether.h>
#include <net/bpf.h>
#include <sys/bus.h>
#include <sys/intr.h>
#include <dev/mii/miivar.h>
#include <dev/ic/aic6915reg.h>
#include <dev/ic/aic6915var.h>
static void sf_start(struct ifnet *);
static void sf_watchdog(struct ifnet *);
static int sf_ioctl(struct ifnet *, u_long, void *);
static int sf_init(struct ifnet *);
static void sf_stop(struct ifnet *, int);
static bool sf_shutdown(device_t, int);
static void sf_txintr(struct sf_softc *);
static void sf_rxintr(struct sf_softc *);
static void sf_stats_update(struct sf_softc *);
static void sf_reset(struct sf_softc *);
static void sf_macreset(struct sf_softc *);
static void sf_rxdrain(struct sf_softc *);
static int sf_add_rxbuf(struct sf_softc *, int);
static uint8_t sf_read_eeprom(struct sf_softc *, int);
static void sf_set_filter(struct sf_softc *);
static int sf_mii_read(device_t, int, int, uint16_t *);
static int sf_mii_write(device_t, int, int, uint16_t);
static void sf_mii_statchg(struct ifnet *);
static void sf_tick(void *);
#define sf_funcreg_read(sc, reg) \
bus_space_read_4((sc)->sc_st, (sc)->sc_sh_func, (reg))
#define sf_funcreg_write(sc, reg, val) \
bus_space_write_4((sc)->sc_st, (sc)->sc_sh_func, (reg), (val))
static inline uint32_t
sf_reg_read(struct sf_softc *sc, bus_addr_t reg)
{
if (__predict_false(sc->sc_iomapped)) {
bus_space_write_4(sc->sc_st, sc->sc_sh, SF_IndirectIoAccess,
reg);
return (bus_space_read_4(sc->sc_st, sc->sc_sh,
SF_IndirectIoDataPort));
}
return (bus_space_read_4(sc->sc_st, sc->sc_sh, reg));
}
static inline void
sf_reg_write(struct sf_softc *sc, bus_addr_t reg, uint32_t val)
{
if (__predict_false(sc->sc_iomapped)) {
bus_space_write_4(sc->sc_st, sc->sc_sh, SF_IndirectIoAccess,
reg);
bus_space_write_4(sc->sc_st, sc->sc_sh, SF_IndirectIoDataPort,
val);
return;
}
bus_space_write_4(sc->sc_st, sc->sc_sh, reg, val);
}
#define sf_genreg_read(sc, reg) \
sf_reg_read((sc), (reg) + SF_GENREG_OFFSET)
#define sf_genreg_write(sc, reg, val) \
sf_reg_write((sc), (reg) + SF_GENREG_OFFSET, (val))
/*
* sf_attach:
*
* Attach a Starfire interface to the system.
*/
void
sf_attach(struct sf_softc *sc)
{
struct ifnet *ifp = &sc->sc_ethercom.ec_if;
struct mii_data * const mii = &sc->sc_mii;
int i, rseg, error;
bus_dma_segment_t seg;
uint8_t enaddr[ETHER_ADDR_LEN];
callout_init(&sc->sc_tick_callout, 0);
callout_setfunc(&sc->sc_tick_callout, sf_tick, sc);
/*
* If we're I/O mapped, the functional register handle is
* the same as the base handle. If we're memory mapped,
* carve off a chunk of the register space for the functional
* registers, to save on arithmetic later.
*/
if (sc->sc_iomapped)
sc->sc_sh_func = sc->sc_sh;
else {
if ((error = bus_space_subregion(sc->sc_st, sc->sc_sh,
SF_GENREG_OFFSET, SF_FUNCREG_SIZE, &sc->sc_sh_func)) != 0) {
aprint_error_dev(sc->sc_dev, "unable to sub-region "
"functional registers, error = %d\n", error);
return;
}
}
/*
* Initialize the transmit threshold for this interface. The
* manual describes the default as 4 * 16 bytes. We start out
* at 10 * 16 bytes, to avoid a bunch of initial underruns on
* several platforms.
*/
sc->sc_txthresh = 10;
/*
* Allocate the control data structures, and create and load the
* DMA map for it.
*/
if ((error = bus_dmamem_alloc(sc->sc_dmat,
sizeof(struct sf_control_data), PAGE_SIZE, 0, &seg, 1, &rseg,
BUS_DMA_NOWAIT)) != 0) {
aprint_error_dev(sc->sc_dev,
"unable to allocate control data, error = %d\n", error);
goto fail_0;
}
if ((error = bus_dmamem_map(sc->sc_dmat, &seg, rseg,
sizeof(struct sf_control_data), (void **)&sc->sc_control_data,
BUS_DMA_NOWAIT | BUS_DMA_COHERENT)) != 0) {
aprint_error_dev(sc->sc_dev,
"unable to map control data, error = %d\n", error);
goto fail_1;
}
if ((error = bus_dmamap_create(sc->sc_dmat,
sizeof(struct sf_control_data), 1,
sizeof(struct sf_control_data), 0, BUS_DMA_NOWAIT,
&sc->sc_cddmamap)) != 0) {
aprint_error_dev(sc->sc_dev, "unable to create control data "
"DMA map, error = %d\n", error);
goto fail_2;
}
if ((error = bus_dmamap_load(sc->sc_dmat, sc->sc_cddmamap,
sc->sc_control_data, sizeof(struct sf_control_data), NULL,
BUS_DMA_NOWAIT)) != 0) {
aprint_error_dev(sc->sc_dev, "unable to load control data "
"DMA map, error = %d\n", error);
goto fail_3;
}
/*
* Create the transmit buffer DMA maps.
