linux-sg2042/drivers/net/pcnet32.c

3063 lines
83 KiB
C
Raw Normal View History

/* pcnet32.c: An AMD PCnet32 ethernet driver for linux. */
/*
* Copyright 1996-1999 Thomas Bogendoerfer
*
* Derived from the lance driver written 1993,1994,1995 by Donald Becker.
*
* Copyright 1993 United States Government as represented by the
* Director, National Security Agency.
*
* This software may be used and distributed according to the terms
* of the GNU General Public License, incorporated herein by reference.
*
* This driver is for PCnet32 and PCnetPCI based ethercards
*/
/**************************************************************************
* 23 Oct, 2000.
* Fixed a few bugs, related to running the controller in 32bit mode.
*
* Carsten Langgaard, carstenl@mips.com
* Copyright (C) 2000 MIPS Technologies, Inc. All rights reserved.
*
*************************************************************************/
#define DRV_NAME "pcnet32"
#define DRV_VERSION "1.35"
#define DRV_RELDATE "21.Apr.2008"
#define PFX DRV_NAME ": "
2006-03-22 05:15:44 +08:00
static const char *const version =
DRV_NAME ".c:v" DRV_VERSION " " DRV_RELDATE " tsbogend@alpha.franken.de\n";
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/ioport.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/pci.h>
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/ethtool.h>
#include <linux/mii.h>
#include <linux/crc32.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/if_ether.h>
#include <linux/skbuff.h>
#include <linux/spinlock.h>
#include <linux/moduleparam.h>
#include <linux/bitops.h>
#include <asm/dma.h>
#include <asm/io.h>
#include <asm/uaccess.h>
#include <asm/irq.h>
/*
* PCI device identifiers for "new style" Linux PCI Device Drivers
*/
static struct pci_device_id pcnet32_pci_tbl[] = {
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_LANCE_HOME), },
{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_LANCE), },
2006-03-22 05:15:44 +08:00
/*
* Adapters that were sold with IBM's RS/6000 or pSeries hardware have
* the incorrect vendor id.
*/
{ PCI_DEVICE(PCI_VENDOR_ID_TRIDENT, PCI_DEVICE_ID_AMD_LANCE),
.class = (PCI_CLASS_NETWORK_ETHERNET << 8), .class_mask = 0xffff00, },
2006-03-22 05:15:44 +08:00
{ } /* terminate list */
};
2006-03-22 05:15:44 +08:00
MODULE_DEVICE_TABLE(pci, pcnet32_pci_tbl);
static int cards_found;
/*
* VLB I/O addresses
*/
static unsigned int pcnet32_portlist[] __initdata =
2006-03-22 05:15:44 +08:00
{ 0x300, 0x320, 0x340, 0x360, 0 };
static int pcnet32_debug = 0;
2006-03-22 05:15:44 +08:00
static int tx_start = 1; /* Mapping -- 0:20, 1:64, 2:128, 3:~220 (depends on chip vers) */
static int pcnet32vlb; /* check for VLB cards ? */
static struct net_device *pcnet32_dev;
static int max_interrupt_work = 2;
static int rx_copybreak = 200;
#define PCNET32_PORT_AUI 0x00
#define PCNET32_PORT_10BT 0x01
#define PCNET32_PORT_GPSI 0x02
#define PCNET32_PORT_MII 0x03
#define PCNET32_PORT_PORTSEL 0x03
#define PCNET32_PORT_ASEL 0x04
#define PCNET32_PORT_100 0x40
#define PCNET32_PORT_FD 0x80
#define PCNET32_DMA_MASK 0xffffffff
#define PCNET32_WATCHDOG_TIMEOUT (jiffies + (2 * HZ))
#define PCNET32_BLINK_TIMEOUT (jiffies + (HZ/4))
/*
* table to translate option values from tulip
* to internal options
*/
2006-03-04 10:33:57 +08:00
static const unsigned char options_mapping[] = {
2006-03-22 05:15:44 +08:00
PCNET32_PORT_ASEL, /* 0 Auto-select */
PCNET32_PORT_AUI, /* 1 BNC/AUI */
PCNET32_PORT_AUI, /* 2 AUI/BNC */
PCNET32_PORT_ASEL, /* 3 not supported */
PCNET32_PORT_10BT | PCNET32_PORT_FD, /* 4 10baseT-FD */
PCNET32_PORT_ASEL, /* 5 not supported */
PCNET32_PORT_ASEL, /* 6 not supported */
PCNET32_PORT_ASEL, /* 7 not supported */
PCNET32_PORT_ASEL, /* 8 not supported */
PCNET32_PORT_MII, /* 9 MII 10baseT */
PCNET32_PORT_MII | PCNET32_PORT_FD, /* 10 MII 10baseT-FD */
PCNET32_PORT_MII, /* 11 MII (autosel) */
PCNET32_PORT_10BT, /* 12 10BaseT */
PCNET32_PORT_MII | PCNET32_PORT_100, /* 13 MII 100BaseTx */
/* 14 MII 100BaseTx-FD */
PCNET32_PORT_MII | PCNET32_PORT_100 | PCNET32_PORT_FD,
PCNET32_PORT_ASEL /* 15 not supported */
};
static const char pcnet32_gstrings_test[][ETH_GSTRING_LEN] = {
2006-03-22 05:15:44 +08:00
"Loopback test (offline)"
};
2006-03-22 05:15:44 +08:00
#define PCNET32_TEST_LEN ARRAY_SIZE(pcnet32_gstrings_test)
#define PCNET32_NUM_REGS 136
2006-03-22 05:15:44 +08:00
#define MAX_UNITS 8 /* More are supported, limit only on options */
static int options[MAX_UNITS];
static int full_duplex[MAX_UNITS];
static int homepna[MAX_UNITS];
/*
* Theory of Operation
*
* This driver uses the same software structure as the normal lance
* driver. So look for a verbose description in lance.c. The differences
* to the normal lance driver is the use of the 32bit mode of PCnet32
* and PCnetPCI chips. Because these chips are 32bit chips, there is no
* 16MB limitation and we don't need bounce buffers.
*/
/*
* Set the number of Tx and Rx buffers, using Log_2(# buffers).
* Reasonable default values are 4 Tx buffers, and 16 Rx buffers.
* That translates to 2 (4 == 2^^2) and 4 (16 == 2^^4).
*/
#ifndef PCNET32_LOG_TX_BUFFERS
#define PCNET32_LOG_TX_BUFFERS 4
#define PCNET32_LOG_RX_BUFFERS 5
#define PCNET32_LOG_MAX_TX_BUFFERS 9 /* 2^9 == 512 */
#define PCNET32_LOG_MAX_RX_BUFFERS 9
#endif
#define TX_RING_SIZE (1 << (PCNET32_LOG_TX_BUFFERS))
#define TX_MAX_RING_SIZE (1 << (PCNET32_LOG_MAX_TX_BUFFERS))
#define RX_RING_SIZE (1 << (PCNET32_LOG_RX_BUFFERS))
#define RX_MAX_RING_SIZE (1 << (PCNET32_LOG_MAX_RX_BUFFERS))
#define PKT_BUF_SKB 1544
/* actual buffer length after being aligned */
#define PKT_BUF_SIZE (PKT_BUF_SKB - NET_IP_ALIGN)
/* chip wants twos complement of the (aligned) buffer length */
#define NEG_BUF_SIZE (NET_IP_ALIGN - PKT_BUF_SKB)
/* Offsets from base I/O address. */
#define PCNET32_WIO_RDP 0x10
#define PCNET32_WIO_RAP 0x12
#define PCNET32_WIO_RESET 0x14
#define PCNET32_WIO_BDP 0x16
#define PCNET32_DWIO_RDP 0x10
#define PCNET32_DWIO_RAP 0x14
#define PCNET32_DWIO_RESET 0x18
#define PCNET32_DWIO_BDP 0x1C
#define PCNET32_TOTAL_SIZE 0x20
#define CSR0 0
#define CSR0_INIT 0x1
#define CSR0_START 0x2
#define CSR0_STOP 0x4
#define CSR0_TXPOLL 0x8
#define CSR0_INTEN 0x40
#define CSR0_IDON 0x0100
#define CSR0_NORMAL (CSR0_START | CSR0_INTEN)
#define PCNET32_INIT_LOW 1
#define PCNET32_INIT_HIGH 2
#define CSR3 3
#define CSR4 4
#define CSR5 5
#define CSR5_SUSPEND 0x0001
#define CSR15 15
#define PCNET32_MC_FILTER 8
#define PCNET32_79C970A 0x2621
/* The PCNET32 Rx and Tx ring descriptors. */
struct pcnet32_rx_head {
__le32 base;
__le16 buf_length; /* two`s complement of length */
__le16 status;
__le32 msg_length;
__le32 reserved;
};
struct pcnet32_tx_head {
__le32 base;
__le16 length; /* two`s complement of length */
__le16 status;
__le32 misc;
__le32 reserved;
};
/* The PCNET32 32-Bit initialization block, described in databook. */
struct pcnet32_init_block {
__le16 mode;
__le16 tlen_rlen;
u8 phys_addr[6];
__le16 reserved;
__le32 filter[2];
2006-03-22 05:15:44 +08:00
/* Receive and transmit ring base, along with extra bits. */
__le32 rx_ring;
__le32 tx_ring;
};
/* PCnet32 access functions */
struct pcnet32_access {
2006-03-22 05:15:44 +08:00
u16 (*read_csr) (unsigned long, int);
void (*write_csr) (unsigned long, int, u16);
u16 (*read_bcr) (unsigned long, int);
void (*write_bcr) (unsigned long, int, u16);
u16 (*read_rap) (unsigned long);
void (*write_rap) (unsigned long, u16);
void (*reset) (unsigned long);
};
/*
* The first field of pcnet32_private is read by the ethernet device
* so the structure should be allocated using pci_alloc_consistent().
*/
struct pcnet32_private {
struct pcnet32_init_block *init_block;
2006-03-22 05:15:44 +08:00
/* The Tx and Rx ring entries must be aligned on 16-byte boundaries in 32bit mode. */
struct pcnet32_rx_head *rx_ring;
struct pcnet32_tx_head *tx_ring;
dma_addr_t init_dma_addr;/* DMA address of beginning of the init block,
returned by pci_alloc_consistent */
struct pci_dev *pci_dev;
const char *name;
2006-03-22 05:15:44 +08:00
/* The saved address of a sent-in-place packet/buffer, for skfree(). */
struct sk_buff **tx_skbuff;
struct sk_buff **rx_skbuff;
dma_addr_t *tx_dma_addr;
dma_addr_t *rx_dma_addr;
struct pcnet32_access a;
spinlock_t lock; /* Guard lock */
unsigned int cur_rx, cur_tx; /* The next free ring entry */
unsigned int rx_ring_size; /* current rx ring size */
unsigned int tx_ring_size; /* current tx ring size */
unsigned int rx_mod_mask; /* rx ring modular mask */
unsigned int tx_mod_mask; /* tx ring modular mask */
unsigned short rx_len_bits;
unsigned short tx_len_bits;
dma_addr_t rx_ring_dma_addr;
dma_addr_t tx_ring_dma_addr;
unsigned int dirty_rx, /* ring entries to be freed. */
dirty_tx;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
struct net_device *dev;
struct napi_struct napi;
char tx_full;
char phycount; /* number of phys found */
int options;
unsigned int shared_irq:1, /* shared irq possible */
dxsuflo:1, /* disable transmit stop on uflo */
mii:1; /* mii port available */
struct net_device *next;
struct mii_if_info mii_if;
struct timer_list watchdog_timer;
struct timer_list blink_timer;
u32 msg_enable; /* debug message level */
2006-03-22 05:15:44 +08:00
/* each bit indicates an available PHY */
u32 phymask;
unsigned short chip_version; /* which variant this is */
};
2006-03-22 05:15:44 +08:00
static int pcnet32_probe_pci(struct pci_dev *, const struct pci_device_id *);
static int pcnet32_probe1(unsigned long, int, struct pci_dev *);
static int pcnet32_open(struct net_device *);
static int pcnet32_init_ring(struct net_device *);
static netdev_tx_t pcnet32_start_xmit(struct sk_buff *,
struct net_device *);
2006-03-22 05:15:44 +08:00
static void pcnet32_tx_timeout(struct net_device *dev);
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t pcnet32_interrupt(int, void *);
2006-03-22 05:15:44 +08:00
static int pcnet32_close(struct net_device *);
static struct net_device_stats *pcnet32_get_stats(struct net_device *);
static void pcnet32_load_multicast(struct net_device *dev);
static void pcnet32_set_multicast_list(struct net_device *);
2006-03-22 05:15:44 +08:00
static int pcnet32_ioctl(struct net_device *, struct ifreq *, int);
static void pcnet32_watchdog(struct net_device *);
static int mdio_read(struct net_device *dev, int phy_id, int reg_num);
2006-03-22 05:15:44 +08:00
static void mdio_write(struct net_device *dev, int phy_id, int reg_num,
int val);
static void pcnet32_restart(struct net_device *dev, unsigned int csr0_bits);
static void pcnet32_ethtool_test(struct net_device *dev,
2006-03-22 05:15:44 +08:00
struct ethtool_test *eth_test, u64 * data);
static int pcnet32_loopback_test(struct net_device *dev, uint64_t * data1);
static int pcnet32_phys_id(struct net_device *dev, u32 data);
static void pcnet32_led_blink_callback(struct net_device *dev);
static int pcnet32_get_regs_len(struct net_device *dev);
static void pcnet32_get_regs(struct net_device *dev, struct ethtool_regs *regs,
2006-03-22 05:15:44 +08:00
void *ptr);
static void pcnet32_purge_tx_ring(struct net_device *dev);
static int pcnet32_alloc_ring(struct net_device *dev, const char *name);
static void pcnet32_free_ring(struct net_device *dev);
static void pcnet32_check_media(struct net_device *dev, int verbose);
2006-03-22 05:15:44 +08:00
static u16 pcnet32_wio_read_csr(unsigned long addr, int index)
{
2006-03-22 05:15:44 +08:00
outw(index, addr + PCNET32_WIO_RAP);
return inw(addr + PCNET32_WIO_RDP);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_wio_write_csr(unsigned long addr, int index, u16 val)
{
2006-03-22 05:15:44 +08:00
outw(index, addr + PCNET32_WIO_RAP);
outw(val, addr + PCNET32_WIO_RDP);
}
2006-03-22 05:15:44 +08:00
static u16 pcnet32_wio_read_bcr(unsigned long addr, int index)
{
2006-03-22 05:15:44 +08:00
outw(index, addr + PCNET32_WIO_RAP);
return inw(addr + PCNET32_WIO_BDP);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_wio_write_bcr(unsigned long addr, int index, u16 val)
{
2006-03-22 05:15:44 +08:00
outw(index, addr + PCNET32_WIO_RAP);
outw(val, addr + PCNET32_WIO_BDP);
}
2006-03-22 05:15:44 +08:00
static u16 pcnet32_wio_read_rap(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
