qemu/hw/net/sunhme.c
Bin Meng 0fe0efc9cd hw/net: sunhme: Remove the logic of padding short frames in the receive path
Now that we have implemented unified short frames padding in the
QEMU networking codes, remove the same logic in the NIC codes.

Signed-off-by: Bin Meng <bmeng@tinylab.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
2023-07-07 16:35:12 +08:00

973 lines
28 KiB
C

/*
* QEMU Sun Happy Meal Ethernet emulation
*
* Copyright (c) 2017 Mark Cave-Ayland
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
#include "qemu/osdep.h"
#include "hw/pci/pci_device.h"
#include "hw/qdev-properties.h"
#include "migration/vmstate.h"
#include "hw/net/mii.h"
#include "net/net.h"
#include "qemu/module.h"
#include "net/checksum.h"
#include "net/eth.h"
#include "sysemu/sysemu.h"
#include "trace.h"
#include "qom/object.h"
#define HME_REG_SIZE 0x8000
#define HME_SEB_REG_SIZE 0x2000
#define HME_SEBI_RESET 0x0
#define HME_SEB_RESET_ETX 0x1
#define HME_SEB_RESET_ERX 0x2
#define HME_SEBI_STAT 0x100
#define HME_SEBI_STAT_LINUXBUG 0x108
#define HME_SEB_STAT_RXTOHOST 0x10000
#define HME_SEB_STAT_NORXD 0x20000
#define HME_SEB_STAT_MIFIRQ 0x800000
#define HME_SEB_STAT_HOSTTOTX 0x1000000
#define HME_SEB_STAT_TXALL 0x2000000
#define HME_SEBI_IMASK 0x104
#define HME_SEBI_IMASK_LINUXBUG 0x10c
#define HME_ETX_REG_SIZE 0x2000
#define HME_ETXI_PENDING 0x0
#define HME_ETXI_RING 0x8
#define HME_ETXI_RING_ADDR 0xffffff00
#define HME_ETXI_RING_OFFSET 0xff
#define HME_ETXI_RSIZE 0x2c
#define HME_ERX_REG_SIZE 0x2000
#define HME_ERXI_CFG 0x0
#define HME_ERX_CFG_RINGSIZE 0x600
#define HME_ERX_CFG_RINGSIZE_SHIFT 9
#define HME_ERX_CFG_BYTEOFFSET 0x38
#define HME_ERX_CFG_BYTEOFFSET_SHIFT 3
#define HME_ERX_CFG_CSUMSTART 0x7f0000
#define HME_ERX_CFG_CSUMSHIFT 16
#define HME_ERXI_RING 0x4
#define HME_ERXI_RING_ADDR 0xffffff00
#define HME_ERXI_RING_OFFSET 0xff
#define HME_MAC_REG_SIZE 0x1000
#define HME_MACI_TXCFG 0x20c
#define HME_MAC_TXCFG_ENABLE 0x1
#define HME_MACI_RXCFG 0x30c
#define HME_MAC_RXCFG_ENABLE 0x1
#define HME_MAC_RXCFG_PMISC 0x40
#define HME_MAC_RXCFG_HENABLE 0x800
#define HME_MACI_MACADDR2 0x318
#define HME_MACI_MACADDR1 0x31c
#define HME_MACI_MACADDR0 0x320
#define HME_MACI_HASHTAB3 0x340
#define HME_MACI_HASHTAB2 0x344
#define HME_MACI_HASHTAB1 0x348
#define HME_MACI_HASHTAB0 0x34c
#define HME_MIF_REG_SIZE 0x20
#define HME_MIFI_FO 0xc
#define HME_MIF_FO_ST 0xc0000000
#define HME_MIF_FO_ST_SHIFT 30
#define HME_MIF_FO_OPC 0x30000000
#define HME_MIF_FO_OPC_SHIFT 28
#define HME_MIF_FO_PHYAD 0x0f800000
#define HME_MIF_FO_PHYAD_SHIFT 23
#define HME_MIF_FO_REGAD 0x007c0000
#define HME_MIF_FO_REGAD_SHIFT 18
#define HME_MIF_FO_TAMSB 0x20000
#define HME_MIF_FO_TALSB 0x10000
#define HME_MIF_FO_DATA 0xffff
#define HME_MIFI_CFG 0x10
#define HME_MIF_CFG_MDI0 0x100
#define HME_MIF_CFG_MDI1 0x200
#define HME_MIFI_IMASK 0x14