*/
for (i = 0; i < SF_NTXDESC; i++) {
if ((error = bus_dmamap_create(sc->sc_dmat, MCLBYTES,
SF_NTXFRAGS, MCLBYTES, 0, BUS_DMA_NOWAIT,
&sc->sc_txsoft[i].ds_dmamap)) != 0) {
aprint_error_dev(sc->sc_dev,
"unable to create tx DMA map %d, error = %d\n", i,
error);
goto fail_4;
}
}
/*
* Create the receive buffer DMA maps.
*/
for (i = 0; i < SF_NRXDESC; i++) {
if ((error = bus_dmamap_create(sc->sc_dmat, MCLBYTES, 1,
MCLBYTES, 0, BUS_DMA_NOWAIT,
&sc->sc_rxsoft[i].ds_dmamap)) != 0) {
aprint_error_dev(sc->sc_dev,
"unable to create rx DMA map %d, error = %d\n", i,
error);
goto fail_5;
}
}
/*
* Reset the chip to a known state.
*/
sf_reset(sc);
/*
* Read the Ethernet address from the EEPROM.
*/
for (i = 0; i < ETHER_ADDR_LEN; i++)
enaddr[i] = sf_read_eeprom(sc, (15 + (ETHER_ADDR_LEN - 1)) - i);
printf("%s: Ethernet address %s\n", device_xname(sc->sc_dev),
ether_sprintf(enaddr));
if (sf_funcreg_read(sc, SF_PciDeviceConfig) & PDC_System64)
printf("%s: 64-bit PCI slot detected\n",
device_xname(sc->sc_dev));
/*
* Initialize our media structures and probe the MII.
*/
mii->mii_ifp = ifp;
mii->mii_readreg = sf_mii_read;
mii->mii_writereg = sf_mii_write;
mii->mii_statchg = sf_mii_statchg;
sc->sc_ethercom.ec_mii = mii;
ifmedia_init(&mii->mii_media, IFM_IMASK, ether_mediachange,
ether_mediastatus);
mii_attach(sc->sc_dev, mii, 0xffffffff, MII_PHY_ANY,
MII_OFFSET_ANY, 0);
if (LIST_FIRST(&mii->mii_phys) == NULL) {
ifmedia_add(&mii->mii_media, IFM_ETHER | IFM_NONE, 0, NULL);
ifmedia_set(&mii->mii_media, IFM_ETHER | IFM_NONE);
} else
ifmedia_set(&mii->mii_media, IFM_ETHER | IFM_AUTO);
strlcpy(ifp->if_xname, device_xname(sc->sc_dev), IFNAMSIZ);
ifp->if_softc = sc;
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
ifp->if_ioctl = sf_ioctl;
ifp->if_start = sf_start;
ifp->if_watchdog = sf_watchdog;
ifp->if_init = sf_init;
ifp->if_stop = sf_stop;
IFQ_SET_READY(&ifp->if_snd);
/*
* Attach the interface.
*/
if_attach(ifp);
if_deferred_start_init(ifp, NULL);
ether_ifattach(ifp, enaddr);
/*
* Make sure the interface is shutdown during reboot.
*/
if (pmf_device_register1(sc->sc_dev, NULL, NULL, sf_shutdown))
pmf_class_network_register(sc->sc_dev, ifp);
else
aprint_error_dev(sc->sc_dev,
"couldn't establish power handler\n");
return;
/*
* Free any resources we've allocated during the failed attach
* attempt. Do this in reverse order an fall through.
*/
fail_5:
for (i = 0; i < SF_NRXDESC; i++) {
if (sc->sc_rxsoft[i].ds_dmamap != NULL)
bus_dmamap_destroy(sc->sc_dmat,
sc->sc_rxsoft[i].ds_dmamap);
}
fail_4:
for (i = 0; i < SF_NTXDESC; i++) {
if (sc->sc_txsoft[i].ds_dmamap != NULL)
bus_dmamap_destroy(sc->sc_dmat,
sc->sc_txsoft[i].ds_dmamap);
}
bus_dmamap_unload(sc->sc_dmat, sc->sc_cddmamap);
fail_3:
bus_dmamap_destroy(sc->sc_dmat, sc->sc_cddmamap);
fail_2:
bus_dmamem_unmap(sc->sc_dmat, (void *) sc->sc_control_data,
sizeof(struct sf_control_data));
fail_1:
bus_dmamem_free(sc->sc_dmat, &seg, rseg);
fail_0:
return;
}
/*
* sf_shutdown:
*
* Shutdown hook -- make sure the interface is stopped at reboot.
*/
static bool
sf_shutdown(device_t self, int howto)
{
struct sf_softc *sc;
sc = device_private(self);
sf_stop(&sc->sc_ethercom.ec_if, 1);
return true;
}
/*
* sf_start: [ifnet interface function]
*
* Start packet transmission on the interface.
*/
static void
sf_start(struct ifnet *ifp)
{
struct sf_softc *sc = ifp->if_softc;
struct mbuf *m0, *m;
struct sf_txdesc0 *txd;
struct sf_descsoft *ds;
bus_dmamap_t dmamap;
int error, producer, last = -1, opending, seg;
/*
* Remember the previous number of pending transmits.
*/
opending = sc->sc_txpending;
/*
* Find out where we're sitting.
*/
producer = SF_TXDINDEX_TO_HOST(
TDQPI_HiPrTxProducerIndex_get(
sf_funcreg_read(sc, SF_TxDescQueueProducerIndex)));
/*
* Loop through the send queue, setting up transmit descriptors
* until we drain the queue, or use up all available transmit
* descriptors. Leave a blank one at the end for sanity's sake.