return inw(addr + PCNET32_WIO_RAP);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_wio_write_rap(unsigned long addr, u16 val)
{
2006-03-22 05:15:44 +08:00
outw(val, addr + PCNET32_WIO_RAP);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_wio_reset(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
inw(addr + PCNET32_WIO_RESET);
}
2006-03-22 05:15:44 +08:00
static int pcnet32_wio_check(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
outw(88, addr + PCNET32_WIO_RAP);
return (inw(addr + PCNET32_WIO_RAP) == 88);
}
static struct pcnet32_access pcnet32_wio = {
2006-03-22 05:15:44 +08:00
.read_csr = pcnet32_wio_read_csr,
.write_csr = pcnet32_wio_write_csr,
.read_bcr = pcnet32_wio_read_bcr,
.write_bcr = pcnet32_wio_write_bcr,
.read_rap = pcnet32_wio_read_rap,
.write_rap = pcnet32_wio_write_rap,
.reset = pcnet32_wio_reset
};
2006-03-22 05:15:44 +08:00
static u16 pcnet32_dwio_read_csr(unsigned long addr, int index)
{
2006-03-22 05:15:44 +08:00
outl(index, addr + PCNET32_DWIO_RAP);
return (inl(addr + PCNET32_DWIO_RDP) & 0xffff);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_dwio_write_csr(unsigned long addr, int index, u16 val)
{
2006-03-22 05:15:44 +08:00
outl(index, addr + PCNET32_DWIO_RAP);
outl(val, addr + PCNET32_DWIO_RDP);
}
2006-03-22 05:15:44 +08:00
static u16 pcnet32_dwio_read_bcr(unsigned long addr, int index)
{
2006-03-22 05:15:44 +08:00
outl(index, addr + PCNET32_DWIO_RAP);
return (inl(addr + PCNET32_DWIO_BDP) & 0xffff);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_dwio_write_bcr(unsigned long addr, int index, u16 val)
{
2006-03-22 05:15:44 +08:00
outl(index, addr + PCNET32_DWIO_RAP);
outl(val, addr + PCNET32_DWIO_BDP);
}
2006-03-22 05:15:44 +08:00
static u16 pcnet32_dwio_read_rap(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
return (inl(addr + PCNET32_DWIO_RAP) & 0xffff);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_dwio_write_rap(unsigned long addr, u16 val)
{
2006-03-22 05:15:44 +08:00
outl(val, addr + PCNET32_DWIO_RAP);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_dwio_reset(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
inl(addr + PCNET32_DWIO_RESET);
}
2006-03-22 05:15:44 +08:00
static int pcnet32_dwio_check(unsigned long addr)
{
2006-03-22 05:15:44 +08:00
outl(88, addr + PCNET32_DWIO_RAP);
return ((inl(addr + PCNET32_DWIO_RAP) & 0xffff) == 88);
}
static struct pcnet32_access pcnet32_dwio = {
2006-03-22 05:15:44 +08:00
.read_csr = pcnet32_dwio_read_csr,
.write_csr = pcnet32_dwio_write_csr,
.read_bcr = pcnet32_dwio_read_bcr,
.write_bcr = pcnet32_dwio_write_bcr,
.read_rap = pcnet32_dwio_read_rap,
.write_rap = pcnet32_dwio_write_rap,
.reset = pcnet32_dwio_reset
};
static void pcnet32_netif_stop(struct net_device *dev)
{
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
struct pcnet32_private *lp = netdev_priv(dev);
dev->trans_start = jiffies;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
napi_disable(&lp->napi);
netif_tx_disable(dev);
}
static void pcnet32_netif_start(struct net_device *dev)
{
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
struct pcnet32_private *lp = netdev_priv(dev);
ulong ioaddr = dev->base_addr;
u16 val;
netif_wake_queue(dev);
val = lp->a.read_csr(ioaddr, CSR3);
val &= 0x00ff;
lp->a.write_csr(ioaddr, CSR3, val);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
napi_enable(&lp->napi);
}
/*
* Allocate space for the new sized tx ring.
* Free old resources
* Save new resources.
* Any failure keeps old resources.
* Must be called with lp->lock held.
*/
static void pcnet32_realloc_tx_ring(struct net_device *dev,
struct pcnet32_private *lp,
unsigned int size)
{
dma_addr_t new_ring_dma_addr;
dma_addr_t *new_dma_addr_list;
struct pcnet32_tx_head *new_tx_ring;
struct sk_buff **new_skb_list;
pcnet32_purge_tx_ring(dev);
new_tx_ring = pci_alloc_consistent(lp->pci_dev,
sizeof(struct pcnet32_tx_head) *
(1 << size),
&new_ring_dma_addr);
if (new_tx_ring == NULL) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Consistent memory allocation failed.\n",
dev->name);
return;
}
memset(new_tx_ring, 0, sizeof(struct pcnet32_tx_head) * (1 << size));
new_dma_addr_list = kcalloc((1 << size), sizeof(dma_addr_t),
GFP_ATOMIC);
if (!new_dma_addr_list) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Memory allocation failed.\n", dev->name);
goto free_new_tx_ring;
}
new_skb_list = kcalloc((1 << size), sizeof(struct sk_buff *),
GFP_ATOMIC);
if (!new_skb_list) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Memory allocation failed.\n", dev->name);
goto free_new_lists;
}
kfree(lp->tx_skbuff);
kfree(lp->tx_dma_addr);
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_tx_head) *
lp->tx_ring_size, lp->tx_ring,
lp->tx_ring_dma_addr);
lp->tx_ring_size = (1 << size);
lp->tx_mod_mask = lp->tx_ring_size - 1;
lp->tx_len_bits = (size << 12);
lp->tx_ring = new_tx_ring;
lp->tx_ring_dma_addr = new_ring_dma_addr;
lp->tx_dma_addr = new_dma_addr_list;
lp->tx_skbuff = new_skb_list;
return;
free_new_lists:
kfree(new_dma_addr_list);
free_new_tx_ring:
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_tx_head) *
(1 << size),
new_tx_ring,
new_ring_dma_addr);
return;
}
/*
* Allocate space for the new sized rx ring.
* Re-use old receive buffers.
* alloc extra buffers
* free unneeded buffers
* free unneeded buffers
* Save new resources.
* Any failure keeps old resources.
* Must be called with lp->lock held.
*/
static void pcnet32_realloc_rx_ring(struct net_device *dev,
struct pcnet32_private *lp,
unsigned int size)
{
dma_addr_t new_ring_dma_addr;
dma_addr_t *new_dma_addr_list;
struct pcnet32_rx_head *new_rx_ring;
struct sk_buff **new_skb_list;
int new, overlap;
new_rx_ring = pci_alloc_consistent(lp->pci_dev,
sizeof(struct pcnet32_rx_head) *
(1 << size),
&new_ring_dma_addr);
if (new_rx_ring == NULL) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Consistent memory allocation failed.\n",
dev->name);
return;
}
memset(new_rx_ring, 0, sizeof(struct pcnet32_rx_head) * (1 << size));
new_dma_addr_list = kcalloc((1 << size), sizeof(dma_addr_t),
GFP_ATOMIC);
if (!new_dma_addr_list) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Memory allocation failed.\n", dev->name);
goto free_new_rx_ring;
}
new_skb_list = kcalloc((1 << size), sizeof(struct sk_buff *),
GFP_ATOMIC);
if (!new_skb_list) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Memory allocation failed.\n", dev->name);
goto free_new_lists;
}
/* first copy the current receive buffers */
overlap = min(size, lp->rx_ring_size);
for (new = 0; new < overlap; new++) {
new_rx_ring[new] = lp->rx_ring[new];
new_dma_addr_list[new] = lp->rx_dma_addr[new];
new_skb_list[new] = lp->rx_skbuff[new];
}
/* now allocate any new buffers needed */
for (; new < size; new++ ) {
struct sk_buff *rx_skbuff;
new_skb_list[new] = dev_alloc_skb(PKT_BUF_SKB);
if (!(rx_skbuff = new_skb_list[new])) {
/* keep the original lists and buffers */
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: pcnet32_realloc_rx_ring dev_alloc_skb failed.\n",
dev->name);
goto free_all_new;
}
skb_reserve(rx_skbuff, NET_IP_ALIGN);
new_dma_addr_list[new] =
pci_map_single(lp->pci_dev, rx_skbuff->data,
PKT_BUF_SIZE, PCI_DMA_FROMDEVICE);
new_rx_ring[new].base = cpu_to_le32(new_dma_addr_list[new]);
new_rx_ring[new].buf_length = cpu_to_le16(NEG_BUF_SIZE);
new_rx_ring[new].status = cpu_to_le16(0x8000);
}
/* and free any unneeded buffers */
for (; new < lp->rx_ring_size; new++) {
if (lp->rx_skbuff[new]) {
pci_unmap_single(lp->pci_dev, lp->rx_dma_addr[new],
PKT_BUF_SIZE, PCI_DMA_FROMDEVICE);
dev_kfree_skb(lp->rx_skbuff[new]);
}
}
kfree(lp->rx_skbuff);
kfree(lp->rx_dma_addr);
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_rx_head) *
lp->rx_ring_size, lp->rx_ring,
lp->rx_ring_dma_addr);
lp->rx_ring_size = (1 << size);
lp->rx_mod_mask = lp->rx_ring_size - 1;
lp->rx_len_bits = (size << 4);
lp->rx_ring = new_rx_ring;
lp->rx_ring_dma_addr = new_ring_dma_addr;
lp->rx_dma_addr = new_dma_addr_list;
lp->rx_skbuff = new_skb_list;
return;
free_all_new:
for (; --new >= lp->rx_ring_size; ) {
if (new_skb_list[new]) {
pci_unmap_single(lp->pci_dev, new_dma_addr_list[new],
PKT_BUF_SIZE, PCI_DMA_FROMDEVICE);
dev_kfree_skb(new_skb_list[new]);
}
}
kfree(new_skb_list);
free_new_lists:
kfree(new_dma_addr_list);
free_new_rx_ring:
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_rx_head) *
(1 << size),
new_rx_ring,
new_ring_dma_addr);
return;
}
static void pcnet32_purge_rx_ring(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
int i;
/* free all allocated skbuffs */
for (i = 0; i < lp->rx_ring_size; i++) {
lp->rx_ring[i].status = 0; /* CPU owns buffer */
wmb(); /* Make sure adapter sees owner change */
if (lp->rx_skbuff[i]) {
pci_unmap_single(lp->pci_dev, lp->rx_dma_addr[i],
PKT_BUF_SIZE, PCI_DMA_FROMDEVICE);
dev_kfree_skb_any(lp->rx_skbuff[i]);
}
lp->rx_skbuff[i] = NULL;
lp->rx_dma_addr[i] = 0;
}
}
#ifdef CONFIG_NET_POLL_CONTROLLER
static void pcnet32_poll_controller(struct net_device *dev)
{
2006-03-22 05:15:44 +08:00
disable_irq(dev->irq);
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
pcnet32_interrupt(0, dev);
2006-03-22 05:15:44 +08:00
enable_irq(dev->irq);
}
#endif
static int pcnet32_get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
int r = -EOPNOTSUPP;
2006-03-22 05:15:44 +08:00
if (lp->mii) {
spin_lock_irqsave(&lp->lock, flags);
mii_ethtool_gset(&lp->mii_if, cmd);
spin_unlock_irqrestore(&lp->lock, flags);
r = 0;
}
return r;
}
static int pcnet32_set_settings(struct net_device *dev, struct ethtool_cmd *cmd)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
int r = -EOPNOTSUPP;
2006-03-22 05:15:44 +08:00
if (lp->mii) {
spin_lock_irqsave(&lp->lock, flags);
r = mii_ethtool_sset(&lp->mii_if, cmd);
spin_unlock_irqrestore(&lp->lock, flags);
}
return r;
}
2006-03-22 05:15:44 +08:00
static void pcnet32_get_drvinfo(struct net_device *dev,
struct ethtool_drvinfo *info)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
strcpy(info->driver, DRV_NAME);
strcpy(info->version, DRV_VERSION);
if (lp->pci_dev)
strcpy(info->bus_info, pci_name(lp->pci_dev));
else
sprintf(info->bus_info, "VLB 0x%lx", dev->base_addr);
}
static u32 pcnet32_get_link(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
int r;
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
if (lp->mii) {
r = mii_link_ok(&lp->mii_if);
} else if (lp->chip_version >= PCNET32_79C970A) {
2006-03-22 05:15:44 +08:00
ulong ioaddr = dev->base_addr; /* card base I/O address */
r = (lp->a.read_bcr(ioaddr, 4) != 0xc0);
} else { /* can not detect link on really old chips */
r = 1;
2006-03-22 05:15:44 +08:00
}
spin_unlock_irqrestore(&lp->lock, flags);
return r;
}
static u32 pcnet32_get_msglevel(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
return lp->msg_enable;
}
static void pcnet32_set_msglevel(struct net_device *dev, u32 value)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
lp->msg_enable = value;
}
static int pcnet32_nway_reset(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
int r = -EOPNOTSUPP;
2006-03-22 05:15:44 +08:00
if (lp->mii) {
spin_lock_irqsave(&lp->lock, flags);
r = mii_nway_restart(&lp->mii_if);
spin_unlock_irqrestore(&lp->lock, flags);
}
return r;
}
2006-03-22 05:15:44 +08:00
static void pcnet32_get_ringparam(struct net_device *dev,
struct ethtool_ringparam *ering)
{
struct pcnet32_private *lp = netdev_priv(dev);
ering->tx_max_pending = TX_MAX_RING_SIZE;
ering->tx_pending = lp->tx_ring_size;
ering->rx_max_pending = RX_MAX_RING_SIZE;
ering->rx_pending = lp->rx_ring_size;
}
2006-03-22 05:15:44 +08:00
static int pcnet32_set_ringparam(struct net_device *dev,
struct ethtool_ringparam *ering)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
unsigned int size;
ulong ioaddr = dev->base_addr;
2006-03-22 05:15:44 +08:00
int i;
if (ering->rx_mini_pending || ering->rx_jumbo_pending)
return -EINVAL;
if (netif_running(dev))
pcnet32_netif_stop(dev);
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP); /* stop the chip */
size = min(ering->tx_pending, (unsigned int)TX_MAX_RING_SIZE);
2006-03-22 05:15:44 +08:00
/* set the minimum ring size to 4, to allow the loopback test to work
* unchanged.