#define HME_MIFI_STAT 0x18
/* Wired HME PHY addresses */
#define HME_PHYAD_INTERNAL 1
#define HME_PHYAD_EXTERNAL 0
#define MII_COMMAND_START 0x1
#define MII_COMMAND_READ 0x2
#define MII_COMMAND_WRITE 0x1
#define TYPE_SUNHME "sunhme"
OBJECT_DECLARE_SIMPLE_TYPE(SunHMEState, SUNHME)
/* Maximum size of buffer */
#define HME_FIFO_SIZE 0x800
/* Size of TX/RX descriptor */
#define HME_DESC_SIZE 0x8
#define HME_XD_OWN 0x80000000
#define HME_XD_OFL 0x40000000
#define HME_XD_SOP 0x40000000
#define HME_XD_EOP 0x20000000
#define HME_XD_RXLENMSK 0x3fff0000
#define HME_XD_RXLENSHIFT 16
#define HME_XD_RXCKSUM 0xffff
#define HME_XD_TXLENMSK 0x00001fff
#define HME_XD_TXCKSUM 0x10000000
#define HME_XD_TXCSSTUFF 0xff00000
#define HME_XD_TXCSSTUFFSHIFT 20
#define HME_XD_TXCSSTART 0xfc000
#define HME_XD_TXCSSTARTSHIFT 14
#define HME_MII_REGS_SIZE 0x20
struct SunHMEState {
/*< private >*/
PCIDevice parent_obj;
NICState *nic;
NICConf conf;
MemoryRegion hme;
MemoryRegion sebreg;
MemoryRegion etxreg;
MemoryRegion erxreg;
MemoryRegion macreg;
MemoryRegion mifreg;
uint32_t sebregs[HME_SEB_REG_SIZE >> 2];
uint32_t etxregs[HME_ETX_REG_SIZE >> 2];
uint32_t erxregs[HME_ERX_REG_SIZE >> 2];
uint32_t macregs[HME_MAC_REG_SIZE >> 2];
uint32_t mifregs[HME_MIF_REG_SIZE >> 2];
uint16_t miiregs[HME_MII_REGS_SIZE];
};
static Property sunhme_properties[] = {
DEFINE_NIC_PROPERTIES(SunHMEState, conf),
DEFINE_PROP_END_OF_LIST(),
};
static void sunhme_reset_tx(SunHMEState *s)
{
/* Indicate TX reset complete */
s->sebregs[HME_SEBI_RESET] &= ~HME_SEB_RESET_ETX;
}
static void sunhme_reset_rx(SunHMEState *s)
{
/* Indicate RX reset complete */
s->sebregs[HME_SEBI_RESET] &= ~HME_SEB_RESET_ERX;
}
static void sunhme_update_irq(SunHMEState *s)
{
PCIDevice *d = PCI_DEVICE(s);
int level;
/* MIF interrupt mask (16-bit) */
uint32_t mifmask = ~(s->mifregs[HME_MIFI_IMASK >> 2]) & 0xffff;
uint32_t mif = s->mifregs[HME_MIFI_STAT >> 2] & mifmask;
/* Main SEB interrupt mask (include MIF status from above) */
uint32_t sebmask = ~(s->sebregs[HME_SEBI_IMASK >> 2]) &
~HME_SEB_STAT_MIFIRQ;
uint32_t seb = s->sebregs[HME_SEBI_STAT >> 2] & sebmask;
if (mif) {
seb |= HME_SEB_STAT_MIFIRQ;
}
level = (seb ? 1 : 0);
trace_sunhme_update_irq(mifmask, mif, sebmask, seb, level);
pci_set_irq(d, level);
}
static void sunhme_seb_write(void *opaque, hwaddr addr,
uint64_t val, unsigned size)
{
SunHMEState *s = SUNHME(opaque);
trace_sunhme_seb_write(addr, val);
/* Handly buggy Linux drivers before 4.13 which have
the wrong offsets for HME_SEBI_STAT and HME_SEBI_IMASK */
switch (addr) {
case HME_SEBI_STAT_LINUXBUG:
addr = HME_SEBI_STAT;
break;
case HME_SEBI_IMASK_LINUXBUG:
addr = HME_SEBI_IMASK;
break;
default:
break;
}
switch (addr) {
case HME_SEBI_RESET:
if (val & HME_SEB_RESET_ETX) {
sunhme_reset_tx(s);
}
if (val & HME_SEB_RESET_ERX) {
sunhme_reset_rx(s);