*/
while (sc->sc_txpending < (SF_NTXDESC - 1)) {
/*
* Grab a packet off the queue.
*/
IFQ_POLL(&ifp->if_snd, m0);
if (m0 == NULL)
break;
m = NULL;
/*
* Get the transmit descriptor.
*/
txd = &sc->sc_txdescs[producer];
ds = &sc->sc_txsoft[producer];
dmamap = ds->ds_dmamap;
/*
* Load the DMA map. If this fails, the packet either
* didn't fit in the allotted number of frags, or we were
* short on resources. In this case, we'll copy and try
* again.
*/
if (bus_dmamap_load_mbuf(sc->sc_dmat, dmamap, m0,
BUS_DMA_WRITE | BUS_DMA_NOWAIT) != 0) {
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL) {
aprint_error_dev(sc->sc_dev,
"unable to allocate Tx mbuf\n");
break;
}
MCLAIM(m, &sc->sc_ethercom.ec_tx_mowner);
if (m0->m_pkthdr.len > MHLEN) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
aprint_error_dev(sc->sc_dev,
"unable to allocate Tx cluster\n");
m_freem(m);
break;
}
}
m_copydata(m0, 0, m0->m_pkthdr.len, mtod(m, void *));
m->m_pkthdr.len = m->m_len = m0->m_pkthdr.len;
error = bus_dmamap_load_mbuf(sc->sc_dmat, dmamap,
m, BUS_DMA_WRITE | BUS_DMA_NOWAIT);
if (error) {
aprint_error_dev(sc->sc_dev,
"unable to load Tx buffer, error = %d\n",
error);
break;
}
}
/*
* WE ARE NOW COMMITTED TO TRANSMITTING THE PACKET.
*/
IFQ_DEQUEUE(&ifp->if_snd, m0);
if (m != NULL) {
m_freem(m0);
m0 = m;
}
/* Initialize the descriptor. */
txd->td_word0 =
htole32(TD_W0_ID | TD_W0_CRCEN | m0->m_pkthdr.len);
if (producer == (SF_NTXDESC - 1))
txd->td_word0 |= TD_W0_END;
txd->td_word1 = htole32(dmamap->dm_nsegs);
for (seg = 0; seg < dmamap->dm_nsegs; seg++) {
txd->td_frags[seg].fr_addr =
htole32(dmamap->dm_segs[seg].ds_addr);
txd->td_frags[seg].fr_len =
htole32(dmamap->dm_segs[seg].ds_len);
}
/* Sync the descriptor and the DMA map. */
SF_CDTXDSYNC(sc, producer, BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(sc->sc_dmat, dmamap, 0, dmamap->dm_mapsize,
BUS_DMASYNC_PREWRITE);
/*
* Store a pointer to the packet so we can free it later.
*/
ds->ds_mbuf = m0;
/* Advance the Tx pointer. */
sc->sc_txpending++;
last = producer;
producer = SF_NEXTTX(producer);
/*
* Pass the packet to any BPF listeners.
*/
bpf_mtap(ifp, m0, BPF_D_OUT);
}
if (sc->sc_txpending != opending) {
KASSERT(last != -1);
/*
* We enqueued packets. Cause a transmit interrupt to
* happen on the last packet we enqueued, and give the
* new descriptors to the chip by writing the new
* producer index.
*/
sc->sc_txdescs[last].td_word0 |= TD_W0_INTR;
SF_CDTXDSYNC(sc, last, BUS_DMASYNC_PREWRITE);
sf_funcreg_write(sc, SF_TxDescQueueProducerIndex,
TDQPI_HiPrTxProducerIndex(SF_TXDINDEX_TO_CHIP(producer)));
/* Set a watchdog timer in case the chip flakes out. */
ifp->if_timer = 5;
}
}
/*
* sf_watchdog: [ifnet interface function]
*
* Watchdog timer handler.
*/
static void
sf_watchdog(struct ifnet *ifp)
{
struct sf_softc *sc = ifp->if_softc;
printf("%s: device timeout\n", device_xname(sc->sc_dev));
if_statinc(ifp, if_oerrors);
(void) sf_init(ifp);
/* Try to get more packets going. */
sf_start(ifp);
}
/*
* sf_ioctl: [ifnet interface function]
*
* Handle control requests from the operator.
*/
static int
sf_ioctl(struct ifnet *ifp, u_long cmd, void *data)
{
struct sf_softc *sc = ifp->if_softc;
int s, error;
s = splnet();
error = ether_ioctl(ifp, cmd, data);
if (error == ENETRESET) {
/*
* Multicast list has changed; set the hardware filter
* accordingly.
*/
if (ifp->if_flags & IFF_RUNNING)
sf_set_filter(sc);
error = 0;
}
/* Try to get more packets going. */
sf_start(ifp);
splx(s);
return (error);
}
/*
* sf_intr:
*
* Interrupt service routine.