*/
for (i = 2; i <= PCNET32_LOG_MAX_TX_BUFFERS; i++) {
if (size <= (1 << i))
2006-03-22 05:15:44 +08:00
break;
}
if ((1 << i) != lp->tx_ring_size)
pcnet32_realloc_tx_ring(dev, lp, i);
size = min(ering->rx_pending, (unsigned int)RX_MAX_RING_SIZE);
2006-03-22 05:15:44 +08:00
for (i = 2; i <= PCNET32_LOG_MAX_RX_BUFFERS; i++) {
if (size <= (1 << i))
2006-03-22 05:15:44 +08:00
break;
}
if ((1 << i) != lp->rx_ring_size)
pcnet32_realloc_rx_ring(dev, lp, i);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
lp->napi.weight = lp->rx_ring_size / 2;
if (netif_running(dev)) {
pcnet32_netif_start(dev);
pcnet32_restart(dev, CSR0_NORMAL);
2006-03-22 05:15:44 +08:00
}
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
if (netif_msg_drv(lp))
printk(KERN_INFO
2006-03-22 05:15:44 +08:00
"%s: Ring Param Settings: RX: %d, TX: %d\n", dev->name,
lp->rx_ring_size, lp->tx_ring_size);
2006-03-22 05:15:44 +08:00
return 0;
}
2006-03-22 05:15:44 +08:00
static void pcnet32_get_strings(struct net_device *dev, u32 stringset,
u8 * data)
{
2006-03-22 05:15:44 +08:00
memcpy(data, pcnet32_gstrings_test, sizeof(pcnet32_gstrings_test));
}
static int pcnet32_get_sset_count(struct net_device *dev, int sset)
{
switch (sset) {
case ETH_SS_TEST:
return PCNET32_TEST_LEN;
default:
return -EOPNOTSUPP;
}
}
static void pcnet32_ethtool_test(struct net_device *dev,
2006-03-22 05:15:44 +08:00
struct ethtool_test *test, u64 * data)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int rc;
if (test->flags == ETH_TEST_FL_OFFLINE) {
rc = pcnet32_loopback_test(dev, data);
if (rc) {
if (netif_msg_hw(lp))
printk(KERN_DEBUG "%s: Loopback test failed.\n",
dev->name);
test->flags |= ETH_TEST_FL_FAILED;
} else if (netif_msg_hw(lp))
printk(KERN_DEBUG "%s: Loopback test passed.\n",
dev->name);
} else if (netif_msg_hw(lp))
2006-03-22 05:15:44 +08:00
printk(KERN_DEBUG
"%s: No tests to run (specify 'Offline' on ethtool).",
dev->name);
} /* end pcnet32_ethtool_test */
2006-03-22 05:15:44 +08:00
static int pcnet32_loopback_test(struct net_device *dev, uint64_t * data1)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
struct pcnet32_access *a = &lp->a; /* access to registers */
ulong ioaddr = dev->base_addr; /* card base I/O address */
struct sk_buff *skb; /* sk buff */
int x, i; /* counters */
int numbuffs = 4; /* number of TX/RX buffers and descs */
u16 status = 0x8300; /* TX ring status */
__le16 teststatus; /* test of ring status */
2006-03-22 05:15:44 +08:00
int rc; /* return code */
int size; /* size of packets */
unsigned char *packet; /* source packet data */
static const int data_len = 60; /* length of source packets */
unsigned long flags;
unsigned long ticks;
rc = 1; /* default to fail */
if (netif_running(dev))
pcnet32_netif_stop(dev);
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP); /* stop the chip */
numbuffs = min(numbuffs, (int)min(lp->rx_ring_size, lp->tx_ring_size));
2006-03-22 05:15:44 +08:00
/* Reset the PCNET32 */
lp->a.reset(ioaddr);
lp->a.write_csr(ioaddr, CSR4, 0x0915); /* auto tx pad */
2006-03-22 05:15:44 +08:00
/* switch pcnet32 to 32bit mode */
lp->a.write_bcr(ioaddr, 20, 2);
/* purge & init rings but don't actually restart */
pcnet32_restart(dev, 0x0000);
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP); /* Set STOP bit */
2006-03-22 05:15:44 +08:00
/* Initialize Transmit buffers. */
size = data_len + 15;
for (x = 0; x < numbuffs; x++) {
if (!(skb = dev_alloc_skb(size))) {
if (netif_msg_hw(lp))
printk(KERN_DEBUG
"%s: Cannot allocate skb at line: %d!\n",
dev->name, __LINE__);
goto clean_up;
} else {
packet = skb->data;
skb_put(skb, size); /* create space for data */
lp->tx_skbuff[x] = skb;
lp->tx_ring[x].length = cpu_to_le16(-skb->len);
2006-03-22 05:15:44 +08:00
lp->tx_ring[x].misc = 0;
/* put DA and SA into the skb */
for (i = 0; i < 6; i++)
*packet++ = dev->dev_addr[i];
for (i = 0; i < 6; i++)
*packet++ = dev->dev_addr[i];
/* type */
*packet++ = 0x08;
*packet++ = 0x06;
/* packet number */
*packet++ = x;
/* fill packet with data */
for (i = 0; i < data_len; i++)
*packet++ = i;
lp->tx_dma_addr[x] =
pci_map_single(lp->pci_dev, skb->data, skb->len,
PCI_DMA_TODEVICE);
lp->tx_ring[x].base = cpu_to_le32(lp->tx_dma_addr[x]);
2006-03-22 05:15:44 +08:00
wmb(); /* Make sure owner changes after all others are visible */
lp->tx_ring[x].status = cpu_to_le16(status);
2006-03-22 05:15:44 +08:00
}
}
x = a->read_bcr(ioaddr, 32); /* set internal loopback in BCR32 */
a->write_bcr(ioaddr, 32, x | 0x0002);
2006-03-22 05:15:44 +08:00
/* set int loopback in CSR15 */
x = a->read_csr(ioaddr, CSR15) & 0xfffc;
lp->a.write_csr(ioaddr, CSR15, x | 0x0044);
2006-03-22 05:15:44 +08:00
teststatus = cpu_to_le16(0x8000);
lp->a.write_csr(ioaddr, CSR0, CSR0_START); /* Set STRT bit */
2006-03-22 05:15:44 +08:00
/* Check status of descriptors */
for (x = 0; x < numbuffs; x++) {
ticks = 0;
rmb();
while ((lp->rx_ring[x].status & teststatus) && (ticks < 200)) {
spin_unlock_irqrestore(&lp->lock, flags);
msleep(1);
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
rmb();
ticks++;
}
if (ticks == 200) {
if (netif_msg_hw(lp))
printk("%s: Desc %d failed to reset!\n",
dev->name, x);
break;
}
}
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP); /* Set STOP bit */
2006-03-22 05:15:44 +08:00
wmb();
if (netif_msg_hw(lp) && netif_msg_pktdata(lp)) {
printk(KERN_DEBUG "%s: RX loopback packets:\n", dev->name);
for (x = 0; x < numbuffs; x++) {
printk(KERN_DEBUG "%s: Packet %d:\n", dev->name, x);
skb = lp->rx_skbuff[x];
for (i = 0; i < size; i++) {
printk("%02x ", *(skb->data + i));
}
printk("\n");
}
}
2006-03-22 05:15:44 +08:00
x = 0;
rc = 0;
while (x < numbuffs && !rc) {
skb = lp->rx_skbuff[x];
packet = lp->tx_skbuff[x]->data;
for (i = 0; i < size; i++) {
if (*(skb->data + i) != packet[i]) {
if (netif_msg_hw(lp))
printk(KERN_DEBUG
"%s: Error in compare! %2x - %02x %02x\n",
dev->name, i, *(skb->data + i),
packet[i]);
rc = 1;
break;
}
}
x++;
}
2006-03-22 05:15:44 +08:00
clean_up:
*data1 = rc;
2006-03-22 05:15:44 +08:00
pcnet32_purge_tx_ring(dev);
x = a->read_csr(ioaddr, CSR15);
a->write_csr(ioaddr, CSR15, (x & ~0x0044)); /* reset bits 6 and 2 */
x = a->read_bcr(ioaddr, 32); /* reset internal loopback */
a->write_bcr(ioaddr, 32, (x & ~0x0002));
2006-03-22 05:15:44 +08:00
if (netif_running(dev)) {
pcnet32_netif_start(dev);
pcnet32_restart(dev, CSR0_NORMAL);
} else {
pcnet32_purge_rx_ring(dev);
lp->a.write_bcr(ioaddr, 20, 4); /* return to 16bit mode */
}
spin_unlock_irqrestore(&lp->lock, flags);
2006-03-22 05:15:44 +08:00
return (rc);
} /* end pcnet32_loopback_test */
static void pcnet32_led_blink_callback(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
struct pcnet32_access *a = &lp->a;
ulong ioaddr = dev->base_addr;
unsigned long flags;
int i;
spin_lock_irqsave(&lp->lock, flags);
for (i = 4; i < 8; i++) {
a->write_bcr(ioaddr, i, a->read_bcr(ioaddr, i) ^ 0x4000);
}
spin_unlock_irqrestore(&lp->lock, flags);
mod_timer(&lp->blink_timer, PCNET32_BLINK_TIMEOUT);
}
static int pcnet32_phys_id(struct net_device *dev, u32 data)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
struct pcnet32_access *a = &lp->a;
ulong ioaddr = dev->base_addr;
unsigned long flags;
int i, regs[4];
if (!lp->blink_timer.function) {
init_timer(&lp->blink_timer);
lp->blink_timer.function = (void *)pcnet32_led_blink_callback;
lp->blink_timer.data = (unsigned long)dev;
}
/* Save the current value of the bcrs */
spin_lock_irqsave(&lp->lock, flags);
for (i = 4; i < 8; i++) {
regs[i - 4] = a->read_bcr(ioaddr, i);
}
spin_unlock_irqrestore(&lp->lock, flags);
mod_timer(&lp->blink_timer, jiffies);
set_current_state(TASK_INTERRUPTIBLE);
/* AV: the limit here makes no sense whatsoever */
2006-03-22 05:15:44 +08:00
if ((!data) || (data > (u32) (MAX_SCHEDULE_TIMEOUT / HZ)))
data = (u32) (MAX_SCHEDULE_TIMEOUT / HZ);
msleep_interruptible(data * 1000);
del_timer_sync(&lp->blink_timer);
/* Restore the original value of the bcrs */
spin_lock_irqsave(&lp->lock, flags);
for (i = 4; i < 8; i++) {
a->write_bcr(ioaddr, i, regs[i - 4]);
}
spin_unlock_irqrestore(&lp->lock, flags);
return 0;
}
/*
* lp->lock must be held.
*/
static int pcnet32_suspend(struct net_device *dev, unsigned long *flags,
int can_sleep)
{
int csr5;
struct pcnet32_private *lp = netdev_priv(dev);
struct pcnet32_access *a = &lp->a;
ulong ioaddr = dev->base_addr;
int ticks;
/* really old chips have to be stopped. */
if (lp->chip_version < PCNET32_79C970A)
return 0;
/* set SUSPEND (SPND) - CSR5 bit 0 */
csr5 = a->read_csr(ioaddr, CSR5);
a->write_csr(ioaddr, CSR5, csr5 | CSR5_SUSPEND);
/* poll waiting for bit to be set */
ticks = 0;
while (!(a->read_csr(ioaddr, CSR5) & CSR5_SUSPEND)) {
spin_unlock_irqrestore(&lp->lock, *flags);
if (can_sleep)
msleep(1);
else
mdelay(1);
spin_lock_irqsave(&lp->lock, *flags);
ticks++;
if (ticks > 200) {
if (netif_msg_hw(lp))
printk(KERN_DEBUG
"%s: Error getting into suspend!\n",
dev->name);
return 0;
}
}
return 1;
}
/*
* process one receive descriptor entry
*/
static void pcnet32_rx_entry(struct net_device *dev,
struct pcnet32_private *lp,
struct pcnet32_rx_head *rxp,
int entry)
{
int status = (short)le16_to_cpu(rxp->status) >> 8;
int rx_in_place = 0;
struct sk_buff *skb;
short pkt_len;
if (status != 0x03) { /* There was an error. */
/*
* There is a tricky error noted by John Murphy,
* <murf@perftech.com> to Russ Nelson: Even with full-sized
* buffers it's possible for a jabber packet to use two
* buffers, with only the last correctly noting the error.