}
val = s->sebregs[HME_SEBI_RESET >> 2];
break;
}
s->sebregs[addr >> 2] = val;
}
static uint64_t sunhme_seb_read(void *opaque, hwaddr addr,
unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t val;
/* Handly buggy Linux drivers before 4.13 which have
the wrong offsets for HME_SEBI_STAT and HME_SEBI_IMASK */
switch (addr) {
case HME_SEBI_STAT_LINUXBUG:
addr = HME_SEBI_STAT;
break;
case HME_SEBI_IMASK_LINUXBUG:
addr = HME_SEBI_IMASK;
break;
default:
break;
}
val = s->sebregs[addr >> 2];
switch (addr) {
case HME_SEBI_STAT:
/* Autoclear status (except MIF) */
s->sebregs[HME_SEBI_STAT >> 2] &= HME_SEB_STAT_MIFIRQ;
sunhme_update_irq(s);
break;
}
trace_sunhme_seb_read(addr, val);
return val;
}
static const MemoryRegionOps sunhme_seb_ops = {
.read = sunhme_seb_read,
.write = sunhme_seb_write,
.endianness = DEVICE_LITTLE_ENDIAN,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
},
};
static void sunhme_transmit(SunHMEState *s);
static void sunhme_etx_write(void *opaque, hwaddr addr,
uint64_t val, unsigned size)
{
SunHMEState *s = SUNHME(opaque);
trace_sunhme_etx_write(addr, val);
switch (addr) {
case HME_ETXI_PENDING:
if (val) {
sunhme_transmit(s);
}
break;
}
s->etxregs[addr >> 2] = val;
}
static uint64_t sunhme_etx_read(void *opaque, hwaddr addr,
unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t val;
val = s->etxregs[addr >> 2];
trace_sunhme_etx_read(addr, val);
return val;
}
static const MemoryRegionOps sunhme_etx_ops = {
.read = sunhme_etx_read,
.write = sunhme_etx_write,
.endianness = DEVICE_LITTLE_ENDIAN,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
},
};
static void sunhme_erx_write(void *opaque, hwaddr addr,
uint64_t val, unsigned size)
{
SunHMEState *s = SUNHME(opaque);
trace_sunhme_erx_write(addr, val);
s->erxregs[addr >> 2] = val;
}
static uint64_t sunhme_erx_read(void *opaque, hwaddr addr,
unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t val;
val = s->erxregs[addr >> 2];
trace_sunhme_erx_read(addr, val);
return val;
}
static const MemoryRegionOps sunhme_erx_ops = {
.read = sunhme_erx_read,
.write = sunhme_erx_write,
.endianness = DEVICE_LITTLE_ENDIAN,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
},
};
static void sunhme_mac_write(void *opaque, hwaddr addr,
uint64_t val, unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t oldval = s->macregs[addr >> 2];
trace_sunhme_mac_write(addr, val);
s->macregs[addr >> 2] = val;
switch (addr) {
case HME_MACI_RXCFG:
if (!(oldval & HME_MAC_RXCFG_ENABLE) &&
(val & HME_MAC_RXCFG_ENABLE)) {
qemu_flush_queued_packets(qemu_get_queue(s->nic));
}
break;
}
}
static uint64_t sunhme_mac_read(void *opaque, hwaddr addr,
unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t val;
val = s->macregs[addr >> 2];
trace_sunhme_mac_read(addr, val);
return val;
}
static const MemoryRegionOps sunhme_mac_ops = {