*/
int
sf_intr(void *arg)
{
struct sf_softc *sc = arg;
uint32_t isr;
int handled = 0, wantinit = 0;
for (;;) {
/* Reading clears all interrupts we're interested in. */
isr = sf_funcreg_read(sc, SF_InterruptStatus);
if ((isr & IS_PCIPadInt) == 0)
break;
handled = 1;
/* Handle receive interrupts. */
if (isr & IS_RxQ1DoneInt)
sf_rxintr(sc);
/* Handle transmit completion interrupts. */
if (isr & (IS_TxDmaDoneInt | IS_TxQueueDoneInt))
sf_txintr(sc);
/* Handle abnormal interrupts. */
if (isr & IS_AbnormalInterrupt) {
/* Statistics. */
if (isr & IS_StatisticWrapInt)
sf_stats_update(sc);
/* DMA errors. */
if (isr & IS_DmaErrInt) {
wantinit = 1;
aprint_error_dev(sc->sc_dev,
"WARNING: DMA error\n");
}
/* Transmit FIFO underruns. */
if (isr & IS_TxDataLowInt) {
if (sc->sc_txthresh < 0xff)
sc->sc_txthresh++;
printf("%s: transmit FIFO underrun, new "
"threshold: %d bytes\n",
device_xname(sc->sc_dev),
sc->sc_txthresh * 16);
sf_funcreg_write(sc, SF_TransmitFrameCSR,
sc->sc_TransmitFrameCSR |
TFCSR_TransmitThreshold(sc->sc_txthresh));
sf_funcreg_write(sc, SF_TxDescQueueCtrl,
sc->sc_TxDescQueueCtrl |
TDQC_TxHighPriorityFifoThreshold(
sc->sc_txthresh));
}
}
}
if (handled) {
/* Reset the interface, if necessary. */
if (wantinit)
sf_init(&sc->sc_ethercom.ec_if);
/* Try and get more packets going. */
if_schedule_deferred_start(&sc->sc_ethercom.ec_if);
}
return (handled);
}
/*
* sf_txintr:
*
* Helper -- handle transmit completion interrupts.
*/
static void
sf_txintr(struct sf_softc *sc)
{
struct ifnet *ifp = &sc->sc_ethercom.ec_if;
struct sf_descsoft *ds;
uint32_t cqci, tcd;
int consumer, producer, txidx;
try_again:
cqci = sf_funcreg_read(sc, SF_CompletionQueueConsumerIndex);
consumer = CQCI_TxCompletionConsumerIndex_get(cqci);
producer = CQPI_TxCompletionProducerIndex_get(
sf_funcreg_read(sc, SF_CompletionQueueProducerIndex));
if (consumer == producer)
return;
while (consumer != producer) {
SF_CDTXCSYNC(sc, consumer, BUS_DMASYNC_POSTREAD);
tcd = le32toh(sc->sc_txcomp[consumer].tcd_word0);
txidx = SF_TCD_INDEX_TO_HOST(TCD_INDEX(tcd));
#ifdef DIAGNOSTIC
if ((tcd & TCD_PR) == 0)
aprint_error_dev(sc->sc_dev,
"Tx queue mismatch, index %d\n", txidx);
#endif
/*
* NOTE: stats are updated later. We're just
* releasing packets that have been DMA'd to
* the chip.
*/
ds = &sc->sc_txsoft[txidx];
SF_CDTXDSYNC(sc, txidx, BUS_DMASYNC_POSTWRITE);
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap,
0, ds->ds_dmamap->dm_mapsize,
BUS_DMASYNC_POSTWRITE);
m_freem(ds->ds_mbuf);
ds->ds_mbuf = NULL;
consumer = SF_NEXTTCD(consumer);
sc->sc_txpending--;
}
/* XXXJRT -- should be KDASSERT() */
KASSERT(sc->sc_txpending >= 0);
/* If all packets are done, cancel the watchdog timer. */
if (sc->sc_txpending == 0)
ifp->if_timer = 0;
/* Update the consumer index. */
sf_funcreg_write(sc, SF_CompletionQueueConsumerIndex,
(cqci & ~CQCI_TxCompletionConsumerIndex(0x7ff)) |
CQCI_TxCompletionConsumerIndex(consumer));
/* Double check for new completions. */
goto try_again;
}
/*
* sf_rxintr:
*
* Helper -- handle receive interrupts.
*/
static void
sf_rxintr(struct sf_softc *sc)
{
struct ifnet *ifp = &sc->sc_ethercom.ec_if;
struct sf_descsoft *ds;
struct sf_rcd_full *rcd;
struct mbuf *m;
uint32_t cqci, word0;
int consumer, producer, bufproducer, rxidx, len;
try_again:
cqci = sf_funcreg_read(sc, SF_CompletionQueueConsumerIndex);
consumer = CQCI_RxCompletionQ1ConsumerIndex_get(cqci);
producer = CQPI_RxCompletionQ1ProducerIndex_get(
sf_funcreg_read(sc, SF_CompletionQueueProducerIndex));
bufproducer = RXQ1P_RxDescQ1Producer_get(
sf_funcreg_read(sc, SF_RxDescQueue1Ptrs));
if (consumer == producer)
return;
while (consumer != producer) {
rcd = &sc->sc_rxcomp[consumer];
SF_CDRXCSYNC(sc, consumer,
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
SF_CDRXCSYNC(sc, consumer,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
word0 = le32toh(rcd->rcd_word0);
rxidx = RCD_W0_EndIndex(word0);
ds = &sc->sc_rxsoft[rxidx];
consumer = SF_NEXTRCD(consumer);
bufproducer = SF_NEXTRX(bufproducer);
if ((word0 & RCD_W0_OK) == 0) {
SF_INIT_RXDESC(sc, rxidx);
continue;
}
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap, 0,
ds->ds_dmamap->dm_mapsize, BUS_DMASYNC_POSTREAD);
/*
* No errors; receive the packet. Note that we have
* configured the Starfire to NOT transfer the CRC
* with the packet.