*/
if (status & 0x01) /* Only count a general error at the */
dev->stats.rx_errors++; /* end of a packet. */
if (status & 0x20)
dev->stats.rx_frame_errors++;
if (status & 0x10)
dev->stats.rx_over_errors++;
if (status & 0x08)
dev->stats.rx_crc_errors++;
if (status & 0x04)
dev->stats.rx_fifo_errors++;
return;
}
pkt_len = (le32_to_cpu(rxp->msg_length) & 0xfff) - 4;
/* Discard oversize frames. */
if (unlikely(pkt_len > PKT_BUF_SIZE)) {
if (netif_msg_drv(lp))
printk(KERN_ERR "%s: Impossible packet size %d!\n",
dev->name, pkt_len);
dev->stats.rx_errors++;
return;
}
if (pkt_len < 60) {
if (netif_msg_rx_err(lp))
printk(KERN_ERR "%s: Runt packet!\n", dev->name);
dev->stats.rx_errors++;
return;
}
if (pkt_len > rx_copybreak) {
struct sk_buff *newskb;
if ((newskb = dev_alloc_skb(PKT_BUF_SKB))) {
skb_reserve(newskb, NET_IP_ALIGN);
skb = lp->rx_skbuff[entry];
pci_unmap_single(lp->pci_dev,
lp->rx_dma_addr[entry],
PKT_BUF_SIZE,
PCI_DMA_FROMDEVICE);
skb_put(skb, pkt_len);
lp->rx_skbuff[entry] = newskb;
lp->rx_dma_addr[entry] =
pci_map_single(lp->pci_dev,
newskb->data,
PKT_BUF_SIZE,
PCI_DMA_FROMDEVICE);
rxp->base = cpu_to_le32(lp->rx_dma_addr[entry]);
rx_in_place = 1;
} else
skb = NULL;
} else {
skb = dev_alloc_skb(pkt_len + NET_IP_ALIGN);
}
if (skb == NULL) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Memory squeeze, dropping packet.\n",
dev->name);
dev->stats.rx_dropped++;
return;
}
if (!rx_in_place) {
skb_reserve(skb, NET_IP_ALIGN);
skb_put(skb, pkt_len); /* Make room */
pci_dma_sync_single_for_cpu(lp->pci_dev,
lp->rx_dma_addr[entry],
pkt_len,
PCI_DMA_FROMDEVICE);
skb_copy_to_linear_data(skb,
(unsigned char *)(lp->rx_skbuff[entry]->data),
pkt_len);
pci_dma_sync_single_for_device(lp->pci_dev,
lp->rx_dma_addr[entry],
pkt_len,
PCI_DMA_FROMDEVICE);
}
dev->stats.rx_bytes += skb->len;
skb->protocol = eth_type_trans(skb, dev);
netif_receive_skb(skb);
dev->stats.rx_packets++;
return;
}
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
static int pcnet32_rx(struct net_device *dev, int budget)
{
struct pcnet32_private *lp = netdev_priv(dev);
int entry = lp->cur_rx & lp->rx_mod_mask;
struct pcnet32_rx_head *rxp = &lp->rx_ring[entry];
int npackets = 0;
/* If we own the next entry, it's a new packet. Send it up. */
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
while (npackets < budget && (short)le16_to_cpu(rxp->status) >= 0) {
pcnet32_rx_entry(dev, lp, rxp, entry);
npackets += 1;
/*
* The docs say that the buffer length isn't touched, but Andrew
* Boyd of QNX reports that some revs of the 79C965 clear it.
*/
rxp->buf_length = cpu_to_le16(NEG_BUF_SIZE);
wmb(); /* Make sure owner changes after others are visible */
rxp->status = cpu_to_le16(0x8000);
entry = (++lp->cur_rx) & lp->rx_mod_mask;
rxp = &lp->rx_ring[entry];
}
return npackets;
}
static int pcnet32_tx(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
unsigned int dirty_tx = lp->dirty_tx;
int delta;
int must_restart = 0;
while (dirty_tx != lp->cur_tx) {
int entry = dirty_tx & lp->tx_mod_mask;
int status = (short)le16_to_cpu(lp->tx_ring[entry].status);
if (status < 0)
break; /* It still hasn't been Txed */
lp->tx_ring[entry].base = 0;
if (status & 0x4000) {
/* There was a major error, log it. */
int err_status = le32_to_cpu(lp->tx_ring[entry].misc);
dev->stats.tx_errors++;
if (netif_msg_tx_err(lp))
printk(KERN_ERR
"%s: Tx error status=%04x err_status=%08x\n",
dev->name, status,
err_status);
if (err_status & 0x04000000)
dev->stats.tx_aborted_errors++;
if (err_status & 0x08000000)
dev->stats.tx_carrier_errors++;
if (err_status & 0x10000000)
dev->stats.tx_window_errors++;
#ifndef DO_DXSUFLO
if (err_status & 0x40000000) {
dev->stats.tx_fifo_errors++;
/* Ackk! On FIFO errors the Tx unit is turned off! */
/* Remove this verbosity later! */
if (netif_msg_tx_err(lp))
printk(KERN_ERR
"%s: Tx FIFO error!\n",
dev->name);
must_restart = 1;
}
#else
if (err_status & 0x40000000) {
dev->stats.tx_fifo_errors++;
if (!lp->dxsuflo) { /* If controller doesn't recover ... */
/* Ackk! On FIFO errors the Tx unit is turned off! */
/* Remove this verbosity later! */
if (netif_msg_tx_err(lp))
printk(KERN_ERR
"%s: Tx FIFO error!\n",
dev->name);
must_restart = 1;
}
}
#endif
} else {
if (status & 0x1800)
dev->stats.collisions++;
dev->stats.tx_packets++;
}
/* We must free the original skb */
if (lp->tx_skbuff[entry]) {
pci_unmap_single(lp->pci_dev,
lp->tx_dma_addr[entry],
lp->tx_skbuff[entry]->
len, PCI_DMA_TODEVICE);
dev_kfree_skb_any(lp->tx_skbuff[entry]);
lp->tx_skbuff[entry] = NULL;
lp->tx_dma_addr[entry] = 0;
}
dirty_tx++;
}
delta = (lp->cur_tx - dirty_tx) & (lp->tx_mod_mask + lp->tx_ring_size);
if (delta > lp->tx_ring_size) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: out-of-sync dirty pointer, %d vs. %d, full=%d.\n",
dev->name, dirty_tx, lp->cur_tx,
lp->tx_full);
dirty_tx += lp->tx_ring_size;
delta -= lp->tx_ring_size;
}
if (lp->tx_full &&
netif_queue_stopped(dev) &&
delta < lp->tx_ring_size - 2) {
/* The ring is no longer full, clear tbusy. */
lp->tx_full = 0;
netif_wake_queue(dev);
}
lp->dirty_tx = dirty_tx;
return must_restart;
}
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
static int pcnet32_poll(struct napi_struct *napi, int budget)
{
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
struct pcnet32_private *lp = container_of(napi, struct pcnet32_private, napi);
struct net_device *dev = lp->dev;
unsigned long ioaddr = dev->base_addr;
unsigned long flags;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
int work_done;
u16 val;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
work_done = pcnet32_rx(dev, budget);
spin_lock_irqsave(&lp->lock, flags);
if (pcnet32_tx(dev)) {
/* reset the chip to clear the error condition, then restart */
lp->a.reset(ioaddr);
lp->a.write_csr(ioaddr, CSR4, 0x0915); /* auto tx pad */
pcnet32_restart(dev, CSR0_START);
netif_wake_queue(dev);
}
spin_unlock_irqrestore(&lp->lock, flags);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
if (work_done < budget) {
spin_lock_irqsave(&lp->lock, flags);
__napi_complete(napi);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
/* clear interrupt masks */
val = lp->a.read_csr(ioaddr, CSR3);
val &= 0x00ff;
lp->a.write_csr(ioaddr, CSR3, val);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
/* Set interrupt enable. */
lp->a.write_csr(ioaddr, CSR0, CSR0_INTEN);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
}
return work_done;
}
#define PCNET32_REGS_PER_PHY 32
#define PCNET32_MAX_PHYS 32
static int pcnet32_get_regs_len(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int j = lp->phycount * PCNET32_REGS_PER_PHY;
2006-03-22 05:15:44 +08:00
return ((PCNET32_NUM_REGS + j) * sizeof(u16));
}
static void pcnet32_get_regs(struct net_device *dev, struct ethtool_regs *regs,
2006-03-22 05:15:44 +08:00
void *ptr)
{
2006-03-22 05:15:44 +08:00
int i, csr0;
u16 *buff = ptr;
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
struct pcnet32_access *a = &lp->a;
ulong ioaddr = dev->base_addr;
unsigned long flags;
spin_lock_irqsave(&lp->lock, flags);
csr0 = a->read_csr(ioaddr, CSR0);
if (!(csr0 & CSR0_STOP)) /* If not stopped */
pcnet32_suspend(dev, &flags, 1);
2006-03-22 05:15:44 +08:00
/* read address PROM */
for (i = 0; i < 16; i += 2)
*buff++ = inw(ioaddr + i);
/* read control and status registers */
for (i = 0; i < 90; i++) {
*buff++ = a->read_csr(ioaddr, i);
}
*buff++ = a->read_csr(ioaddr, 112);
*buff++ = a->read_csr(ioaddr, 114);
2006-03-22 05:15:44 +08:00
/* read bus configuration registers */
for (i = 0; i < 30; i++) {
*buff++ = a->read_bcr(ioaddr, i);
}
*buff++ = 0; /* skip bcr30 so as not to hang 79C976 */
for (i = 31; i < 36; i++) {
*buff++ = a->read_bcr(ioaddr, i);
}
/* read mii phy registers */
if (lp->mii) {
int j;
for (j = 0; j < PCNET32_MAX_PHYS; j++) {
if (lp->phymask & (1 << j)) {
for (i = 0; i < PCNET32_REGS_PER_PHY; i++) {
lp->a.write_bcr(ioaddr, 33,
(j << 5) | i);
*buff++ = lp->a.read_bcr(ioaddr, 34);
}
}
}
}
if (!(csr0 & CSR0_STOP)) { /* If not stopped */
int csr5;
2006-03-22 05:15:44 +08:00
/* clear SUSPEND (SPND) - CSR5 bit 0 */
csr5 = a->read_csr(ioaddr, CSR5);
a->write_csr(ioaddr, CSR5, csr5 & (~CSR5_SUSPEND));
2006-03-22 05:15:44 +08:00
}
spin_unlock_irqrestore(&lp->lock, flags);
}
static const struct ethtool_ops pcnet32_ethtool_ops = {
2006-03-22 05:15:44 +08:00
.get_settings = pcnet32_get_settings,
.set_settings = pcnet32_set_settings,
.get_drvinfo = pcnet32_get_drvinfo,
.get_msglevel = pcnet32_get_msglevel,
.set_msglevel = pcnet32_set_msglevel,
.nway_reset = pcnet32_nway_reset,
.get_link = pcnet32_get_link,
.get_ringparam = pcnet32_get_ringparam,
.set_ringparam = pcnet32_set_ringparam,
.get_strings = pcnet32_get_strings,
.self_test = pcnet32_ethtool_test,
.phys_id = pcnet32_phys_id,
.get_regs_len = pcnet32_get_regs_len,
.get_regs = pcnet32_get_regs,
.get_sset_count = pcnet32_get_sset_count,
};
/* only probes for non-PCI devices, the rest are handled by
* pci_register_driver via pcnet32_probe_pci */
static void __devinit pcnet32_probe_vlbus(unsigned int *pcnet32_portlist)
{
2006-03-22 05:15:44 +08:00
unsigned int *port, ioaddr;
/* search for PCnet32 VLB cards at known addresses */
for (port = pcnet32_portlist; (ioaddr = *port); port++) {
if (request_region
(ioaddr, PCNET32_TOTAL_SIZE, "pcnet32_probe_vlbus")) {
/* check if there is really a pcnet chip on that ioaddr */
if ((inb(ioaddr + 14) == 0x57) &&
(inb(ioaddr + 15) == 0x57)) {
2006-03-22 05:15:44 +08:00
pcnet32_probe1(ioaddr, 0, NULL);
} else {
release_region(ioaddr, PCNET32_TOTAL_SIZE);
}
}
}
}
static int __devinit
pcnet32_probe_pci(struct pci_dev *pdev, const struct pci_device_id *ent)
{
2006-03-22 05:15:44 +08:00
unsigned long ioaddr;
int err;
err = pci_enable_device(pdev);
if (err < 0) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX
"failed to enable device -- err=%d\n", err);
return err;
}
pci_set_master(pdev);
ioaddr = pci_resource_start(pdev, 0);
if (!ioaddr) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX
"card has no PCI IO resources, aborting\n");
return -ENODEV;
}
2006-03-22 05:15:44 +08:00
if (!pci_dma_supported(pdev, PCNET32_DMA_MASK)) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX
"architecture does not support 32bit PCI busmaster DMA\n");
return -ENODEV;
}
if (request_region(ioaddr, PCNET32_TOTAL_SIZE, "pcnet32_probe_pci") ==
NULL) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX
"io address range already allocated\n");
return -EBUSY;
}
2006-03-22 05:15:44 +08:00
err = pcnet32_probe1(ioaddr, 1, pdev);
if (err < 0) {
pci_disable_device(pdev);
}
return err;
}
static const struct net_device_ops pcnet32_netdev_ops = {
.ndo_open = pcnet32_open,
.ndo_stop = pcnet32_close,
.ndo_start_xmit = pcnet32_start_xmit,
.ndo_tx_timeout = pcnet32_tx_timeout,
.ndo_get_stats = pcnet32_get_stats,
.ndo_set_multicast_list = pcnet32_set_multicast_list,
.ndo_do_ioctl = pcnet32_ioctl,
.ndo_change_mtu = eth_change_mtu,
.ndo_set_mac_address = eth_mac_addr,
.ndo_validate_addr = eth_validate_addr,
#ifdef CONFIG_NET_POLL_CONTROLLER
.ndo_poll_controller = pcnet32_poll_controller,
#endif
};
/* pcnet32_probe1
* Called from both pcnet32_probe_vlbus and pcnet_probe_pci.
* pdev will be NULL when called from pcnet32_probe_vlbus.