.read = sunhme_mac_read,
.write = sunhme_mac_write,
.endianness = DEVICE_LITTLE_ENDIAN,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
},
};
static void sunhme_mii_write(SunHMEState *s, uint8_t reg, uint16_t data)
{
trace_sunhme_mii_write(reg, data);
switch (reg) {
case MII_BMCR:
if (data & MII_BMCR_RESET) {
/* Autoclear reset bit, enable auto negotiation */
data &= ~MII_BMCR_RESET;
data |= MII_BMCR_AUTOEN;
}
if (data & MII_BMCR_ANRESTART) {
/* Autoclear auto negotiation restart */
data &= ~MII_BMCR_ANRESTART;
/* Indicate negotiation complete */
s->miiregs[MII_BMSR] |= MII_BMSR_AN_COMP;
if (!qemu_get_queue(s->nic)->link_down) {
s->miiregs[MII_ANLPAR] |= MII_ANLPAR_TXFD;
s->miiregs[MII_BMSR] |= MII_BMSR_LINK_ST;
}
}
break;
}
s->miiregs[reg] = data;
}
static uint16_t sunhme_mii_read(SunHMEState *s, uint8_t reg)
{
uint16_t data = s->miiregs[reg];
trace_sunhme_mii_read(reg, data);
return data;
}
static void sunhme_mif_write(void *opaque, hwaddr addr,
uint64_t val, unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint8_t cmd, reg;
uint16_t data;
trace_sunhme_mif_write(addr, val);
switch (addr) {
case HME_MIFI_CFG:
/* Mask the read-only bits */
val &= ~(HME_MIF_CFG_MDI0 | HME_MIF_CFG_MDI1);
val |= s->mifregs[HME_MIFI_CFG >> 2] &
(HME_MIF_CFG_MDI0 | HME_MIF_CFG_MDI1);
break;
case HME_MIFI_FO:
/* Detect start of MII command */
if ((val & HME_MIF_FO_ST) >> HME_MIF_FO_ST_SHIFT
!= MII_COMMAND_START) {
val |= HME_MIF_FO_TALSB;
break;
}
/* Internal phy only */
if ((val & HME_MIF_FO_PHYAD) >> HME_MIF_FO_PHYAD_SHIFT
!= HME_PHYAD_INTERNAL) {
val |= HME_MIF_FO_TALSB;
break;
}
cmd = (val & HME_MIF_FO_OPC) >> HME_MIF_FO_OPC_SHIFT;
reg = (val & HME_MIF_FO_REGAD) >> HME_MIF_FO_REGAD_SHIFT;
data = (val & HME_MIF_FO_DATA);
switch (cmd) {
case MII_COMMAND_WRITE:
sunhme_mii_write(s, reg, data);
break;
case MII_COMMAND_READ:
val &= ~HME_MIF_FO_DATA;
val |= sunhme_mii_read(s, reg);
break;
}
val |= HME_MIF_FO_TALSB;
break;
}
s->mifregs[addr >> 2] = val;
}
static uint64_t sunhme_mif_read(void *opaque, hwaddr addr,
unsigned size)
{
SunHMEState *s = SUNHME(opaque);
uint64_t val;
val = s->mifregs[addr >> 2];
switch (addr) {
case HME_MIFI_STAT:
/* Autoclear MIF interrupt status */
s->mifregs[HME_MIFI_STAT >> 2] = 0;
sunhme_update_irq(s);
break;
}
trace_sunhme_mif_read(addr, val);
return val;
}
static const MemoryRegionOps sunhme_mif_ops = {
.read = sunhme_mif_read,
.write = sunhme_mif_write,
.endianness = DEVICE_LITTLE_ENDIAN,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
},
};
static void sunhme_transmit_frame(SunHMEState *s, uint8_t *buf, int size)
{
qemu_send_packet(qemu_get_queue(s->nic), buf, size);
}
static inline int sunhme_get_tx_ring_count(SunHMEState *s)
{
return (s->etxregs[HME_ETXI_RSIZE >> 2] + 1) << 4;
}
static inline int sunhme_get_tx_ring_nr(SunHMEState *s)
{