*/
len = RCD_W0_Length(word0);
#ifdef __NO_STRICT_ALIGNMENT
/*
* Allocate a new mbuf cluster. If that fails, we are
* out of memory, and must drop the packet and recycle
* the buffer that's already attached to this descriptor.
*/
m = ds->ds_mbuf;
if (sf_add_rxbuf(sc, rxidx) != 0) {
if_statinc(ifp, if_ierrors);
SF_INIT_RXDESC(sc, rxidx);
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap, 0,
ds->ds_dmamap->dm_mapsize, BUS_DMASYNC_PREREAD);
continue;
}
#else
/*
* The Starfire's receive buffer must be 4-byte aligned.
* But this means that the data after the Ethernet header
* is misaligned. We must allocate a new buffer and
* copy the data, shifted forward 2 bytes.
*/
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL) {
dropit:
if_statinc(ifp, if_ierrors);
SF_INIT_RXDESC(sc, rxidx);
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap, 0,
ds->ds_dmamap->dm_mapsize, BUS_DMASYNC_PREREAD);
continue;
}
MCLAIM(m, &sc->sc_ethercom.ec_rx_mowner);
if (len > (MHLEN - 2)) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
m_freem(m);
goto dropit;
}
}
m->m_data += 2;
/*
* Note that we use cluster for incoming frames, so the
* buffer is virtually contiguous.
*/
memcpy(mtod(m, void *), mtod(ds->ds_mbuf, void *), len);
/* Allow the receive descriptor to continue using its mbuf. */
SF_INIT_RXDESC(sc, rxidx);
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap, 0,
ds->ds_dmamap->dm_mapsize, BUS_DMASYNC_PREREAD);
#endif /* __NO_STRICT_ALIGNMENT */
m_set_rcvif(m, ifp);
m->m_pkthdr.len = m->m_len = len;
/* Pass it on. */
if_percpuq_enqueue(ifp->if_percpuq, m);
}
/* Update the chip's pointers. */
sf_funcreg_write(sc, SF_CompletionQueueConsumerIndex,
(cqci & ~CQCI_RxCompletionQ1ConsumerIndex(0x7ff)) |
CQCI_RxCompletionQ1ConsumerIndex(consumer));
sf_funcreg_write(sc, SF_RxDescQueue1Ptrs,
RXQ1P_RxDescQ1Producer(bufproducer));
/* Double-check for any new completions. */
goto try_again;
}
/*
* sf_tick:
*
* One second timer, used to tick the MII and update stats.
*/
static void
sf_tick(void *arg)
{
struct sf_softc *sc = arg;
int s;
s = splnet();
mii_tick(&sc->sc_mii);
sf_stats_update(sc);
splx(s);
callout_schedule(&sc->sc_tick_callout, hz);
}
/*
* sf_stats_update:
*
* Read the statitistics counters.
*/
static void
sf_stats_update(struct sf_softc *sc)
{
struct sf_stats stats;
struct ifnet *ifp = &sc->sc_ethercom.ec_if;
uint32_t *p;
u_int i;
p = &stats.TransmitOKFrames;
for (i = 0; i < (sizeof(stats) / sizeof(uint32_t)); i++) {
*p++ = sf_genreg_read(sc,
SF_STATS_BASE + (i * sizeof(uint32_t)));
sf_genreg_write(sc, SF_STATS_BASE + (i * sizeof(uint32_t)), 0);
}
net_stat_ref_t nsr = IF_STAT_GETREF(ifp);
if_statadd_ref(nsr, if_opackets, stats.TransmitOKFrames);
if_statadd_ref(nsr, if_collisions,
stats.SingleCollisionFrames +
stats.MultipleCollisionFrames);
if_statadd_ref(nsr, if_oerrors,
stats.TransmitAbortDueToExcessiveCollisions +
stats.TransmitAbortDueToExcessingDeferral +
stats.FramesLostDueToInternalTransmitErrors);
if_statadd_ref(nsr, if_ierrors,
stats.ReceiveCRCErrors + stats.AlignmentErrors +
stats.ReceiveFramesTooLong + stats.ReceiveFramesTooShort +
stats.ReceiveFramesJabbersError +
stats.FramesLostDueToInternalReceiveErrors);
IF_STAT_PUTREF(ifp);
}
/*
* sf_reset:
*
* Perform a soft reset on the Starfire.
*/
static void
sf_reset(struct sf_softc *sc)
{
int i;
sf_funcreg_write(sc, SF_GeneralEthernetCtrl, 0);
sf_macreset(sc);
sf_funcreg_write(sc, SF_PciDeviceConfig, PDC_SoftReset);
for (i = 0; i < 1000; i++) {
delay(10);
if ((sf_funcreg_read(sc, SF_PciDeviceConfig) &
PDC_SoftReset) == 0)
break;
}
if (i == 1000) {
aprint_error_dev(sc->sc_dev, "reset failed to complete\n");
sf_funcreg_write(sc, SF_PciDeviceConfig, 0);
}
delay(1000);
}
/*
* sf_macreset:
*
* Reset the MAC portion of the Starfire.
*/
static void
sf_macreset(struct sf_softc *sc)
{
sf_genreg_write(sc, SF_MacConfig1, sc->sc_MacConfig1 | MC1_SoftRst);
delay(1000);
sf_genreg_write(sc, SF_MacConfig1, sc->sc_MacConfig1);
}
/*
* sf_init: [ifnet interface function]
*
* Initialize the interface. Must be called at splnet().
*/
static int
sf_init(struct ifnet *ifp)
{
struct sf_softc *sc = ifp->if_softc;
struct sf_descsoft *ds;
int error = 0;
u_int i;
/*
* Cancel any pending I/O.