*/
static int __devinit
pcnet32_probe1(unsigned long ioaddr, int shared, struct pci_dev *pdev)
{
2006-03-22 05:15:44 +08:00
struct pcnet32_private *lp;
int i, media;
int fdx, mii, fset, dxsuflo;
int chip_version;
char *chipname;
struct net_device *dev;
struct pcnet32_access *a = NULL;
u8 promaddr[6];
int ret = -ENODEV;
/* reset the chip */
pcnet32_wio_reset(ioaddr);
/* NOTE: 16-bit check is first, otherwise some older PCnet chips fail */
if (pcnet32_wio_read_csr(ioaddr, 0) == 4 && pcnet32_wio_check(ioaddr)) {
a = &pcnet32_wio;
} else {
pcnet32_dwio_reset(ioaddr);
if (pcnet32_dwio_read_csr(ioaddr, 0) == 4 &&
pcnet32_dwio_check(ioaddr)) {
2006-03-22 05:15:44 +08:00
a = &pcnet32_dwio;
} else {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX "No access methods\n");
2006-03-22 05:15:44 +08:00
goto err_release_region;
}
2006-03-22 05:15:44 +08:00
}
chip_version =
a->read_csr(ioaddr, 88) | (a->read_csr(ioaddr, 89) << 16);
if ((pcnet32_debug & NETIF_MSG_PROBE) && (pcnet32_debug & NETIF_MSG_HW))
printk(KERN_INFO " PCnet chip version is %#x.\n",
chip_version);
if ((chip_version & 0xfff) != 0x003) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_INFO PFX "Unsupported chip version.\n");
goto err_release_region;
}
/* initialize variables */
fdx = mii = fset = dxsuflo = 0;
chip_version = (chip_version >> 12) & 0xffff;
switch (chip_version) {
case 0x2420:
chipname = "PCnet/PCI 79C970"; /* PCI */
break;
case 0x2430:
if (shared)
chipname = "PCnet/PCI 79C970"; /* 970 gives the wrong chip id back */
else
chipname = "PCnet/32 79C965"; /* 486/VL bus */
break;
case 0x2621:
chipname = "PCnet/PCI II 79C970A"; /* PCI */
fdx = 1;
break;
case 0x2623:
chipname = "PCnet/FAST 79C971"; /* PCI */
fdx = 1;
mii = 1;
fset = 1;
break;
case 0x2624:
chipname = "PCnet/FAST+ 79C972"; /* PCI */
fdx = 1;
mii = 1;
fset = 1;
break;
case 0x2625:
chipname = "PCnet/FAST III 79C973"; /* PCI */
fdx = 1;
mii = 1;
break;
case 0x2626:
chipname = "PCnet/Home 79C978"; /* PCI */
fdx = 1;
/*
* This is based on specs published at www.amd.com. This section
* assumes that a card with a 79C978 wants to go into standard
* ethernet mode. The 79C978 can also go into 1Mb HomePNA mode,
* and the module option homepna=1 can select this instead.
*/
media = a->read_bcr(ioaddr, 49);
media &= ~3; /* default to 10Mb ethernet */
if (cards_found < MAX_UNITS && homepna[cards_found])
media |= 1; /* switch to home wiring mode */
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_DEBUG PFX "media set to %sMbit mode.\n",
(media & 1) ? "1" : "10");
a->write_bcr(ioaddr, 49, media);
break;
case 0x2627:
chipname = "PCnet/FAST III 79C975"; /* PCI */
fdx = 1;
mii = 1;
break;
case 0x2628:
chipname = "PCnet/PRO 79C976";
fdx = 1;
mii = 1;
break;
default:
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_INFO PFX
"PCnet version %#x, no PCnet32 chip.\n",
chip_version);
goto err_release_region;
}
/*
2006-03-22 05:15:44 +08:00
* On selected chips turn on the BCR18:NOUFLO bit. This stops transmit
* starting until the packet is loaded. Strike one for reliability, lose
* one for latency - although on PCI this isnt a big loss. Older chips
* have FIFO's smaller than a packet, so you can't do this.
* Turn on BCR18:BurstRdEn and BCR18:BurstWrEn.
*/
2006-03-22 05:15:44 +08:00
if (fset) {
a->write_bcr(ioaddr, 18, (a->read_bcr(ioaddr, 18) | 0x0860));
a->write_csr(ioaddr, 80,
(a->read_csr(ioaddr, 80) & 0x0C00) | 0x0c00);
dxsuflo = 1;
}
dev = alloc_etherdev(sizeof(*lp));
2006-03-22 05:15:44 +08:00
if (!dev) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX "Memory allocation failed.\n");
ret = -ENOMEM;
goto err_release_region;
}
if (pdev)
SET_NETDEV_DEV(dev, &pdev->dev);
2006-03-22 05:15:44 +08:00
if (pcnet32_debug & NETIF_MSG_PROBE)
2006-03-22 05:15:44 +08:00
printk(KERN_INFO PFX "%s at %#3lx,", chipname, ioaddr);
/* In most chips, after a chip reset, the ethernet address is read from the
* station address PROM at the base address and programmed into the
* "Physical Address Registers" CSR12-14.
* As a precautionary measure, we read the PROM values and complain if
* they disagree with the CSRs. If they miscompare, and the PROM addr
* is valid, then the PROM addr is used.
2006-03-22 05:15:44 +08:00
*/
for (i = 0; i < 3; i++) {
unsigned int val;
val = a->read_csr(ioaddr, i + 12) & 0x0ffff;
/* There may be endianness issues here. */
dev->dev_addr[2 * i] = val & 0x0ff;
dev->dev_addr[2 * i + 1] = (val >> 8) & 0x0ff;
}
/* read PROM address and compare with CSR address */
for (i = 0; i < 6; i++)
2006-03-22 05:15:44 +08:00
promaddr[i] = inb(ioaddr + i);
if (memcmp(promaddr, dev->dev_addr, 6) ||
!is_valid_ether_addr(dev->dev_addr)) {
2006-03-22 05:15:44 +08:00
if (is_valid_ether_addr(promaddr)) {
if (pcnet32_debug & NETIF_MSG_PROBE) {
printk(" warning: CSR address invalid,\n");
printk(KERN_INFO
" using instead PROM address of");
}
memcpy(dev->dev_addr, promaddr, 6);
}
}
memcpy(dev->perm_addr, dev->dev_addr, dev->addr_len);
/* if the ethernet address is not valid, force to 00:00:00:00:00:00 */
if (!is_valid_ether_addr(dev->perm_addr))
memset(dev->dev_addr, 0, ETH_ALEN);
2006-03-22 05:15:44 +08:00
if (pcnet32_debug & NETIF_MSG_PROBE) {
printk(" %pM", dev->dev_addr);
2006-03-22 05:15:44 +08:00
/* Version 0x2623 and 0x2624 */
if (((chip_version + 1) & 0xfffe) == 0x2624) {
i = a->read_csr(ioaddr, 80) & 0x0C00; /* Check tx_start_pt */
printk(KERN_INFO " tx_start_pt(0x%04x):", i);
2006-03-22 05:15:44 +08:00
switch (i >> 10) {
case 0:
printk(KERN_CONT " 20 bytes,");
2006-03-22 05:15:44 +08:00
break;
case 1:
printk(KERN_CONT " 64 bytes,");
2006-03-22 05:15:44 +08:00
break;
case 2:
printk(KERN_CONT " 128 bytes,");
2006-03-22 05:15:44 +08:00
break;
case 3:
printk(KERN_CONT "~220 bytes,");
2006-03-22 05:15:44 +08:00
break;
}
i = a->read_bcr(ioaddr, 18); /* Check Burst/Bus control */
printk(KERN_CONT " BCR18(%x):", i & 0xffff);
2006-03-22 05:15:44 +08:00
if (i & (1 << 5))
printk(KERN_CONT "BurstWrEn ");
2006-03-22 05:15:44 +08:00
if (i & (1 << 6))
printk(KERN_CONT "BurstRdEn ");
2006-03-22 05:15:44 +08:00
if (i & (1 << 7))
printk(KERN_CONT "DWordIO ");
2006-03-22 05:15:44 +08:00
if (i & (1 << 11))
printk(KERN_CONT "NoUFlow ");
2006-03-22 05:15:44 +08:00
i = a->read_bcr(ioaddr, 25);
printk(KERN_INFO " SRAMSIZE=0x%04x,", i << 8);
2006-03-22 05:15:44 +08:00
i = a->read_bcr(ioaddr, 26);
printk(KERN_CONT " SRAM_BND=0x%04x,", i << 8);
2006-03-22 05:15:44 +08:00
i = a->read_bcr(ioaddr, 27);
if (i & (1 << 14))
printk(KERN_CONT "LowLatRx");
2006-03-22 05:15:44 +08:00
}
}
dev->base_addr = ioaddr;
lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
/* pci_alloc_consistent returns page-aligned memory, so we do not have to check the alignment */
if ((lp->init_block =
pci_alloc_consistent(pdev, sizeof(*lp->init_block), &lp->init_dma_addr)) == NULL) {
2006-03-22 05:15:44 +08:00
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_ERR PFX
"Consistent memory allocation failed.\n");
ret = -ENOMEM;
goto err_free_netdev;
}
lp->pci_dev = pdev;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
lp->dev = dev;
2006-03-22 05:15:44 +08:00
spin_lock_init(&lp->lock);
lp->name = chipname;
lp->shared_irq = shared;
lp->tx_ring_size = TX_RING_SIZE; /* default tx ring size */
lp->rx_ring_size = RX_RING_SIZE; /* default rx ring size */
lp->tx_mod_mask = lp->tx_ring_size - 1;
lp->rx_mod_mask = lp->rx_ring_size - 1;
lp->tx_len_bits = (PCNET32_LOG_TX_BUFFERS << 12);
lp->rx_len_bits = (PCNET32_LOG_RX_BUFFERS << 4);
lp->mii_if.full_duplex = fdx;
lp->mii_if.phy_id_mask = 0x1f;
lp->mii_if.reg_num_mask = 0x1f;
lp->dxsuflo = dxsuflo;
lp->mii = mii;
lp->chip_version = chip_version;
2006-03-22 05:15:44 +08:00
lp->msg_enable = pcnet32_debug;
if ((cards_found >= MAX_UNITS) ||
(options[cards_found] >= sizeof(options_mapping)))
2006-03-22 05:15:44 +08:00
lp->options = PCNET32_PORT_ASEL;
else
lp->options = options_mapping[options[cards_found]];
lp->mii_if.dev = dev;
lp->mii_if.mdio_read = mdio_read;
lp->mii_if.mdio_write = mdio_write;
/* napi.weight is used in both the napi and non-napi cases */
lp->napi.weight = lp->rx_ring_size / 2;
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
netif_napi_add(dev, &lp->napi, pcnet32_poll, lp->rx_ring_size / 2);
2006-03-22 05:15:44 +08:00
if (fdx && !(lp->options & PCNET32_PORT_ASEL) &&
((cards_found >= MAX_UNITS) || full_duplex[cards_found]))
lp->options |= PCNET32_PORT_FD;
lp->a = *a;
/* prior to register_netdev, dev->name is not yet correct */
if (pcnet32_alloc_ring(dev, pci_name(lp->pci_dev))) {
ret = -ENOMEM;
goto err_free_ring;
}
/* detect special T1/E1 WAN card by checking for MAC address */
if (dev->dev_addr[0] == 0x00 && dev->dev_addr[1] == 0xe0 &&
dev->dev_addr[2] == 0x75)
2006-03-22 05:15:44 +08:00
lp->options = PCNET32_PORT_FD | PCNET32_PORT_GPSI;
lp->init_block->mode = cpu_to_le16(0x0003); /* Disable Rx and Tx. */
lp->init_block->tlen_rlen =
cpu_to_le16(lp->tx_len_bits | lp->rx_len_bits);
2006-03-22 05:15:44 +08:00
for (i = 0; i < 6; i++)
lp->init_block->phys_addr[i] = dev->dev_addr[i];
lp->init_block->filter[0] = 0x00000000;
lp->init_block->filter[1] = 0x00000000;
lp->init_block->rx_ring = cpu_to_le32(lp->rx_ring_dma_addr);
lp->init_block->tx_ring = cpu_to_le32(lp->tx_ring_dma_addr);
2006-03-22 05:15:44 +08:00
/* switch pcnet32 to 32bit mode */
a->write_bcr(ioaddr, 20, 2);
a->write_csr(ioaddr, 1, (lp->init_dma_addr & 0xffff));
a->write_csr(ioaddr, 2, (lp->init_dma_addr >> 16));
2006-03-22 05:15:44 +08:00
if (pdev) { /* use the IRQ provided by PCI */
dev->irq = pdev->irq;
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(" assigned IRQ %d.\n", dev->irq);
} else {
unsigned long irq_mask = probe_irq_on();
/*
* To auto-IRQ we enable the initialization-done and DMA error
* interrupts. For ISA boards we get a DMA error, but VLB and PCI
* boards will work.