return s->etxregs[HME_ETXI_RING >> 2] & HME_ETXI_RING_OFFSET;
}
static inline void sunhme_set_tx_ring_nr(SunHMEState *s, int i)
{
uint32_t ring = s->etxregs[HME_ETXI_RING >> 2] & ~HME_ETXI_RING_OFFSET;
ring |= i & HME_ETXI_RING_OFFSET;
s->etxregs[HME_ETXI_RING >> 2] = ring;
}
static void sunhme_transmit(SunHMEState *s)
{
PCIDevice *d = PCI_DEVICE(s);
dma_addr_t tb, addr;
uint32_t intstatus, status, buffer, sum = 0;
int cr, nr, len, xmit_pos, csum_offset = 0, csum_stuff_offset = 0;
uint16_t csum = 0;
uint8_t xmit_buffer[HME_FIFO_SIZE];
tb = s->etxregs[HME_ETXI_RING >> 2] & HME_ETXI_RING_ADDR;
nr = sunhme_get_tx_ring_count(s);
cr = sunhme_get_tx_ring_nr(s);
pci_dma_read(d, tb + cr * HME_DESC_SIZE, &status, 4);
pci_dma_read(d, tb + cr * HME_DESC_SIZE + 4, &buffer, 4);
xmit_pos = 0;
while (status & HME_XD_OWN) {
trace_sunhme_tx_desc(buffer, status, cr, nr);
/* Copy data into transmit buffer */
addr = buffer;
len = status & HME_XD_TXLENMSK;
if (xmit_pos + len > HME_FIFO_SIZE) {
len = HME_FIFO_SIZE - xmit_pos;
}
pci_dma_read(d, addr, &xmit_buffer[xmit_pos], len);
xmit_pos += len;
/* Detect start of packet for TX checksum */
if (status & HME_XD_SOP) {
sum = 0;
csum_offset = (status & HME_XD_TXCSSTART) >> HME_XD_TXCSSTARTSHIFT;
csum_stuff_offset = (status & HME_XD_TXCSSTUFF) >>
HME_XD_TXCSSTUFFSHIFT;
}
if (status & HME_XD_TXCKSUM) {
/* Only start calculation from csum_offset */
if (xmit_pos - len <= csum_offset && xmit_pos > csum_offset) {
sum += net_checksum_add(xmit_pos - csum_offset,
xmit_buffer + csum_offset);
trace_sunhme_tx_xsum_add(csum_offset, xmit_pos - csum_offset);
} else {
sum += net_checksum_add(len, xmit_buffer + xmit_pos - len);
trace_sunhme_tx_xsum_add(xmit_pos - len, len);
}
}
/* Detect end of packet for TX checksum */
if (status & HME_XD_EOP) {
/* Stuff the checksum if required */
if (status & HME_XD_TXCKSUM) {
csum = net_checksum_finish(sum);
stw_be_p(xmit_buffer + csum_stuff_offset, csum);
trace_sunhme_tx_xsum_stuff(csum, csum_stuff_offset);
}
if (s->macregs[HME_MACI_TXCFG >> 2] & HME_MAC_TXCFG_ENABLE) {
sunhme_transmit_frame(s, xmit_buffer, xmit_pos);
trace_sunhme_tx_done(xmit_pos);
}
}
/* Update status */
status &= ~HME_XD_OWN;
pci_dma_write(d, tb + cr * HME_DESC_SIZE, &status, 4);
/* Move onto next descriptor */
cr++;
if (cr >= nr) {
cr = 0;
}
sunhme_set_tx_ring_nr(s, cr);
pci_dma_read(d, tb + cr * HME_DESC_SIZE, &status, 4);
pci_dma_read(d, tb + cr * HME_DESC_SIZE + 4, &buffer, 4);
/* Indicate TX complete */
intstatus = s->sebregs[HME_SEBI_STAT >> 2];
intstatus |= HME_SEB_STAT_HOSTTOTX;
s->sebregs[HME_SEBI_STAT >> 2] = intstatus;
/* Autoclear TX pending */
s->etxregs[HME_ETXI_PENDING >> 2] = 0;
sunhme_update_irq(s);
}
/* TX FIFO now clear */
intstatus = s->sebregs[HME_SEBI_STAT >> 2];
intstatus |= HME_SEB_STAT_TXALL;
s->sebregs[HME_SEBI_STAT >> 2] = intstatus;
sunhme_update_irq(s);
}