*/
sf_stop(ifp, 0);
/*
* Reset the Starfire to a known state.
*/
sf_reset(sc);
/* Clear the stat counters. */
for (i = 0; i < sizeof(struct sf_stats); i += sizeof(uint32_t))
sf_genreg_write(sc, SF_STATS_BASE + i, 0);
/*
* Initialize the transmit descriptor ring.
*/
memset(sc->sc_txdescs, 0, sizeof(sc->sc_txdescs));
sf_funcreg_write(sc, SF_TxDescQueueHighAddr, 0);
sf_funcreg_write(sc, SF_HiPrTxDescQueueBaseAddr, SF_CDTXDADDR(sc, 0));
sf_funcreg_write(sc, SF_LoPrTxDescQueueBaseAddr, 0);
/*
* Initialize the transmit completion ring.
*/
for (i = 0; i < SF_NTCD; i++) {
sc->sc_txcomp[i].tcd_word0 = TCD_DMA_ID;
SF_CDTXCSYNC(sc, i, BUS_DMASYNC_PREREAD |BUS_DMASYNC_PREWRITE);
}
sf_funcreg_write(sc, SF_CompletionQueueHighAddr, 0);
sf_funcreg_write(sc, SF_TxCompletionQueueCtrl, SF_CDTXCADDR(sc, 0));
/*
* Initialize the receive descriptor ring.
*/
for (i = 0; i < SF_NRXDESC; i++) {
ds = &sc->sc_rxsoft[i];
if (ds->ds_mbuf == NULL) {
if ((error = sf_add_rxbuf(sc, i)) != 0) {
aprint_error_dev(sc->sc_dev,
"unable to allocate or map rx buffer %d, "
"error = %d\n", i, error);
/*
* XXX Should attempt to run with fewer receive
* XXX buffers instead of just failing.
*/
sf_rxdrain(sc);
goto out;
}
} else
SF_INIT_RXDESC(sc, i);
}
sf_funcreg_write(sc, SF_RxDescQueueHighAddress, 0);
sf_funcreg_write(sc, SF_RxDescQueue1LowAddress, SF_CDRXDADDR(sc, 0));
sf_funcreg_write(sc, SF_RxDescQueue2LowAddress, 0);
/*
* Initialize the receive completion ring.
*/
for (i = 0; i < SF_NRCD; i++) {
sc->sc_rxcomp[i].rcd_word0 = RCD_W0_ID;
sc->sc_rxcomp[i].rcd_word1 = 0;
sc->sc_rxcomp[i].rcd_word2 = 0;
sc->sc_rxcomp[i].rcd_timestamp = 0;
SF_CDRXCSYNC(sc, i, BUS_DMASYNC_PREREAD |BUS_DMASYNC_PREWRITE);
}
sf_funcreg_write(sc, SF_RxCompletionQueue1Ctrl, SF_CDRXCADDR(sc, 0) |
RCQ1C_RxCompletionQ1Type(3));
sf_funcreg_write(sc, SF_RxCompletionQueue2Ctrl, 0);
/*
* Initialize the Tx CSR.
*/
sc->sc_TransmitFrameCSR = 0;
sf_funcreg_write(sc, SF_TransmitFrameCSR,
sc->sc_TransmitFrameCSR |
TFCSR_TransmitThreshold(sc->sc_txthresh));
/*
* Initialize the Tx descriptor control register.
*/
sc->sc_TxDescQueueCtrl = TDQC_SkipLength(0) |
TDQC_TxDmaBurstSize(4) | /* default */
TDQC_MinFrameSpacing(3) | /* 128 bytes */
TDQC_TxDescType(0);
sf_funcreg_write(sc, SF_TxDescQueueCtrl,
sc->sc_TxDescQueueCtrl |
TDQC_TxHighPriorityFifoThreshold(sc->sc_txthresh));
/*
* Initialize the Rx descriptor control registers.
*/
sf_funcreg_write(sc, SF_RxDescQueue1Ctrl,
RDQ1C_RxQ1BufferLength(MCLBYTES) |
RDQ1C_RxDescSpacing(0));
sf_funcreg_write(sc, SF_RxDescQueue2Ctrl, 0);
/*
* Initialize the Tx descriptor producer indices.
*/
sf_funcreg_write(sc, SF_TxDescQueueProducerIndex,
TDQPI_HiPrTxProducerIndex(0) |
TDQPI_LoPrTxProducerIndex(0));
/*
* Initialize the Rx descriptor producer indices.
*/
sf_funcreg_write(sc, SF_RxDescQueue1Ptrs,
RXQ1P_RxDescQ1Producer(SF_NRXDESC - 1));
sf_funcreg_write(sc, SF_RxDescQueue2Ptrs,
RXQ2P_RxDescQ2Producer(0));
/*
* Initialize the Tx and Rx completion queue consumer indices.
*/
sf_funcreg_write(sc, SF_CompletionQueueConsumerIndex,
CQCI_TxCompletionConsumerIndex(0) |
CQCI_RxCompletionQ1ConsumerIndex(0));
sf_funcreg_write(sc, SF_RxHiPrCompletionPtrs, 0);
/*
* Initialize the Rx DMA control register.
*/
sf_funcreg_write(sc, SF_RxDmaCtrl,
RDC_RxHighPriorityThreshold(6) | /* default */
RDC_RxBurstSize(4)); /* default */
/*
* Set the receive filter.
*/
sc->sc_RxAddressFilteringCtl = 0;
sf_set_filter(sc);
/*
* Set MacConfig1. When we set the media, MacConfig1 will
* actually be written and the MAC part reset.