*/
/* Trigger an initialization just for the interrupt. */
a->write_csr(ioaddr, CSR0, CSR0_INTEN | CSR0_INIT);
2006-03-22 05:15:44 +08:00
mdelay(1);
dev->irq = probe_irq_off(irq_mask);
if (!dev->irq) {
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(", failed to detect IRQ line.\n");
ret = -ENODEV;
goto err_free_ring;
}
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(", probed IRQ %d.\n", dev->irq);
}
2006-03-22 05:15:44 +08:00
/* Set the mii phy_id so that we can query the link state */
if (lp->mii) {
/* lp->phycount and lp->phymask are set to 0 by memset above */
lp->mii_if.phy_id = ((lp->a.read_bcr(ioaddr, 33)) >> 5) & 0x1f;
/* scan for PHYs */
for (i = 0; i < PCNET32_MAX_PHYS; i++) {
unsigned short id1, id2;
id1 = mdio_read(dev, i, MII_PHYSID1);
if (id1 == 0xffff)
continue;
id2 = mdio_read(dev, i, MII_PHYSID2);
if (id2 == 0xffff)
continue;
if (i == 31 && ((chip_version + 1) & 0xfffe) == 0x2624)
continue; /* 79C971 & 79C972 have phantom phy at id 31 */
lp->phycount++;
lp->phymask |= (1 << i);
lp->mii_if.phy_id = i;
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_INFO PFX
"Found PHY %04x:%04x at address %d.\n",
id1, id2, i);
}
lp->a.write_bcr(ioaddr, 33, (lp->mii_if.phy_id) << 5);
if (lp->phycount > 1) {
lp->options |= PCNET32_PORT_MII;
}
}
2006-03-22 05:15:44 +08:00
init_timer(&lp->watchdog_timer);
lp->watchdog_timer.data = (unsigned long)dev;
lp->watchdog_timer.function = (void *)&pcnet32_watchdog;
/* The PCNET32-specific entries in the device structure. */
dev->netdev_ops = &pcnet32_netdev_ops;
2006-03-22 05:15:44 +08:00
dev->ethtool_ops = &pcnet32_ethtool_ops;
dev->watchdog_timeo = (5 * HZ);
2006-03-22 05:15:44 +08:00
/* Fill in the generic fields of the device structure. */
if (register_netdev(dev))
goto err_free_ring;
if (pdev) {
pci_set_drvdata(pdev, dev);
} else {
lp->next = pcnet32_dev;
pcnet32_dev = dev;
}
if (pcnet32_debug & NETIF_MSG_PROBE)
printk(KERN_INFO "%s: registered as %s\n", dev->name, lp->name);
cards_found++;
/* enable LED writes */
a->write_bcr(ioaddr, 2, a->read_bcr(ioaddr, 2) | 0x1000);
2006-03-22 05:15:44 +08:00
return 0;
err_free_ring:
2006-03-22 05:15:44 +08:00
pcnet32_free_ring(dev);
pci_free_consistent(lp->pci_dev, sizeof(*lp->init_block),
lp->init_block, lp->init_dma_addr);
err_free_netdev:
2006-03-22 05:15:44 +08:00
free_netdev(dev);
err_release_region:
2006-03-22 05:15:44 +08:00
release_region(ioaddr, PCNET32_TOTAL_SIZE);
return ret;
}
/* if any allocation fails, caller must also call pcnet32_free_ring */
static int pcnet32_alloc_ring(struct net_device *dev, const char *name)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
lp->tx_ring = pci_alloc_consistent(lp->pci_dev,
sizeof(struct pcnet32_tx_head) *
lp->tx_ring_size,
&lp->tx_ring_dma_addr);
if (lp->tx_ring == NULL) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Consistent memory allocation failed.\n",
name);
return -ENOMEM;
}
2006-03-22 05:15:44 +08:00
lp->rx_ring = pci_alloc_consistent(lp->pci_dev,
sizeof(struct pcnet32_rx_head) *
lp->rx_ring_size,
&lp->rx_ring_dma_addr);
if (lp->rx_ring == NULL) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Consistent memory allocation failed.\n",
name);
return -ENOMEM;
}
lp->tx_dma_addr = kcalloc(lp->tx_ring_size, sizeof(dma_addr_t),
2006-03-22 05:15:44 +08:00
GFP_ATOMIC);
if (!lp->tx_dma_addr) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Memory allocation failed.\n", name);
return -ENOMEM;
}
lp->rx_dma_addr = kcalloc(lp->rx_ring_size, sizeof(dma_addr_t),
2006-03-22 05:15:44 +08:00
GFP_ATOMIC);
if (!lp->rx_dma_addr) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Memory allocation failed.\n", name);
return -ENOMEM;
}
lp->tx_skbuff = kcalloc(lp->tx_ring_size, sizeof(struct sk_buff *),
2006-03-22 05:15:44 +08:00
GFP_ATOMIC);
if (!lp->tx_skbuff) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Memory allocation failed.\n", name);
return -ENOMEM;
}
lp->rx_skbuff = kcalloc(lp->rx_ring_size, sizeof(struct sk_buff *),
2006-03-22 05:15:44 +08:00
GFP_ATOMIC);
if (!lp->rx_skbuff) {
if (netif_msg_drv(lp))
printk(KERN_ERR PFX
2006-03-22 05:15:44 +08:00
"%s: Memory allocation failed.\n", name);
return -ENOMEM;
}
return 0;
}
static void pcnet32_free_ring(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
kfree(lp->tx_skbuff);
lp->tx_skbuff = NULL;
2006-03-22 05:15:44 +08:00
kfree(lp->rx_skbuff);
lp->rx_skbuff = NULL;
2006-03-22 05:15:44 +08:00
kfree(lp->tx_dma_addr);
lp->tx_dma_addr = NULL;
2006-03-22 05:15:44 +08:00
kfree(lp->rx_dma_addr);
lp->rx_dma_addr = NULL;
2006-03-22 05:15:44 +08:00
if (lp->tx_ring) {
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_tx_head) *
lp->tx_ring_size, lp->tx_ring,
lp->tx_ring_dma_addr);
lp->tx_ring = NULL;
}
2006-03-22 05:15:44 +08:00
if (lp->rx_ring) {
pci_free_consistent(lp->pci_dev,
sizeof(struct pcnet32_rx_head) *
lp->rx_ring_size, lp->rx_ring,
lp->rx_ring_dma_addr);
lp->rx_ring = NULL;
}
}
2006-03-22 05:15:44 +08:00
static int pcnet32_open(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
struct pci_dev *pdev = lp->pci_dev;
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
u16 val;
int i;
int rc;
unsigned long flags;
if (request_irq(dev->irq, pcnet32_interrupt,
lp->shared_irq ? IRQF_SHARED : 0, dev->name,
2006-03-22 05:15:44 +08:00
(void *)dev)) {
return -EAGAIN;
}
spin_lock_irqsave(&lp->lock, flags);
/* Check for a valid station address */
if (!is_valid_ether_addr(dev->dev_addr)) {
rc = -EINVAL;
goto err_free_irq;
}
/* Reset the PCNET32 */
lp->a.reset(ioaddr);
/* switch pcnet32 to 32bit mode */
lp->a.write_bcr(ioaddr, 20, 2);
if (netif_msg_ifup(lp))
printk(KERN_DEBUG
"%s: pcnet32_open() irq %d tx/rx rings %#x/%#x init %#x.\n",
dev->name, dev->irq, (u32) (lp->tx_ring_dma_addr),
(u32) (lp->rx_ring_dma_addr),
(u32) (lp->init_dma_addr));
2006-03-22 05:15:44 +08:00
/* set/reset autoselect bit */
val = lp->a.read_bcr(ioaddr, 2) & ~2;
if (lp->options & PCNET32_PORT_ASEL)
val |= 2;
2006-03-22 05:15:44 +08:00
lp->a.write_bcr(ioaddr, 2, val);
/* handle full duplex setting */
if (lp->mii_if.full_duplex) {
val = lp->a.read_bcr(ioaddr, 9) & ~3;
if (lp->options & PCNET32_PORT_FD) {
val |= 1;
if (lp->options == (PCNET32_PORT_FD | PCNET32_PORT_AUI))
val |= 2;
} else if (lp->options & PCNET32_PORT_ASEL) {
/* workaround of xSeries250, turn on for 79C975 only */
if (lp->chip_version == 0x2627)
2006-03-22 05:15:44 +08:00
val |= 3;
}
lp->a.write_bcr(ioaddr, 9, val);
}
/* set/reset GPSI bit in test register */
val = lp->a.read_csr(ioaddr, 124) & ~0x10;
if ((lp->options & PCNET32_PORT_PORTSEL) == PCNET32_PORT_GPSI)
val |= 0x10;
lp->a.write_csr(ioaddr, 124, val);
/* Allied Telesyn AT 2700/2701 FX are 100Mbit only and do not negotiate */
if (pdev && pdev->subsystem_vendor == PCI_VENDOR_ID_AT &&
(pdev->subsystem_device == PCI_SUBDEVICE_ID_AT_2700FX ||
pdev->subsystem_device == PCI_SUBDEVICE_ID_AT_2701FX)) {
if (lp->options & PCNET32_PORT_ASEL) {
2006-03-22 05:15:44 +08:00
lp->options = PCNET32_PORT_FD | PCNET32_PORT_100;
if (netif_msg_link(lp))
printk(KERN_DEBUG
"%s: Setting 100Mb-Full Duplex.\n",
dev->name);
}
}
if (lp->phycount < 2) {
/*
* 24 Jun 2004 according AMD, in order to change the PHY,
* DANAS (or DISPM for 79C976) must be set; then select the speed,
* duplex, and/or enable auto negotiation, and clear DANAS
*/
if (lp->mii && !(lp->options & PCNET32_PORT_ASEL)) {
lp->a.write_bcr(ioaddr, 32,
lp->a.read_bcr(ioaddr, 32) | 0x0080);
/* disable Auto Negotiation, set 10Mpbs, HD */
val = lp->a.read_bcr(ioaddr, 32) & ~0xb8;
if (lp->options & PCNET32_PORT_FD)
val |= 0x10;
if (lp->options & PCNET32_PORT_100)
val |= 0x08;
lp->a.write_bcr(ioaddr, 32, val);
} else {
if (lp->options & PCNET32_PORT_ASEL) {
lp->a.write_bcr(ioaddr, 32,
lp->a.read_bcr(ioaddr,
32) | 0x0080);
/* enable auto negotiate, setup, disable fd */
val = lp->a.read_bcr(ioaddr, 32) & ~0x98;
val |= 0x20;
lp->a.write_bcr(ioaddr, 32, val);
}
}
} else {
int first_phy = -1;
u16 bmcr;
u32 bcr9;
struct ethtool_cmd ecmd;
/*
* There is really no good other way to handle multiple PHYs
* other than turning off all automatics
*/
val = lp->a.read_bcr(ioaddr, 2);
lp->a.write_bcr(ioaddr, 2, val & ~2);
val = lp->a.read_bcr(ioaddr, 32);
lp->a.write_bcr(ioaddr, 32, val & ~(1 << 7)); /* stop MII manager */
if (!(lp->options & PCNET32_PORT_ASEL)) {
/* setup ecmd */
ecmd.port = PORT_MII;
ecmd.transceiver = XCVR_INTERNAL;
ecmd.autoneg = AUTONEG_DISABLE;
ecmd.speed =
lp->
options & PCNET32_PORT_100 ? SPEED_100 : SPEED_10;
bcr9 = lp->a.read_bcr(ioaddr, 9);
if (lp->options & PCNET32_PORT_FD) {
ecmd.duplex = DUPLEX_FULL;
bcr9 |= (1 << 0);
} else {
ecmd.duplex = DUPLEX_HALF;
bcr9 |= ~(1 << 0);
}
lp->a.write_bcr(ioaddr, 9, bcr9);
}
2006-03-22 05:15:44 +08:00
for (i = 0; i < PCNET32_MAX_PHYS; i++) {
if (lp->phymask & (1 << i)) {
/* isolate all but the first PHY */
bmcr = mdio_read(dev, i, MII_BMCR);
if (first_phy == -1) {
first_phy = i;
mdio_write(dev, i, MII_BMCR,
bmcr & ~BMCR_ISOLATE);
} else {
mdio_write(dev, i, MII_BMCR,
bmcr | BMCR_ISOLATE);
}
/* use mii_ethtool_sset to setup PHY */
lp->mii_if.phy_id = i;
ecmd.phy_address = i;
if (lp->options & PCNET32_PORT_ASEL) {
mii_ethtool_gset(&lp->mii_if, &ecmd);
ecmd.autoneg = AUTONEG_ENABLE;
}
mii_ethtool_sset(&lp->mii_if, &ecmd);
}
}
lp->mii_if.phy_id = first_phy;
if (netif_msg_link(lp))
printk(KERN_INFO "%s: Using PHY number %d.\n",
dev->name, first_phy);
}
#ifdef DO_DXSUFLO
2006-03-22 05:15:44 +08:00
if (lp->dxsuflo) { /* Disable transmit stop on underflow */
val = lp->a.read_csr(ioaddr, CSR3);
2006-03-22 05:15:44 +08:00
val |= 0x40;
lp->a.write_csr(ioaddr, CSR3, val);
2006-03-22 05:15:44 +08:00
}
#endif
lp->init_block->mode =
cpu_to_le16((lp->options & PCNET32_PORT_PORTSEL) << 7);
2006-03-22 05:15:44 +08:00
pcnet32_load_multicast(dev);
if (pcnet32_init_ring(dev)) {
rc = -ENOMEM;
goto err_free_ring;
}
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
napi_enable(&lp->napi);
2006-03-22 05:15:44 +08:00
/* Re-initialize the PCNET32, and start it when done. */
lp->a.write_csr(ioaddr, 1, (lp->init_dma_addr & 0xffff));
lp->a.write_csr(ioaddr, 2, (lp->init_dma_addr >> 16));
2006-03-22 05:15:44 +08:00
lp->a.write_csr(ioaddr, CSR4, 0x0915); /* auto tx pad */
lp->a.write_csr(ioaddr, CSR0, CSR0_INIT);
2006-03-22 05:15:44 +08:00
netif_start_queue(dev);
if (lp->chip_version >= PCNET32_79C970A) {
/* Print the link status and start the watchdog */
pcnet32_check_media(dev, 1);
mod_timer(&lp->watchdog_timer, PCNET32_WATCHDOG_TIMEOUT);
}
2006-03-22 05:15:44 +08:00
i = 0;
while (i++ < 100)
if (lp->a.read_csr(ioaddr, CSR0) & CSR0_IDON)
2006-03-22 05:15:44 +08:00
break;
/*
* We used to clear the InitDone bit, 0x0100, here but Mark Stockton
* reports that doing so triggers a bug in the '974.
*/
lp->a.write_csr(ioaddr, CSR0, CSR0_NORMAL);
2006-03-22 05:15:44 +08:00
if (netif_msg_ifup(lp))
printk(KERN_DEBUG
"%s: pcnet32 open after %d ticks, init block %#x csr0 %4.4x.\n",
dev->name, i,
(u32) (lp->init_dma_addr),
lp->a.read_csr(ioaddr, CSR0));
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
return 0; /* Always succeed */
err_free_ring:
/* free any allocated skbuffs */
pcnet32_purge_rx_ring(dev);
2006-03-22 05:15:44 +08:00
/*
* Switch back to 16bit mode to avoid problems with dumb
* DOS packet driver after a warm reboot
*/
lp->a.write_bcr(ioaddr, 20, 4);
err_free_irq:
spin_unlock_irqrestore(&lp->lock, flags);
free_irq(dev->irq, dev);
return rc;
}
/*
* The LANCE has been halted for one reason or another (busmaster memory
* arbitration error, Tx FIFO underflow, driver stopped it to reconfigure,
* etc.). Modern LANCE variants always reload their ring-buffer
* configuration when restarted, so we must reinitialize our ring
* context before restarting. As part of this reinitialization,
* find all packets still on the Tx ring and pretend that they had been
* sent (in effect, drop the packets on the floor) - the higher-level
* protocols will time out and retransmit. It'd be better to shuffle
* these skbs to a temp list and then actually re-Tx them after
* restarting the chip, but I'm too lazy to do so right now. dplatt@3do.com
*/
2006-03-22 05:15:44 +08:00
static void pcnet32_purge_tx_ring(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int i;
2006-03-22 05:15:44 +08:00
for (i = 0; i < lp->tx_ring_size; i++) {
lp->tx_ring[i].status = 0; /* CPU owns buffer */
wmb(); /* Make sure adapter sees owner change */
if (lp->tx_skbuff[i]) {
pci_unmap_single(lp->pci_dev, lp->tx_dma_addr[i],
lp->tx_skbuff[i]->len,
PCI_DMA_TODEVICE);
dev_kfree_skb_any(lp->tx_skbuff[i]);
}
lp->tx_skbuff[i] = NULL;
lp->tx_dma_addr[i] = 0;
}
}
/* Initialize the PCNET32 Rx and Tx rings. */
2006-03-22 05:15:44 +08:00
static int pcnet32_init_ring(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int i;
lp->tx_full = 0;
lp->cur_rx = lp->cur_tx = 0;
lp->dirty_rx = lp->dirty_tx = 0;
for (i = 0; i < lp->rx_ring_size; i++) {
struct sk_buff *rx_skbuff = lp->rx_skbuff[i];
if (rx_skbuff == NULL) {
if (!