static bool sunhme_can_receive(NetClientState *nc)
{
SunHMEState *s = qemu_get_nic_opaque(nc);
return !!(s->macregs[HME_MACI_RXCFG >> 2] & HME_MAC_RXCFG_ENABLE);
}
static void sunhme_link_status_changed(NetClientState *nc)
{
SunHMEState *s = qemu_get_nic_opaque(nc);
if (nc->link_down) {
s->miiregs[MII_ANLPAR] &= ~MII_ANLPAR_TXFD;
s->miiregs[MII_BMSR] &= ~MII_BMSR_LINK_ST;
} else {
s->miiregs[MII_ANLPAR] |= MII_ANLPAR_TXFD;
s->miiregs[MII_BMSR] |= MII_BMSR_LINK_ST;
}
/* Exact bits unknown */
s->mifregs[HME_MIFI_STAT >> 2] = 0xffff;
sunhme_update_irq(s);
}
static inline int sunhme_get_rx_ring_count(SunHMEState *s)
{
uint32_t rings = (s->erxregs[HME_ERXI_CFG >> 2] & HME_ERX_CFG_RINGSIZE)
>> HME_ERX_CFG_RINGSIZE_SHIFT;
switch (rings) {
case 0:
return 32;
case 1:
return 64;
case 2:
return 128;
case 3:
return 256;
}
return 0;
}
static inline int sunhme_get_rx_ring_nr(SunHMEState *s)
{
return s->erxregs[HME_ERXI_RING >> 2] & HME_ERXI_RING_OFFSET;
}
static inline void sunhme_set_rx_ring_nr(SunHMEState *s, int i)
{
uint32_t ring = s->erxregs[HME_ERXI_RING >> 2] & ~HME_ERXI_RING_OFFSET;
ring |= i & HME_ERXI_RING_OFFSET;
s->erxregs[HME_ERXI_RING >> 2] = ring;
}
static ssize_t sunhme_receive(NetClientState *nc, const uint8_t *buf,
size_t size)
{
SunHMEState *s = qemu_get_nic_opaque(nc);
PCIDevice *d = PCI_DEVICE(s);
dma_addr_t rb, addr;
uint32_t intstatus, status, buffer, buffersize, sum;
uint16_t csum;
int nr, cr, len, rxoffset, csum_offset;
trace_sunhme_rx_incoming(size);
/* Do nothing if MAC RX disabled */
if (!(s->macregs[HME_MACI_RXCFG >> 2] & HME_MAC_RXCFG_ENABLE)) {
return 0;
}
trace_sunhme_rx_filter_destmac(buf[0], buf[1], buf[2],
buf[3], buf[4], buf[5]);
/* Check destination MAC address */
if (!(s->macregs[HME_MACI_RXCFG >> 2] & HME_MAC_RXCFG_PMISC)) {
/* Try and match local MAC address */
if (((s->macregs[HME_MACI_MACADDR0 >> 2] & 0xff00) >> 8) == buf[0] &&
(s->macregs[HME_MACI_MACADDR0 >> 2] & 0xff) == buf[1] &&
((s->macregs[HME_MACI_MACADDR1 >> 2] & 0xff00) >> 8) == buf[2] &&
(s->macregs[HME_MACI_MACADDR1 >> 2] & 0xff) == buf[3] &&
((s->macregs[HME_MACI_MACADDR2 >> 2] & 0xff00) >> 8) == buf[4] &&
(s->macregs[HME_MACI_MACADDR2 >> 2] & 0xff) == buf[5]) {
/* Matched local MAC address */
trace_sunhme_rx_filter_local_match();
} else if (buf[0] == 0xff && buf[1] == 0xff && buf[2] == 0xff &&
buf[3] == 0xff && buf[4] == 0xff && buf[5] == 0xff) {
/* Matched broadcast address */
trace_sunhme_rx_filter_bcast_match();
} else if (s->macregs[HME_MACI_RXCFG >> 2] & HME_MAC_RXCFG_HENABLE) {
/* Didn't match local address, check hash filter */
int mcast_idx = net_crc32_le(buf, ETH_ALEN) >> 26;
if (!(s->macregs[(HME_MACI_HASHTAB0 >> 2) - (mcast_idx >> 4)] &
(1 << (mcast_idx & 0xf)))) {
/* Didn't match hash filter */
trace_sunhme_rx_filter_hash_nomatch();
trace_sunhme_rx_filter_reject();
return -1;
} else {