*/
sc->sc_MacConfig1 = MC1_PadEn;
/*
* Set the media.
*/
if ((error = ether_mediachange(ifp)) != 0)
goto out;
/*
* Initialize the interrupt register.
*/
sc->sc_InterruptEn = IS_PCIPadInt | IS_RxQ1DoneInt |
IS_TxQueueDoneInt | IS_TxDmaDoneInt | IS_DmaErrInt |
IS_StatisticWrapInt;
sf_funcreg_write(sc, SF_InterruptEn, sc->sc_InterruptEn);
sf_funcreg_write(sc, SF_PciDeviceConfig, PDC_IntEnable |
PDC_PCIMstDmaEn | (1 << PDC_FifoThreshold_SHIFT));
/*
* Start the transmit and receive processes.
*/
sf_funcreg_write(sc, SF_GeneralEthernetCtrl,
GEC_TxDmaEn | GEC_RxDmaEn | GEC_TransmitEn | GEC_ReceiveEn);
/* Start the on second clock. */
callout_schedule(&sc->sc_tick_callout, hz);
/*
* Note that the interface is now running.
*/
ifp->if_flags |= IFF_RUNNING;
out:
if (error) {
ifp->if_flags &= ~IFF_RUNNING;
ifp->if_timer = 0;
printf("%s: interface not running\n", device_xname(sc->sc_dev));
}
return (error);
}
/*
* sf_rxdrain:
*
* Drain the receive queue.
*/
static void
sf_rxdrain(struct sf_softc *sc)
{
struct sf_descsoft *ds;
int i;
for (i = 0; i < SF_NRXDESC; i++) {
ds = &sc->sc_rxsoft[i];
if (ds->ds_mbuf != NULL) {
bus_dmamap_unload(sc->sc_dmat, ds->ds_dmamap);
m_freem(ds->ds_mbuf);
ds->ds_mbuf = NULL;
}
}
}
/*
* sf_stop: [ifnet interface function]
*
* Stop transmission on the interface.
*/
static void
sf_stop(struct ifnet *ifp, int disable)
{
struct sf_softc *sc = ifp->if_softc;
struct sf_descsoft *ds;
int i;
/* Stop the one second clock. */
callout_stop(&sc->sc_tick_callout);
/* Down the MII. */
mii_down(&sc->sc_mii);
/* Disable interrupts. */
sf_funcreg_write(sc, SF_InterruptEn, 0);
/* Stop the transmit and receive processes. */
sf_funcreg_write(sc, SF_GeneralEthernetCtrl, 0);
/*
* Release any queued transmit buffers.
*/
for (i = 0; i < SF_NTXDESC; i++) {
ds = &sc->sc_txsoft[i];
if (ds->ds_mbuf != NULL) {
bus_dmamap_unload(sc->sc_dmat, ds->ds_dmamap);
m_freem(ds->ds_mbuf);
ds->ds_mbuf = NULL;
}
}
/*
* Mark the interface down and cancel the watchdog timer.
*/
ifp->if_flags &= ~IFF_RUNNING;
ifp->if_timer = 0;
if (disable)
sf_rxdrain(sc);
}
/*
* sf_read_eeprom:
*
* Read from the Starfire EEPROM.
*/
static uint8_t
sf_read_eeprom(struct sf_softc *sc, int offset)
{
uint32_t reg;
reg = sf_genreg_read(sc, SF_EEPROM_BASE + (offset & ~3));
return ((reg >> (8 * (offset & 3))) & 0xff);
}
/*
* sf_add_rxbuf:
*
* Add a receive buffer to the indicated descriptor.
*/
static int
sf_add_rxbuf(struct sf_softc *sc, int idx)
{
struct sf_descsoft *ds = &sc->sc_rxsoft[idx];
struct mbuf *m;
int error;
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
return (ENOBUFS);
MCLAIM(m, &sc->sc_ethercom.ec_rx_mowner);
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
m_freem(m);
return (ENOBUFS);
}
if (ds->ds_mbuf != NULL)
bus_dmamap_unload(sc->sc_dmat, ds->ds_dmamap);
ds->ds_mbuf = m;
error = bus_dmamap_load(sc->sc_dmat, ds->ds_dmamap,
m->m_ext.ext_buf, m->m_ext.ext_size, NULL,
BUS_DMA_READ | BUS_DMA_NOWAIT);
if (error) {
aprint_error_dev(sc->sc_dev,
"can't load rx DMA map %d, error = %d\n", idx, error);
panic("sf_add_rxbuf"); /* XXX */
}
bus_dmamap_sync(sc->sc_dmat, ds->ds_dmamap, 0,
ds->ds_dmamap->dm_mapsize, BUS_DMASYNC_PREREAD);
SF_INIT_RXDESC(sc, idx);
return (0);
}
static void
sf_set_filter_perfect(struct sf_softc *sc, int slot, const uint8_t *enaddr)
{
uint32_t reg0, reg1, reg2;
reg0 = enaddr[5] | (enaddr[4] << 8);
reg1 = enaddr[3] | (enaddr[2] << 8);
reg2 = enaddr[1] | (enaddr[0] << 8);
sf_genreg_write(sc, SF_PERFECT_BASE + (slot * 0x10) + 0, reg0);
sf_genreg_write(sc, SF_PERFECT_BASE + (slot * 0x10) + 4, reg1);
sf_genreg_write(sc, SF_PERFECT_BASE + (slot * 0x10) + 8, reg2);
}
static void
sf_set_filter_hash(struct sf_softc *sc, uint8_t *enaddr)
{
uint32_t hash, slot, reg;
hash = ether_crc32_be(enaddr, ETHER_ADDR_LEN) >> 23;
slot = hash >> 4;
reg = sf_genreg_read(sc, SF_HASH_BASE + (slot * 0x10));
reg |= 1 << (hash & 0xf);
sf_genreg_write(sc, SF_HASH_BASE + (slot * 0x10), reg);
}
/*
* sf_set_filter:
*
* Set the Starfire receive filter.