(rx_skbuff = lp->rx_skbuff[i] =
dev_alloc_skb(PKT_BUF_SKB))) {
2006-03-22 05:15:44 +08:00
/* there is not much, we can do at this point */
if (netif_msg_drv(lp))
2006-03-22 05:15:44 +08:00
printk(KERN_ERR
"%s: pcnet32_init_ring dev_alloc_skb failed.\n",
dev->name);
return -1;
}
skb_reserve(rx_skbuff, NET_IP_ALIGN);
2006-03-22 05:15:44 +08:00
}
rmb();
if (lp->rx_dma_addr[i] == 0)
lp->rx_dma_addr[i] =
pci_map_single(lp->pci_dev, rx_skbuff->data,
PKT_BUF_SIZE, PCI_DMA_FROMDEVICE);
lp->rx_ring[i].base = cpu_to_le32(lp->rx_dma_addr[i]);
lp->rx_ring[i].buf_length = cpu_to_le16(NEG_BUF_SIZE);
2006-03-22 05:15:44 +08:00
wmb(); /* Make sure owner changes after all others are visible */
lp->rx_ring[i].status = cpu_to_le16(0x8000);
2006-03-22 05:15:44 +08:00
}
/* The Tx buffer address is filled in as needed, but we do need to clear
* the upper ownership bit. */
for (i = 0; i < lp->tx_ring_size; i++) {
lp->tx_ring[i].status = 0; /* CPU owns buffer */
wmb(); /* Make sure adapter sees owner change */
lp->tx_ring[i].base = 0;
lp->tx_dma_addr[i] = 0;
}
lp->init_block->tlen_rlen =
cpu_to_le16(lp->tx_len_bits | lp->rx_len_bits);
2006-03-22 05:15:44 +08:00
for (i = 0; i < 6; i++)
lp->init_block->phys_addr[i] = dev->dev_addr[i];
lp->init_block->rx_ring = cpu_to_le32(lp->rx_ring_dma_addr);
lp->init_block->tx_ring = cpu_to_le32(lp->tx_ring_dma_addr);
2006-03-22 05:15:44 +08:00
wmb(); /* Make sure all changes are visible */
return 0;
}
/* the pcnet32 has been issued a stop or reset. Wait for the stop bit
* then flush the pending transmit operations, re-initialize the ring,
* and tell the chip to initialize.
*/
2006-03-22 05:15:44 +08:00
static void pcnet32_restart(struct net_device *dev, unsigned int csr0_bits)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
int i;
2006-03-22 05:15:44 +08:00
/* wait for stop */
for (i = 0; i < 100; i++)
if (lp->a.read_csr(ioaddr, CSR0) & CSR0_STOP)
2006-03-22 05:15:44 +08:00
break;
2006-03-22 05:15:44 +08:00
if (i >= 100 && netif_msg_drv(lp))
printk(KERN_ERR
"%s: pcnet32_restart timed out waiting for stop.\n",
dev->name);
2006-03-22 05:15:44 +08:00
pcnet32_purge_tx_ring(dev);
if (pcnet32_init_ring(dev))
return;
2006-03-22 05:15:44 +08:00
/* ReInit Ring */
lp->a.write_csr(ioaddr, CSR0, CSR0_INIT);
2006-03-22 05:15:44 +08:00
i = 0;
while (i++ < 1000)
if (lp->a.read_csr(ioaddr, CSR0) & CSR0_IDON)
2006-03-22 05:15:44 +08:00
break;
lp->a.write_csr(ioaddr, CSR0, csr0_bits);
}
2006-03-22 05:15:44 +08:00
static void pcnet32_tx_timeout(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr, flags;
spin_lock_irqsave(&lp->lock, flags);
/* Transmitter timeout, serious problems. */
if (pcnet32_debug & NETIF_MSG_DRV)
printk(KERN_ERR
"%s: transmit timed out, status %4.4x, resetting.\n",
dev->name, lp->a.read_csr(ioaddr, CSR0));
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP);
dev->stats.tx_errors++;
2006-03-22 05:15:44 +08:00
if (netif_msg_tx_err(lp)) {
int i;
printk(KERN_DEBUG
" Ring data dump: dirty_tx %d cur_tx %d%s cur_rx %d.",
lp->dirty_tx, lp->cur_tx, lp->tx_full ? " (full)" : "",
lp->cur_rx);
for (i = 0; i < lp->rx_ring_size; i++)
printk("%s %08x %04x %08x %04x", i & 1 ? "" : "\n ",
le32_to_cpu(lp->rx_ring[i].base),
(-le16_to_cpu(lp->rx_ring[i].buf_length)) &
0xffff, le32_to_cpu(lp->rx_ring[i].msg_length),
le16_to_cpu(lp->rx_ring[i].status));
for (i = 0; i < lp->tx_ring_size; i++)
printk("%s %08x %04x %08x %04x", i & 1 ? "" : "\n ",
le32_to_cpu(lp->tx_ring[i].base),
(-le16_to_cpu(lp->tx_ring[i].length)) & 0xffff,
le32_to_cpu(lp->tx_ring[i].misc),
le16_to_cpu(lp->tx_ring[i].status));
printk("\n");
}
pcnet32_restart(dev, CSR0_NORMAL);
2006-03-22 05:15:44 +08:00
dev->trans_start = jiffies;
netif_wake_queue(dev);
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
}
static netdev_tx_t pcnet32_start_xmit(struct sk_buff *skb,
struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
u16 status;
int entry;
unsigned long flags;
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
2006-03-22 05:15:44 +08:00
if (netif_msg_tx_queued(lp)) {
printk(KERN_DEBUG
"%s: pcnet32_start_xmit() called, csr0 %4.4x.\n",
dev->name, lp->a.read_csr(ioaddr, CSR0));
2006-03-22 05:15:44 +08:00
}
2006-03-22 05:15:44 +08:00
/* Default status -- will not enable Successful-TxDone
* interrupt when that option is available to us.
*/
status = 0x8300;
2006-03-22 05:15:44 +08:00
/* Fill in a Tx ring entry */
2006-03-22 05:15:44 +08:00
/* Mask to ring buffer boundary. */
entry = lp->cur_tx & lp->tx_mod_mask;
2006-03-22 05:15:44 +08:00
/* Caution: the write order is important here, set the status
* with the "ownership" bits last. */
lp->tx_ring[entry].length = cpu_to_le16(-skb->len);
2006-03-22 05:15:44 +08:00
lp->tx_ring[entry].misc = 0x00000000;
2006-03-22 05:15:44 +08:00
lp->tx_skbuff[entry] = skb;
lp->tx_dma_addr[entry] =
pci_map_single(lp->pci_dev, skb->data, skb->len, PCI_DMA_TODEVICE);
lp->tx_ring[entry].base = cpu_to_le32(lp->tx_dma_addr[entry]);
2006-03-22 05:15:44 +08:00
wmb(); /* Make sure owner changes after all others are visible */
lp->tx_ring[entry].status = cpu_to_le16(status);
2006-03-22 05:15:44 +08:00
lp->cur_tx++;
dev->stats.tx_bytes += skb->len;
2006-03-22 05:15:44 +08:00
/* Trigger an immediate send poll. */
lp->a.write_csr(ioaddr, CSR0, CSR0_INTEN | CSR0_TXPOLL);
2006-03-22 05:15:44 +08:00
dev->trans_start = jiffies;
2006-03-22 05:15:44 +08:00
if (lp->tx_ring[(entry + 1) & lp->tx_mod_mask].base != 0) {
lp->tx_full = 1;
netif_stop_queue(dev);
}
spin_unlock_irqrestore(&lp->lock, flags);
return NETDEV_TX_OK;
}
/* The PCNET32 interrupt handler. */
static irqreturn_t
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
pcnet32_interrupt(int irq, void *dev_id)
{
2006-03-22 05:15:44 +08:00
struct net_device *dev = dev_id;
struct pcnet32_private *lp;
unsigned long ioaddr;
u16 csr0;
2006-03-22 05:15:44 +08:00
int boguscnt = max_interrupt_work;
ioaddr = dev->base_addr;
lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
spin_lock(&lp->lock);
csr0 = lp->a.read_csr(ioaddr, CSR0);
while ((csr0 & 0x8f00) && --boguscnt >= 0) {
2006-03-22 05:15:44 +08:00
if (csr0 == 0xffff) {
break; /* PCMCIA remove happened */
}
/* Acknowledge all of the current interrupt sources ASAP. */
lp->a.write_csr(ioaddr, CSR0, csr0 & ~0x004f);
2006-03-22 05:15:44 +08:00
if (netif_msg_intr(lp))
printk(KERN_DEBUG
"%s: interrupt csr0=%#2.2x new csr=%#2.2x.\n",
dev->name, csr0, lp->a.read_csr(ioaddr, CSR0));
2006-03-22 05:15:44 +08:00
/* Log misc errors. */
if (csr0 & 0x4000)
dev->stats.tx_errors++; /* Tx babble. */
2006-03-22 05:15:44 +08:00
if (csr0 & 0x1000) {
/*
* This happens when our receive ring is full. This
* shouldn't be a problem as we will see normal rx
* interrupts for the frames in the receive ring. But
* there are some PCI chipsets (I can reproduce this
* on SP3G with Intel saturn chipset) which have
* sometimes problems and will fill up the receive
* ring with error descriptors. In this situation we
* don't get a rx interrupt, but a missed frame
* interrupt sooner or later.
2006-03-22 05:15:44 +08:00
*/
dev->stats.rx_errors++; /* Missed a Rx frame. */
2006-03-22 05:15:44 +08:00
}
if (csr0 & 0x0800) {
if (netif_msg_drv(lp))
printk(KERN_ERR
"%s: Bus master arbitration failure, status %4.4x.\n",
dev->name, csr0);
/* unlike for the lance, there is no restart needed */
}
if (napi_schedule_prep(&lp->napi)) {
u16 val;
/* set interrupt masks */
val = lp->a.read_csr(ioaddr, CSR3);
val |= 0x5f00;
lp->a.write_csr(ioaddr, CSR3, val);
__napi_schedule(&lp->napi);
break;
}
csr0 = lp->a.read_csr(ioaddr, CSR0);
2006-03-22 05:15:44 +08:00
}
if (netif_msg_intr(lp))
printk(KERN_DEBUG "%s: exiting interrupt, csr0=%#4.4x.\n",
dev->name, lp->a.read_csr(ioaddr, CSR0));
2006-03-22 05:15:44 +08:00
spin_unlock(&lp->lock);
return IRQ_HANDLED;
}
2006-03-22 05:15:44 +08:00
static int pcnet32_close(struct net_device *dev)
{
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
2006-03-22 05:15:44 +08:00
del_timer_sync(&lp->watchdog_timer);
2006-03-22 05:15:44 +08:00
netif_stop_queue(dev);
[NET]: Make NAPI polling independent of struct net_device objects. Several devices have multiple independant RX queues per net device, and some have a single interrupt doorbell for several queues. In either case, it's easier to support layouts like that if the structure representing the poll is independant from the net device itself. The signature of the ->poll() call back goes from: int foo_poll(struct net_device *dev, int *budget) to int foo_poll(struct napi_struct *napi, int budget) The caller is returned the number of RX packets processed (or the number of "NAPI credits" consumed if you want to get abstract). The callee no longer messes around bumping dev->quota, *budget, etc. because that is all handled in the caller upon return. The napi_struct is to be embedded in the device driver private data structures. Furthermore, it is the driver's responsibility to disable all NAPI instances in it's ->stop() device close handler. Since the napi_struct is privatized into the driver's private data structures, only the driver knows how to get at all of the napi_struct instances it may have per-device. With lots of help and suggestions from Rusty Russell, Roland Dreier, Michael Chan, Jeff Garzik, and Jamal Hadi Salim. Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra, Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan. [ Ported to current tree and all drivers converted. Integrated Stephen's follow-on kerneldoc additions, and restored poll_list handling to the old style to fix mutual exclusion issues. -DaveM ] Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-10-04 07:41:36 +08:00
napi_disable(&lp->napi);
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
dev->stats.rx_missed_errors = lp->a.read_csr(ioaddr, 112);
2006-03-22 05:15:44 +08:00
if (netif_msg_ifdown(lp))
printk(KERN_DEBUG
"%s: Shutting down ethercard, status was %2.2x.\n",
dev->name, lp->a.read_csr(ioaddr, CSR0));
2006-03-22 05:15:44 +08:00
/* We stop the PCNET32 here -- it occasionally polls memory if we don't. */
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP);
2006-03-22 05:15:44 +08:00
/*
* Switch back to 16bit mode to avoid problems with dumb
* DOS packet driver after a warm reboot
*/
lp->a.write_bcr(ioaddr, 20, 4);
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
2006-03-22 05:15:44 +08:00
free_irq(dev->irq, dev);
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
pcnet32_purge_rx_ring(dev);
pcnet32_purge_tx_ring(dev);
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
2006-03-22 05:15:44 +08:00
return 0;
}
2006-03-22 05:15:44 +08:00
static struct net_device_stats *pcnet32_get_stats(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
unsigned long flags;
spin_lock_irqsave(&lp->lock, flags);
dev->stats.rx_missed_errors = lp->a.read_csr(ioaddr, 112);
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
return &dev->stats;
}
/* taken from the sunlance driver, which it took from the depca driver */
2006-03-22 05:15:44 +08:00
static void pcnet32_load_multicast(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
volatile struct pcnet32_init_block *ib = lp->init_block;
volatile __le16 *mcast_table = (__le16 *)ib->filter;
2006-03-22 05:15:44 +08:00
struct dev_mc_list *dmi = dev->mc_list;
unsigned long ioaddr = dev->base_addr;
2006-03-22 05:15:44 +08:00
char *addrs;
int i;
u32 crc;
/* set all multicast bits */
if (dev->flags & IFF_ALLMULTI) {
ib->filter[0] = cpu_to_le32(~0U);
ib->filter[1] = cpu_to_le32(~0U);
lp->a.write_csr(ioaddr, PCNET32_MC_FILTER, 0xffff);
lp->a.write_csr(ioaddr, PCNET32_MC_FILTER+1, 0xffff);
lp->a.write_csr(ioaddr, PCNET32_MC_FILTER+2, 0xffff);
lp->a.write_csr(ioaddr, PCNET32_MC_FILTER+3, 0xffff);
2006-03-22 05:15:44 +08:00
return;
}
/* clear the multicast filter */
ib->filter[0] = 0;
ib->filter[1] = 0;
/* Add addresses */
for (i = 0; i < dev->mc_count; i++) {
addrs = dmi->dmi_addr;
dmi = dmi->next;
/* multicast address? */
if (!(*addrs & 1))
continue;
crc = ether_crc_le(6, addrs);
crc = crc >> 26;
mcast_table[crc >> 4] |= cpu_to_le16(1 << (crc & 0xf));
2006-03-22 05:15:44 +08:00
}
for (i = 0; i < 4; i++)
lp->a.write_csr(ioaddr, PCNET32_MC_FILTER + i,
le16_to_cpu(mcast_table[i]));
return;
}
/*
* Set or clear the multicast filter for this adaptor.