trace_sunhme_rx_filter_hash_match();
}
} else {
/* Not for us */
trace_sunhme_rx_filter_reject();
return -1;
}
} else {
trace_sunhme_rx_filter_promisc_match();
}
trace_sunhme_rx_filter_accept();
rb = s->erxregs[HME_ERXI_RING >> 2] & HME_ERXI_RING_ADDR;
nr = sunhme_get_rx_ring_count(s);
cr = sunhme_get_rx_ring_nr(s);
pci_dma_read(d, rb + cr * HME_DESC_SIZE, &status, 4);
pci_dma_read(d, rb + cr * HME_DESC_SIZE + 4, &buffer, 4);
/* If we don't own the current descriptor then indicate overflow error */
if (!(status & HME_XD_OWN)) {
s->sebregs[HME_SEBI_STAT >> 2] |= HME_SEB_STAT_NORXD;
sunhme_update_irq(s);
trace_sunhme_rx_norxd();
return -1;
}
rxoffset = (s->erxregs[HME_ERXI_CFG >> 2] & HME_ERX_CFG_BYTEOFFSET) >>
HME_ERX_CFG_BYTEOFFSET_SHIFT;
addr = buffer + rxoffset;
buffersize = (status & HME_XD_RXLENMSK) >> HME_XD_RXLENSHIFT;
/* Detect receive overflow */
len = size;
if (size > buffersize) {
status |= HME_XD_OFL;
len = buffersize;
}
pci_dma_write(d, addr, buf, len);
trace_sunhme_rx_desc(buffer, rxoffset, status, len, cr, nr);
/* Calculate the receive checksum */
csum_offset = (s->erxregs[HME_ERXI_CFG >> 2] & HME_ERX_CFG_CSUMSTART) >>
HME_ERX_CFG_CSUMSHIFT << 1;
sum = 0;
sum += net_checksum_add(len - csum_offset, (uint8_t *)buf + csum_offset);
csum = net_checksum_finish(sum);
trace_sunhme_rx_xsum_calc(csum);
/* Update status */
status &= ~HME_XD_OWN;
status &= ~HME_XD_RXLENMSK;
status |= len << HME_XD_RXLENSHIFT;
status &= ~HME_XD_RXCKSUM;
status |= csum;
pci_dma_write(d, rb + cr * HME_DESC_SIZE, &status, 4);
cr++;
if (cr >= nr) {
cr = 0;
}
sunhme_set_rx_ring_nr(s, cr);
/* Indicate RX complete */
intstatus = s->sebregs[HME_SEBI_STAT >> 2];
intstatus |= HME_SEB_STAT_RXTOHOST;
s->sebregs[HME_SEBI_STAT >> 2] = intstatus;
sunhme_update_irq(s);
return len;
}
static NetClientInfo net_sunhme_info = {
.type = NET_CLIENT_DRIVER_NIC,
.size = sizeof(NICState),
.can_receive = sunhme_can_receive,
.receive = sunhme_receive,
.link_status_changed = sunhme_link_status_changed,
};
static void sunhme_realize(PCIDevice *pci_dev, Error **errp)
{
SunHMEState *s = SUNHME(pci_dev);
DeviceState *d = DEVICE(pci_dev);
uint8_t *pci_conf;
pci_conf = pci_dev->config;
pci_conf[PCI_INTERRUPT_PIN] = 1; /* interrupt pin A */
memory_region_init(&s->hme, OBJECT(pci_dev), "sunhme", HME_REG_SIZE);
pci_register_bar(pci_dev, 0, PCI_BASE_ADDRESS_SPACE_MEMORY, &s->hme);
memory_region_init_io(&s->sebreg, OBJECT(pci_dev), &sunhme_seb_ops, s,
"sunhme.seb", HME_SEB_REG_SIZE);
memory_region_add_subregion(&s->hme, 0, &s->sebreg);
memory_region_init_io(&s->etxreg, OBJECT(pci_dev), &sunhme_etx_ops, s,
"sunhme.etx", HME_ETX_REG_SIZE);
memory_region_add_subregion(&s->hme, 0x2000, &s->etxreg);
memory_region_init_io(&s->erxreg, OBJECT(pci_dev), &sunhme_erx_ops, s,
"sunhme.erx", HME_ERX_REG_SIZE);
memory_region_add_subregion(&s->hme, 0x4000, &s->erxreg);