*/
static void
sf_set_filter(struct sf_softc *sc)
{
struct ethercom *ec = &sc->sc_ethercom;
struct ifnet *ifp = &sc->sc_ethercom.ec_if;
struct ether_multi *enm;
struct ether_multistep step;
int i;
/* Start by clearing the perfect and hash tables. */
for (i = 0; i < SF_PERFECT_SIZE; i += sizeof(uint32_t))
sf_genreg_write(sc, SF_PERFECT_BASE + i, 0);
for (i = 0; i < SF_HASH_SIZE; i += sizeof(uint32_t))
sf_genreg_write(sc, SF_HASH_BASE + i, 0);
/*
* Clear the perfect and hash mode bits.
*/
sc->sc_RxAddressFilteringCtl &=
~(RAFC_PerfectFilteringMode(3) | RAFC_HashFilteringMode(3));
if (ifp->if_flags & IFF_BROADCAST)
sc->sc_RxAddressFilteringCtl |= RAFC_PassBroadcast;
else
sc->sc_RxAddressFilteringCtl &= ~RAFC_PassBroadcast;
if (ifp->if_flags & IFF_PROMISC) {
sc->sc_RxAddressFilteringCtl |= RAFC_PromiscuousMode;
goto allmulti;
} else
sc->sc_RxAddressFilteringCtl &= ~RAFC_PromiscuousMode;
/*
* Set normal perfect filtering mode.
*/
sc->sc_RxAddressFilteringCtl |= RAFC_PerfectFilteringMode(1);
/*
* First, write the station address to the perfect filter
* table.
*/
sf_set_filter_perfect(sc, 0, CLLADDR(ifp->if_sadl));
/*
* Now set the hash bits for each multicast address in our
* list.
*/
ETHER_LOCK(ec);
ETHER_FIRST_MULTI(step, ec, enm);
if (enm == NULL) {
ETHER_UNLOCK(ec);
goto done;
}
while (enm != NULL) {
if (memcmp(enm->enm_addrlo, enm->enm_addrhi, ETHER_ADDR_LEN)) {
/*
* We must listen to a range of multicast addresses.
* For now, just accept all multicasts, rather than
* trying to set only those filter bits needed to match
* the range. (At this time, the only use of address
* ranges is for IP multicast routing, for which the
* range is big enough to require all bits set.)
*/
ETHER_UNLOCK(ec);
goto allmulti;
}
sf_set_filter_hash(sc, enm->enm_addrlo);
ETHER_NEXT_MULTI(step, enm);
}
ETHER_UNLOCK(ec);
/*
* Set "hash only multicast dest, match regardless of VLAN ID".
*/
sc->sc_RxAddressFilteringCtl |= RAFC_HashFilteringMode(2);
goto done;
allmulti:
/*
* XXX RAFC_PassMulticast is sub-optimal if using VLAN mode.
*/
sc->sc_RxAddressFilteringCtl |= RAFC_PassMulticast;
ifp->if_flags |= IFF_ALLMULTI;
done:
sf_funcreg_write(sc, SF_RxAddressFilteringCtl,
sc->sc_RxAddressFilteringCtl);
}
/*
* sf_mii_read: [mii interface function]
*
* Read from the MII.
*/
static int
sf_mii_read(device_t self, int phy, int reg, uint16_t *data)
{
struct sf_softc *sc = device_private(self);
uint32_t v;
int i;
for (i = 0; i < 1000; i++) {
v = sf_genreg_read(sc, SF_MII_PHY_REG(phy, reg));
if (v & MiiDataValid)
break;
delay(1);
}
if ((v & MiiDataValid) == 0)
return -1;
if (MiiRegDataPort(v) == 0xffff)
return -1;
*data = MiiRegDataPort(v);
return 0;
}
/*
* sf_mii_write: [mii interface function]
*
* Write to the MII.
*/
static int
sf_mii_write(device_t self, int phy, int reg, uint16_t val)
{
struct sf_softc *sc = device_private(self);
int i;
sf_genreg_write(sc, SF_MII_PHY_REG(phy, reg), val);
for (i = 0; i < 1000; i++) {
if ((sf_genreg_read(sc, SF_MII_PHY_REG(phy, reg)) &
MiiBusy) == 0)
return 0;
delay(1);
}
printf("%s: MII write timed out\n", device_xname(sc->sc_dev));
return ETIMEDOUT;
}
/*
* sf_mii_statchg: [mii interface function]
*
* Callback from the PHY when the media changes.
*/
static void
sf_mii_statchg(struct ifnet *ifp)
{
struct sf_softc *sc = ifp->if_softc;
uint32_t ipg;
if (sc->sc_mii.mii_media_active & IFM_FDX) {
sc->sc_MacConfig1 |= MC1_FullDuplex;
ipg = 0x15;
} else {
sc->sc_MacConfig1 &= ~MC1_FullDuplex;
ipg = 0x11;
}
sf_genreg_write(sc, SF_MacConfig1, sc->sc_MacConfig1);
sf_macreset(sc);
sf_genreg_write(sc, SF_BkToBkIPG, ipg);
}