*/
static void pcnet32_set_multicast_list(struct net_device *dev)
{
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr, flags;
struct pcnet32_private *lp = netdev_priv(dev);
int csr15, suspended;
2006-03-22 05:15:44 +08:00
spin_lock_irqsave(&lp->lock, flags);
suspended = pcnet32_suspend(dev, &flags, 0);
csr15 = lp->a.read_csr(ioaddr, CSR15);
2006-03-22 05:15:44 +08:00
if (dev->flags & IFF_PROMISC) {
/* Log any net taps. */
if (netif_msg_hw(lp))
printk(KERN_INFO "%s: Promiscuous mode enabled.\n",
dev->name);
lp->init_block->mode =
cpu_to_le16(0x8000 | (lp->options & PCNET32_PORT_PORTSEL) <<
2006-03-22 05:15:44 +08:00
7);
lp->a.write_csr(ioaddr, CSR15, csr15 | 0x8000);
2006-03-22 05:15:44 +08:00
} else {
lp->init_block->mode =
cpu_to_le16((lp->options & PCNET32_PORT_PORTSEL) << 7);
lp->a.write_csr(ioaddr, CSR15, csr15 & 0x7fff);
2006-03-22 05:15:44 +08:00
pcnet32_load_multicast(dev);
}
if (suspended) {
int csr5;
/* clear SUSPEND (SPND) - CSR5 bit 0 */
csr5 = lp->a.read_csr(ioaddr, CSR5);
lp->a.write_csr(ioaddr, CSR5, csr5 & (~CSR5_SUSPEND));
} else {
lp->a.write_csr(ioaddr, CSR0, CSR0_STOP);
pcnet32_restart(dev, CSR0_NORMAL);
netif_wake_queue(dev);
}
2006-03-22 05:15:44 +08:00
spin_unlock_irqrestore(&lp->lock, flags);
}
/* This routine assumes that the lp->lock is held */
static int mdio_read(struct net_device *dev, int phy_id, int reg_num)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
u16 val_out;
2006-03-22 05:15:44 +08:00
if (!lp->mii)
return 0;
2006-03-22 05:15:44 +08:00
lp->a.write_bcr(ioaddr, 33, ((phy_id & 0x1f) << 5) | (reg_num & 0x1f));
val_out = lp->a.read_bcr(ioaddr, 34);
2006-03-22 05:15:44 +08:00
return val_out;
}
/* This routine assumes that the lp->lock is held */
static void mdio_write(struct net_device *dev, int phy_id, int reg_num, int val)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long ioaddr = dev->base_addr;
2006-03-22 05:15:44 +08:00
if (!lp->mii)
return;
2006-03-22 05:15:44 +08:00
lp->a.write_bcr(ioaddr, 33, ((phy_id & 0x1f) << 5) | (reg_num & 0x1f));
lp->a.write_bcr(ioaddr, 34, val);
}
static int pcnet32_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int rc;
unsigned long flags;
2006-03-22 05:15:44 +08:00
/* SIOC[GS]MIIxxx ioctls */
if (lp->mii) {
spin_lock_irqsave(&lp->lock, flags);
rc = generic_mii_ioctl(&lp->mii_if, if_mii(rq), cmd, NULL);
spin_unlock_irqrestore(&lp->lock, flags);
} else {
rc = -EOPNOTSUPP;
}
2006-03-22 05:15:44 +08:00
return rc;
}
static int pcnet32_check_otherphy(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
struct mii_if_info mii = lp->mii_if;
u16 bmcr;
int i;
2006-03-22 05:15:44 +08:00
for (i = 0; i < PCNET32_MAX_PHYS; i++) {
if (i == lp->mii_if.phy_id)
continue; /* skip active phy */
if (lp->phymask & (1 << i)) {
mii.phy_id = i;
if (mii_link_ok(&mii)) {
/* found PHY with active link */
if (netif_msg_link(lp))
printk(KERN_INFO
"%s: Using PHY number %d.\n",
dev->name, i);
/* isolate inactive phy */
bmcr =
mdio_read(dev, lp->mii_if.phy_id, MII_BMCR);
mdio_write(dev, lp->mii_if.phy_id, MII_BMCR,
bmcr | BMCR_ISOLATE);
/* de-isolate new phy */
bmcr = mdio_read(dev, i, MII_BMCR);
mdio_write(dev, i, MII_BMCR,
bmcr & ~BMCR_ISOLATE);
/* set new phy address */
lp->mii_if.phy_id = i;
return 1;
}
}
}
2006-03-22 05:15:44 +08:00
return 0;
}
/*
* Show the status of the media. Similar to mii_check_media however it
* correctly shows the link speed for all (tested) pcnet32 variants.
* Devices with no mii just report link state without speed.
*
* Caller is assumed to hold and release the lp->lock.
*/
static void pcnet32_check_media(struct net_device *dev, int verbose)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
int curr_link;
int prev_link = netif_carrier_ok(dev) ? 1 : 0;
u32 bcr9;
if (lp->mii) {
2006-03-22 05:15:44 +08:00
curr_link = mii_link_ok(&lp->mii_if);
} else {
2006-03-22 05:15:44 +08:00
ulong ioaddr = dev->base_addr; /* card base I/O address */
curr_link = (lp->a.read_bcr(ioaddr, 4) != 0xc0);
}
if (!curr_link) {
if (prev_link || verbose) {
netif_carrier_off(dev);
if (netif_msg_link(lp))
printk(KERN_INFO "%s: link down\n", dev->name);
}
if (lp->phycount > 1) {
curr_link = pcnet32_check_otherphy(dev);
prev_link = 0;
}
} else if (verbose || !prev_link) {
netif_carrier_on(dev);
if (lp->mii) {
if (netif_msg_link(lp)) {
struct ethtool_cmd ecmd;
mii_ethtool_gset(&lp->mii_if, &ecmd);
printk(KERN_INFO
"%s: link up, %sMbps, %s-duplex\n",
dev->name,
(ecmd.speed == SPEED_100) ? "100" : "10",
(ecmd.duplex ==
DUPLEX_FULL) ? "full" : "half");
}
bcr9 = lp->a.read_bcr(dev->base_addr, 9);
if ((bcr9 & (1 << 0)) != lp->mii_if.full_duplex) {
if (lp->mii_if.full_duplex)
bcr9 |= (1 << 0);
else
bcr9 &= ~(1 << 0);
lp->a.write_bcr(dev->base_addr, 9, bcr9);
}
} else {
if (netif_msg_link(lp))
printk(KERN_INFO "%s: link up\n", dev->name);
}
}
}
/*
* Check for loss of link and link establishment.
* Can not use mii_check_media because it does nothing if mode is forced.
*/
static void pcnet32_watchdog(struct net_device *dev)
{
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unsigned long flags;
2006-03-22 05:15:44 +08:00
/* Print the link status if it has changed */
spin_lock_irqsave(&lp->lock, flags);
pcnet32_check_media(dev, 0);
spin_unlock_irqrestore(&lp->lock, flags);
mod_timer(&lp->watchdog_timer, round_jiffies(PCNET32_WATCHDOG_TIMEOUT));
}
static int pcnet32_pm_suspend(struct pci_dev *pdev, pm_message_t state)
{
struct net_device *dev = pci_get_drvdata(pdev);
if (netif_running(dev)) {
netif_device_detach(dev);
pcnet32_close(dev);
}
pci_save_state(pdev);
pci_set_power_state(pdev, pci_choose_state(pdev, state));
return 0;
}
static int pcnet32_pm_resume(struct pci_dev *pdev)
{
struct net_device *dev = pci_get_drvdata(pdev);
pci_set_power_state(pdev, PCI_D0);
pci_restore_state(pdev);
if (netif_running(dev)) {
pcnet32_open(dev);
netif_device_attach(dev);
}
return 0;
}
static void __devexit pcnet32_remove_one(struct pci_dev *pdev)
{
2006-03-22 05:15:44 +08:00
struct net_device *dev = pci_get_drvdata(pdev);
if (dev) {
struct pcnet32_private *lp = netdev_priv(dev);
2006-03-22 05:15:44 +08:00
unregister_netdev(dev);
pcnet32_free_ring(dev);
release_region(dev->base_addr, PCNET32_TOTAL_SIZE);
pci_free_consistent(lp->pci_dev, sizeof(*lp->init_block),
lp->init_block, lp->init_dma_addr);
2006-03-22 05:15:44 +08:00
free_netdev(dev);
pci_disable_device(pdev);
pci_set_drvdata(pdev, NULL);
}
}
static struct pci_driver pcnet32_driver = {
2006-03-22 05:15:44 +08:00
.name = DRV_NAME,
.probe = pcnet32_probe_pci,
.remove = __devexit_p(pcnet32_remove_one),
.id_table = pcnet32_pci_tbl,
.suspend = pcnet32_pm_suspend,
.resume = pcnet32_pm_resume,
};
/* An additional parameter that may be passed in... */
static int debug = -1;
static int tx_start_pt = -1;
static int pcnet32_have_pci;
module_param(debug, int, 0);
MODULE_PARM_DESC(debug, DRV_NAME " debug level");
module_param(max_interrupt_work, int, 0);
2006-03-22 05:15:44 +08:00
MODULE_PARM_DESC(max_interrupt_work,
DRV_NAME " maximum events handled per interrupt");
module_param(rx_copybreak, int, 0);
2006-03-22 05:15:44 +08:00
MODULE_PARM_DESC(rx_copybreak,
DRV_NAME " copy breakpoint for copy-only-tiny-frames");
module_param(tx_start_pt, int, 0);
MODULE_PARM_DESC(tx_start_pt, DRV_NAME " transmit start point (0-3)");
module_param(pcnet32vlb, int, 0);
MODULE_PARM_DESC(pcnet32vlb, DRV_NAME " Vesa local bus (VLB) support (0/1)");
module_param_array(options, int, NULL, 0);
MODULE_PARM_DESC(options, DRV_NAME " initial option setting(s) (0-15)");
module_param_array(full_duplex, int, NULL, 0);
MODULE_PARM_DESC(full_duplex, DRV_NAME " full duplex setting(s) (1)");
/* Module Parameter for HomePNA cards added by Patrick Simmons, 2004 */
module_param_array(homepna, int, NULL, 0);
2006-03-22 05:15:44 +08:00
MODULE_PARM_DESC(homepna,
DRV_NAME
" mode for 79C978 cards (1 for HomePNA, 0 for Ethernet, default Ethernet");
MODULE_AUTHOR("Thomas Bogendoerfer");
MODULE_DESCRIPTION("Driver for PCnet32 and PCnetPCI based ethercards");
MODULE_LICENSE("GPL");
#define PCNET32_MSG_DEFAULT (NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_LINK)
static int __init pcnet32_init_module(void)
{
2006-03-22 05:15:44 +08:00
printk(KERN_INFO "%s", version);
2006-03-22 05:15:44 +08:00
pcnet32_debug = netif_msg_init(debug, PCNET32_MSG_DEFAULT);
2006-03-22 05:15:44 +08:00
if ((tx_start_pt >= 0) && (tx_start_pt <= 3))
tx_start = tx_start_pt;
2006-03-22 05:15:44 +08:00
/* find the PCI devices */
if (!pci_register_driver(&pcnet32_driver))
2006-03-22 05:15:44 +08:00
pcnet32_have_pci = 1;
2006-03-22 05:15:44 +08:00
/* should we find any remaining VLbus devices ? */
if (pcnet32vlb)
pcnet32_probe_vlbus(pcnet32_portlist);
2006-03-22 05:15:44 +08:00
if (cards_found && (pcnet32_debug & NETIF_MSG_PROBE))
printk(KERN_INFO PFX "%d cards_found.\n", cards_found);
2006-03-22 05:15:44 +08:00
return (pcnet32_have_pci + cards_found) ? 0 : -ENODEV;
}
static void __exit pcnet32_cleanup_module(void)
{
2006-03-22 05:15:44 +08:00
struct net_device *next_dev;
while (pcnet32_dev) {
struct pcnet32_private *lp = netdev_priv(pcnet32_dev);
2006-03-22 05:15:44 +08:00
next_dev = lp->next;
unregister_netdev(pcnet32_dev);
pcnet32_free_ring(pcnet32_dev);
release_region(pcnet32_dev->base_addr, PCNET32_TOTAL_SIZE);
pci_free_consistent(lp->pci_dev, sizeof(*lp->init_block),
lp->init_block, lp->init_dma_addr);
2006-03-22 05:15:44 +08:00
free_netdev(pcnet32_dev);
pcnet32_dev = next_dev;
}
2006-03-22 05:15:44 +08:00
if (pcnet32_have_pci)
pci_unregister_driver(&pcnet32_driver);
}
module_init(pcnet32_init_module);
module_exit(pcnet32_cleanup_module);
/*
* Local variables:
* c-indent-level: 4
* tab-width: 8
* End:
*/