memory_region_init_io(&s->macreg, OBJECT(pci_dev), &sunhme_mac_ops, s,
"sunhme.mac", HME_MAC_REG_SIZE);
memory_region_add_subregion(&s->hme, 0x6000, &s->macreg);
memory_region_init_io(&s->mifreg, OBJECT(pci_dev), &sunhme_mif_ops, s,
"sunhme.mif", HME_MIF_REG_SIZE);
memory_region_add_subregion(&s->hme, 0x7000, &s->mifreg);
qemu_macaddr_default_if_unset(&s->conf.macaddr);
s->nic = qemu_new_nic(&net_sunhme_info, &s->conf,
object_get_typename(OBJECT(d)), d->id, s);
qemu_format_nic_info_str(qemu_get_queue(s->nic), s->conf.macaddr.a);
}
static void sunhme_instance_init(Object *obj)
{
SunHMEState *s = SUNHME(obj);
device_add_bootindex_property(obj, &s->conf.bootindex,
"bootindex", "/ethernet-phy@0",
DEVICE(obj));
}
static void sunhme_reset(DeviceState *ds)
{
SunHMEState *s = SUNHME(ds);
/* Configure internal transceiver */
s->mifregs[HME_MIFI_CFG >> 2] |= HME_MIF_CFG_MDI0;
/* Advetise auto, 100Mbps FD */
s->miiregs[MII_ANAR] = MII_ANAR_TXFD;
s->miiregs[MII_BMSR] = MII_BMSR_AUTONEG | MII_BMSR_100TX_FD |
MII_BMSR_AN_COMP;
if (!qemu_get_queue(s->nic)->link_down) {
s->miiregs[MII_ANLPAR] |= MII_ANLPAR_TXFD;
s->miiregs[MII_BMSR] |= MII_BMSR_LINK_ST;
}
/* Set manufacturer */
s->miiregs[MII_PHYID1] = DP83840_PHYID1;
s->miiregs[MII_PHYID2] = DP83840_PHYID2;
/* Configure default interrupt mask */
s->mifregs[HME_MIFI_IMASK >> 2] = 0xffff;
s->sebregs[HME_SEBI_IMASK >> 2] = 0xff7fffff;
}
static const VMStateDescription vmstate_hme = {
.name = "sunhme",
.version_id = 0,
.minimum_version_id = 0,
.fields = (VMStateField[]) {
VMSTATE_PCI_DEVICE(parent_obj, SunHMEState),
VMSTATE_MACADDR(conf.macaddr, SunHMEState),
VMSTATE_UINT32_ARRAY(sebregs, SunHMEState, (HME_SEB_REG_SIZE >> 2)),
VMSTATE_UINT32_ARRAY(etxregs, SunHMEState, (HME_ETX_REG_SIZE >> 2)),
VMSTATE_UINT32_ARRAY(erxregs, SunHMEState, (HME_ERX_REG_SIZE >> 2)),
VMSTATE_UINT32_ARRAY(macregs, SunHMEState, (HME_MAC_REG_SIZE >> 2)),
VMSTATE_UINT32_ARRAY(mifregs, SunHMEState, (HME_MIF_REG_SIZE >> 2)),
VMSTATE_UINT16_ARRAY(miiregs, SunHMEState, HME_MII_REGS_SIZE),
VMSTATE_END_OF_LIST()
}
};
static void sunhme_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
k->realize = sunhme_realize;
k->vendor_id = PCI_VENDOR_ID_SUN;
k->device_id = PCI_DEVICE_ID_SUN_HME;
k->class_id = PCI_CLASS_NETWORK_ETHERNET;
dc->vmsd = &vmstate_hme;
dc->reset = sunhme_reset;
device_class_set_props(dc, sunhme_properties);
set_bit(DEVICE_CATEGORY_NETWORK, dc->categories);
}
static const TypeInfo sunhme_info = {
.name = TYPE_SUNHME,
.parent = TYPE_PCI_DEVICE,
.class_init = sunhme_class_init,
.instance_size = sizeof(SunHMEState),
.instance_init = sunhme_instance_init,
.interfaces = (InterfaceInfo[]) {
{ INTERFACE_CONVENTIONAL_PCI_DEVICE },
{ }
}
};
static void sunhme_register_types(void)
{
type_register_static(&sunhme_info);
}
type_init(sunhme_register_types)